mfaktc: a CUDA program for Mersenne prefactoring - Page 318

James Heinrich · 2021-03-19, 21:47

If you look at GPU-Z monitoring clockspeed and GPU utilization, do you notice any relevant correlation when you open/minimize Chrome?

ZacHFX · 2021-03-20, 13:14

Quote:

Originally Posted by James Heinrich

If you look at GPU-Z monitoring clockspeed and GPU utilization, do you notice any relevant correlation when you open/minimize Chrome?

Not at all. Utilization stays steady at 99%, and clockspeed stays consistent within 50MHz.

lalera · 2021-03-20, 15:20

win10 - poweroptions ?

Wargs · 2021-03-23, 17:49

New here and been reading a lot about how to get this started. I have a new 3070 and wanted to use that over my i7-10700K. I will continue to read the posts and how to get this working but I am having more trouble than the CPU version by far.

James Heinrich · 2021-03-23, 21:49

Quote:

Originally Posted by Wargs

I have a new 3070

Are you using the CUDA 11.2 version posted on the previous page of this thread (post #3481)?

ZacHFX · 2021-03-24, 14:09

Quote:

Originally Posted by ZacHFX

Not at all. Utilization stays steady at 99%, and clockspeed stays consistent within 50MHz.

I figured out the problem. I use a live wallpaper app that was using significantly more GPU cycles than I would have thought. Submitting an updated benchmark to James that is much more accurate as soon as I get through a TF run.

pinhodecarlos · 2021-04-24, 17:09

From SRBase:

NVIDIA NVIDIA GeForce RTX 3090 (4095MB) driver: 466.11 OpenCL: 3.0
Microsoft Windows 10
Professional x64 Edition, (10.00.19042.00)
(https://srbase.my-firewall.org/sr5/r...?hostid=191961)

Date Time | class Pct | time ETA | GHz-d/day Sieve Wait
Apr 24 13:03 | 4515 98.0% | 0.022 n.a. | 5711.23 82485 n.a.%
Apr 24 13:03 | 4519 98.1% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4536 98.2% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4539 98.3% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4540 98.4% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4551 98.5% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4555 98.6% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4564 98.8% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4567 98.9% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4572 99.0% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4575 99.1% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4576 99.2% | 0.022 n.a. | 5711.23 82485 n.a.%
Apr 24 13:03 | 4579 99.3% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4584 99.4% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4591 99.5% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4596 99.6% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4599 99.7% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4600 99.8% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4611 99.9% | 0.021 n.a. | 5983.19 82485 n.a.%
Apr 24 13:03 | 4612 100.0% | 0.021 n.a. | 5983.19 82485 n.a.%
no factor for M685141049 from 2^72 to 2^73 [mfaktc 0.21 barrett76_mul32_gs]
tf(): total time spent: 20.735s

NVIDIA NVIDIA GeForce RTX 3060 Ti (4095MB) driver: 466.11 OpenCL: 3.0
Microsoft Windows 10
Professional x64 Edition, (10.00.19042.00)
(https://srbase.my-firewall.org/sr5/r...?hostid=212591)

Date Time | class Pct | time ETA | GHz-d/day Sieve Wait
Apr 24 19:03 | 4523 98.0% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4524 98.1% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4535 98.2% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4536 98.3% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4539 98.4% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4544 98.5% | 0.044 n.a. | 2861.80 82485 n.a.%
Apr 24 19:03 | 4548 98.6% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4556 98.8% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4559 98.9% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4560 99.0% | 0.044 n.a. | 2861.80 82485 n.a.%
Apr 24 19:03 | 4563 99.1% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4571 99.2% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4580 99.3% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4583 99.4% | 0.045 n.a. | 2798.20 82485 n.a.%
Apr 24 19:03 | 4599 99.5% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4604 99.6% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4608 99.7% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4611 99.8% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4616 99.9% | 0.043 n.a. | 2928.35 82485 n.a.%
Apr 24 19:03 | 4619 100.0% | 0.043 n.a. | 2928.35 82485 n.a.%
no factor for M683660221 from 2^72 to 2^73 [mfaktc 0.21 barrett76_mul32_gs]
tf(): total time spent: 41.986s

More benchmarks at https://srbase.my-firewall.org/sr5/top_hosts.php.
1) First check how many GPU's are allocated per host. If more than X then the host is running X-wus in parallel, GHz-d/day will be variable.
2) On computer Info click "Tasks" per ID
3) Click "Valid" on the top
4) Click any result under "Task"
5) Scroll down to see results. Sometimes the client hides the host details so you won't see it. Go to 2) and repeat until you find a host with all output open to outside.

Xyzzy · 2021-05-28, 20:59

It is entirely possible that we are "Doing it Wrong" trying to get mfaktc to work on RHEL 8, but here is what works for us.

If there is an easier way, please let us know! (Shouldn't it just work OOTB?)

1 - Install the proprietary Nvidia driver.

https://developer.nvidia.com/blog/st...arity-streams/

2 - Install cuda.

sudo dnf install cuda-nvcc-11-3

We chose 11.3 because it is the newest available.

