PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Proth Prime Search : Running PPS sieve under Mesa3D 17.3.0rcX (Message 111943)
Posted 1334 days ago by Profile matszpk
Stupid question: Can I fix my the first post in this thread? Or, can anyone fix that stupid bug in first post? I can't find option edit my first post.
2) Message boards : Proth Prime Search : Running PPS sieve under Mesa3D 17.3.0rcX (Message 111942)
Posted 1334 days ago by Profile matszpk
Oh, no. actually, i was omitting res value. but was initialized, so wrong patch disables disk cache for shaders.
new corrected path (in file 'gallium/drivers/radeon/r600_pipe_common.c' (at r600_disk_cache_create function):
/*res = asprintf(&timestamp_str, "%u_%u", mesa_timestamp, llvm_timestamp);*/ timestamp_str = malloc(100); res = sprintf(timestamp_str, "%u_%u", mesa_timestamp, llvm_timestamp);


patch:

diff --git a/src/gallium/drivers/radeon/r600_pipe_common.c b/src/gallium/drivers/radeon/r600_pipe_common.c index b77d85901f..2e83c83754 100644 --- a/src/gallium/drivers/radeon/r600_pipe_common.c +++ b/src/gallium/drivers/radeon/r600_pipe_common.c @@ -914,7 +914,10 @@ static void r600_disk_cache_create(struct r600_common_screen *rscreen) if (disk_cache_get_function_timestamp(LLVMInitializeAMDGPUTargetInfo, &llvm_timestamp)) { - res = asprintf(&timestamp_str, "%u_%u", + /*res = asprintf(&timestamp_str, "%u_%u", + mesa_timestamp, llvm_timestamp);*/ + timestamp_str = malloc(100); + res = sprintf(timestamp_str, "%u_%u", mesa_timestamp, llvm_timestamp); }
3) Message boards : Proth Prime Search : Running PPS sieve under Mesa3D 17.3.0rcX (Message 111929)
Posted 1335 days ago by Profile matszpk
I made small fix for Mesa3D that fixes segfault on PPS Sieve OpenCL application:

file 'gallium/drivers/radeon/r600_pipe_common.c' (at r600_disk_cache_create function):
if (disk_cache_get_function_timestamp(LLVMInitializeAMDGPUTargetInfo, &llvm_timestamp)) { /*res = asprintf(&timestamp_str, "%u_%u", mesa_timestamp, llvm_timestamp);*/ timestamp_str = malloc(100); sprintf(timestamp_str, "%u_%u", mesa_timestamp, llvm_timestamp); }


diff:
diff --git a/src/gallium/drivers/radeon/r600_pipe_common.c b/src/gallium/drivers/radeon/r600_pipe_common.c index b77d85901f..40c713a422 100644 --- a/src/gallium/drivers/radeon/r600_pipe_common.c +++ b/src/gallium/drivers/radeon/r600_pipe_common.c @@ -914,7 +914,10 @@ static void r600_disk_cache_create(struct r600_common_screen *rscreen) if (disk_cache_get_function_timestamp(LLVMInitializeAMDGPUTargetInfo, &llvm_timestamp)) { - res = asprintf(&timestamp_str, "%u_%u", + /*res = asprintf(&timestamp_str, "%u_%u", + mesa_timestamp, llvm_timestamp);*/ + timestamp_str = malloc(100); + sprintf(timestamp_str, "%u_%u", mesa_timestamp, llvm_timestamp); }


works good on my RX 480 and RX VEGA (but slower than on propietary AMD OpenCL drivers).
4) Message boards : Number crunching : No workunit for AMD GPU on Linux (Message 93427)
Posted 1947 days ago by Profile matszpk
Thank you for advice.

