gpuOwL: an OpenCL program for Mersenne primality testing - Page 242

storm5510 · 2020-12-22, 19:19

Quote:

Originally Posted by DrobinsonPE

GTX 1650 Super...

I have a GTX 1650, but not "Super." It actually surprised me with what it could do. It will run a consistent 950 GHz-d/day with mfaktc at factory settings. I was not expecting much above 500. However, it really struggles with gpuOwl 6.x, enough so that I don't run this on it. For the cost and low energy consumption, I am quite pleased with it even if it does not do certain things very well.

moebius · 2020-12-22, 19:29

Quote:

Originally Posted by storm5510

I have a GTX 1650, but not "Super."

I would also rather do TF on the card, but you can still create a benchmark for me if you have the time. The GTX 1650 uses the TU117 chipset, the GTX 1650 Super TU116. Possibly you can still do LL double checks with gpuowl with this one in an acceptable time.

Zenzoma · 2020-12-22, 21:57

I bought a Radeon Pro VII from B&H last month. It's a new system with an AMD 3700X.

Here's a recent DC:

Code:

2020-12-13 09:49:47 gfx906-0 84296071 FFT: 4.50M 1K:9:256 (17.86 bpw)
2020-12-13 09:49:47 gfx906-0 84296071 OpenCL args "-DEXP=84296071u -DWIDTH=1024u -DSMALL_HEIGHT=256u -DMIDDLE=9u -DAMDGPU=1 -DWEIGHT_STEP_MINUS_1=0.098346913950080425 -DIWEIGHT_STEP_MINUS_1=-0.089540847887837977 -DIWEIGHTS={0,-0.089540847887837977,-0.17106413233520301,-0.24528775277054959,-0.31286532729881017,-0.37439194850560675,-0.43040942388187292,-0.48141104701641219,-0.055691883143657521,-0.14024603259434201,-0.21722913181077724,-0.28731909905033931,-0.35113315119484023,-0.40923323900316377,-0.46213099568677224,-0.020584484771131676,}  -cl-std=CL2.0 -cl-finite-math-only "
2020-12-13 09:49:47 gfx906-0 84296071 ASM compilation failed, retrying compilation using NO_ASM
2020-12-13 09:49:49 gfx906-0 84296071 OpenCL compilation in 2.39 s
2020-12-13 09:49:49 gfx906-0 84296071 maxAlloc: 15.0 GB
2020-12-13 09:49:49 gfx906-0 84296071 P1(0) 0 bits
2020-12-13 09:49:49 gfx906-0 84296071 PRP starting from beginning
2020-12-13 09:49:50 gfx906-0 84296071 OK         0 on-load: blockSize 400, 0000000000000003
2020-12-13 09:49:50 gfx906-0 84296071 validating proof residues for power 8
2020-12-13 09:49:50 gfx906-0 84296071 Proof using power 8
2020-12-13 09:49:51 gfx906-0 84296071 OK       800   0.00% 37d10eab5ea9325d  636 us/it + check 0.29s + save 0.10s; ETA 14:53
2020-12-13 09:49:56 gfx906-0 84296071        10000   0.01% 56dfc915b7d9b1a9  637 us/it
2020-12-13 09:50:03 gfx906-0 84296071        20000   0.02% 7ead874b22960f01  639 us/it
2020-12-13 09:50:09 gfx906-0 84296071        30000   0.04% 95a5fa6a25f1938a  639 us/it
2020-12-13 09:50:16 gfx906-0 84296071        40000   0.05% 3dfac0fb9cdc046d  640 us/it
2020-12-13 09:50:22 gfx906-0 84296071        50000   0.06% d74b3a862f8bc1e9  640 us/it
2020-12-13 09:50:28 gfx906-0 84296071        60000   0.07% cae747451d88d5b0  640 us/it
2020-12-13 09:50:35 gfx906-0 84296071        70000   0.08% bce7c61fd4990794  640 us/it
2020-12-13 09:50:41 gfx906-0 84296071        80000   0.09% ddab7fe2666699b7  640 us/it
2020-12-13 09:50:48 gfx906-0 84296071        90000   0.11% c4b82ccbdb8e4c88  640 us/it

