gpuOwL on RX-480
 

[CODE]gpuOwL v2.0- GPU Mersenne primality checker
Ellesmere-36x1266-@28:0.0 Radeon (TM) RX 480 Graphics
Note: using long carry and fused tail kernels
OpenCL compilation in 1606 ms, with " -DEXP=83871443u -I. -cl-fast-relaxed-math -cl-kernel-arg-info "
PRP-3: FFT 5000K (625 * 4096 * 2) of 83871443 (16.38 bits/word) [2018-05-22 18:30:12 Central Daylight Time]
Starting at iteration 3391500
OK 3391500 / 83871443 [ 4.04%], 0.00 ms/it [0.00, 0.00], check 4.80s; ETA 0d 00:00; 5188fd46df3cb580 [18:30:18]
OK 3392000 / 83871443 [ 4.04%], 5.18 ms/it [5.18, 5.18], check 4.73s; ETA 4d 19:45; 71a58007ca04e5f0 [18:30:26]
OK 3394000 / 83871443 [ 4.05%], 5.14 ms/it [5.12, 5.15], check 4.73s; ETA 4d 18:54; b54a931ac82e17f4 [18:30:41]
OK 3395000 / 83871443 [ 4.05%], 5.16 ms/it [5.15, 5.18], check 4.73s; ETA 4d 19:25; 2ecd625c5ac3d167 [18:30:51]
OK 3400000 / 83871443 [ 4.05%], 5.26 ms/it [5.15, 6.21], check 4.71s; ETA 4d 21:35; 395e3747e24d13c1 [18:31:22]
OK 3410000 / 83871443 [ 4.07%], 5.21 ms/it [5.16, 6.21], check 4.71s; ETA 4d 20:33; 4ed4f275d67bab43 [18:32:18]
[/CODE][CODE]
gpuOwL v1.9- GPU Mersenne primality checker
Radeon (TM) RX 480 Graphics 36 @28:0.0, Ellesmere 1266MHz

OpenCL compilation in 2948 ms, with "-I. -cl-fast-relaxed-math -cl-std=CL2.0 -DEXP=152500021u -DWIDTH=2048u -DHEIGHT=2048u -DLOG_NWORDS=23u -DFP_DP=1 "
PRP-3: FFT 8M (2048 * 2048 * 2) of 152500021 (18.18 bits/word) [2018-05-22 18:05:31 Central Daylight Time]
Starting at iteration 33568000
OK 33568000 / 152500021 [22.01%], 0.00 ms/it; ETA 0d 00:00; b68cd3e37955c5c6 [18:05:39]
OK 33569000 / 152500021 [22.01%], 6.46 ms/it [6.46, 6.46] CV 0.0%, check 7.53s; ETA 8d 21:21; eb70e9f8dc16ca48 [18:05:53]
OK 33570000 / 152500021 [22.01%], 6.47 ms/it [6.46, 6.49] CV 0.3%, check 6.68s; ETA 8d 21:51; e52284b737f8d3e7 [18:06:07]
OK 33575000 / 152500021 [22.02%], 6.48 ms/it [6.46, 6.52] CV 0.3%, check 6.88s; ETA 8d 22:02; 25fb1cd868b9cd06 [18:06:46]
OK 33580000 / 152500021 [22.02%], 6.62 ms/it [6.49, 7.67] CV 5.6%, check 6.71s; ETA 9d 02:33; 95456516edd7574e [18:07:26][/CODE]For comparison, RX-550 timings were 17.4 and 23.2 ms/it respectively.

When trying gpuowl on Windoze 10 with an Nvidia card, it reports Opencl v 1.2 but gives me an error -46 (fftw)
and points to line 205 of clwrap.h which is weird because there is no assert via the check or check2 macro near there .
I am using the executable made by user kracker.
Any ideas?

Thanks.

Also, as a side note, the AMD site has had its link to the AMD OpenCL SDK broken for about the last 3 weeks !!

[QUOTE=tServo;488231]When trying gpuowl on Windoze 10 with an Nvidia card, it reports Opencl v 1.2 but gives me an error -46 (fftw)
and points to line 205 of clwrap.h which is weird because there is no assert via the check or check2 macro near there .
I am using the executable made by user kracker.
Any ideas?
[/QUOTE]

I bet gpuowl requires OpenCL 2, which was released in Nov 2013. Nvidia has been dragging their feet on supporting new versions.

[QUOTE=Mark Rose;488233]I bet gpuowl requires OpenCL 2, which was released in Nov 2013. Nvidia has been dragging their feet on supporting new versions.[/QUOTE]

It just so happens that I have a Windoze 10 box with an AMD Rx580 in it with OpenCL 2.0 and it did the same thing.

