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 : Problems and Help : Recognizing GPU (Message 132227)
Posted 111 days ago by Profile mfl0pProject donor
yes, probably easier to reinstall and don't pick service mode during install
2) Message boards : Problems and Help : Recognizing GPU (Message 132173)
Posted 112 days ago by Profile mfl0pProject donor
check the tools->event log tab you should see something like this when starting BOINC

8/20/2019 8:15:38 PM | | Starting BOINC client version 7.14.2 for windows_x86_64
8/20/2019 8:15:38 PM | | log flags: file_xfer, sched_ops, task
8/20/2019 8:15:38 PM | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2g zlib/1.2.8
8/20/2019 8:15:38 PM | | Data directory: C:\ProgramData\BOINC
8/20/2019 8:15:38 PM | | CUDA: NVIDIA GPU 0: GeForce GTX 1070 Ti (driver version 431.36, CUDA version 10.1, compute capability 6.1, 4096MB, 3552MB available, 8186 GFLOPS peak)
8/20/2019 8:15:38 PM | | OpenCL: NVIDIA GPU 0: GeForce GTX 1070 Ti (driver version 431.36, device version OpenCL 1.2 CUDA, 8192MB, 3552MB available, 8186 GFLOPS peak)

and yes, gpus will certainly work in BOINC on VMs with proper passthru
3) Message boards : AP26 - AP27 Search : CPU speeds vs GPU for Oktoberfest Challenge (Message 132094)
Posted 115 days ago by Profile mfl0pProject donor
We can HT AP?


No, you cannot multithread the AP CPU app.
4) Message boards : AP26 - AP27 Search : CPU speeds vs GPU for Oktoberfest Challenge (Message 132044)
Posted 119 days ago by Profile mfl0pProject donor
In preparing for the AP27 Oktoberfest Challenge some of you may be wondering "How many more points can I get by running my CPU also?" I did some testing and here are the results.

Core i7 7700k Kaby Lake 4 core
completes 4 workunits in about 47000 seconds
equivalent to 1 workunit done every 11750 seconds

Core i7 3770k Ivy Bridge 4 core
completes 4 workunits in about 62000 seconds
equivalent to 1 workunit done every 15500 seconds

Ok, now let's compare these CPUs to a 5+ year old Maxwell generation Nvidia GTX 750ti. Not exactly a high end GPU, but very power efficient when released.
It completes a workunit in 6800 seconds.
So, as you see the lowly 750ti is more than twice as fast as the Ivy Bridge CPU and still much faster than even the Kaby Lake!

Newer GPUs such as RTX 2080 Ti can be over 20 times faster than the Kaby Lake and the fastest current GPU, a Tesla V100 is over 45 times faster!
5) Message boards : AP26 - AP27 Search : New feature with latest AP27 gpu application (Message 131759)
Posted 127 days ago by Profile mfl0pProject donor
try

opencl_ati_AP27

refer to https://www.primegrid.com/apps.php
6) Message boards : General discussion : Can Anyone Recommend A Good Book To Understand This Prime Math (Message 131417)
Posted 140 days ago by Profile mfl0pProject donor
not a book, but Jens Kruse Andersen's webpage at http://primerecords.dk/aprecords.htm has some info relating to APs and the history of searching for record APs. At the bottom there are links to more info on APs.
7) Message boards : AP26 - AP27 Search : 32-bit AP search? (Message 130326)
Posted 183 days ago by Profile mfl0pProject donor
and, 64 bit is faster on AP search
8) Message boards : Number crunching : LLR 3.8.23 Testing (AVX-512) (Message 129711)
Posted 202 days ago by Profile mfl0pProject donor
There is another factor to consider. The processors maximum clock speed is reduced when running avx512 work.

For example the Xeon 8124M will only reach 2900Mhz compared to 3300Mhz with AVX2 and 3400Mhz with all other workloads with all cores loaded.

So there are many things at play here, number of avx512 units per processor and the reduced clock speeds.
9) Message boards : Number crunching : LLR 3.8.23 Testing (AVX-512) (Message 129710)
Posted 202 days ago by Profile mfl0pProject donor
Could someone with an AVX-512 capable CPU please try running the first couple of iterations of this test on llr 3.8.21 vs 3.8.23:



Xeon Platinum 8124M running 9 threads per test

3.8.21
Starting Proth prime test of 19249*2^13018586+1
Using all-complex FMA3 FFT length 1152K, Pass1=384, Pass2=3K, 9 threads, a = 3
^C249*2^13018586+1, bit: 160000 / 13018600 [1.22%]. Time per bit: 0.799 ms.

3.8.23
Starting Proth prime test of 19249*2^13018586+1
Using all-complex AVX-512 FFT length 1152K, Pass1=1152, Pass2=1K, clm=2, 9 threads, a = 3
^C249*2^13018586+1, bit: 170000 / 13018600 [1.30%]. Time per bit: 0.594 ms.
10) Message boards : Aggie The Pew message board (Message 126346)
Posted 308 days ago by Profile mfl0pProject donor
found one today


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2019 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.09, 3.24, 2.61
Generated 11 Dec 2019 | 7:02:54 UTC