Advanced search

Message boards : Graphics cards (GPUs) : Nov2010> New ACEMD application

Author Message
Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19063 - Posted: 27 Oct 2010 | 9:41:10 UTC
Last modified: 27 Oct 2010 | 12:56:25 UTC

Hi,
we would like to update the ACEMD application with a newer version which is faster and offers some extra features for the science.

We will try to keep only 4 versions:
compiled for CUDA3.2 for windows and Linux
compiled for CUDA2.2 for windows and Linux

However, we are not sure that the app will compile for CUDA2.2. I will confirm in the next few days.

gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19080 - Posted: 28 Oct 2010 | 20:33:55 UTC - in response to Message 19063.

Confirmed.
We will start to come out next week with cuda2.2 applications for windows and linux in beta.

If all goes well we will go on with the cuda3.2. All cards will be able to receive cuda3.2 apps.

gdf

Profile Fred J. Verster
Send message
Joined: 1 Apr 09
Posts: 58
Credit: 35,833,978
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19081 - Posted: 28 Oct 2010 | 20:46:39 UTC - in response to Message 19080.
Last modified: 28 Oct 2010 | 20:54:01 UTC

Couldn't help saying this, when crunching for SETI, MB tasks, CUDA 2.3 appeared to be much faster then CUDA 2.2, why choose for CUDA 2.2?
(Or is the app already compiled?) :)

Well maybe I'm compairing apples and oranges......
____________

Knight Who Says Ni N!

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19082 - Posted: 28 Oct 2010 | 20:54:55 UTC - in response to Message 19081.
Last modified: 28 Oct 2010 | 20:57:30 UTC

For us it's the same speed.

By the way, we will deprecate cuda3.1 and 3.0 applications due to several bugs in the CUDA libs. So you need to change driver to the latest driver to keep receiving cuda3 applications. This will be done slowly and notified in the news.

gdf

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19084 - Posted: 28 Oct 2010 | 23:23:38 UTC - in response to Message 19082.

For us it's the same speed.

By the way, we will deprecate cuda3.1 and 3.0 applications due to several bugs in the CUDA libs. So you need to change driver to the latest driver to keep receiving cuda3 applications. This will be done slowly and notified in the news.

gdf

What driver version do we need for CUDA 3.2? The v260.89 and v260.99 drivers are badly flawed.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19087 - Posted: 29 Oct 2010 | 1:20:12 UTC - in response to Message 19084.
Last modified: 29 Oct 2010 | 1:21:17 UTC

GDF, can you confirm that the latest driver will only be needed for Fermi cards?

The v260.89 and v260.99 drivers are bad for the 200Series cards, and probably earlier cards - the frequency drops and stays low.

I have not noticed any drop in frequency for my GTX470's with the latest driver, so I expect 260.99 works for Fermis.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19088 - Posted: 29 Oct 2010 | 7:25:27 UTC - in response to Message 19087.

I assume that if you have a g200 cards then you are still using the cuda2.2 application at the moment. In this case, you don't need to change driver.
You will need to change driver to receive a CUDA3 application. A 3.1 driver will not be enough when we remove the 3.1 app.

gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19093 - Posted: 29 Oct 2010 | 11:04:51 UTC - in response to Message 19088.

Thanks.

I guess this might in itself bring Linux support to the GTX460, GTS450 and still unreleased GTX475 (basically just a GTX460 with a full complement of shaders and cores) as it needs the 260.99 driver, but I don’t know what is under the bonnet of the latest ACEMD tool, so we might still need to wait on the 3.2 dev app to move from RC to final version?

Whatever the case, if it improves reliability or increases speed it will be very welcome.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19095 - Posted: 29 Oct 2010 | 13:15:35 UTC - in response to Message 19084.

For us it's the same speed.

By the way, we will deprecate cuda3.1 and 3.0 applications due to several bugs in the CUDA libs. So you need to change driver to the latest driver to keep receiving cuda3 applications. This will be done slowly and notified in the news.

gdf

What driver version do we need for CUDA 3.2? The v260.89 and v260.99 drivers are badly flawed.

Still, no answer to the question. What driver level is needed for cuda 3.2 support?

GPUGRID Role account
Send message
Joined: 15 Feb 07
Posts: 134
Credit: 1,349,535,983
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19098 - Posted: 29 Oct 2010 | 14:20:58 UTC - in response to Message 19095.

What driver level is needed for cuda 3.2 support?


Since 3.2 is not yet released, that is an unanswerable question.

MJH

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19099 - Posted: 29 Oct 2010 | 14:45:07 UTC - in response to Message 19098.

What driver level is needed for cuda 3.2 support?

Since 3.2 is not yet released, that is an unanswerable question.

MJH

Found out myself:

nVidia has released the Forceware 261.00 driver. This driver is a special release intended primarily for developers, providing support for the CUDA Toolkit 3.2 RC2, so it's best used for experimental purposes.

It seems the fist OFFICIAL cuda 3.2 drivers will be the 265.xx series...

Here's some info on cuda 3.2:

http://developer.nvidia.com/object/cuda_3_2_toolkit_rc.html

There was a link to the v261.00 drivers but it has been removed.



Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19103 - Posted: 29 Oct 2010 | 23:47:59 UTC - in response to Message 19099.

For CUDA 2.2 you don't need to update your driver.

For CUDA 3.2, who knows, but this only applies to Fermi's and you will need the latest driver - whatever that will be tomorrow?

I don't know why I am surprised NVidia said 265.xx, it's not as if they are conformists when it comes to time tables, design, testing, series... Still it is a bit of a curve ball. If they do manage to ship a GTX475 by the end of Nov, the driver scene will be very messy. Chaos looms if they inflict a GF110 upon us.

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19107 - Posted: 30 Oct 2010 | 5:36:10 UTC - in response to Message 19099.

What driver level is needed for cuda 3.2 support?

Since 3.2 is not yet released, that is an unanswerable question.

MJH

Found out myself:

nVidia has released the Forceware 261.00 driver. This driver is a special release intended primarily for developers, providing support for the CUDA Toolkit 3.2 RC2, so it's best used for experimental purposes.

It seems the fist OFFICIAL cuda 3.2 drivers will be the 265.xx series...

Here's some info on cuda 3.2:

http://developer.nvidia.com/object/cuda_3_2_toolkit_rc.html

There was a link to the v261.00 drivers but it has been removed.


The 260.89 onwards support cuda 3.2. It was released as whql so i'd call it "official". It has since been superceeded by 260.99 which is also whql. Both have the down-clocking issue on non-Fermi cards though, so you might want to wait.
____________
BOINC blog

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19111 - Posted: 30 Oct 2010 | 10:16:18 UTC - in response to Message 19107.

Developer Drivers for WinXP (261.00) <Coming Soon!>
Developer Drivers for WinVista and Win7 (261.00) <Coming Soon!>
Saw it listed here

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19113 - Posted: 30 Oct 2010 | 11:47:54 UTC - in response to Message 19111.
Last modified: 30 Oct 2010 | 11:48:12 UTC

I have changed the server policy.
Now all cards with the appropriate driver should be able to get the CUDA3.1 application. G200 cards installed together with Fermi cards will not work properly until we will release a cuda3.2 app.

This allows many old cuda2.2 bugs to be fixed on g200 cards if you just update the driver to a recent one. Change only if you have problems with the cuda2.2 app.

gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19118 - Posted: 30 Oct 2010 | 14:22:22 UTC - in response to Message 19113.
Last modified: 30 Oct 2010 | 15:10:52 UTC

To test if this frequency bug effects all 200series cards or just some I installed 260.99 for a GT240 (DDR3 1GB version) on a W7x64 system.
Boinc downloaded the CUDA3.1 app and is now running a 6.11 TONI_KKi4 task.

It's possible it will not cause problems on the 40nm GT240s and may only cause problems on specific makes of cards, depending on their Firmware.

GPU details:
Core Clock at 597MHz,
Shaders at 1595MHz,
RAM at 800MHz (x2)
Fan speed 60%
GPU temp 49 deg C.
GPU usage 89 to 94 %, mostly 91%.

No problems so far; the GT240 has not slowed any after 1h.

Picked up a 6.11 CUDA 3010 task on another GT240 with the 258.96 driver, so we dont need to use the 260.99 driver for 200 series cards.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19122 - Posted: 30 Oct 2010 | 15:24:47 UTC - in response to Message 19118.

I installed 260.99 for a GT240 (DDR3 1GB version) on a W7x64 system.
Boinc downloaded the CUDA3.1 app and is now running a 6.11 TONI_KKi4 task.
No problems so far, but I want to test if this frequency bug effects all 200series cards or just some; it's possible it will not cause problems on the GT240s and may only cause problems on specific makes of cards, depending on their Firmware.

GPU details:
Core Clock at 597MHz,
Shaders at 1595MHz,
RAM at 800MHz (x2)
Fan speed 60%
GPU temp 49 deg C.
GPU usage 89 to 94 %, mostly 91%.

