WU 13456

[PaulTV]

Nah, I won't be holding my breath, but I will add tracking the core sub version to my job monitoring

[JohnChodera]

We're attempting to build core22 0.0.16 with updated CUDA 11.2, which should restore CUDA functionality for RTX 30x0s. Will aim to test later this week if we don't run into trouble with the automated build.

Surprisingly, 0.0.13---which @Neil-B and others report _did_ successfully use CUDA on RTX 30x0s---was built using CUDA 9.2, while 0.0.14 and 0.0.15 used our new automated build system with CUDA 10.2.

~ John Chodera // MSKCC

[Crunchtimer]

Hi Folders!

I happy to contribute to all science, however as of now I'm very much into assisting the success of the Covid Moonshot, hence Sprint 10 and WU 13456-7 are of main interest.
7Unfortunately I'm not seeing any GPU WUs on other than 0.0.13 and the below is pretty much what I*m getting lately, so no 13456-7 are turning up in the logs.

I'm running GTX1070s and is this a problem or am I missing anything else, any upgrade needed?

Thanks for the assistance!

Code: Select all

FS01:0x22:Project: 18018 (Run 55, Clone 92, Gen 17)
FS01:0x22:Project: 17601 (Run 260, Clone 4, Gen 98)
FS01:0x22:Project: 18018 (Run 94, Clone 47, Gen 15)
FS02:0x22:Project: 18018 (Run 93, Clone 52, Gen 27)
FS02:0x22:Project: 16469 (Run 0, Clone 257, Gen 148)
FS02:0x22:Project: 17806 (Run 27, Clone 47, Gen 105)
FS02:0x22:Project: 16468 (Run 0, Clone 124, Gen 145)

[aetch]

Preferred cause is just a preference, it's not a guarantee you'll always get work units for that cause.
It does mean the assignment server will try to serve you up Covid-19 projects if it can but it does depend on the projects being configured to use your gpu and suitable work units being available when you ask for new work.
Failing that it will fall back to work units that are suitable for your hardware, regardless of cause.

Personally, I joined on the back of Covid but I left my preference as "ANY".
I came with the view that all research is valid and if some folders want to donate their folding resources to specific causes then I'll quite happily take up the slack in other places.

[PaulTV]

...And core 0x22-0.0.16 has landed, with RTX 30x0 using CUDA as well, or so I see in the logging. Indeed with a nice performance boost!

[Tashgan]

With Core 22.0.0.16 cuda works again for my RTX 3070Ti. More than 25% faster compared to OpenCL with nearly identical power draw. A good efficiency boost. Thanks for the bugfix.

[Smookin_Joe]

Hi
I just noticed one of my many folding computers is chewing really hard on a 13456 which came from 54.157.202.86.
I am not close to being an expert...but I try and do what I can.
I put things together, install software, post issues and rely on the software to look after itself.
Sorry but true.
Is there anything I should/could be doing to help this computer not have an issue with the 13456 WU?
Joe

[Smookin_Joe]

Geforce RTX 2060 XC Ultra running at 5% on the 13456 wu
ETA is 1.44 days
54400 base credit
54400 Estimated Credit
Est PPD 29579

[Smookin_Joe]

17:03:39:WU00:FS02:0x22:Attempting to create CUDA context:
17:03:39:WU00:FS02:0x22: Configuring platform CUDA
17:03:39:WU01:FS00:0xa8:Completed 1 out of 125000 steps (0%)
17:04:22:WU00:FS02:0x22:Failed to create CUDA context:
17:04:22:WU00:FS02:0x22:Error loading CUDA module: CUDA_ERROR_UNSUPPORTED_PTX_VERSION (222)
17:04:22:WU00:FS02:0x22:Attempting to create OpenCL context:
17:04:22:WU00:FS02:0x22: Configuring platform OpenCL
17:04:37:FS00:Finishing
17:04:40:18:127.0.0.1:New Web session
17:04:57:WU00:FS02:0x22: Using OpenCL on platformId 1 and gpu 0
17:04:58:WU00:FS02:0x22:Completed 200000 out of 1000000 steps (20%)

[Neil-B]

You might want to try (re)installing the latest drivers (471.68) and then rebooting ... I believe this has cleared this for others ... post below explains some background - some people have found the drivers may need to be "latest" rather than just >=456.38 (which is what "the book" says).

From @toTOW:

Since core 22 v0.0.16 has been compiled with a newer version of CUDA (11.x), make sure that you have the latest NV drivers for your GPU installed on your system.

Windows drivers must be >=456.38
Linux drivers must be >= 450.80.02

Too old drivers will show these messages in the log :

Code: Select all

12:55:28:WU01:FS01:0x22:Attempting to create CUDA context:
12:55:28:WU01:FS01:0x22:  Configuring platform CUDA
12:55:29:WU01:FS01:0x22:Failed to create CUDA context:
12:55:29:WU01:FS01:0x22:Error loading CUDA module: CUDA_ERROR_UNSUPPORTED_PTX_VERSION (222)

And then fallback to OpenCL ...

