[QUOTE=ixfd64;529877]The changes are now live in the official mfakto repository. Thanks to kracker for reviewing my pull request. :smile:
...

Any feedback would be appreciated![/QUOTE]If you haven't yet, please modify it like mfaktc has been modified to allow gpusievesize up to 2047 (Mib), or even 4095 if converting to unsigned int. See [URL]https://www.mersenneforum.org/showpost.php?p=508680&postcount=116[/URL] On NVIDIA the increase to 2047 max allowed several percent throughput increase. [URL]https://www.mersenneforum.org/showpost.php?p=526899&postcount=8[/URL]

And post mfakto Windows executables with that change, please.

I'm afraid that's beyond my current knowledge. I don't think it's a matter of simply changing some variables. See: [url]https://mersenneforum.org/showthread.php?p=525178#post525178[/url]

[QUOTE]WARNING: Unknown GPU name, assuming GCN. Please post the device name "gfx1012 (Advanced Micro Devices, Inc.)[/QUOTE]


AMD Sapphire 5500XT 8GB

runtimes:
96s - AMD 7950 (Catalyst driver)

1h estimated - AMD 5500XT (Adrenalin driver)

Is there a problem with latest drivers? The 5500XT should be faster than the 7950.
BOINC is reporting

18.02.2020 19:58:36 OpenCL: AMD/ATI GPU 0: Radeon RX 5500 XT (driver version 3004.8 (PAL,LC), device version OpenCL 2.0 AMD-APP (3004.8), 8176MB, 8176MB available, 2446 GFLOPS peak)
if this helps

[QUOTE=rebirther;537859]AMD Sapphire 5500XT 8GB
Is there a problem with latest drivers? The 5500XT should be faster than the 7950.[/QUOTE]If you get sane numbers I would dearly love to get some [URL="https://www.mersenne.ca/mfaktc.php#benchmark"]mfakto[/URL] and [URL="https://www.mersenne.ca/cudalucas.php"]gpuowl[/URL] benchmarks from you, I have very very few benchmarks (as in, one of each) for the RX 5x00 series of RDNA1 GPUs.

But based on the limited data I do have, the 5500 XT should be ~20% faster than the 7950:
[url]https://www.mersenne.ca/mfaktc.php?filter=5500|7950[/url]

[QUOTE=James Heinrich;537860]If you get sane numbers I would dearly love to get some [URL="https://www.mersenne.ca/mfaktc.php#benchmark"]mfakto[/URL] and [URL="https://www.mersenne.ca/cudalucas.php"]gpuowl[/URL] benchmarks from you, I have very very few benchmarks (as in, one of each) for the RX 5x00 series of RDNA1 GPUs.

But based on the limited data I do have, the 5500 XT should be ~20% faster than the 7950:
[URL]https://www.mersenne.ca/mfaktc.php?filter=5500|7950[/URL][/QUOTE]


From the BOINC wrapper test:


got assignment: exp=100832399 bit_min=73 bit_max=74 (18.97 GHz-days)
Starting trial factoring M100832399 from 2^73 to 2^74 (18.97GHz-days)
Using GPU kernel "cl_barrett15_82_gs_2"
Date Time | class Pct | time ETA | GHz-d/day Sieve Wait
Feb 18 20:12 | 0 0.1% | 3.769 1h00m | 453.04 82485 0.00%


I have reduced the vector size from 4 to 2

Ignore my last post. All is good, the number was different, here are the correct estimated runtimes:


7950: 1,5h
RX5500XT: 1h

[QUOTE=rebirther;537869]All is good, the number was different, here are the correct estimated runtimes[/QUOTE]When you have completed a test (preferably uninterrupted to maximize accuracy) can you please submit the benchmark data here:
[url]https://www.mersenne.ca/mfaktc.php#benchmark[/url]

[QUOTE=James Heinrich;537875]When you have completed a test (preferably uninterrupted to maximize accuracy) can you please submit the benchmark data here:
[URL]https://www.mersenne.ca/mfaktc.php#benchmark[/URL][/QUOTE]


submitted a factor found test but blacklisted

[QUOTE=rebirther;538012]submitted a factor found test but blacklisted[/QUOTE]Sorry about the blacklist error message, it got triggered when it should not have been.

Also, benchmarks should be submitted for no-factor-found tests that ran the entire bitlevel, please.

[QUOTE=James Heinrich;538014]Blacklisted? What do you mean?

Also, benchmarks should be submitted for no-factor-found tests that ran the entire bitlevel, please.[/QUOTE]


yes, it was for benchmarks. I will run another test later.