Flakey AMD/ATI GPUs, including RX 5700 XT, Cross Validating, polluting the Database

Message boards : Number crunching : Flakey AMD/ATI GPUs, including RX 5700 XT, Cross Validating, polluting the Database
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 17 · 18 · 19 · 20

AuthorMessage
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2034271 - Posted: 27 Feb 2020, 20:36:16 UTC - in response to Message 2034267.  

im still not sure why he got credit for it then. i've seen plenty of WUs invalidate for less.
You're only able to look at the raw signal count from out here. The actual validation is done against the 20KB - 30KB result file which is uploaded at the end of processing - just one of those signals can look like:

<gaussian>
  <peak_power>5.7957577705383</peak_power>
  <mean_power>0.57813739776611</mean_power>
  <time>2458905.5078695</time>
  <ra>6.5102765875108</ra>
  <decl>8.7759971597974</decl>
  <q_pix>0</q_pix>
  <freq>1418912409.544</freq>
  <detection_freq>1418914503.6237</detection_freq>
  <barycentric_freq>0</barycentric_freq>
  <fft_len>16384</fft_len>
  <chirp_rate>42.310804859724</chirp_rate>
  <rfi_checked>0</rfi_checked>
  <rfi_found>0</rfi_found>
  <reserved>0</reserved>
  <sigma>1.5730149745941</sigma>
  <chisqr>1.3424669504166</chisqr>
  <null_chisqr>2.281421661377</null_chisqr>
  <score>0</score>
  <max_power>11.661991119385</max_power>
  <pot length=181 encoding="x-csv">
    16,3,7,4,2,8,21,20,3,8,30,10,6,0,18,6,13,5,2,4,17,38,4,1,54,11,55,7,33,
    255,198,28,76,41,1,25,12,16,13,13,3,2,5,9,20,45,5,17,22,31,22,1,4,23,3,
    4,21,1,26,4,1,10,5,28
  </pot>
</gaussian>
ID: 2034271 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2034272 - Posted: 27 Feb 2020, 20:39:02 UTC - in response to Message 2034271.  

yes, but even looking at the raw signal counts, if one result has wildly different counts than the other, can they really be similar?

i found 3 apples
i found 5 apples

hmm, but 2 of them look kind of the same, good enough!
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2034272 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2034274 - Posted: 27 Feb 2020, 20:45:02 UTC - in response to Message 2034272.  

"weakly similar" is really very, very, weak.
ID: 2034274 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2034276 - Posted: 27 Feb 2020, 20:49:58 UTC - in response to Message 2034274.  

my point is less about a canonical result. and that's good that it's not being used.

but it's giving this user the false impression that their system is all good because it's getting "mostly valid" results, but they aren't really. if they invalidated, the user would be faced with the realization that something needs to be rectified.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2034276 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2034286 - Posted: 27 Feb 2020, 22:20:01 UTC - in response to Message 2034276.  

I think this dates back to the very early days of SETI@Home, when the whole concept of 'distributed computing' was weird and nobody knew whether it would take off. The idea was to give users a warm, fuzzy feeling for a 'good try'.

Unfortunately, in this case, it seems to have succeeded only too well...
ID: 2034286 · Report as offensive     Reply Quote
4-C

Send message
Joined: 19 Jun 05
Posts: 6
Credit: 4,069,157
RAC: 6
United States
Message 2034291 - Posted: 27 Feb 2020, 22:24:50 UTC - in response to Message 2034276.  

So there are two issues.
One, why are invalid results being marked as valid?
Two, why is it returning invalid results?
The installed GPU drivers are the most current available and have been on since early February.
I will disable the GPU from receiving work but that doesn't fix the issue which I would think is more wide spread than just my system.
ID: 2034291 · Report as offensive     Reply Quote
Profile Kissagogo27 Special Project $75 donor
Avatar

Send message
Joined: 6 Nov 99
Posts: 716
Credit: 8,032,827
RAC: 62
France
Message 2037090 - Posted: 10 Mar 2020, 9:59:42 UTC

Find this 5700 with lot of invalids

8896142
ID: 2037090 · Report as offensive     Reply Quote
Profile lunkerlander
Avatar

Send message
Joined: 23 Jul 18
Posts: 82
Credit: 1,353,232
RAC: 4
United States
Message 2038012 - Posted: 15 Mar 2020, 3:32:08 UTC

I found another 5700 with lots of invalids https://setiathome.berkeley.edu/results.php?hostid=8854997
ID: 2038012 · Report as offensive     Reply Quote
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36769
Credit: 261,360,520
RAC: 489
Australia
Message 2038027 - Posted: 15 Mar 2020, 6:38:48 UTC

It's just a shame that BOINC can't work out and display those AMD driver versions used, but I'd say that they're just using the drivers that were supplied at the time of intro and don't have those with the proper OpenCL support.

Cheers.
ID: 2038027 · Report as offensive     Reply Quote
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5126
Credit: 276,046,078
RAC: 462
Message 2047387 - Posted: 28 Apr 2020, 12:29:22 UTC

What is the status on Linux drivers for the RX 5700 XT? Still broken?

I think I understand that the Windows 10 drivers have been fixed?

Tom M
A proud member of the OFA (Old Farts Association).
ID: 2047387 · Report as offensive     Reply Quote
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2047423 - Posted: 28 Apr 2020, 16:50:42 UTC - in response to Message 2047387.  

What is the status on Linux drivers for the RX 5700 XT? Still broken?

I think I understand that the Windows 10 drivers have been fixed?

Tom M

If you read further back in this thread, the latest AMD drivers have fixed the issue with the 5700* cards.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2047423 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2047424 - Posted: 28 Apr 2020, 16:53:16 UTC - in response to Message 2047423.  

Yes, for Windows. But a driver change in a windows package doesn’t necessarily get reflected in the Linux package.

I don’t know if there is/was any real confirmation of RX5700+Linux being fixed.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2047424 · Report as offensive     Reply Quote
Paul

Send message
Joined: 17 May 99
Posts: 72
Credit: 42,977,964
RAC: 43
United States
Message 2047460 - Posted: 28 Apr 2020, 21:32:53 UTC - in response to Message 2047424.  

I don't know, I don't have the 5700 and I don't run amdgpu-pro. But, I can report that amdgpu drivers v20.10 were just released. It's hard to follow the versioning; release version is different than internal version. I'm assuming the new Linux drivers include the patches that were in the last Windows driver update. In any case, there are new drivers, so it is reasonable to re-evaluate the situation.
ID: 2047460 · Report as offensive     Reply Quote
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5126
Credit: 276,046,078
RAC: 462
Message 2047466 - Posted: 28 Apr 2020, 22:51:36 UTC - in response to Message 2047460.  

TY Paul.
A proud member of the OFA (Old Farts Association).
ID: 2047466 · Report as offensive     Reply Quote
Previous · 1 . . . 17 · 18 · 19 · 20

Message boards : Number crunching : Flakey AMD/ATI GPUs, including RX 5700 XT, Cross Validating, polluting the Database


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.