3 - Download and extract mfaktc.

https://download.mersenne.ca/mfaktc/...ize2047.tar.gz

4 - Download a "safe" libcudart file and extract it into the mfaktc directory.

http://mirrors.kernel.org/ubuntu/poo...43-3_amd64.deb

Code:

$ ls -l libcudart.so.10.1*
lrwxrwxrwx. 1 m m     21 May 28 15:33 libcudart.so.10.1 -> libcudart.so.10.1.243
-rw-r--r--. 1 m m 504480 Apr 11  2020 libcudart.so.10.1.243

5 - Set a weird environment variable.

export LD_LIBRARY_PATH=/usr/local/cuda-11.3/lib64/:$LD_LIBRARY_PATH

6 - Profit!

Code:

$ ./mfaktc.exe
mfaktc v0.21 (64bit built)

Compiletime options
  THREADS_PER_BLOCK         256
  SIEVE_SIZE_LIMIT          32kiB
  SIEVE_SIZE                193154bits
  SIEVE_SPLIT               250
  MORE_CLASSES              enabled

Runtime options
  SievePrimes               25000
  SievePrimesAdjust         1
  SievePrimesMin            5000
  SievePrimesMax            100000
  NumStreams                3
  CPUStreams                3
  GridSize                  3
  GPU Sieving               enabled
  GPUSievePrimes            82486
  GPUSieveSize              2047Mi bits
  GPUSieveProcessSize       32Ki bits
  Checkpoints               enabled
  CheckpointDelay           30s
  WorkFileAddDelay          600s
  Stages                    enabled
  StopAfterFactor           bitlevel
  PrintMode                 full
  V5UserID                  (none)
  ComputerID                (none)
  AllowSleep                no
  TimeStampInResults        no

CUDA version info
  binary compiled for CUDA  10.10
  CUDA runtime version      10.10
  CUDA driver version       11.30

CUDA device info
  name                      NVIDIA Quadro RTX 8000
  compute capability        7.5
  max threads per block     1024
  max shared memory per MP  65536 byte
  number of multiprocessors 72
  clock rate (CUDA cores)   1770MHz
  memory clock rate:        7001MHz
  memory bus width:         384 bit

Automatic parameters
  threads per grid          589824
  GPUSievePrimes (adjusted) 82486
  GPUsieve minimum exponent 1055144

running a simple selftest...
Selftest statistics
  number of tests           107
  successfull tests         107

selftest PASSED!

Can't open workfile worktodo.txt
ERROR: get_next_assignment(): can't open "worktodo.txt"

We think it is very weird that the libcudart file isn't part of the proprietary driver or the cuda package.

vasyannyasha · 2021-06-05, 08:05

Hello everyone! Cant start mfaktc due to error. Anyone know how to fix this?

Code:

mfaktc v0.21 (64bit built)
 
Compiletime options
  THREADS_PER_BLOCK         256
  SIEVE_SIZE_LIMIT          32kiB
  SIEVE_SIZE                193154bits
  SIEVE_SPLIT               250
  MORE_CLASSES              enabled

Runtime options
  SievePrimes               25000
  SievePrimesAdjust         1
  SievePrimesMin            5000
  SievePrimesMax            100000
  NumStreams                3
  CPUStreams                3
  GridSize                  3
  GPU Sieving               enabled
  GPUSievePrimes            82486
  GPUSieveSize              64Mi bits
  GPUSieveProcessSize       16Ki bits
  Checkpoints               enabled
  CheckpointDelay           30s
  WorkFileAddDelay          600s
  Stages                    enabled
  StopAfterFactor           bitlevel
  PrintMode                 full
  V5UserID                  (none)
  ComputerID                (none)
  AllowSleep                no
  TimeStampInResults        no

CUDA version info
  binary compiled for CUDA  6.50
  CUDA runtime version      6.50
  CUDA driver version       11.20

CUDA device info
  name                      GeForce RTX 3060 Laptop GPU
  compute capability        8.6
  max threads per block     1024
  max shared memory per MP  102400 byte
  number of multiprocessors 30
  clock rate (CUDA cores)   1425MHz
  memory clock rate:        7001MHz
  memory bus width:         192 bit

Automatic parameters
  threads per grid          983040
  GPUSievePrimes (adjusted) 82486
  GPUsieve minimum exponent 1055144

########## testcase 1/2867 ##########
Starting trial factoring M50804297 from 2^67 to 2^68 (0.59 GHz-days)
Using GPU kernel "75bit_mul32_gs"
Date    Time | class   Pct |   time     ETA | GHz-d/day    Sieve     Wait
Jun 05 11:01 | 3387   0.1% |  0.001    n.a. |      n.a.    82485    n.a.%
ERROR: cudaGetLastError() returned 8: invalid device function

Viliam Furik · 2021-06-05, 10:12

Quote:

Originally Posted by vasyannyasha

CUDA version info
binary compiled for CUDA 6.50
CUDA runtime version 6.50
CUDA driver version 11.20

There you go. You have an old mfaktc version, not compatible with your GPU CUDA version.