This means it's either Windoze or how I am running gpuowl.

My worktodo file is 1 line as so:
Test=<lots of hex chars>,exponent

and I use no command line params.

Is there a necessary .ini file or some command line params I should be using?

[QUOTE=tServo;488262]It just so happens that I have a Windoze 10 box with an AMD Rx580 in it with OpenCL 2.0 and it did the same thing.

This means it's either Windoze or how I am running gpuowl.

My worktodo file is 1 line as so:
Test=<lots of hex chars>,exponent

and I use no command line params.

Is there a necessary .ini file or some command line params I should be using?[/QUOTE]
I use

set opts=-legacy
gpuowl -user kriesel -cpu condorella-rx550 -device 3 -verbosity 2 %opts%

Do you have gpuowl.cl in the working directory? The M61 transform would require nttshared.h also.

Hi,


I've tried to run gpuowl on debian with a GTX 1050. I think i've installed all necessary packets, i'm using the backports-drivers 390.48.
It compiles fine, but when I run it, I get an error:


[CODE]gpuOwL v2.1-a1df37b GPU Mersenne primality checker
error -1001
gpuowl: clwrap.h:36: int getDeviceIDs(bool, size_t, _cl_device_id**): Assertion `check(clGetPlatformIDs(8, platforms, &nPlatforms))' failed.
Aborted
[/CODE]Any help is appreciated.

Upgraded to debian testing and at least it detects now the GPU, but still no success:

[CODE]
gpuOwL v2.1-a1df37b GPU Mersenne primality checker
GeForce GTX 1050-5x1455-
Note: using long carry and fused tail kernels


OpenCL compilation in 844 ms, with " -DEXP=43920223u -DWIDTH=2048u -DNW=8u -DHEIGHT=2048u -DNH=8u -I. -cl-fast-relaxed-math -cl-kernel-arg-info "
PRP-3: FFT 8192K (2048 * 2048 * 2) of 43920223 (5.24 bits/word) [2018-05-25 19:06:34 CEST]
Starting at iteration 0
error -9999 (fftW)
gpuowl: clwrap.h:267: void run(cl_queue, cl_kernel, size_t, size_t, const string&): Assertion `check(clEnqueueNDRangeKernel(queue, kernel, 1, __null, &workSize, &groupSize, 0, __null, __null), name.c_str())' failed.
Aborted[/CODE]

[QUOTE=heliosh;488357]Upgraded to debian testing and at least it detects now the GPU, but still no success:

[CODE]
gpuOwL v2.1-a1df37b GPU Mersenne primality checker
GeForce GTX 1050-5x1455-
Note: using long carry and fused tail kernels


OpenCL compilation in 844 ms, with " -DEXP=43920223u -DWIDTH=2048u -DNW=8u -DHEIGHT=2048u -DNH=8u -I. -cl-fast-relaxed-math -cl-kernel-arg-info "
PRP-3: FFT 8192K (2048 * 2048 * 2) of 43920223 (5.24 bits/word) [2018-05-25 19:06:34 CEST]
Starting at iteration 0
error -9999 (fftW)
gpuowl: clwrap.h:267: void run(cl_queue, cl_kernel, size_t, size_t, const string&): Assertion `check(clEnqueueNDRangeKernel(queue, kernel, 1, __null, &workSize, &groupSize, 0, __null, __null), name.c_str())' failed.
Aborted[/CODE][/QUOTE]


Hello,
You have to use kernel 4.15.3 because the kernel 4.16 contains a regression that causes amdgpu_job_timedout errors.

for nvidia [url]https://wiki.debian.org/NvidiaGraphicsDrivers[/url]

I've tried 4.15.0-3 with the same outcome.

[url]https://packages.debian.org/sid/linux-image-4.15.0-3-amd64[/url]

[QUOTE=heliosh;488388]I've tried 4.15.0-3 with the same outcome.

[URL]https://packages.debian.org/sid/linux-image-4.15.0-3-amd64[/URL][/QUOTE]


As far as I know gpuowl is using OpenCL and not CUDA ...

[url]https://wiki.tiker.net/CudaVsOpenCL[/url]

Yes, I have the opencl drivers installed and gpuowl appears to detect the GTX 1050.
[CODE]ii nvidia-opencl-common 390.48-3 amd64 NVIDIA OpenCL driver - common files
ii nvidia-opencl-dev:amd64 9.1.85-4+b1 amd64 NVIDIA OpenCL development files
[/CODE]
But as far as I've read in this thread, nvidia GPUs aren't supported.