Welcome to MilkyWay@home

Errors on new 5970


Advanced search

Message boards : Number crunching : Errors on new 5970
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Thamir Ghaslan

Send message
Joined: 31 Mar 08
Posts: 61
Credit: 18,325,284
RAC: 0
10 million credit badge10 year member badge
Message 39718 - Posted: 14 May 2010, 20:45:16 UTC - in response to Message 39706.  

ID: 39718 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileThe Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
200 million credit badge10 year member badge
Message 39724 - Posted: 14 May 2010, 23:07:03 UTC - in response to Message 39718.  

I was under the impression that a 5970 would be doulbe the speed of a 4870x2, turns out that one GPU in a 5970 is roughly 3 times as fast as a 4870 GPU, so overall that should give out a rough 6x increase.

Based on my results I'm seeing each OC'd GPU on the 5970 doing just over twice the throughput of my 4870. See here.
ID: 39724 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Thamir Ghaslan

Send message
Joined: 31 Mar 08
Posts: 61
Credit: 18,325,284
RAC: 0
10 million credit badge10 year member badge
Message 39727 - Posted: 15 May 2010, 5:50:43 UTC - in response to Message 39724.  
Last modified: 15 May 2010, 5:52:40 UTC

I was under the impression that a 5970 would be doulbe the speed of a 4870x2, turns out that one GPU in a 5970 is roughly 3 times as fast as a 4870 GPU, so overall that should give out a rough 6x increase.

Based on my results I'm seeing each OC'd GPU on the 5970 doing just over twice the throughput of my 4870. See here.


Strange!

Doing almost 100 second for 5800s, 300 seconds for 4800s. Same ratio hold in other projects!

http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=148531&offset=0&show_names=0&state=3

I'm suspecious of your PCI slot setups, maybe it would help if your device 0 and 1 were the 5800s?

Running a gigabyte motherboard, GA-EP45-DQ6, not sure if pci-e slot speed plays a role here, if yours are both full 16x or not?

Or it might be somethign else all together? :/

the 4800 and 5800 cards are both saphire, no OC.
ID: 39727 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileThe Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
200 million credit badge10 year member badge
Message 39731 - Posted: 15 May 2010, 11:11:01 UTC - in response to Message 39727.  

ID: 39731 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Thamir Ghaslan

Send message
Joined: 31 Mar 08
Posts: 61
Credit: 18,325,284
RAC: 0
10 million credit badge10 year member badge
Message 39732 - Posted: 15 May 2010, 11:57:29 UTC - in response to Message 39731.  

ID: 39732 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileThe Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
200 million credit badge10 year member badge
Message 39741 - Posted: 15 May 2010, 17:30:00 UTC - in response to Message 39732.  

ID: 39741 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Zydor
Avatar

Send message
Joined: 24 Feb 09
Posts: 620
Credit: 100,587,625
RAC: 0
100 million credit badge10 year member badgeextraordinary contributions badge
Message 39747 - Posted: 15 May 2010, 21:55:39 UTC - in response to Message 39741.  
Last modified: 15 May 2010, 22:08:47 UTC

..Obviously I am missing something here.


Yours are fine - 5970 pumps them out two at a time, so in reality the WU throughput per card (on your current settings) is every 45 seconds.

The BOINC TFlops figure is usually generous.. 5970 TFlops is officially 4.64TFlops per card, so is BOINC expectation and the 4870 is 1.2TFlops per card, 2.4TFlops for two of them

Even a 5970 will work happily over a x8 PCIe slot, it takes a hell of an application to fill that bandwidth, let alone a x16, BOINC does not have one that big. The only likely issue with 5970s is a slight slowing, very small, if the cpu is not a high end one, as the card cycles waiting for free cpu time. i7s are ok, slight cpu bound - very minor - with Phenom2s. Below that the cpu delay builds more.

Your 4870 & 5970 working fine.

Regards
Zy
ID: 39747 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilekashi

Send message
Joined: 30 Dec 07
Posts: 311
Credit: 149,490,184
RAC: 0
100 million credit badge10 year member badge
Message 39752 - Posted: 16 May 2010, 4:20:04 UTC

I've noticed I'm getting a few errors since I installed my 5970. The errors are on both cards though, the 5970 and the 5870. Errors are before the task has finished its usual runtime and give error message: "exit code -1 (0xffffffff)"

I can't tell how many errors I'm getting daily because the results get cleared quickly. The 3 errors that are currently showing are about 3 hours apart so if that is representative I'm getting about 8 per day. That is about 0.29% of the total processed daily so it's very minor.

Still like to know what it is though. The only thing I'm doing differently now is I have removed the f20 parameter so the task iterations are now split into the default 5 parts for my cards instead of 3 parts.
ID: 39752 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileThe Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
200 million credit badge10 year member badge
Message 39818 - Posted: 19 May 2010, 2:44:19 UTC

Eeek! Getting quite a few invalid wu's again. Dammit, I'm not at home to see what the problem is!
ID: 39818 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilekashi

Send message
Joined: 30 Dec 07
Posts: 311
Credit: 149,490,184
RAC: 0
100 million credit badge10 year member badge
Message 39822 - Posted: 19 May 2010, 4:55:35 UTC - in response to Message 39818.  

Yes all on your 5970 too. You have one error that is not a validate error. The error message for that one is usually associated with those runing multiple GPUs on a computer that has a relatively small amount of system memory. Perhaps it is also related to your validate errors. When you get home you could try running for a while with no CPU projects and see how that affects it. If it something to do with that you may need to choose CPU projects that use minimal amounts of ram or only run 1 task concurrently on your GPUs.

When I'm in doubt I usually blame AQUA because the long AQUA tasks have consistently given problems on my computer when processing GPU tasks.

My GPUs mucked up yesterday when I was fiddling around trying to be able to turn off the monitor. Tasks wouldn't crunch at all and kept giving blue screens and I had to reinstall the Catalyst driver. Been good since then and I haven't had any errors for a while now, well none that I have noticed anyway.

I was reading on another site that Catalyst 10.4 that I am using is causing problems for some gamers and/or those with multiple monitors because it is more aggressive in shutting down the second core than previous drivers. The fix is to change some settings in the registry to disable ULPS (Ultra Low Power State). I will give that a try tomorrow.
ID: 39822 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileThe Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
200 million credit badge10 year member badge
Message 39827 - Posted: 19 May 2010, 11:07:16 UTC

I've disabled turning the monitor off after x minutes and just set it to go black. I'll just turn the monitor off when I'm done. I've also done a few things to try and get the heat away from my HAF932 case. It may be good for air flow, but like any case it just blows it out allowing it to accumulate around the case to be sucked in again through a different entry.
ID: 39827 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfilePaul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
100 million credit badge10 year member badge
Message 39829 - Posted: 19 May 2010, 11:48:13 UTC - in response to Message 39818.  

Eeek! Getting quite a few invalid wu's again.

Aqua is using the MP feature of BOINC and I note that there was a change-set to change the way that MP scheduling is being handled. So, Kashi's note on Aqua is well taken, you may be seeing a collision with MP tasks and the GPU tasks ... or, a simple running out of main memory ...

That is the problem with the bleeding edge ... Ow! :)
ID: 39829 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Number crunching : Errors on new 5970

©2020 Astroinformatics Group