Author |
Message |
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Hello, just added a new workstation with an nvidia card and the current driver for it (GTX 750 and 335.23). On the server status page over 400 long run work units and 800 short run work units are reported as available.
When I try to get no work, the report via BOINC is no work available.
|
|
|
TJSend message
Joined: 26 Jun 09 Posts: 815 Credit: 1,470,385,294 RAC: 0 Level
Scientific publications
|
Checking your computers I don't see the GTX750, but one PC has work done and has work in progress, so that works. Perhaps you need to check the settings on the other PC if it may run on the GPU.
____________
Greetings from TJ |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Thanks for the response.
Last night I tried a full uninstall and resinstall of BOINC on that system -- including first deleting all traces of of BOINC in the Program Data directory.
The reinstall was clean. Didn't solve the problem.
The computer which gets GPUGrid GPU workunits is running Windows 7 64 bit,
a GT640 video card, BOINC 7.2.42, nvidea driver 335.23.
The computer which isn't getting GPUGrid GPU workunits is running Windows 7 64 bit, a GTX650 video card, BOINC 7.2.42, nvidea driver 335.23.
So the only variable I can see is the GT640 versus the GTX650.
I can download and process GPU work units from Prime Grid, and Collatz with either system. So the other variable if you will is GPUGrid.
Hopefully someone will be able to enlighten here.
|
|
|
|
Hi,
Is that computer set to the same location profile in GPUgrids preferences as the one that is working?
...regards |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Yes -- actually simply set to default -- I haven't configured separate groups.
I have a computer set to drop off due to it no longer running BOINC -- it was getting work with a very similar configuration (earlier sub versions of BOINC and the Nvidea driver, but also a GTX-650.
|
|
|
|
Could you copy & paste the first 30 or so lines of the Boinc event log just after it starts. |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Here you go -- this is from my restart a few minutes ago:
4/29/2014 12:41:27 PM | | cc_config.xml not found - using defaults
4/29/2014 12:41:27 PM | | Starting BOINC client version 7.2.42 for windows_x86_64
4/29/2014 12:41:27 PM | | log flags: file_xfer, sched_ops, task
4/29/2014 12:41:27 PM | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
4/29/2014 12:41:27 PM | | Data directory: C:\ProgramData\BOINC
4/29/2014 12:41:27 PM | | Running under account Barry
4/29/2014 12:41:27 PM | | CUDA: NVIDIA GPU 0: GeForce GTX 650 (driver version 335.23, CUDA version 6.0, compute capability 1.1, 2048MB, 1983MB available, 396 GFLOPS peak)
4/29/2014 12:41:27 PM | | OpenCL: NVIDIA GPU 0: GeForce GTX 650 (driver version 335.23, device version OpenCL 1.0 CUDA, 2048MB, 1983MB available, 396 GFLOPS peak)
4/29/2014 12:41:27 PM | | OpenCL CPU: AMD Phenom(tm) II X4 955 Processor (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 2.0 (sse2), device version OpenCL 1.2 AMD-APP (938.2))
4/29/2014 12:41:27 PM | | Host name: BSC-GA970-955
4/29/2014 12:41:27 PM | | Processor: 4 AuthenticAMD AMD Phenom(tm) II X4 955 Processor [Family 16 Model 4 Stepping 3]
4/29/2014 12:41:27 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 htt pni cx16 popcnt syscall nx lm svm sse4a osvw ibs skinit wdt page1gb rdtscp 3dnowext 3dnow
4/29/2014 12:41:27 PM | | OS: Microsoft Windows 7: Home Premium x64 Edition, Service Pack 1, (06.01.7601.00)
4/29/2014 12:41:27 PM | | Memory: 7.98 GB physical, 15.96 GB virtual
4/29/2014 12:41:27 PM | | Disk: 111.57 GB total, 75.24 GB free
4/29/2014 12:41:27 PM | | Local time is UTC -7 hours
4/29/2014 12:41:27 PM | Poem@Home | URL http://boinc.fzk.de/poem/; Computer ID 300353; resource share 2000
4/29/2014 12:41:27 PM | GPUGRID | URL http://www.gpugrid.net/; Computer ID 173624; resource share 8000
4/29/2014 12:41:27 PM | PrimeGrid | URL http://www.primegrid.com/; Computer ID 431391; resource share 4000
4/29/2014 12:41:27 PM | | General prefs: from http://moowrap.net/ (last modified 20-Jun-2011 09:11:42)
4/29/2014 12:41:27 PM | | Host location: none
4/29/2014 12:41:27 PM | | General prefs: using your defaults
4/29/2014 12:41:27 PM | | Reading preferences override file
4/29/2014 12:41:27 PM | | Preferences:
4/29/2014 12:41:27 PM | | max memory usage when active: 8172.66MB
4/29/2014 12:41:27 PM | | max memory usage when idle: 8172.66MB
4/29/2014 12:41:27 PM | | max disk usage: 25.00GB
4/29/2014 12:41:27 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
4/29/2014 12:41:27 PM | | Not using a proxy
4/29/2014 12:41:28 PM | GPUGRID | Sending scheduler request: To fetch work.