The GT240 did not slow down after 40min.

Picked up a 6.11 CUDA 3010 task on another GT240 with the 258.96 driver, so we dont need to use the 260.99 driver for 200 series cards.

The slowdown occurred in XP64 with both v260.89 & v460.99 on 5 cards. Two were GTX 260 and three were GT 240. Win7 uses a different driver so your experience may be better, but then you have to put up with the ~ 20% Win7 penalty...

I also picked up 6.11 tasks on both of my GTX 260 cards with 258.96 drivers. So far running OK but appear to be a little slower than the 6.05 app.


Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19123 - Posted: 30 Oct 2010 | 16:58:35 UTC - in response to Message 19122.

Thanks Beyond, that paints a clearer picture.
It looks like the 260.89 and 260.99 drivers may work for 200series cards on Win7 and Vista, but not on XP. I suggest people only use the 260.99 driver, if they must, but we really need a better driver before going to CUDA 3.2 only, if that is the plan. Hopefully NVidia will bring out a fully functional 261.xx soon.

I will update to the 260.99 driver on my Vista system this evening (UK) and give it a go. It has 4 GT240's.
For some still unknown reason my GTX260 failed all tasks on Vista and W7 the last time I tried, but I may try it on XP X86 in a day or so (if no new drivers turn up). It's sitting on my desk awaiting a test system.

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19130 - Posted: 31 Oct 2010 | 5:48:23 UTC

Would the updated application solve the problem regarding on the 65nm GTC260s?

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19131 - Posted: 31 Oct 2010 | 7:17:01 UTC - in response to Message 19130.

It might, but other updates did not, so I'm not holding out much hope. Give it a go in a few days and see.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19133 - Posted: 31 Oct 2010 | 9:15:38 UTC - in response to Message 19130.

Would the updated application solve the problem regarding on the 65nm GTC260s?



Even the current cuda3.1 one should. Please test.

gdf

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19135 - Posted: 31 Oct 2010 | 13:45:03 UTC - in response to Message 19122.

I also picked up 6.11 tasks on both of my GTX 260 cards with 258.96 drivers.

The 6.11 app so far is running about the same speed on the GTX 260, but all 3 of my GT 240 cards have slowed down about 60%. GPU usage is still above 90% but the WUs are running so slow that it's not worth keeping them on the project with the 6.11 app. Can we PLEASE go back to 6.05?

Profile Sutaru Tsureku
Send message
Joined: 1 May 09
Posts: 13
Credit: 3,655,193
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 19137 - Posted: 31 Oct 2010 | 14:00:59 UTC
Last modified: 31 Oct 2010 | 14:05:18 UTC

I hope GPUGRID will have also in future a CUDA V2.2 app for Windows.
I need to stay at nVIDIA driver 190.38, the fastest driver for/on my systems.


(I tested with app_info.xml file, CUDA V2.3 .dll's the GPUGRID CUDA V2.2 app.
It's look like it was little bit slower then (but wasn't a real bench-WU test). But maybe because of the app wasn't compiled with CUDA V2.3 ..)


But I have from time to time probs:
Number crunching : No progress change during calculation?

Where is the prob?

This are manufacturer OCed GTX260-216 55nm graphic cards. 4x EVGA SSC.
To now the other one, a GIGABYTE SOC (also GTX260-216 55nm) had no probs. But maybe because of too less WUs crunched.

I read now here, the old 65nm card have probs? Which kind of?
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19139 - Posted: 31 Oct 2010 | 14:15:40 UTC - in response to Message 19135.

I also picked up 6.11 tasks on both of my GTX 260 cards with 258.96 drivers.

The 6.11 app so far is running about the same speed on the GTX 260, but all 3 of my GT 240 cards have slowed down about 60%. GPU usage is still above 90% but the WUs are running so slow that it's not worth keeping them on the project with the 6.11 app. Can we PLEASE go back to 6.05?


Yes, move back to a CUDA2.2 driver.

gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19140 - Posted: 31 Oct 2010 | 14:17:32 UTC - in response to Message 19137.

Yes, there will be a CUDA2.2 app for Windows. In fact, there is already one as long as you have a cuda2.2 driver.

gdf

I hope GPUGRID will have also in future a CUDA V2.2 app for Windows.
I need to stay at nVIDIA driver 190.38, the fastest driver for/on my systems.


(I tested with app_info.xml file, CUDA V2.3 .dll's the GPUGRID CUDA V2.2 app.
It's look like it was little bit slower then (but wasn't a real bench-WU test). But maybe because of the app wasn't compiled with CUDA V2.3 ..)


But I have from time to time probs:
Number crunching : No progress change during calculation?

Where is the prob?

This are manufacturer OCed GTX260-216 55nm graphic cards. 4x EVGA SSC.
To now the other one, a GIGABYTE SOC (also GTX260-216 55nm) had no probs. But maybe because of too less WUs crunched.

I read now here, the old 65nm card have probs? Which kind of?

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19141 - Posted: 31 Oct 2010 | 14:38:00 UTC - in response to Message 19139.

I also picked up 6.11 tasks on both of my GTX 260 cards with 258.96 drivers.

The 6.11 app so far is running about the same speed on the GTX 260, but all 3 of my GT 240 cards have slowed down about 60%. GPU usage is still above 90% but the WUs are running so slow that it's not worth keeping them on the project with the 6.11 app. Can we PLEASE go back to 6.05?

Yes, move back to a CUDA2.2 driver.

gdf

You mean I have to downgrade the video drivers to go back to 6.05? Arg. What version do I have to downgrade to in order to avoid 6.11?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19144 - Posted: 31 Oct 2010 | 18:36:18 UTC - in response to Message 19141.
Last modified: 31 Oct 2010 | 18:36:46 UTC

You can choose any driver before cuda3. Look at the suggested driver here.
http://developer.nvidia.com/object/cuda_archive.html

190 would be fine.

gdf

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19146 - Posted: 31 Oct 2010 | 19:26:21 UTC - in response to Message 19133.

Would the updated application solve the problem regarding on the 65nm GTC260s?



Even the current cuda3.1 one should. Please test.

gdf


I don't really understand this, please clarify. If I want to test it, do I just check the "Run test applications" box on the preferences?

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19148 - Posted: 31 Oct 2010 | 21:22:36 UTC - in response to Message 19146.

lkiller123, if you have a GTX260-192 you could test it with the CUDA 3.1 application on XP using the 258.96 driver, still available from NVidia (achieved drivers).

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19149 - Posted: 31 Oct 2010 | 21:27:54 UTC - in response to Message 19148.

I actually have the 65nm of the GTX260-216. I have been trying to get them to work for two months already. I will try to test it again ASAP.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19150 - Posted: 31 Oct 2010 | 21:44:29 UTC - in response to Message 19148.

To run the CUDA 2.2 app on Windows you need a driver between 185.85 and 197.57 (inclusive).

To run the CUDA 3.1 app drivers from 257.21 through to 258.96 are required.

To run the CUDA 3.2 app (not presently in use, but soon to be tested) only the 260.89 or 260.99 drivers can be used.

http://www.nvidia.com/Download/Find.aspx

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19151 - Posted: 31 Oct 2010 | 22:11:27 UTC

I have Win7 64bit, does it make any difference?

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19152 - Posted: 31 Oct 2010 | 22:18:23 UTC - in response to Message 19150.

To run the CUDA 2.2 app on Windows you need a driver between 185.85 and 197.57 (inclusive).

To run the CUDA 3.1 app drivers from 257.21 through to 258.96 are required.

To run the CUDA 3.2 app (not presently in use, but soon to be tested) only the 260.89 or 260.99 drivers can be used.

http://www.nvidia.com/Download/Find.aspx

I see your GT 240 cards are also performing horribly with 6.11. Things here were running so well lately, I was wondering when the other shoe would drop. To make matters worse the GT 240 cards under 6.11 are so slow that they also miss the 1 day bonus. What reason was there to make this change? Are you rolling back to earlier drivers?

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19153 - Posted: 31 Oct 2010 | 22:21:34 UTC - in response to Message 19151.

Use the Win7 64bit driver version. There are some differences between W7 and XP drivers but the app would really be what you are testing rather than the driver.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19156 - Posted: 31 Oct 2010 | 23:20:42 UTC - in response to Message 19152.

I'm still testing 260.99 but it's a messy picture, at least for the GT240's.

It seems to work on the Win7 x64 system and on one of two Vista systems, however on the Vista x64 system I have with 4 GPU's three of the cards are running at the correct speed, but one card is much slower, 405MHz rather than 579MHz and worse, the shaders on that card seem to be synced to the core at 810Mhz. The strange thing is that the card that dropped in frequency is identical to two other cards that did not drop in frequency.
I might try to install EVGA precision again but after I try a restart; I was able to change power management to maximum power, but it made no immediate impact.