However I enabled GPU computing (ATI) and I enabled ATI GPU app on PPS sieve and Genefer on Primegrid prefs (I did it on 'work' and 'default' host place). Still, no workunit fetched. In my boinc logs just print:
25-Mar-2016 17:00:33 [PrimeGrid] project resumed by user 25-Mar-2016 17:00:36 [PrimeGrid] update requested by user 25-Mar-2016 17:00:41 [PrimeGrid] Sending scheduler request: Requested by user. 25-Mar-2016 17:00:41 [PrimeGrid] Requesting new tasks for ATI 25-Mar-2016 17:00:42 [PrimeGrid] Scheduler request completed: got 0 new tasks
.

My host is in 'work' place, so that should be working.
5) Message boards : Number crunching : No workunit for AMD GPU on Linux (Message 93422)
Posted 1947 days ago by Profile matszpk
Thank you for answer. I was trying to fetch to PPS Sieve and Genefer. BOINC ofcourse detects correctly two my graphics cards:
25-Mar-2016 15:28:41 [---] OpenCL: AMD/ATI GPU 0: Pitcairn (driver version 1912.5 (VM), device versi on OpenCL 1.2 AMD-APP (1912.5), 964MB, 964MB available, 1293 GFLOPS peak) 25-Mar-2016 15:28:41 [---] OpenCL: AMD/ATI GPU 1: Bonaire (driver version 1912.5 (VM), device versio n OpenCL 2.0 AMD-APP (1912.5), 1983MB, 1983MB available, 1037 GFLOPS peak)


However, in site of any project, that config has been recognized as two Bonaire graphics cards (in Primegrid too). Any other project just give available WU's to this computer (Einstein, Collatz, Moo!, excepts Milkyway) for GPU. I had some problems with milkyway due to incorrect recognizing my new graphics card (R7 360). Whether similar problems would be occurred while trying to be fetching WU's from Primegrid?
6) Message boards : Number crunching : No workunit for AMD GPU on Linux (Message 93409)
Posted 1947 days ago by Profile matszpk
I added new graphics card to my computer (AMD Radeon R7 360). I am trying to fetch any workunit to crunching on a GPU. After many trials, no workunits sent. In the logs boinc i found message:

Tasks for NVIDIA GPU are available, but your preferences are set to not accept them.


What is mean? Only WU for NVIDIA GPU available currently? May be my project settings doesn't allow to accept AMD WU's? Ofcourse, I enabled OpenCL app in primegrid prefs. Where is error made by me?
Can you help me or explain current behaviour of primegrid?

my host is: http://www.primegrid.com/show_host_detail.php?hostid=413271[/url]
7) Message boards : Generalized Fermat Prime Search : Gefener 3.03 fails on slightly overclocked GPU (Message 79413)
Posted 2509 days ago by Profile matszpk
Temperature of GPU core was OK (not hotter than 63C) , but doesnt exist method to check GPU memory temp (ofcourse).
gfx card was working good even its temp was greater than 70C (when I was torturing it with using matrix multiplication: sgemm :) ).
Unfortunatelly genefer WU with 'trick' failed on slighty overclocked mem (1255MHz) :(.
Maybe depends on average number of operations on GPU mem (faster=hotter).
8) Message boards : Generalized Fermat Prime Search : Gefener 3.03 fails on slightly overclocked GPU (Message 79405)
Posted 2510 days ago by Profile matszpk
I did know about this trick, however I was crunching on the standard setup for OpenCL.
Ok. I can try.
EDIT: I confirmed. This is weird, but without optimizations is 10%-13% faster. I will try without changing setup and I will check stability.
Thank you for any advance.
9) Message boards : Generalized Fermat Prime Search : Gefener 3.03 fails on slightly overclocked GPU (Message 79401)
Posted 2510 days ago by Profile matszpk
Thank you for advance
Nevertheless my graphics card successfuly crunched one Genefer WU on mem=1255MHz, core=1000MHz.
10) Message boards : Generalized Fermat Prime Search : Gefener 3.03 fails on slightly overclocked GPU (Message 79316)
Posted 2512 days ago by Profile matszpk
After 7 hours of crunching error was encountered again :(. Currently I stopped crunching of genefer.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2021 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 0.00, 0.00, 0.00
Generated 25 Jul 2021 | 2:44:08 UTC