4/29/2014 12:41:28 PM | GPUGRID | Requesting new tasks for NVIDIA
4/29/2014 12:41:31 PM | GPUGRID | Scheduler request completed: got 0 new tasks
4/29/2014 12:41:31 PM | GPUGRID | No tasks sent
4/29/2014 12:41:31 PM | GPUGRID | No tasks are available for Short runs (2-3 hours on fastest card)
4/29/2014 12:41:31 PM | GPUGRID | No tasks are available for ACEMD beta version
4/29/2014 12:41:31 PM | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
4/29/2014 12:41:31 PM | GPUGRID | Tasks for CPU are available, but your preferences are set to not accept them
|
|
|
TJSend message
Joined: 26 Jun 09 Posts: 815 Credit: 1,470,385,294 RAC: 0 Level
Scientific publications
|
Two things I can think off.
1 You have had to many error lately, then you are banned for new work for 24 hours. I don't know the message about that in BOINC as I have never seen it. I don't think this is the case with your PC.
2 You have other work running on the GPU that has a long time to finish. You are not getting new work then. But normally the message is then "no new work, not needed".
Have you a CPU core free?
____________
Greetings from TJ |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Thanks for the follow up. I am doing other BOINC work on that system, but I'd get a not needed in that situation. Been there done that.
There are available CPU -- currently running either POEM or PrimeGrid. Current PrimeGrid is using the GPU as it had no problem with this configuration.
As to errors and lockout -- don't think this applies here. After all, it isn't error messages, just a curious 'no work sent'.
Now I have seen that before with other applications on BOINC, but in this case I encountered this with a fresh clean install of BOINC and no other project attached to begin with.
|
|
|
|
For some reason, the GPU is being reported as compute capability 1.1, which I suspect is why it's not getting work. I would try an earlier driver.
What does your 640 report as? |
|
|
|
For some reason, the GPU is being reported as compute capability 1.1, which I suspect is why it's not getting work. I would try an earlier driver.
What does your 640 report as?
Absolutely - CC 1.1 is dead wrong for that card.
In the opening post of this thread, the problematic card was a GTX 750, which should report as CC 5.0. I forget the exact value for a GTX 650, but IIRC it was 3.0. CC 1.1 would take you back about 5 years, to the 9xxxGT series that I've just junked - they certainly wouldn't have run current GPUGrid work. |
|
|
|
Regarding the GTX 650 erroneously reporting as Compute Capability 1.1 ....
... are we looking at a BOINC bug here, or an NVIDIA bug? Is there any way to tell? |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
OK -- thanks, I will try a driver uninstall, then install an older driver -- say from about 6 months ago.
I'll report back tomorrow.
|
|
|
|
Regarding the GTX 650 erroneously reporting as Compute Capability 1.1 ....
... are we looking at a BOINC bug here, or an NVIDIA bug? Is there any way to tell?
I don't think it's BOINC. BOINC just calls the functions in an NVidia DLL, and reports on the answers it gets. Look at this combo:
18-Apr-2014 11:14:56 [---] Starting BOINC client version 6.12.34 for windows_intelx86
18-Apr-2014 11:14:56 [---] NVIDIA GPU 0: GeForce GTX 750 Ti (driver version 33528, CUDA version 6000, compute capability 5.0, 2048MB, 410 GFLOPS peak)
Even a three year old version of BOINC can report CC 5.0 accurately - though it goes on to be completely confused about GFLOPs peak (that would be shown as 1640 by an alpha BOINC that has the correct cores per SM for a Maxwell). BarryAZ is using BOINC v7.2.42 - we'd know all about it if that version had a bug as gross as that.
I don't know which other GPU tools report compute capability, but I'm sure somebody here will.