It also looks like the 3.1 performance may not be as good as with 2.2 app for the 200series, but it will take a few more tasks to come in before I can be sure of that, and only for GT240’s.

For now I will leave the 260.99 on but if it's not as fast I may have go back to 197.57 - cant remember if I need dummy plugs for that one or not, but I have a spare omnicube.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19157 - Posted: 31 Oct 2010 | 23:36:24 UTC - in response to Message 19156.
Last modified: 1 Nov 2010 | 10:24:07 UTC

In my view, after seen as Nvidia works, once a new architecture is out they don't care anymore about the older one. This happened for the g80 when the g200 came out, and to the g200 now that the fermi are out. The best driver is the one just before the change in GPU core: For g200 is the cuda2.2/2.3, and for fermi usually the latest.

That's why we give two applications out. We would like to support at least two generations, the G200 and Fermi now.

gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19162 - Posted: 1 Nov 2010 | 2:01:45 UTC - in response to Message 19157.
Last modified: 1 Nov 2010 | 2:04:38 UTC

This is why I don’t think that the latest app and driver will fix the bug for the 60nm GTX260-192; they don't make a GTX260-192 any more.

After changing the power management to maximum power, I restarted the quad GT240 system and the card is now at the desired frequency (579MHz core and 1600MHz shaders). It has stayed at that for 2h 25min, so it looks good.

This GTX295 on XP x86 seems to be doing well with the 260.99 drivers; task completion times are the same for 3.1 as for 2.2, so the frequency can't be dropping off.

Profile Sutaru Tsureku
Send message
Joined: 1 May 09
Posts: 13
Credit: 3,655,193
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 19163 - Posted: 1 Nov 2010 | 4:02:47 UTC

On my system wuid=2028574 ended in calculation error.
But on an old GTX260-192 it was fine.

If I look to the tasks overview of hostid=62699, I guess he found the trick.. ;-) (Win7 64bit, nVIDIA driver 260.63)
6.05 and 6.11 app, all WUs fine.

____________

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19164 - Posted: 1 Nov 2010 | 5:14:47 UTC - in response to Message 19163.

A bit sad to announce that 6.11 is still not working for the 65nm GTX260-216, it works, however, on the 55nm version.

Driver is 258.96
WUs that errored:
http://www.gpugrid.net/result.php?resultid=3220894
http://www.gpugrid.net/result.php?resultid=3220882

There is another 55nm 260-216 running in the same rig which works fine.

Am I doing something wrong here?

Ivailo Bonev
Avatar
Send message
Joined: 14 Jun 10
Posts: 5
Credit: 300,613
RAC: 0
Level

Scientific publications
watwat
Message 19165 - Posted: 1 Nov 2010 | 8:49:42 UTC

I don't see any differences in performance of my 9800GT between CUDA2.2 and CUDA3.1. KASHIF tasks are going well with new app, thanks.
____________

Windows: "Where do you want to go today?"
Linux: "Where do you want to go tomorrow?"
FreeBSD: "Are you guys coming, or what?"

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 670
Credit: 2,498,095,550
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19166 - Posted: 1 Nov 2010 | 8:55:43 UTC

GT240 really a lot slower with new app. I can't hit the 1 day deadlin anymore and computer freezes periodically. i can't say this is due to the new app and card driver but it started after I changed.
____________
Radio Caroline, the world's most famous offshore pirate radio station.
Great music since April 1964. Support Radio Caroline Team -
Radio Caroline

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19169 - Posted: 1 Nov 2010 | 9:32:59 UTC - in response to Message 19166.
Last modified: 1 Nov 2010 | 10:35:45 UTC

Betting Slip, you have only ran the slower IBUCH tasks since moving to CUDA 3.1. On my Fermi’s these are basically half speed tasks, using between 45% and 55% of the GPU. Wait until you have run a few faster WU's and you might want to test the new app when it comes out, just in case it turns out to be faster for 200series cards.

Sutaru Tsureku, you seem to have found a rare and endangered species - a working GTX260-192 (65nm)
one of its tasks

lkiller123, your clock speeds for both of those GTX260's is high, up from 1.24 to 1.51 and 1.49, and they are in the same rig. What are the GPU temperatures? All 3 failures came from device 1, device 0 has however not returned any completed tasks. Device 1 is at 1.49 GHz; is this the 65nm version? I would be happier to see those cards running in different systems. Putting a second card into a system usually increases the temperature of both cards, especially if you are crunching on your X4 940 as well. You might want to increase the fan speeds of both (I use EVGA Precision to do that).

All 4 of my GT240's (in the one rig) are still at the correct clocks, but it may be tomorrow before they return tasks that were running on a stable rig. So I will not be able to confirm performance differences of that Vista x64 system before then.
On my GTX470's the change in driver to 260.99 made no difference in terms of task performance; completion times were identical for same task types.

Anyone with Vista or Win7 and experiencing speed (frequency) problems should right click on the desktop, click NVidia Control Panel, on the left pane click Manage 3D Settings, on the right pane click the Program Settings Tab, make sure the Power Management Mode is set to Prefer Maximum Performance. If you had to change this you will need to restart the system to make it take effect.

On Win XP, there is no option to change from Adaptive Power Management to Prefer Maximum Performance. If anyone knows of an alternative method to force this please post it up. While this does not effect Fermi's (GTX470's at least) it does mean that some 200series cards go into power saving mode. Firmware updates might solve this, but just try getting the hold of one. I suspect the absence of a Maximum Performance option is a deliberate ploy by NVidia to undermine older generations of cards (they have done this in the past). It's like going into a garage to have your car checked over, and finding they topped up the windscreen washer reservoir but let two tyres down.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19176 - Posted: 1 Nov 2010 | 16:52:48 UTC - in response to Message 19166.
Last modified: 1 Nov 2010 | 17:00:31 UTC

GT240 really a lot slower with new app. I can't hit the 1 day deadlin anymore and computer freezes periodically. i can't say this is due to the new app and card driver but it started after I changed.

Betting Slip, all my GT 240 cards slowed at least 60% with the 6.11 app.

Betting Slip, you have only ran the slower IBUCH tasks since moving to CUDA 3.1. On my Fermi’s these are basically half speed tasks, using between 45% and 55% of the GPU. Wait until you have run a few faster WU's and you might want to test the new app when it comes out, just in case it turns out to be faster for 200series cards.

SK, don't know why you're covering up this performance loss but here's your own GT 240 machines:

http://www.gpugrid.net/results.php?hostid=51747&offset=0&show_names=1&state=3

(71,990.56 seconds for TONI_KKi4 in 6.05 versus 128,498.32 seconds for TONI_KKi4 in 6.11)

http://www.gpugrid.net/results.php?hostid=33731&offset=0&show_names=1&state=3

(66,627.82 seconds for IBUCH__pYEEI in 6.05 versus 104,038.66 seconds for -IBUCH__pYEEI in 6.11)

SK, your machines are showing even more than a 60% slowdown. Your 3rd GT 240 machine is also showing a huge slowdown but has no EXACTLY comparable WUs to post. To make matters even worse, with 6.11 they are almost always missing the 1 day bonus deadline so the credit/day is approximately (even more than) cut in half :(

Betting Slip, to get the GT 240 cards running 6.05 again I had to revert to older drivers. So far 197.45 seems to be working the best for a combination of stability and reasonable speed. Would like to hear results from others. I tried 160.38 and it was fast at first but maybe because it doesn't officially support the GT 240 it started having strange problems after running for 6-8 hours.

Had to move one machine to Collatz as it had to have newer drivers to support another application and 6.11 is too slow on the GT 240 to even bother with.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19179 - Posted: 1 Nov 2010 | 18:09:53 UTC - in response to Message 19176.

On my four GT240 card system I have 3 cards at 1599MHz and one at stock, 1340MHz. So I have to be careful to make sure I am comparing the same card.
The problem I have on comparing tasks on that system is that one of the cards dropped back to 405MHz, and I can't be sure the others did not at some stage. I also got 2 of the slower IBUCH tasks. I was not getting any of these slow IBUCH tasks with 6.05. Until now only Fermi's ran the 3.1 app, so the longer tasks are harder to finish in time for the full bonus. I expect you are right about the GT240's being slower, certainly unable to finish for the full bonus, but I just don’t have enough to go on to be sure about the performace; I have only completed 4 tasks. In a day or two I will know, and I want to wait and see how the 3.2 app works out for GT240's.

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19180 - Posted: 1 Nov 2010 | 19:20:44 UTC - in response to Message 19169.
Last modified: 1 Nov 2010 | 19:23:40 UTC

Sometimes the WUs will just error out by itself. But most of the errors come from Windows when it says "ACEMD2 6.11....... has stopped working." When I just ignore that message and let it kept crunching, the WU just hangs. So I will be forced to abort the WU through Windows.

By the way, most errors are from the 65nm card.

