mfakto: an OpenCL program for Mersenne prefactoring - Page 95

Bdot · 2014-04-12, 09:45

Thanks to you all for your testing. Everything seems to be successful - I'll try to release 0.14 next week.

The lag can be reduced a bit by lowering the GPUSieveSize. Your Linux Catalyst version also seems to be much lower (938.2) than your win version (1124.2) - and they are all quite old, Catalyst 13.12 has version 1348.5. Anyway it is good to know that mfakto 0.14 also works with older drivers.

Did anyone (besides me) run the selftest with the LessClasses mode (MoreClasses=0)?

kracker · 2014-04-12, 14:45

Quote:

Originally Posted by Bdot

Did anyone (besides me) run the selftest with the LessClasses mode (MoreClasses=0)?

On it on GCN

kracker · 2014-04-12, 23:27

st2 with lessclasses, 287351 tests passed

Bdot · 2014-04-17, 20:22

I have released mfakto 0.14.

As most people seem to use the GPU sieve, I no longer created the versions with different (CPU-) sieve sizes. If anyone still needs them, just let me know.

Thanks to all testers!

Here again what's new:

--perftest enhancements including GPU sieve evaluation (for optimizing GPUSievePrimes etc.)
successfully resync when the working directory was temporarily lost (ejected USB device or interrupted network drive)
save and reload compiled OpenCL kernels => reduce startup time (UseBinfile config variable)
MoreClasses config variable to allow for a "less classes" version for very short assignments (GPU sieve only)
BugFix: enforce GPUSieveSize being a multiple of GPUSieveProcessSize
FlushInterval config variable to fine tune the number of kernels in the GPU queue => address high CPU load issue of newer AMD drivers
MinGW build (thanks to kracker)
slight performance improvement for the montgomery kernels
improved English wording in program output, ini file etc. (thanks to kracker)
recognition of new GPUs (8xxx, R9, new APUs) (thanks to kracker)
added a warning when using VectorSize=1 (AMD driver issue (?), http://devgurus.amd.com/thread/167571)
fix for a small memory leak (~0.5kB per assignment)
works on Win8.1

kracker · 2014-04-17, 22:24

I guess my OS is too old

Code:

./mfakto-x64: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.15' not found (required by ./mfakto-x64)
./mfakto-x64: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.14' not found (required by ./mfakto-x64)

LaurV · 2014-04-18, 02:48

I don't like it:

there is no double speed improvement
there is no triple speed improvement
it can not play minesweeper! (thanks to kracker)...

good work guys! thanks a lot!

kracker · 2014-04-18, 02:54

Quote:

Originally Posted by LaurV

I don't like it:

there is no double speed improvement
there is no triple speed improvement
it can not play minesweeper! (thanks to kracker)...

[COLOR==White]good work guys! thanks a lot![//COLOR]

...

Axelsson · 2014-04-18, 04:09

Is there any practical difference in the results from the pre4 version to the release? I discovered that I left the pre4 version running assignments for a couple of days. If needed I'll just rerun them but I rather run new assignments if I can.

/Göran

Bdot · 2014-04-18, 06:21

Quote:

Originally Posted by Axelsson

Is there any practical difference in the results from the pre4 version to the release? I discovered that I left the pre4 version running assignments for a couple of days. If needed I'll just rerun them but I rather run new assignments if I can.

/Göran

No, the different version string is the biggest difference

No need to rerun anything.

Bdot · 2014-04-18, 06:28

Quote:

Originally Posted by kracker

I guess my OS is too old

Code:

./mfakto-x64: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.15' not found (required by ./mfakto-x64)
./mfakto-x64: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.14' not found (required by ./mfakto-x64)

Oops, I did not notice this new requirement when I switched to gcc 4.9

Can you install the 4.9 version or should I try to build on older gcc for you?

I need to update the requirements section ...

kracker · 2014-04-18, 14:22

Quote:

Originally Posted by Bdot

Oops, I did not notice this new requirement when I switched to gcc 4.9

Can you install the 4.9 version or should I try to build on older gcc for you?

I need to update the requirements section ...

Don't worry, I built it myself again

2014-04-17, 20:22	#1038
Bdot Nov 2010 Germany 3×199 Posts	mfakto 0.14 available I have released mfakto 0.14. As most people seem to use the GPU sieve, I no longer created the versions with different (CPU-) sieve sizes. If anyone still needs them, just let me know. Thanks to all testers! Here again what's new: --perftest enhancements including GPU sieve evaluation (for optimizing GPUSievePrimes etc.) successfully resync when the working directory was temporarily lost (ejected USB device or interrupted network drive) save and reload compiled OpenCL kernels => reduce startup time (UseBinfile config variable) MoreClasses config variable to allow for a "less classes" version for very short assignments (GPU sieve only) BugFix: enforce GPUSieveSize being a multiple of GPUSieveProcessSize FlushInterval config variable to fine tune the number of kernels in the GPU queue => address high CPU load issue of newer AMD drivers MinGW build (thanks to kracker) slight performance improvement for the montgomery kernels improved English wording in program output, ini file etc. (thanks to kracker) recognition of new GPUs (8xxx, R9, new APUs) (thanks to kracker) added a warning when using VectorSize=1 (AMD driver issue (?), http://devgurus.amd.com/thread/167571) fix for a small memory leak (~0.5kB per assignment) works on Win8.1 Last fiddled with by Bdot on 2014-04-17 at 20:31 Reason: sent off too early ... 2nd edit: changes

2014-04-18, 02:48	#1040
LaurV Romulan Interpreter Jun 2011 Thailand 2×3×1,609 Posts	I don't like it: there is no double speed improvement there is no triple speed improvement it can not play minesweeper! (thanks to kracker)... good work guys! thanks a lot! Last fiddled with by LaurV on 2014-04-18 at 02:52

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
gpuOwL: an OpenCL program for Mersenne primality testing	preda	GpuOwl	2718	2021-07-06 18:30
mfaktc: a CUDA program for Mersenne prefactoring	TheJudger	GPU Computing	3497	2021-06-05 12:27
LL with OpenCL	msft	GPU Computing	433	2019-06-23 21:11
OpenCL for FPGAs	TObject	GPU Computing	2	2013-10-12 21:09
Program to TF Mersenne numbers with more than 1 sextillion digits?	Stargate38	Factoring	24	2011-11-03 00:34

2014-04-12, 09:45	#1035
Bdot Nov 2010 Germany 3×199 Posts	Thanks to you all for your testing. Everything seems to be successful - I'll try to release 0.14 next week. The lag can be reduced a bit by lowering the GPUSieveSize. Your Linux Catalyst version also seems to be much lower (938.2) than your win version (1124.2) - and they are all quite old, Catalyst 13.12 has version 1348.5. Anyway it is good to know that mfakto 0.14 also works with older drivers. Did anyone (besides me) run the selftest with the LessClasses mode (MoreClasses=0)?

2014-04-12, 23:27	#1037
kracker "Mr. Meeseeks" Jan 2012 California, USA 4170₈ Posts	st2 with lessclasses, 287351 tests passed

2014-04-18, 04:09	#1042
Axelsson Jul 2012 Sweden 2×3×7 Posts	Is there any practical difference in the results from the pre4 version to the release? I discovered that I left the pre4 version running assignments for a couple of days. If needed I'll just rerun them but I rather run new assignments if I can. /Göran