@ BarryAZ - I don't think the age of the driver matters: just choose one certified for your card, and go through the 'clean install' procedure to remove any stray components from some older installation. |
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
Sounds like a driver issue - possibly a failure. Suggest downloading the drivers again (try the latest Beta) and performing an advanced clean re-installation of NVidia drivers, under an appropriate account.
Wouldn't hurt to reinstall Boinc afterwards, making sure that its not a service installation (should the drivers not fix the problem).
GPU Caps Viewer is a tool that tells you the GPU's Compute Capability, among many other things. Look under the CUDA Tab.
You can find it at http://www.softpedia.com.
The GK104, GK106 and GK107 cards (most GeForce 600 cards) are CC3.0. GK110 is CC3.5. There are a few Fermi models at the lower end of the GeForce rage but the GTX650 and larger cards are all Kepler models.
____________
FAQ's
HOW TO:
- Opt out of Beta Tests
- Ask for Help |
|
|
|
Regarding the GTX 650 erroneously reporting as Compute Capability 1.1 ....
... are we looking at a BOINC bug here, or an NVIDIA bug? Is there any way to tell?
I don't think it's BOINC. BOINC just calls the functions in an NVidia DLL, and reports on the answers it gets. Look at this combo:
18-Apr-2014 11:14:56 [---] Starting BOINC client version 6.12.34 for windows_intelx86
18-Apr-2014 11:14:56 [---] NVIDIA GPU 0: GeForce GTX 750 Ti (driver version 33528, CUDA version 6000, compute capability 5.0, 2048MB, 410 GFLOPS peak)
Even a three year old version of BOINC can report CC 5.0 accurately - though it goes on to be completely confused about GFLOPs peak (that would be shown as 1640 by an alpha BOINC that has the correct cores per SM for a Maxwell). BarryAZ is using BOINC v7.2.42 - we'd know all about it if that version had a bug as gross as that.
I don't know which other GPU tools report compute capability, but I'm sure somebody here will.
@ BarryAZ - I don't think the age of the driver matters: just choose one certified for your card, and go through the 'clean install' procedure to remove any stray components from some older installation.
BarryAZ:
I would also like to see the first 20-or-so lines of BOINC, reacting to the latest NVIDIA Beta Drivers (either 337.50 or 337.61). At some point, could you install one of those driver versions, and post the BOINC startup logs? Could you also provide what GPU Caps Viewer says using those latest drivers? If the most-recent driver is still bugged, I can try to contact NVIDIA for you. |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Folks, thanks for all the suggestions -- no joy here.
I have done several full uninstall/reinstall cycles of BOINC and various different Nvidea drivers. All yield precisely the same frustrating result. For the edification of the collective here I am including the last of these efforts:
For some reason, it seems that GPUGrid or Nvidea doesn't like this particular card. Curiously, other BOINC applications will feed GPU work to this card -- Collatz, MooWrap, PrimeGrid. I suspect that POEM, once they get their application running will work as well.
I don't know that there is any other 'monkey/keyboard/shakespeer tests I can do to get this to work at this point, so, absent a silver bullet, I think I will desist for now from the 'try this, try that, sequence for now. I will set this system up to process GPU work units for one or two of the other projects which doesn't have this curious problem.
4/30/2014 9:48:59 AM | | cc_config.xml not found - using defaults
4/30/2014 9:48:59 AM | | Starting BOINC client version 7.2.42 for windows_x86_64
4/30/2014 9:48:59 AM | | log flags: file_xfer, sched_ops, task
4/30/2014 9:48:59 AM | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
4/30/2014 9:48:59 AM | | Data directory: C:\BOINC2
4/30/2014 9:48:59 AM | | Running under account Barry
4/30/2014 9:48:59 AM | | CUDA: NVIDIA GPU 0: GeForce GTX 650 (driver version 337.50, CUDA version 6.0, compute capability 1.1, 2048MB, 1983MB available, 396 GFLOPS peak)
4/30/2014 9:48:59 AM | | OpenCL: NVIDIA GPU 0: GeForce GTX 650 (driver version 337.50, device version OpenCL 1.0 CUDA, 2048MB, 1983MB available, 396 GFLOPS peak)
4/30/2014 9:48:59 AM | | Host name: BSC-GA970-955