Maybe that might help a bit for troubleshooting.


lkiller123, your clock speeds for both of those GTX260's is high, up from 1.24 to 1.51 and 1.49, and they are in the same rig. What are the GPU temperatures? All 3 failures came from device 1, device 0 has however not returned any completed tasks. Device 1 is at 1.49 GHz; is this the 65nm version? I would be happier to see those cards running in different systems. Putting a second card into a system usually increases the temperature of both cards, especially if you are crunching on your X4 940 as well. You might want to increase the fan speeds of both (I use EVGA Precision to do that).


I can assure you that the clock speed will be okay to crunch with, I've done Folding@Home and Collatz Conjecture with those clock speeds for about a month already, no problems. However I will try to lower the clock a bit to see what will happen.

The temperature is fine. The cards run at about 60 degrees under GPUGRID load.

I will try to swap the cards around to see what happens.


I will try to test it for another day to see what happens.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19181 - Posted: 1 Nov 2010 | 19:28:50 UTC - in response to Message 19179.

I expect you are right about the GT240's being slower, certainly unable to finish for the full bonus, but I just don’t have enough to go on to be sure about the performace; I have only completed 4 tasks. In a day or two I will know, and I want to wait and see how the 3.2 app works out for GT240's.

Given the experience of the 3 of us (Betting Slip, you and I: with a total of over 10 GT 240 cards) I'd say the conclusion is clear. For some reason 6.11 is EXTREMELY slow on GT 240 cards. One strange thing is that GPU usage can stay above 90%, and the core / shader clocks remain high, yet the completion times are abysmal.

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19182 - Posted: 1 Nov 2010 | 19:40:18 UTC

More findings:

Started out with two brand new WUs.

Check out the weird pattern of GPU Usage.

GPU1 is the GTX260-216 65nm
GPU2 is the GTX260-216 55nm

GPU1 is the one that had been giving out errors.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19184 - Posted: 1 Nov 2010 | 22:11:48 UTC - in response to Message 19182.

More findings:

Started out with two brand new WUs.

Check out the weird pattern of GPU Usage.

I've seen this pattern before depending on driver version and what CPU projects are running. The solution for me is to boost the GPUGRID priority to high. I use eFMer Priority, works great:

http://www.efmer.eu/forum_tt/index.php?topic=198.0

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19185 - Posted: 2 Nov 2010 | 0:26:57 UTC

But the weird thing is, only one GPU happens to be like that. Anything wrong there?

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19186 - Posted: 2 Nov 2010 | 0:39:34 UTC - in response to Message 19185.

But the weird thing is, only one GPU happens to be like that. Anything wrong there?

Try my solution above, see if it works. Let us know.

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19187 - Posted: 2 Nov 2010 | 1:07:59 UTC
Last modified: 2 Nov 2010 | 1:10:47 UTC

Well, I aborted that particular WU and it went fine for all other WUs.

After a while of crunching, the WU which the 65nm was crunching failed. Like I said above, it crashed and Windows wants to end it.
GPU1 is the 65nm GTX260-216

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19188 - Posted: 2 Nov 2010 | 1:30:42 UTC - in response to Message 19187.

It looks like the cuda bug still effects your 65nm card, so you should stop crunching with it on GPUGrid. If the 3.2app fairs no better I think it is safe to say your 65nm card will never work here.

Sometimes a system restart can recover from that Windows error. Just keep the error message open and power down or restart the system.

Beyond, I would say I'm seeing a 10% drop in GPU utilization (Vista x64), but I would still like these tasks to finish before I compare them to my previous 6.05 tasks.

lkiller123
Send message
Joined: 24 Dec 09
Posts: 22
Credit: 15,875,809
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 19189 - Posted: 2 Nov 2010 | 1:42:27 UTC - in response to Message 19188.

Will a BIOS flash help with the situation?

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19191 - Posted: 2 Nov 2010 | 8:20:31 UTC - in response to Message 19189.

Not that we know. Apparently there's been sopme subtle hardware changes along with the die shrink.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19194 - Posted: 2 Nov 2010 | 13:18:08 UTC - in response to Message 19191.
Last modified: 2 Nov 2010 | 13:40:06 UTC

ACEMD beta 6.37 cuda2.2 for Windows and Linux is now out. Of course, these should be for g200 cards only and should not work for fermi.

Please accept beta work.

gdf

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19195 - Posted: 2 Nov 2010 | 13:18:19 UTC - in response to Message 19188.

Beyond, I would say I'm seeing a 10% drop in GPU utilization (Vista x64), but I would still like these tasks to finish before I compare them to my previous 6.05 tasks.


Well here ya go, on your Vista machine with a single GT 240:

http://www.gpugrid.net/results.php?hostid=84713&offset=0&show_names=1&state=0

Your first 6.11 WU took 63% longer than a virtually identical IBUCH_xx_PQpYEEIPI 6.05 WU. To add insult to injury you lost the 1 day bonus and received only 9,556.61 credits for the 6.11 WU compared to 11,467.93 credits for the 6.05 WU. You MUST be getting the picture by now?

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19197 - Posted: 2 Nov 2010 | 15:18:26 UTC - in response to Message 19195.
Last modified: 2 Nov 2010 | 15:25:58 UTC

For GT240's the picture is simple; the CUDA3.1 app is slower than the CUDA2.2 app. and the 260.99 drivers are crap for XP. So people with GT240s should stay with the CUDA2.2 app, and not update to the latest drivers.

That ref. card was left at Adaptive power performance (now changed to Max). The reason for leaving it was to see if the frequency would drop or stay high; I wanted to get an idea of which cards dropped the frequency under which Operating System. On my four GT240 system one cards frequency dropped, but not the others. It also occasionally went back up to full performance only to drop again (a messy picture), and sometimes on a restart stayed high for a while (same as you observed). I wanted to work out if this was also happening on single card Vista systems and W7, but this is now irrelevant as it is clear that the CUDA3.1 app is simply too slow for GT240s.

Not sure why there are beta WUs out for CUDA2.2, thought tasks for the CUDA3.2 app were going to Beta instead, hence I kept the 260.99 drivers on? It will be tomorrow before I can go back to old drivers and get CUDA2.2 tasks again.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19199 - Posted: 2 Nov 2010 | 16:09:45 UTC - in response to Message 19197.

There was a problem with the server that it was distributing beta cuda2.2 to every cards as there was no cuda3 app. Now it's fixed.

gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19200 - Posted: 2 Nov 2010 | 21:44:38 UTC - in response to Message 19199.

Linux and Windows app seem to work well for cuda2.2.
So we can think of updating the official ones.

gdf

fractal
Send message
Joined: 16 Aug 08
Posts: 87
Credit: 1,248,879,715
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19204 - Posted: 2 Nov 2010 | 23:11:03 UTC

Will this release address the Linux CPU utilization bug reported so many months ago http://www.gpugrid.net/forum_thread.php?id=2118 ?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19206 - Posted: 3 Nov 2010 | 14:53:10 UTC - in response to Message 19204.

It was not a bug, it was fix to a bug as some linux users would hang the workunits.

gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19207 - Posted: 3 Nov 2010 | 14:53:56 UTC - in response to Message 19206.

Application 6.12 cuda2.2 for linux and windows are now out.

gdf

Profile nenym
Send message
Joined: 31 Mar 09
Posts: 137
Credit: 1,308,230,581
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19208 - Posted: 3 Nov 2010 | 15:39:43 UTC - in response to Message 19207.

How to disable recieving standard tasks that aren§t friendly to GTX260 65nm rev.A2 and get beta tasks only?
A few month ago it could be set, now I do not how to do it.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19210 - Posted: 3 Nov 2010 | 16:21:54 UTC - in response to Message 19208.

The are no unfriendly tasks. Beta work is only on for a few days before any release anyway.

gdf

How to disable recieving standard tasks that aren§t friendly to GTX260 65nm rev.A2 and get beta tasks only?
A few month ago it could be set, now I do not how to do it.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19213 - Posted: 3 Nov 2010 | 16:36:38 UTC - in response to Message 19207.

Application 6.12 cuda2.2 for linux and windows are now out.

gdf

The first 6.12 app I received preempted the 6.05 task on the machine. The 6.05 task suspended at 98% completion. Luckily I caught it and susprnded the 6.12 app. Something to fix on future app releases though...

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19214 - Posted: 3 Nov 2010 | 16:48:12 UTC - in response to Message 19213.

Application 6.12 cuda2.2 for linux and windows are now out.

gdf

The first 6.12 app I received preempted the 6.05 task on the machine. The 6.05 task suspended at 98% completion. Luckily I caught it and susprnded the 6.12 app. Something to fix on future app releases though...

I've just had a v6.11 task preempt a v6.11 task......

BOINC thinks that the second task (newly downloaded) is going to miss deadline and needs High Priority, even though the estimated runtime (23 hours 48 minutes) and deadline (8 November) look normal for the machine.

