Advanced search

Message boards : Server and website : Crunching power grows with new project

Author Message
WPrion
Send message
Joined: 30 Apr 13
Posts: 77
Credit: 1,034,112,811
RAC: 609
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 53715 - Posted: 21 Feb 2020 | 17:21:59 UTC
Last modified: 21 Feb 2020 | 17:22:36 UTC

On the Server Status page there's an item: "current GigaFLOPS" (gF), which appears to be an estimate of the total crunching power in progress. When this new batch of WU's first appeared after a long dry spell, the value was around 800,000 - 900,000 gF (IIRC). Now it's over 1,357,000 Gf, or 1.3 petaFLOPS (pF) and growing.

I'm guessing those who wandered off during the quiet times are finding their way back. Plus, there are dozens of "new registrations" every day.

For comparison (note: I don't know how well these estimated translate between distributed and non-distributed computers) the most powerful supercomputer is IBM's Summit at Oak Ridge rated at 150-200 pF.

1.3 pF isn't bad for a bunch of volunteers!

Profile ServicEnginIC
Avatar
Send message
Joined: 24 Sep 10
Posts: 199
Credit: 1,458,715,434
RAC: 823,206
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 53728 - Posted: 21 Feb 2020 | 23:02:11 UTC - in response to Message 53715.
Last modified: 21 Feb 2020 | 23:03:44 UTC

When this new batch of WU's first appeared after a long dry spell, the value was around 800,000 - 900,000 gF (IIRC). Now it's over 1,357,000 Gf, or 1.3 petaFLOPS (pF) and growing.

We will have to work hard to beat the 4 PetaFLOPS record!
Now we have fuel for our racing car(d)s...

Profile ServicEnginIC
Avatar
Send message
Joined: 24 Sep 10
Posts: 199
Credit: 1,458,715,434
RAC: 823,206
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 53829 - Posted: 3 Mar 2020 | 11:15:54 UTC

Today we're half the way to beat previous record: 2,03 PetaFLOPs, as currently shown at Server Status page.
Come on!

Erich56
Send message
Joined: 1 Jan 15
Posts: 697
Credit: 3,295,149,283
RAC: 329,689
Level
Arg
Scientific publications
watwatwatwatwatwat
Message 53830 - Posted: 3 Mar 2020 | 11:58:56 UTC - in response to Message 53715.

On the Server Status page ...

talking about the Server Status page - a side question: does anyone have an idea why recently it takes up to several minutes for this page and also others from the GPUGRID website until it gets refreshed?

Jim1348
Send message
Joined: 28 Jul 12
Posts: 733
Credit: 1,478,749,566
RAC: 96,002
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 53831 - Posted: 3 Mar 2020 | 12:48:43 UTC - in response to Message 53830.
Last modified: 3 Mar 2020 | 12:51:21 UTC

Good work. But I am more impressed that they have managed to keep the unsent work units at about 310,000.
I wonder what we are doing? I might put more on it if I knew.

WPrion
Send message
Joined: 30 Apr 13
Posts: 77
Credit: 1,034,112,811
RAC: 609
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 53832 - Posted: 3 Mar 2020 | 14:55:57 UTC - in response to Message 53831.
Last modified: 3 Mar 2020 | 14:57:39 UTC

Good work. But I am more impressed that they have managed to keep the unsent work units at about 310,000.
I wonder what we are doing? I might put more on it if I knew.


It looks like that number is finally slowly decreasing. My guess is that each WU is crunched multiple times by different users for validation. As the large list gets "saturated" some WU's are finally completely finished for good. As time goes on the rate of decrease will approach the rate of individual task completion since more tasks are only awaiting their final crunch.

Unless, of course, they are still adding new WUs?!?

Profile ServicEnginIC
Avatar
Send message
Joined: 24 Sep 10
Posts: 199
Credit: 1,458,715,434
RAC: 823,206
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 53834 - Posted: 3 Mar 2020 | 15:06:50 UTC

Good work. But I am more impressed that they have managed to keep the unsent work units at about 310,000.

+1

talking about the Server Status page - a side question: does anyone have an idea why recently it takes up to several minutes for this page and also others from the GPUGRID website until it gets refreshed?

I guess it keeps relationship with current number of workunits to manage (unsent, in process, processed...) >300000. If server is configured to make inventory with every request,it would explain this retard in answering...

Toni
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 9 Dec 08
Posts: 947
Credit: 4,353,973
RAC: 48
Level
Ala
Scientific publications
watwatwatwat
Message 53835 - Posted: 3 Mar 2020 | 17:00:46 UTC - in response to Message 53832.

New WUs are generated from successful returned ones in a 1:1 ratio, until the programmed number of steps is reached. This will keep a "steady state" around 300k WUs for a while.

Good work. But I am more impressed that they have managed to keep the unsent work units at about 310,000.
I wonder what we are doing? I might put more on it if I knew.


It looks like that number is finally slowly decreasing. My guess is that each WU is crunched multiple times by different users for validation. As the large list gets "saturated" some WU's are finally completely finished for good. As time goes on the rate of decrease will approach the rate of individual task completion since more tasks are only awaiting their final crunch.

Unless, of course, they are still adding new WUs?!?

Post to thread

Message boards : Server and website : Crunching power grows with new project