4/30/2014 9:48:59 AM | | Processor: 4 AuthenticAMD AMD Phenom(tm) II X4 955 Processor [Family 16 Model 4 Stepping 3]
4/30/2014 9:48:59 AM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 htt pni cx16 popcnt syscall nx lm svm sse4a osvw ibs skinit wdt page1gb rdtscp 3dnowext 3dnow
4/30/2014 9:48:59 AM | | OS: Microsoft Windows 7: Home Premium x64 Edition, Service Pack 1, (06.01.7601.00)
4/30/2014 9:48:59 AM | | Memory: 7.98 GB physical, 15.96 GB virtual
4/30/2014 9:48:59 AM | | Disk: 111.57 GB total, 73.98 GB free
4/30/2014 9:48:59 AM | | Local time is UTC -7 hours
4/30/2014 9:48:59 AM | GPUGRID | URL http://www.gpugrid.net/; Computer ID 173730; resource share 8000
4/30/2014 9:48:59 AM | | General prefs: from http://moowrap.net/ (last modified 20-Jun-2011 09:11:42)
4/30/2014 9:48:59 AM | | Host location: none
4/30/2014 9:48:59 AM | | General prefs: using your defaults
4/30/2014 9:48:59 AM | | Reading preferences override file
4/30/2014 9:48:59 AM | | Preferences:
4/30/2014 9:48:59 AM | | max memory usage when active: 4086.33MB
4/30/2014 9:48:59 AM | | max memory usage when idle: 7355.40MB
4/30/2014 9:48:59 AM | | max disk usage: 10.00GB
4/30/2014 9:48:59 AM | | (to change preferences, visit a project web site or select Preferences in the Manager)
4/30/2014 9:48:59 AM | | Not using a proxy
4/30/2014 9:49:00 AM | GPUGRID | Sending scheduler request: To fetch work.
4/30/2014 9:49:00 AM | GPUGRID | Requesting new tasks for NVIDIA
4/30/2014 9:49:02 AM | GPUGRID | Scheduler request completed: got 0 new tasks
4/30/2014 9:49:02 AM | GPUGRID | No tasks sent
4/30/2014 9:49:02 AM | GPUGRID | No tasks are available for Short runs (2-3 hours on fastest card)
4/30/2014 9:49:02 AM | GPUGRID | No tasks are available for ACEMD beta version
4/30/2014 9:49:02 AM | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
4/30/2014 9:49:02 AM | GPUGRID | Tasks for CPU are available, but your preferences are set to not accept them
|
|
|
|
First of all, the issue of "not getting work here" definitely feels like it is the result of that erroneous 1.1 compute capability.
For reference, I have posted your problem to NVIDIA's 337.50 driver feedback thread.
https://forums.geforce.com/default/topic/720880/geforce-drivers/official-nvidia-337-50-beta-performance-display-driver-feedback-thread-4-7-14-/post/4206175/#4206175
I'll chime back in if I get a reply from NVIDIA. An experienced user did reply saying he thought the drivers were erroneously reporting OpenCL capability, instead of Compute capability, for your GPU. But we don't know for sure.
Like I said, if I hear more, I'll reply back. |
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
Would also be of interest to see what GPU Caps Viewer reports, just in case the issue isn't with the driver.
Sounds like something you might see if the card was badly flashed. Might be worth checking if the manufacturers have a GPU firmware update, or trying their bespoke drivers (CD).
____________
FAQ's
HOW TO:
- Opt out of Beta Tests
- Ask for Help |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
OK -- so for some reason the card is showing a compute capability of 1 instead of 3.
I will research further
|
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Actually Compute Capability 1.1 - still clearly something is amiss.
I posted a query on the Nvidea boards.
I purchased the card via Ebay, card looks new, and reports properly at the bios level as a GTX250. Nvidea drivers believe it to be a GTX250. It may be simply a well made knockoff (no specific manufacturing info on the card to speak of).
I am not sure what performance benchmark software I could download and use to verify overall performance.
For what its worth, Windows 7 shows 5.7 for graphics and 6.4 for gaming graphics
|
|
|
|
Actually Compute Capability 1.1 - still clearly something is amiss.
I posted a query on the Nvidea boards.
I purchased the card via Ebay, card looks new, and reports properly at the bios level as a GTX250. Nvidea drivers believe it to be a GTX250. It may be simply a well made knockoff (no specific manufacturing info on the card to speak of).
I am not sure what performance benchmark software I could download and use to verify overall performance.
For what its worth, Windows 7 shows 5.7 for graphics and 6.4 for gaming graphics
I thought we were talking about a GTX 650 -- Why are you now mentioning GTX 250??