I'm in the process of preparing logs and a screenshot for submitting to BOINC for analysis - I'll let you know what I get back.

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 19215 - Posted: 3 Nov 2010 | 18:05:10 UTC
Last modified: 3 Nov 2010 | 18:06:17 UTC

sorry for off-topic, but:
right now i've got GTX275 running on 195.30 driver cudatoolkit 2.3 (linux). should I downgrade to 190.38 driver cudatoolkit 2.2? which one is faster - 2.3 or 2.2?
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19216 - Posted: 3 Nov 2010 | 18:36:20 UTC - in response to Message 19215.

it's fine what you have.it should be the same speed.
gdf

sorry for off-topic, but:
right now i've got GTX275 running on 195.30 driver cudatoolkit 2.3 (linux). should I downgrade to 190.38 driver cudatoolkit 2.2? which one is faster - 2.3 or 2.2?

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 19220 - Posted: 3 Nov 2010 | 19:13:51 UTC - in response to Message 19216.

it's fine what you have.it should be the same speed.
gdf

thanks a lot :-)
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19222 - Posted: 3 Nov 2010 | 21:54:24 UTC - in response to Message 19220.

Due to the fact that cuda3.2 is still immature, we are coming out with the new application for cuda3.1 instead.
Tomorrow beta units.

gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 19223 - Posted: 3 Nov 2010 | 23:17:03 UTC

first WU (IBUCH) on 6.12 app finished. I'm running it on GTX275, 195.30 driver cudatoolkit 2.3 (linux). GPUGRID got dedicated core and runs with nice -10

1. looks it's 8-9% faster then the same WU on 6.04 app. That's amaizing :-)
2. it uses up to 20% of the core. that's awesome :-) looks I can use that core with nice 0 for CPU crunching.

the only thing - BOINC do not want to get or the project not sending the 2nd WU at all. and basically it starts to download new WU only when it finish to crunch. this means somewhat couple of minutes of idling.
____________

Profile Mattmon
Send message
Joined: 29 May 10
Posts: 2
Credit: 846,761
RAC: 0
Level
Gly
Scientific publications
watwat
Message 19226 - Posted: 4 Nov 2010 | 9:16:30 UTC

Version 6.12 is causing my work unit to run HUGELY slower than previously, when I was using version 6.04. At version 6.04, I was able to complete a work unit every 1 1/2 days, and get the 25% bonus. Now under 6.12, I have one work unit that has run 13 hours and is 9.328% done, so it is estimated to take 140 hours or 5.8 days to complete. That is almost FOUR TIMES slower. Now, forget about trying to get the bonus, I might not even make the deadline. I'm using Ubuntu Linux with a GT220. I also notice that 6.04 was using lots of CPU power, and 6.12 is not.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19229 - Posted: 4 Nov 2010 | 10:07:06 UTC - in response to Message 19228.
Last modified: 4 Nov 2010 | 11:42:09 UTC

ACEMDbeta 6.37 cuda3.1 for windows and linux are now available.
As usual you need to accept beta units.

gdf

Ivailo Bonev
Avatar
Send message
Joined: 14 Jun 10
Posts: 5
Credit: 300,613
RAC: 0
Level

Scientific publications
watwat
Message 19233 - Posted: 4 Nov 2010 | 10:59:01 UTC

All units that I get now are ok, but strange to me is that new app 6.11 uses 50% of my dual core CPU, and I see 25%-50% slowdown on my CPU projects. Anyone else experience this?
____________

Windows: "Where do you want to go today?"
Linux: "Where do you want to go tomorrow?"
FreeBSD: "Are you guys coming, or what?"

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19241 - Posted: 4 Nov 2010 | 15:05:57 UTC - in response to Message 19226.
Last modified: 4 Nov 2010 | 15:06:46 UTC

Version 6.12 is causing my work unit to run HUGELY slower than previously, when I was using version 6.04. At version 6.04, I was able to complete a work unit every 1 1/2 days, and get the 25% bonus. Now under 6.12, I have one work unit that has run 13 hours and is 9.328% done, so it is estimated to take 140 hours or 5.8 days to complete. That is almost FOUR TIMES slower. Now, forget about trying to get the bonus, I might not even make the deadline. I'm using Ubuntu Linux with a GT220. I also notice that 6.04 was using lots of CPU power, and 6.12 is not.

In WinXP-64 so far 6.12 is running about 10% slower than 6.05 on my systems, not a step forward either. In addition the new IBUCH_x_TRYP WUs are taking about the same time as earlier 13k credit WUs but yield only 10k WUs, so a double net loss.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19245 - Posted: 4 Nov 2010 | 16:56:06 UTC - in response to Message 19244.

How is it going with the cuda3.1 beta versions?

gdf
PS: This thread is for the new applications.

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19247 - Posted: 4 Nov 2010 | 17:41:42 UTC
Last modified: 4 Nov 2010 | 17:42:55 UTC

I just got one of the 6.12, and it will not finish in time, it will perhaps finish before the second one is sent again but I'm not sure.
Strange side effect: the application of all former WUs suddenly became "Not in DB" instead of "ACEMD2: GPU molecular dynamics v6.04 (cuda)".

I did old WUs of the same kind in about 60Ksec, that's less than 17h, enough time to get it back in time.
Now I'm at 10% after 16:40h, that just ridiculous.

How can I get back the good application and get rid of the junk one?

Edith likes to ask:
Why is this number crunching thread in the hardware related part of the forum?
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

BetelgeuseFive
Send message
Joined: 27 Jul 10
Posts: 1
Credit: 4,019,476
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwatwat
Message 19248 - Posted: 4 Nov 2010 | 17:46:18 UTC

Hi folks,

I just noticed that processing new tasks is a lot slower than it used to be. I'm running Linux using a GT240 and it typically took me 12 hours to complete a unit using acemd2 version 604. Now I am processing a new task using acemd2 version 612 and more than 6 hours have passed while I am not even at 20 percent. Another major change is the CPU usage. The old acemd2 executable used nearly 100 percent CPU time but the new one uses less than 5 percent. Does anyone know what changed is this new version ?

Thanks, Tom

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19250 - Posted: 4 Nov 2010 | 18:49:17 UTC

OK, I agree that I did a double post, but why is the whole thread moved from the proper software part of the forum, "Number crunching" to this hardware part "Graphic Cards"? You wouldn't expect software talk here.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19251 - Posted: 4 Nov 2010 | 19:15:40 UTC - in response to Message 19250.
Last modified: 4 Nov 2010 | 19:19:16 UTC

Re cuda3.1 beta versions,

i7-920, XP x86, Boinc 6.10.58, 2 x GTX470, 260.99:

04/11/2010 18:58:18 GPUGRID Scheduler request completed: got 0 new tasks
04/11/2010 18:58:18 GPUGRID Message from server: No work sent
04/11/2010 18:58:18 GPUGRID Message from server: No work is available for ACEMD beta version
04/11/2010 18:58:18 GPUGRID Message from server: Fermi-class GPU not supported by cuda2.2

- Just picking up normal (non-beta) work now.

Martin Aliger
Send message
Joined: 16 Jul 10
Posts: 7
Credit: 35,198,028
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19252 - Posted: 4 Nov 2010 | 19:21:55 UTC - in response to Message 19251.

same here (except I got 1x GTX470)

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19255 - Posted: 4 Nov 2010 | 20:26:20 UTC

As the current one on my computer is absolutely not runnable, probably on any machine, is there any hope to get a normal one if I abort it or will I get again the unusable rubbish I just got?
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19256 - Posted: 4 Nov 2010 | 20:29:41 UTC

@Sänger: well.. let's just say it's tradition. for most problems hard- and software go hand in hand, so in the end most stuff ended up being posted here. I guess that's why something "important for everyone" gets posted here rather than over there. Not exactly correct, though.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19257 - Posted: 4 Nov 2010 | 20:30:21 UTC - in response to Message 19255.

If your p11-IBUCH_1_opt01_pYEEI_101027-3-20-RND2304 task is not running just abort it, but let it try to run for a few minutes and please post some details.

Thanks,

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19258 - Posted: 4 Nov 2010 | 20:35:58 UTC - in response to Message 19257.

It was running for 20h and was still at 11%, no f***ing way to get it done in time.
It was all fine with 6.04, it's just the 6.12 that's behaving in this disgusting way. And as I don't do Betas, it's probably the same for all (Linux-?)users. Wasn't this tested in some Beta-set-up?
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19259 - Posted: 4 Nov 2010 | 21:08:52 UTC - in response to Message 19258.
Last modified: 4 Nov 2010 | 21:15:11 UTC

Looks like the techs moved the Betas to Live too soon, unless opt represents some sort of test WU.

Perhaps its more of an issue with IBUCH tasks than others, but I would suggest you abort it.

- I see you did. I'm worried that the next one will not be any better.