See post #3481 for downloadable .exe file. It should do the trick.

In case of any problems, ~~please contact our customer support~~ (just kidding, there is no customer support. unless you consider yourself a customer, then the forum is the support. anyway, you're in the right place) feel free to ask in this thread again.

vasyannyasha · 2021-06-05, 12:27

Thanks! worked out pretty well.

2021-03-23, 17:49	#3491
Wargs Mar 2021 1 Posts	3070 Needs assistance New here and been reading a lot about how to get this started. I have a new 3070 and wanted to use that over my i7-10700K. I will continue to read the posts and how to get this working but I am having more trouble than the CPU version by far.

2021-04-24, 17:09	#3494
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 3×17×97 Posts	From SRBase: NVIDIA NVIDIA GeForce RTX 3090 (4095MB) driver: 466.11 OpenCL: 3.0 Microsoft Windows 10 Professional x64 Edition, (10.00.19042.00) (https://srbase.my-firewall.org/sr5/r...?hostid=191961) Date Time \| class Pct \| time ETA \| GHz-d/day Sieve Wait Apr 24 13:03 \| 4515 98.0% \| 0.022 n.a. \| 5711.23 82485 n.a.% Apr 24 13:03 \| 4519 98.1% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4536 98.2% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4539 98.3% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4540 98.4% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4551 98.5% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4555 98.6% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4564 98.8% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4567 98.9% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4572 99.0% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4575 99.1% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4576 99.2% \| 0.022 n.a. \| 5711.23 82485 n.a.% Apr 24 13:03 \| 4579 99.3% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4584 99.4% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4591 99.5% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4596 99.6% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4599 99.7% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4600 99.8% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4611 99.9% \| 0.021 n.a. \| 5983.19 82485 n.a.% Apr 24 13:03 \| 4612 100.0% \| 0.021 n.a. \| 5983.19 82485 n.a.% no factor for M685141049 from 2^72 to 2^73 [mfaktc 0.21 barrett76_mul32_gs] tf(): total time spent: 20.735s NVIDIA NVIDIA GeForce RTX 3060 Ti (4095MB) driver: 466.11 OpenCL: 3.0 Microsoft Windows 10 Professional x64 Edition, (10.00.19042.00) (https://srbase.my-firewall.org/sr5/r...?hostid=212591) Date Time \| class Pct \| time ETA \| GHz-d/day Sieve Wait Apr 24 19:03 \| 4523 98.0% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4524 98.1% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4535 98.2% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4536 98.3% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4539 98.4% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4544 98.5% \| 0.044 n.a. \| 2861.80 82485 n.a.% Apr 24 19:03 \| 4548 98.6% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4556 98.8% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4559 98.9% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4560 99.0% \| 0.044 n.a. \| 2861.80 82485 n.a.% Apr 24 19:03 \| 4563 99.1% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4571 99.2% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4580 99.3% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4583 99.4% \| 0.045 n.a. \| 2798.20 82485 n.a.% Apr 24 19:03 \| 4599 99.5% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4604 99.6% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4608 99.7% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4611 99.8% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4616 99.9% \| 0.043 n.a. \| 2928.35 82485 n.a.% Apr 24 19:03 \| 4619 100.0% \| 0.043 n.a. \| 2928.35 82485 n.a.% no factor for M683660221 from 2^72 to 2^73 [mfaktc 0.21 barrett76_mul32_gs] tf(): total time spent: 41.986s More benchmarks at https://srbase.my-firewall.org/sr5/top_hosts.php. 1) First check how many GPU's are allocated per host. If more than X then the host is running X-wus in parallel, GHz-d/day will be variable. 2) On computer Info click "Tasks" per ID 3) Click "Valid" on the top 4) Click any result under "Task" 5) Scroll down to see results. Sometimes the client hides the host details so you won't see it. Go to 2) and repeat until you find a host with all output open to outside. Last fiddled with by pinhodecarlos on 2021-04-24 at 17:20

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
mfakto: an OpenCL program for Mersenne prefactoring	Bdot	GPU Computing	1676	2021-06-30 21:23
The P-1 factoring CUDA program	firejuggler	GPU Computing	753	2020-12-12 18:07
gr-mfaktc: a CUDA program for generalized repunits prefactoring	MrRepunit	GPU Computing	32	2020-11-11 19:56
mfaktc 0.21 - CUDA runtime wrong	keisentraut	Software	2	2020-08-18 07:03
World's second-dumbest CUDA program	fivemack	Programming	112	2015-02-12 22:51

2021-03-19, 21:47	#3488
James Heinrich "James Heinrich" May 2004 ex-Northern Ontario 3405₁₀ Posts	If you look at GPU-Z monitoring clockspeed and GPU utilization, do you notice any relevant correlation when you open/minimize Chrome?

2021-03-20, 15:20	#3490
lalera Jul 2003 13×47 Posts	win10 - poweroptions ?

2021-06-05, 12:27	#3498
vasyannyasha "Vasiliy" Apr 2017 Ukraine 3²×7 Posts	Thanks! worked out pretty well.