4/30/2014 9:48:59 AM | | cc_config.xml not found - using defaults
4/30/2014 9:48:59 AM | | Starting BOINC client version 7.2.42 for windows_x86_64
4/30/2014 9:48:59 AM | | log flags: file_xfer, sched_ops, task
4/30/2014 9:48:59 AM | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
4/30/2014 9:48:59 AM | | Data directory: C:\BOINC2
4/30/2014 9:48:59 AM | | Running under account Barry
4/30/2014 9:48:59 AM | | CUDA: NVIDIA GPU 0: GeForce GTX 650 (driver version 337.50, CUDA version 6.0, compute capability 1.1, 2048MB, 1983MB available, 396 GFLOPS peak)
4/30/2014 9:48:59 AM | | OpenCL: NVIDIA GPU 0: GeForce GTX 650 (driver version 337.50, device version OpenCL 1.0 CUDA, 2048MB, 1983MB available, 396 GFLOPS peak)
4/30/2014 9:48:59 AM | | Host name: BSC-GA970-955
4/30/2014 9:48:59 AM | | Processor: 4 AuthenticAMD AMD Phenom(tm) II X4 955 Processor [Family 16 Model 4 Stepping 3]
4/30/2014 9:48:59 AM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 htt pni cx16 popcnt syscall nx lm svm sse4a osvw ibs skinit wdt page1gb rdtscp 3dnowext 3dnow
4/30/2014 9:48:59 AM | | OS: Microsoft Windows 7: Home Premium x64 Edition, Service Pack 1, (06.01.7601.00)
|
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
My error(I do not see an edit function here for messages).
The card is a GTX-650 with 2G of DDR3 memory.
Something is clearly not right with the card - I just checked my other system, which runs properly, with the GT-640 - in theory a lower performance card, which does run GPUGrid -- the Windows ratings are 6.6/6.6 compared to the 5.7/5.9 for this GTX-650,
Guess the card was too good a deal <sigh>
|
|
|
|
We're trying to solve it. We don't know the cause just yet. ManuelG from NVIDIA is trying to reproduce the issue right now.
It'd really help if you could run GPU Caps Viewer, and tell us the value it shows for compute capability, on that GTX 650.
Also, perhaps you could install GPU-Z, and report to us the BIOS Version, and the Subvendor. GPU-Z can also be used to extract the BIOS (there's a button that shows a microchip with a green arrow), which is something ManuelG might want later. |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
Caps viewer confirms the 1.1 value. I suspect the card is something of a 'good looking knockoff' here.
I sort of doubt there is any way to 'fix' this -- but could be wrong.
By the way, getting that Caps viewer file lit up my ESET AV report.
|
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
I've started a back and forth with the folks at nvidia. I'll report back here. Seriously, I suspect that I'll end up replacing that card with something I purchase direct from NewEgg -- fully 'vetted' as it were.
|
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
After back and forth with Nvidia, we've reached the conclusion that this is a well manufactured knock off card.
It looks well built. The system BIOS and diagnostic software report it to be a GTX-650. But it clearly is not in terms of performance (not just the lack of getting GPU work here, also the poor performance in processing GPU work from other projects.
I've referred the problem back to the Ebay seller for resolution.
|
|
|
|
Any chance we could get a screenshot of GPU-Z's "Graphics Card" tab, as well as a photo of the GPU itself, just to satisfy our curiosity? :)
[Sorry for your grief, if it is a knockoff] |
|
|
BarryAZSend message
Joined: 16 Apr 09 Posts: 163 Credit: 920,875,294 RAC: 0 Level
Scientific publications
|
It's pretty apparent the card was a knock-off (and apparently a pretty good imitation as it 'fooled' the system BIOS. When I protested this to the seller on EBay -- I had two options -- return for a full refund (with me paying shipping), or taking a 25% credit driving my out of pocket down to about $40.
I chose the latter as 'return for credit' is a bit of a hassle with shipping and the like, and the card is functional at about a GT-210 level.
I replaced the card with a new Asus GTX-650 with 1Gb of memory and it was recognized properly, I've already completed on work unit with it.
I'd note I worked with the Nvidia tech support folks -- one or two emails back and for for each of several days before confirming the card as a knockoff.
I will take a photo of the card at some point and post it. Aside from the performance issue and the confirmation of 1.1 versus 3.0 for computing capability, along with the lower Win 7 performance numbers, the other key here is there is no manufacturing information on the card. There is the 'Nvidia' name and logo on the fan. On the back of the card, the 'manufacturing' sticker doesn't mention a manufacturer -- just GeForceGTX650Graphic, 2G, DDR3.
I'd note the Asus card I now have has 1G, DDR5 memory.
|
|
|
|
This is a very edifying story. |
|
|