Crystal Pellet
Send message
Joined: 25 Mar 10
Posts: 18
Credit: 2,568,073
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 19260 - Posted: 4 Nov 2010 | 22:08:22 UTC - in response to Message 19245.

How is it going with the cuda3.1 beta versions?

gdf
PS: This thread is for the new applications.

The new live 6.12 version is not an improvement on my GT240 (GT215/40nm):
Longer runtime and using more cpu-time:

3234682 2040830 3 Nov 2010 19:49:05 UTC 4 Nov 2010 13:41:31 UTC Completed and validated 63,919.69 21,131.15 7,079.92 10,619.87 ACEMD2: GPU molecular dynamics v6.12 (cuda)
3228593 2037157 2 Nov 2010 15:52:37 UTC 3 Nov 2010 10:02:11 UTC Completed and validated 58,862.93 16,774.91 7,079.92 10,619.87 Not in DB <=== (v6.05)

I didn't change driver (197.13), so only change in application could cause this.
I saved the older acemd2_6.05_windows_intelx86__cuda.exe and will try to use that with an app_info.xml if this decrease in performance will not be repaired.

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19261 - Posted: 4 Nov 2010 | 22:36:12 UTC - in response to Message 19260.

I saved the older acemd2_6.05_windows_intelx86__cuda.exe and will try to use that with an app_info.xml if this decrease in performance will not be repaired.

I havn't saved it anywhere, there usually is no need to do so if you don't do Beta-stuff.
Is the right one the acemd2_6.04_x86_64-pc-linux-gnu__cuda available at http://www.gpugrid.net/download/?
The current one is at 0.77% after 1h, so I will abort it as well before too much GPU-time is wasted and rejoin DNETC again, although I don't like that science.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19263 - Posted: 4 Nov 2010 | 23:27:34 UTC

GDF wrote:

PS: This thread is for the new applications.

ExtraTerrestrial Apes wrote:

@Sänger: well.. let's just say it's tradition. for most problems hard- and software go hand in hand, so in the end most stuff ended up being posted here.


So can I post BOINC tips here, or should I start a new thread? (Answer tomorrow, approaching bedtime here) ;-)

Crystal Pellet
Send message
Joined: 25 Mar 10
Posts: 18
Credit: 2,568,073
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 19264 - Posted: 5 Nov 2010 | 9:10:30 UTC - in response to Message 19261.
Last modified: 5 Nov 2010 | 9:16:49 UTC

Sänger wrote:
Is the right one the acemd2_6.04_x86_64-pc-linux-gnu__cuda available at http://www.gpugrid.net/download/?

I don't know. The version I mentioned was for Windows.

Yet another example that the latest version is slower:

p16-IBUCH_11_PQpYEEIPI_101019-9-40-RND7287_0 2044054 4 Nov 2010 14:27:17 UTC 5 Nov 2010 08:56:36 UTC Completed and validated 66,265.58 21,205.96 7,645.29 11,467.93 ACEMD2: GPU molecular dynamics v6.12 (cuda)
p47-IBUCH_30_PQpYEEIPI_101019-9-40-RND4304_1 2033762 1 Nov 2010 22:56:28 UTC 2 Nov 2010 15:52:37 UTC Completed and validated 59,713.49 17,081.64 7,645.29 11,467.93 Not in DB

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19265 - Posted: 5 Nov 2010 | 9:54:49 UTC - in response to Message 19264.

I am going to check it locally as well.

gdf


Sänger wrote:
Is the right one the acemd2_6.04_x86_64-pc-linux-gnu__cuda available at http://www.gpugrid.net/download/?

I don't know. The version I mentioned was for Windows.

Yet another example that the latest version is slower:

p16-IBUCH_11_PQpYEEIPI_101019-9-40-RND7287_0 2044054 4 Nov 2010 14:27:17 UTC 5 Nov 2010 08:56:36 UTC Completed and validated 66,265.58 21,205.96 7,645.29 11,467.93 ACEMD2: GPU molecular dynamics v6.12 (cuda)
p47-IBUCH_30_PQpYEEIPI_101019-9-40-RND4304_1 2033762 1 Nov 2010 22:56:28 UTC 2 Nov 2010 15:52:37 UTC Completed and validated 59,713.49 17,081.64 7,645.29 11,467.93 Not in DB

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19266 - Posted: 5 Nov 2010 | 9:55:16 UTC - in response to Message 19265.

What about the cuda3.1 app in beta now?
gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19268 - Posted: 5 Nov 2010 | 10:20:42 UTC - in response to Message 19263.

So can I post BOINC tips here, or should I start a new thread? (Answer tomorrow, approaching bedtime here)

Richard, please start a new thread for Boinc Tips in the Number Crunching area.

I created a general discussion thread for Boinc Testing (Number Crunching) and moved the relevant threads from here into it. Such a discussion deserves its own thread and your contribution is much appreciated, so do continue there.

Thanks,

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19269 - Posted: 5 Nov 2010 | 10:30:34 UTC - in response to Message 19266.

What about the cuda3.1 app in beta now?
gdf

GPUGRID Message from server: Fermi-class GPU not supported by cuda2.2

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19273 - Posted: 5 Nov 2010 | 10:41:16 UTC - in response to Message 19269.

What about the cuda3.1 app in beta now?
gdf

GPUGRID Message from server: Fermi-class GPU not supported by cuda2.2

Please see Message 19234 in the server area. It's an inappropriate BOINC server response to a request for CPU work, not available from this project.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19276 - Posted: 5 Nov 2010 | 12:47:25 UTC - in response to Message 19273.

It looks like I did not manage to pick up any of the 36 Betas.

We use to be able to select which ACEMD apps to crunch, and if I remember correctly, be able to crunch Betas only by unchecking both ACEMD apps.

Good luck,

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19277 - Posted: 5 Nov 2010 | 13:40:23 UTC - in response to Message 19264.

Sänger wrote:
Is the right one the acemd2_6.04_x86_64-pc-linux-gnu__cuda available at http://www.gpugrid.net/download/?

I don't know. The version I mentioned was for Windows.

Yep, yours was windows, mine linux, I just wanted to know whether those are the right ones in that place.

And I'd like to have an app-info.xml posted by someone here, as I'm no programmer and I don't know what has to be included to get the good app running proper and not be replaced by 6.12.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19278 - Posted: 5 Nov 2010 | 18:58:25 UTC

If I look at my aborted WUs, and have a peek at my actual running one, it seems that it is behaving now as it proclaims, i.e. doesn't use any CPU, but unfortunately that doesn't look like a good set-up, as I won't finish it in time.

The CPU-time is next to 0 compared to run-time, and the CPU-use in the system monitor in my Linux is as well 0% (+4% for Xorg), the other parallel WUs on the CPU get the right percentage.

OK, I've asked for something like that in some older posts, but as it seems to be a not-working set-up, I think my second proposal, taking officially a whole CPU, would be better.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19280 - Posted: 5 Nov 2010 | 21:41:09 UTC - in response to Message 19278.

taking officially a whole CPU, would be better.


Whenever this is done (on purpose or not) people get mad at the devs, no matter the actual GPU-Grid application performance. So far the solution has been to let power users configure full throttle use by creating the environment variable (don't know where to do this in linux) "SWAN_SYNC" and setting its value to 0. That might restore your speed.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19291 - Posted: 6 Nov 2010 | 9:36:13 UTC

Here I found an old app_info.xml from Collatz (which imho has even more useless science than DNETC):

<app_info>
<app>
<name>collatz</name>
<user_friendly_name>collatz</user_friendly_name>
</app>
<file_info>
<name>collatz_2.02_x86_64-pc-linux-gnu__cuda22</name>
<executable/>
</file_info>
<file_info>
<name>libcudart.so.2</name>
<executable/>
</file_info>
<app_version>
<app_name>collatz</app_name>
<version_num>202</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>0.05</max_ncpus>
<coproc>
<type>CUDA</type>
<count>1.0</count>
</coproc>
<file_ref>
<file_name>collatz_2.02_x86_64-pc-linux-gnu__cuda22</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libcudart.so.2</file_name>
</file_ref>
</app_version>
</app_info>


If I replace all stuff collatz-related and put in some GPUGrid stuff, i.e. replace the following, will it work?

Would look like this:
<app_info>
<app>
<name>GPUGrid</name>
<user_friendly_name>GPUGrid</user_friendly_name>
</app>
<file_info>
<name>acemd2_6.04_x86_64-pc-linux-gnu__cuda</name>
<executable/>
</file_info>
<file_info>
<name>libcudart.so.2</name>
<executable/>
</file_info>
<app_version>
<app_name>GPUGrid</app_name>
<version_num>604</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>0.05</max_ncpus>
<coproc>
<type>CUDA</type>
<count>1.0</count>
</coproc>
<file_ref>
<file_name>acemd2_6.04_x86_64-pc-linux-gnu__cuda</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libcudart.so.2</file_name>
</file_ref>
</app_version>
</app_info>


I won't test it before someone with more insight in such stuff says something about it, because I simply changed some names and kept things that looked kind of universal.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19292 - Posted: 6 Nov 2010 | 9:57:48 UTC - in response to Message 19291.

