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 . . . 12 · 13 · 14 · 15 · 16 · 17 · 18 . . . 20 · Next

AuthorMessage
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 2027503 - Posted: 12 Jan 2020, 20:02:41 UTC - in response to Message 2027502.  

Wow fast. I'm still used to running APs on a CPU in 4 hours. If that. ;-)
ID: 2027503 · Report as offensive     Reply Quote
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 2027511 - Posted: 12 Jan 2020, 21:15:24 UTC

This RX 5700 seems to be producing Incorrect results with the New Driver and App, https://setiweb.ssl.berkeley.edu/beta/results.php?hostid=89080
????
ID: 2027511 · Report as offensive     Reply Quote
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22540
Credit: 416,307,556
RAC: 380
United Kingdom
Message 2027515 - Posted: 12 Jan 2020, 21:29:45 UTC - in response to Message 2027511.  

Eh????
The majority of the errors are "user aborts" on both CPU & GPU, there are a couple of invalids, but they are with version 8.22, not 8.24.
Meanwhile his 8.24 results all appear to be "pending".
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 2027515 · Report as offensive     Reply Quote
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 2027518 - Posted: 12 Jan 2020, 21:44:24 UTC - in response to Message 2027515.  

Look again, most of them are "Completed, validation inconclusive". Then compare the 5700 results against what his Wingman reported, https://setiweb.ssl.berkeley.edu/beta/results.php?hostid=89080
ID: 2027518 · Report as offensive     Reply Quote
StrayCat
Avatar

Send message
Joined: 15 May 99
Posts: 3
Credit: 11,701,496
RAC: 151
United States
Message 2027520 - Posted: 12 Jan 2020, 22:38:24 UTC - in response to Message 2027485.  

My GPU load was pretty low, below 50%. Then reading down I saw TBar's comments on CPU usage and realized I forgot to lower my CPU usage from100 when I started running tasks in Beta.

I now have one core reserved for the GPU and my GPU load is much higher, between 75-99%.
ID: 2027520 · Report as offensive     Reply Quote
Rob

Send message
Joined: 7 Apr 12
Posts: 9
Credit: 951,019
RAC: 0
Germany
Message 2027522 - Posted: 12 Jan 2020, 22:59:23 UTC
Last modified: 12 Jan 2020, 23:00:02 UTC

Hi, computer 89080 is me. Indeed the errors were from me being a little stupid. As I set up the BETA project after reading the need for 8.24 testers, I forgot to turn off CPU workunits and only saw 8.22 GPU ones coming in, so I thought I had misconfigured and think I hit reset I think before realizing I'd abort a whole lot of work. Sorry for the confusion this has brought up!

I had received some 8.22 workunits before the 8.24 deployment was fixed so they're both in there as well. All invalids (so far at least) are restricted to 8.22.
ID: 2027522 · Report as offensive     Reply Quote
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3807
Credit: 1,114,826,392
RAC: 3,319
Canada
Message 2027524 - Posted: 12 Jan 2020, 23:10:52 UTC
Last modified: 12 Jan 2020, 23:11:42 UTC

Just returned from a week of completely offline vacation, so I have a lot of catching up to do in this thread... of course I miss the most significant developments, namely new drivers and an app. to go with them. Once both are tested I can re-contact all the people who replied earlier (and perhaps even the ones who didn't) and let them know there's a solution. It's also excellent to see that some of the people I reached out to are now helping with the testing... thank you. :^)
ID: 2027524 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 2027525 - Posted: 12 Jan 2020, 23:25:37 UTC - in response to Message 2027485.  

I was looking at StrayCat’s results over at beta, to get an idea of relative performance.
Judging based on the WU times compared to the cards it’s being stacked up against on the validates tasks, it looks similar in performance to a GTX 1060 or so. That seems a bit disappointing, I think it should be about 2x that.


. . Is that a 1060 running SoG or on the sauce ...

Stephen
??
ID: 2027525 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 2027526 - Posted: 12 Jan 2020, 23:40:45 UTC - in response to Message 2027502.  

“Wow” fast? Or “Wow” Slow?

My 2070s routinely run them about that fast (sometimes faster, sometimes slower), when running 2 WUs at a time.


. . My 1060s take about 20 mins running as singles. So definitely faster than a 1060 :)

Stephen

:)
ID: 2027526 · 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 2027527 - Posted: 12 Jan 2020, 23:41:07 UTC - in response to Message 2027525.  

I was comparing to cards running the SoG apps.

But it looks like his issue was CPU use. So he should be crunching faster now.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2027527 · Report as offensive     Reply Quote
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 2027536 - Posted: 13 Jan 2020, 0:29:42 UTC - in response to Message 2027502.  