[toTOW]

Since core 22 v0.0.16 has been compiled with a newer version of CUDA (11.x), make sure that you have the latest nVidia drivers for your GPU installed on your system.

Windows drivers must be >=456.38
Linux drivers must be >= 450.80.02

Too old drivers will show these messages in the log :

Code: Select all

12:55:28:WU01:FS01:0x22:Attempting to create CUDA context:
12:55:28:WU01:FS01:0x22:  Configuring platform CUDA
12:55:29:WU01:FS01:0x22:Failed to create CUDA context:
12:55:29:WU01:FS01:0x22:Error loading CUDA module: CUDA_ERROR_UNSUPPORTED_PTX_VERSION (222)

And then fallback to OpenCL ...

[Smookin_Joe]

***Update***
Before both of you responded I searched(honestly) through the other computers for which Nvidea drivers they were running
Noticed all were newer versions..
Ran the clean install of driver pkg 471 on problem computer
rebooted
Started Precision X1 checked settings
Moved power from 90% to 100%(gpu was stuck using 5%)
fired up fah client
Things started ramping up...YEAH...I fixed it!

Gpu using 70%+
Then I noticed you guys told me to do all of the above...lol
Thank You!
Give you part credit...

[Smookin_Joe]

18:58:10:WU00:FS02:0x22: Global context and integrator variables write interval: 25000 steps (2.5%) [40 total]
18:58:10:WU00:FS02:0x22:There are 4 platforms available.
18:58:10:WU00:FS02:0x22:Platform 0: Reference
18:58:10:WU00:FS02:0x22:Platform 1: CPU
18:58:10:WU00:FS02:0x22:Platform 2: OpenCL
18:58:10:WU00:FS02:0x22: opencl-device 0 specified
18:58:10:WU00:FS02:0x22:Platform 3: CUDA
18:58:10:WU00:FS02:0x22: cuda-device 0 specified
18:58:23:WU00:FS02:0x22:Attempting to create CUDA context:
18:58:23:WU00:FS02:0x22: Configuring platform CUDA
18:58:26:WU01:FS00:0xa8:Completed 45000 out of 125000 steps (36%)
18:58:34:WU00:FS02:0x22: Using CUDA and gpu 0
18:58:36:WU00:FS02:0x22:Completed 0 out of 1000000 steps (0%)
18:58:37:WU00:FS02:0x22:Checkpoint completed at step 0
18:58:40:Saving configuration to config.xml
18:58:40:<config>

[Smookin_Joe]

19:01:10:WU01:FS00:0xa8:Completed 47500 out of 125000 steps (38%)
19:01:33:WU00:FS02:0x22:Completed 20000 out of 1000000 steps (2%)
19:02:31:WU01:FS00:0xa8:Completed 48750 out of 125000 steps (39%)
19:03:00:WU00:FS02:0x22:Completed 30000 out of 1000000 steps (3%)
19:03:53:WU01:FS00:0xa8:Completed 50000 out of 125000 steps (40%)
19:04:28:WU00:FS02:0x22:Completed 40000 out of 1000000 steps (4%)
19:05:15:WU01:FS00:0xa8:Completed 51250 out of 125000 steps (41%)
19:05:56:WU00:FS02:0x22:Completed 50000 out of 1000000 steps (5%)
19:05:56:WU00:FS02:0x22:Checkpoint completed at step 50000
19:06:37:WU01:FS00:0xa8:Completed 52500 out of 125000 steps (42%)
19:07:24:WU00:FS02:0x22:Completed 60000 out of 1000000 steps (6%)
19:07:59:WU01:FS00:0xa8:Completed 53750 out of 125000 steps (43%)
19:08:52:WU00:FS02:0x22:Completed 70000 out of 1000000 steps (7%)
19:09:20:WU01:FS00:0xa8:Completed 55000 out of 125000 steps (44%)
19:10:20:WU00:FS02:0x22:Completed 80000 out of 1000000 steps (8%)
19:10:40:WU01:FS00:0xa8:Completed 56250 out of 125000 steps (45%)
19:11:47:WU00:FS02:0x22:Completed 90000 out of 1000000 steps (9%)
19:12:01:WU01:FS00:0xa8:Completed 57500 out of 125000 steps (46%)
19:13:15:WU00:FS02:0x22:Completed 100000 out of 1000000 steps (10%)
19:13:16:WU00:FS02:0x22:Checkpoint completed at step 100000
19:13:26:WU01:FS00:0xa8:Completed 58750 out of 125000 steps (47%)
19:14:43:WU00:FS02:0x22:Completed 110000 out of 1000000 steps (11%)
19:14:47:WU01:FS00:0xa8:Completed 60000 out of 125000 steps (48%)
19:16:09:WU01:FS00:0xa8:Completed 61250 out of 125000 steps (49%)
19:16:11:WU00:FS02:0x22:Completed 120000 out of 1000000 steps (12%)

[Smookin_Joe]

Eta on the 13456 wu has dropped to 2 hrs from day and a half
Thanks for your time guys