You don't need <user_friendly_name>, but you might need a <plan_class>.

Have a check in the documentation.

Crystal Pellet
Send message
Joined: 25 Mar 10
Posts: 18
Credit: 2,568,073
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 19295 - Posted: 6 Nov 2010 | 12:09:37 UTC - in response to Message 19265.
Last modified: 6 Nov 2010 | 12:10:54 UTC

I am going to check it locally as well.

gdf


I tested the former 6.05 windows-application with an app_info file and the result is fine: p9-IBUCH_1_opt01_pYEEI_101027-4-20-RND8089_0
and the speed is as it was before the version change.

The BOINC client behaviour is still a bit strange, because the app_info.xml isn't perfect. I will test this further.
Can you or skgiven tell me what are the right dll's to use with v6.05.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19299 - Posted: 6 Nov 2010 | 14:03:19 UTC - in response to Message 19295.

I think it’s down to the Scientists/Techs to determine and use the best app and driver combos available for the present ranges of GPUs, even if it means using the older apps, and I don't appreciate investing in Fermi's only to see them run at 50% GPU Utilization, when they could be and have been running at 98%.

I see little point in crunchers overclocking and optimising their input only to have their contribution crippled by being forced to use poor app/driver combos. So to spend time trying to DIY an app that we already had makes little sense.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19303 - Posted: 6 Nov 2010 | 15:55:00 UTC - in response to Message 19291.

Here I found an old app_info.xml from Collatz (which imho has even more useless science than DNETC):

While the value of DNETC may in question, I think it's short sighted to discount research projects like Collatz that deal with math theory. Without the advances in mathematics over the last centuries modern science would be impossible. Added to that Collatz is IMO one of the best run distributed projects around. Maintained primarily by one person (with a little help from his friends) it was largely responsible for getting ATI GPUs natively integrated into BOINC. Many platforms are supported. The clients are rock solid. Client upgrades are actually upgrades and the changes are fully explained. Project communication is amazing and problems are dealt with quickly. Those who prefer other types of research don't have to run it. Math science aside, as a model of a well run project it earns accolades.

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19305 - Posted: 6 Nov 2010 | 16:31:53 UTC - in response to Message 19299.
Last modified: 6 Nov 2010 | 16:36:53 UTC

So to spend time trying to DIY an app that we already had makes little sense.

As much as I agree that the science team has to deliver some runnable apps, in the meantime while they sent out not runnable apps and have good ones still in stock just not delivered, it's a better utilisation of my GPU as the current set-up with another project of dubious benefit.

As long as the the current app is sent to clients it's a giant waste of resources, and everything that helps to contain that waste has to be considered "good".

Edith says some OT things ;)
I think there are good mathematical projects a) possible and b) already running, and I think there are worse project benefit-wise than DNETC and Collatz, every single Primeproject is imho definitely worse ;) But unfortunately for my GT240 there are no much other projects available under Linux.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Profile Mattmon
Send message
Joined: 29 May 10
Posts: 2
Credit: 846,761
RAC: 0
Level
Gly
Scientific publications
watwat
Message 19307 - Posted: 6 Nov 2010 | 17:42:55 UTC - in response to Message 19305.

I have uncovered the reason why Linux version 6.12 is running so slow. I figured out how to get the GPU utilization.

$ nvidia-smi -a -l
This shows that the GPU utilization fluctuates between 2% and 40%. This explains very well why it is taking four times longer.

Crystal Pellet
Send message
Joined: 25 Mar 10
Posts: 18
Credit: 2,568,073
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 19308 - Posted: 6 Nov 2010 | 17:52:26 UTC - in response to Message 19305.
Last modified: 6 Nov 2010 | 18:19:31 UTC

The new Windows 6.12(cuda) application is (several times stated in this thread)
not performing as well as the older 6.05 version. On my GT240 runtimes are about
10% longer and the new application uses 25% more CPU-time.

Disclaimer: Use this on your on risk

For Windows users, with a single card (from the G200-series), who wants to run the better performing 6.05 version you need following files:

acemd2_6.05_windows_intelx86__cuda.exe
app_info.xml
cudart.dll
cufft.dll
logops3grid.png
project_1.png
project_2.png
project_3.png
slideshow_cellmd_00
slideshow_cellmd_01
slideshow_cellmd_02
slideshow_ga_00
slideshow_ga_01
slideshow_ga_02
stat_icon
tcl85.dll

Procedure:

1. If you are out of work, detach from GPUGRID.
2. Create under projects the www.gpugrid.net directory.
3. Copy above files into that directory.
4. Attach to GPUGRID again.

I didn't have experience until now what is happening if new work is requested when a GPUGRID task is running.

Contents of app_info.xml:

<app_info>
<app>
<name>acemd2</name>
</app>
<file_info>
<name>acemd2_6.05_windows_intelx86__cuda.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>acemd2</app_name>
<version_num>605</version_num>
<plan_class>cuda</plan_class>
<avg_ncpus>0.260000</avg_ncpus>
<max_ncpus>0.260000</max_ncpus>
<flops>2923397447.700002</flops>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>acemd2_6.05_windows_intelx86__cuda.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>


Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19309 - Posted: 6 Nov 2010 | 19:27:21 UTC - in response to Message 19308.

The new Windows 6.12(cuda) application is (several times stated in this thread)
not performing as well as the older 6.05 version. On my GT240 runtimes are about
10% longer and the new application uses 25% more CPU-time.

Disclaimer: Use this on your on risk

For Windows users, with a single card (from the G200-series), who wants to run the better performing 6.05 version you need following files:

acemd2_6.05_windows_intelx86__cuda.exe
cudart.dll
cufft.dll
tcl85.dll
....

If you need those three DLLs (away from my rigs at the moment, can't check directly), then they should be referenced in app_info.xml - twice each, once as <file_info>, marked </executable> like the main program is, and again as a <file_ref> in the app_version> section. You don't need to worry about that for the eye-candy files.

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19310 - Posted: 6 Nov 2010 | 20:01:43 UTC - in response to Message 19307.

I have uncovered the reason why Linux version 6.12 is running so slow. I figured out how to get the GPU utilization.
$ nvidia-smi -a -l
This shows that the GPU utilization fluctuates between 2% and 40%. This explains very well why it is taking four times longer.

And the old one took 1 complete CPU core, now it's running on the CPU as claimed in the description (0.15CPU+1GPU) but never did in reality.

Any comments about my proposed app_info?
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Crystal Pellet
Send message
Joined: 25 Mar 10
Posts: 18
Credit: 2,568,073
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 19314 - Posted: 6 Nov 2010 | 20:49:15 UTC - in response to Message 19309.

Richard wrote:
If you need those three DLLs (away from my rigs at the moment, can't check directly), then they should be referenced in app_info.xml - twice each, once as <file_info>, marked </executable> like the main program is, and again as a <file_ref> in the app_version> section. You don't need to worry about that for the eye-candy files.

I know the other files aren't needed really and would be downloaded from the server.
As I described, without the dll's in the app_info, it's working now.

I tried with the dll's in the app_info file, but after starting the boinc-client they disappeared immediately. The main program needs the dll's.
Probably I put them only once in the app_info. After I'm sure that requesting new work and starting the new task is functioning well, I'll try your proposal.

Werkstatt
Send message
Joined: 23 May 09
Posts: 121
Credit: 321,525,386
RAC: 177,358
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19316 - Posted: 7 Nov 2010 | 0:03:45 UTC - in response to Message 19305.


As long as the the current app is sent to clients it's a giant waste of resources, and everything that helps to contain that waste has to be considered "good".

There is one thing you can do to speed up crunching a little bit ( ~10% )
http://www.gpugrid.net/workunit.php?wuid=2043240
This wu was crunched with default setting
http://www.gpugrid.net/workunit.php?wuid=2044504
http://www.gpugrid.net/workunit.php?wuid=2044547
These two did run together with the setting 0.5 GPU's (done in client_state.xml)
Should be working with an app_info, setting <count>0.5</count>
GPU-usage increased from ~65% to ~79% (Afterburner)
Memory-usage was ~760MB (hard on limit!)
I'll check it tomorrow with the app_info, it's too late now for me.
Alexander

Werkstatt
Send message
Joined: 23 May 09
Posts: 121
Credit: 321,525,386
RAC: 177,358
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19320 - Posted: 7 Nov 2010 | 8:10:35 UTC

Looks good at the moment.





mem-usage ~605MB

cristipurdel
Send message
Joined: 31 Mar 10
Posts: 45
Credit: 103,429,292
RAC: 0
Level
Cys
Scientific publications
watwatwatwat
Message 19321 - Posted: 7 Nov 2010 | 8:49:35 UTC - in response to Message 19320.