An AP in 6m29s. I'm happy with this card. Now if only I could ever get some AP at the main project, where it counts.
ID: 2027536 · Report as offensive     Reply Quote
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 2027571 - Posted: 13 Jan 2020, 14:27:15 UTC

Finally some non-APs, the 8.24 MBs.
https://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=12328994
35297160 	89104 	13 Jan 2020, 0:48:13 UTC 	13 Jan 2020, 13:13:00 UTC 	Completed and validated 	425.73 	115.11 	103.72 	SETI@home v8 v8.24 (opencl_ati5_SoG_nocal)
windows_intelx86
35297161 	78473 	13 Jan 2020, 1:44:00 UTC 	13 Jan 2020, 5:17:11 UTC 	Completed and validated 	997.71 	979.11 	103.72 	SETI@home v8 v8.16 (opencl_nvidia_sah)
windows_intelx86
35297162 	89062 	13 Jan 2020, 1:36:53 UTC 	13 Jan 2020, 8:44:08 UTC 	Completed and validated 	886.92 	849.48 	103.72 	SETI@home v8 v8.22 (opencl_nvidia_SoG)
windows_intelx86

ID: 2027571 · Report as offensive     Reply Quote
Bluerazor

Send message
Joined: 22 May 99
Posts: 15
Credit: 3,889,427
RAC: 12
United States
Message 2027778 - Posted: 15 Jan 2020, 22:57:58 UTC

I went ahead and activated my RX 5700 on Beta and am letting it crunch away. AMD released another new driver revision pn 1/13, updated to the current. Initial AP workunits came up validation inconclusive, but at least it's not overflowing on MB.
ID: 2027778 · Report as offensive     Reply Quote
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22540
Credit: 416,307,556
RAC: 380
United Kingdom
Message 2027786 - Posted: 15 Jan 2020, 23:09:35 UTC

Just a quick sanity check - what's the version number for the driver AMD released on Jan 13th?
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 2027786 · Report as offensive     Reply Quote
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 2027797 - Posted: 15 Jan 2020, 23:32:35 UTC

8.24 made it to main for AMD GPUs: https://setiathome.berkeley.edu/apps.php
ID: 2027797 · 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 2027837 - Posted: 16 Jan 2020, 1:11:55 UTC - in response to Message 2027786.  

Just a quick sanity check - what's the version number for the driver AMD released on Jan 13th?

The original one for the Adrenaline 20.1.1 drivers that fixed the Seti overflow problem had an internal driver number of:
Name: gfx1010
Vendor: Advanced Micro Devices, Inc.
Driver version: 3004.8 (PAL,LC)
when the gpu detect module reports in the Event Log startup. But I don't know what the internal driver number is for the drivers from the 13th.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2027837 · Report as offensive     Reply Quote
Rob

Send message
Joined: 7 Apr 12
Posts: 9
Credit: 951,019
RAC: 0
Germany
Message 2027842 - Posted: 16 Jan 2020, 1:34:59 UTC
Last modified: 16 Jan 2020, 1:35:17 UTC

Got two invalids on RX 5700 with 8.24 beta, in case it is relevant

https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=35288672
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=35301492


Just checked, the 20.1.2 drivers do seem to still report the same OpenCL version for me. clinfo says

Platform Version: OpenCL 2.1 AMD-APP (3004.8)
ID: 2027842 · Report as offensive     Reply Quote
Bluerazor

Send message
Joined: 22 May 99
Posts: 15
Credit: 3,889,427
RAC: 12
United States
Message 2027843 - Posted: 16 Jan 2020, 1:35:52 UTC - in response to Message 2027786.  

Just a quick sanity check - what's the version number for the driver AMD released on Jan 13th?


I'm not sure which version number you're looking for. The Radeon Software version is 20.1.2. For my Windows 10 system, the software reports the following related numbers:
Software Version 2020.0113.1626.29577
Driver Version 19.50.11.10-200113a-350865E-RadeonSoftwareAdrenalin2020

However I don't know where to see that "internal" version mentioned by another poster.
ID: 2027843 · 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 2027847 - Posted: 16 Jan 2020, 1:44:55 UTC

The number that is reported in the Event Log at startup when the gpu detect module probes for gpu hardware. Or in the stderr.txt output of a completed task run by the 5700XT gpu and the latest AMD Adrenaline drivers 20.1.1 or greater.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2027847 · Report as offensive     Reply Quote
Bluerazor

Send message
Joined: 22 May 99
Posts: 15
Credit: 3,889,427
RAC: 12
United States
Message 2027852 - Posted: 16 Jan 2020, 2:10:00 UTC - in response to Message 2027847.  

The number that is reported in the Event Log at startup when the gpu detect module probes for gpu hardware. Or in the stderr.txt output of a completed task run by the 5700XT gpu and the latest AMD Adrenaline drivers 20.1.1 or greater.


I see it now. Same as poster "Rob", it is reporting 3004.8 with this version 20.1.2.
ID: 2027852 · Report as offensive     Reply Quote
Previous · 1 . . . 12 · 13 · 14 · 15 · 16 · 17 · 18 . . . 20 · Next

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.