Looks good at the moment.





mem-usage ~605MB

WUprop with nci, sweet.
Can it wotk side by side with freehal, or is the boinc manager limited to just one nci project?

Werkstatt
Send message
Joined: 23 May 09
Posts: 121
Credit: 321,525,386
RAC: 177,358
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19323 - Posted: 7 Nov 2010 | 9:33:04 UTC - in response to Message 19321.


WUprop with nci, sweet.
Can it wotk side by side with freehal, or is the boinc manager limited to just one nci project?

Sorry, no idea!
I found a hint in a forum (planet3D???), they are collecting statistical data, cpu-load is very low, so why not?
Alexander

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19324 - Posted: 7 Nov 2010 | 9:37:19 UTC - in response to Message 19321.
Last modified: 7 Nov 2010 | 9:39:36 UTC

WUprop with nci, sweet.
Can it wotk side by side with freehal, or is the boinc manager limited to just one nci project?

It can work with as much nci-projects as are available, currently WUProp and QCN, only FreeHAL ain't one, it's just a CPU-intensive project with lots of useless pauses.

Edith says:
Of course most GPU projects are nci-projects as well.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Crystal Pellet
Send message
Joined: 25 Mar 10
Posts: 18
Credit: 2,568,073
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 19339 - Posted: 7 Nov 2010 | 20:50:37 UTC - in response to Message 19314.

Richard wrote:
If you need those three DLLs (away from my rigs at the moment, can't check directly), then they should be referenced in app_info.xml - twice each, once as <file_info>, marked </executable> like the main program is, and again as a <file_ref> in the app_version> section. You don't need to worry about that for the eye-candy files.

I know the other files aren't needed really and would be downloaded from the server.
As I described, without the dll's in the app_info, it's working now.

I tried with the dll's in the app_info file, but after starting the boinc-client they disappeared immediately. The main program needs the dll's.
Probably I put them only once in the app_info. After I'm sure that requesting new work and starting the new task is functioning well, I'll try your proposal.

My former setup with v6.05 and an app_info.xml was working well.
For correctness I changed my app_info.xml into:

<app_info>
<app>
<name>acemd2</name>
</app>
<file_info>
<name>acemd2_6.05_windows_intelx86__cuda.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart.dll</name>
<executable/>
</file_info>
<file_info>
<name>cufft.dll</name>
<executable/>
</file_info>
<file_info>
<name>tcl85.dll</name>
<executable/>
</file_info>
<app_version>
<app_name>acemd2</app_name>
<version_num>605</version_num>
<plan_class>cuda</plan_class>
<avg_ncpus>0.260000</avg_ncpus>
<max_ncpus>0.260000</max_ncpus>
<flops>2923397447.700002</flops>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>acemd2_6.05_windows_intelx86__cuda.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft.dll</file_name>
</file_ref>
<file_ref>
<file_name>tcl85.dll</file_name>
</file_ref>
</app_version>
</app_info>

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19349 - Posted: 8 Nov 2010 | 22:03:38 UTC

Any news from the project team?
Why is there still the rubbish app available?
Are they not interested in a usable one?
The treatment we get from them feels like a kick in the gut.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19353 - Posted: 8 Nov 2010 | 23:01:23 UTC - in response to Message 19349.

Yes, I get annoyed too. We have to remember this is a small team with limited resources and time constraints. To reconfigure the server with the old app and work on a replacement might take more time than just working on the replacement app and uploading that when it’s ready. GDF said they are looking into the problems and hope to see something out this week.

Good Luck,

Profile Saenger
Avatar
Send message
Joined: 20 Jul 08
Posts: 134
Credit: 23,657,183
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 19356 - Posted: 9 Nov 2010 | 5:26:52 UTC

They just have to reinstall the 6.04 again, no real rocket science. Keeping that completely broken stuff online for even an hour longer is wilful neglect and making the middle finger curse to us.

If they let those alpha-stuff loose on unsuspecting non-testers they have to take it back asap after it's clear it's not working. and that's clear since a whole week.
____________
Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19363 - Posted: 9 Nov 2010 | 12:06:59 UTC - in response to Message 19339.

this is just a waste of time as the new workunits will not work on old applications.

gdf

Richard wrote:
If you need those three DLLs (away from my rigs at the moment, can't check directly), then they should be referenced in app_info.xml - twice each, once as <file_info>, marked </executable> like the main program is, and again as a <file_ref> in the app_version> section. You don't need to worry about that for the eye-candy files.

I know the other files aren't needed really and would be downloaded from the server.
As I described, without the dll's in the app_info, it's working now.

I tried with the dll's in the app_info file, but after starting the boinc-client they disappeared immediately. The main program needs the dll's.
Probably I put them only once in the app_info. After I'm sure that requesting new work and starting the new task is functioning well, I'll try your proposal.

My former setup with v6.05 and an app_info.xml was working well.
For correctness I changed my app_info.xml into:

<app_info>
<app>
<name>acemd2</name>
</app>
<file_info>
<name>acemd2_6.05_windows_intelx86__cuda.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart.dll</name>
<executable/>
</file_info>
<file_info>
<name>cufft.dll</name>
<executable/>
</file_info>
<file_info>
<name>tcl85.dll</name>
<executable/>
</file_info>
<app_version>
<app_name>acemd2</app_name>
<version_num>605</version_num>
<plan_class>cuda</plan_class>
<avg_ncpus>0.260000</avg_ncpus>
<max_ncpus>0.260000</max_ncpus>
<flops>2923397447.700002</flops>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>acemd2_6.05_windows_intelx86__cuda.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft.dll</file_name>
</file_ref>
<file_ref>
<file_name>tcl85.dll</file_name>
</file_ref>
</app_version>
</app_info>

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19365 - Posted: 9 Nov 2010 | 12:56:13 UTC - in response to Message 19363.

this is just a waste of time as the new workunits will not work on old applications.

gdf

This is good to know. But looking through recent posts, I don't think you've previously made it clear that the workunits were changing too - all the talk was about applications. In fact, message 18984 gives rather the opposite impression.

It might have helped avoid some of the confusion if there was somewhere where you could give an overview of the intention behind the whole batch of app changes, and the implications - or if that's too time-consuming, brief the moderators, and let them pass it on to the rest of us.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 19366 - Posted: 9 Nov 2010 | 13:42:25 UTC - in response to Message 19365.

The reason to update the cuda2.2 applications is that the new ones have a new algorithm which should be 30% faster for any card. This feature must be activated with new workunits and will crash on old apps. Even without this feature, new applications should be equally fast on G200 cards and slightly faster on Fermi cards.

I have now created a more specific thread to comment on them.

gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19372 - Posted: 9 Nov 2010 | 16:54:09 UTC - in response to Message 19366.
Last modified: 9 Nov 2010 | 17:04:06 UTC

Think of one task type for all cards, and a separate Fermi application and non-Fermi application for Windows platforms and again separate Fermi and non-Fermi applications for Linux platforms.

6.12 (cuda for non-fermi cards), 6.11 (cuda 3.1 for Fermi card) and the old 6.04 and 6.05 are examples of windows application types.

In reality there are project based task separations (IBUCH, HIV...), but an IBUCH or HIV task type can crunch on both 6.11 (the Fermi app) and 6.12 (the non-Fermi app), ditto for the two Linux apps.

The ACEMD2/3 CUDA applications utilize routines and files within NVidias Cuda developers kit, and the cards depend on NVidia drivers. So changes in drivers or the developers kit can improve application/task performance or hinder it, depending on the card.

The scientists have to facilitate the newest and fastest cards in order to expedite the research which is twofold; developing and optimizing GPU research for Molecular simulations, and the actual research itself, into folding pathways for example.

Hope that makes things clearer.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,603,361,851
RAC: 8,788,289
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19374 - Posted: 9 Nov 2010 | 17:29:33 UTC - in response to Message 19372.

Think of one task type for all cards, and separate ... application(s)

That was my understanding as well, but

GDF wrote:
the new workunits will not work on old applications

rather contradicts our shared assumption.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 19394 - Posted: 10 Nov 2010 | 0:05:17 UTC - in response to Message 19374.

It is difficult trying to second guess, predict or retrospectively interpret changes, especially if you are use to project management.

I think the present set of applications are interoperable WRT running tasks, but just not fully backward compatible; a task for the 6.11 app is the same as a task for the 6.12 app, but neither is guaranteed to work with the old app that ran 6.05 tasks. The apps just make the task work on different cards, subject to using an app compatible driver.

The move to new applications is to utilize new subroutines only available within the newer CUDA apps, and to facilitate new cards, such as the GTX580 (for which there is a 262.99 driver, released today, along with the card).
GPUGrid is limited in that it cannot optimize to individual cards; to do so would require multiple servers and more staff.

Post to thread

Message boards : Graphics cards (GPUs) : Nov2010> New ACEMD application

//