DrobinsonPE · 2020-12-22, 22:10

Quote:

Originally Posted by storm5510

I have a GTX 1650, but not "Super." It actually surprised me with what it could do. It will run a consistent 950 GHz-d/day with mfaktc at factory settings. I was not expecting much above 500. However, it really struggles with gpuOwl 6.x, enough so that I don't run this on it. For the cost and low energy consumption, I am quite pleased with it even if it does not do certain things very well.

Yes. This card will do 1460 GHz-d/day with mfaktc and is already back doing TF. I just wanted to see what it could do running gpuowl. It did better than I thought. The gpuowl timing works out to about 40 GHz-d/day.

moebius · 2020-12-22, 23:53

Quote:

Originally Posted by MattL

I bought a Radeon Pro VII from B&H last month. It's a new system with an AMD 3700X.Here's a recent DC:

This performance probably roughly corresponds to the values of the non Pro version. Can you please give me the value for one instance prp 77936867 so that I can enter it in the comparison list. The value for a simple Raedon VII for this exponent is 525 us / it (optimized and with memory OC).Thx.

Zenzoma · 2020-12-23, 00:17

Quote:

Originally Posted by moebius

This performance probably roughly corresponds to the values of the non Pro version. Can you please give me the value for -one instance prp 77936867 so that I can enter it in the comparison list. Thx.

Sure. Here's the log for that exponent. GPUOwl version is v7.2-21-g28dbf88. The Pro card is "stock". No overclock or anything.

Code:

2020-12-22 19:03:19 GpuOwl VERSION v7.2-21-g28dbf88
2020-12-22 19:03:19 Note: not found 'config.txt'
2020-12-22 19:03:19 config: -maxAlloc 15G -prp 77936867 
2020-12-22 19:03:19 device 0, unique id ''
2020-12-22 19:03:19 gfx906-0 77936867 FFT: 4M 1K:8:256 (18.58 bpw)
2020-12-22 19:03:19 gfx906-0 77936867 OpenCL args "-DEXP=77936867u -DWIDTH=1024u -DSMALL_HEIGHT=256u -DMIDDLE=8u -DAMDGPU=1 -DMM_CHAIN=1u -DMM2_CHAIN=2u -DMAX_ACCURACY=1 -DWEIGHT_STEP_MINUS_1=0.33644726404543274 -DIWEIGHT_STEP_MINUS_1=-0.25174750481886216 -DIWEIGHTS={0,-0.25174750481886216,-0.44011820345520131,-0.16213409745771243,-0.37306474779553728,-0.061788266441989627,-0.29798072935699788,-0.47471232907613115,-0.21390437908665341,-0.41180199020062258,-0.11975874301407295,-0.3413572830988989,-0.014337887291734644,-0.26247586476052853,-0.44814572555075455,-0.17414732433395128,}  -cl-std=CL2.0 -cl-finite-math-only "
2020-12-22 19:03:19 gfx906-0 77936867 ASM compilation failed, retrying compilation using NO_ASM
2020-12-22 19:03:21 gfx906-0 77936867 OpenCL compilation in 2.09 s
2020-12-22 19:03:21 gfx906-0 77936867 maxAlloc: 15.0 GB
2020-12-22 19:03:21 gfx906-0 77936867 P1(0) 0 bits
2020-12-22 19:03:21 gfx906-0 77936867 PRP starting from beginning
2020-12-22 19:03:21 gfx906-0 77936867 OK         0 on-load: blockSize 400, 0000000000000003
2020-12-22 19:03:21 gfx906-0 77936867 validating proof residues for power 8
2020-12-22 19:03:21 gfx906-0 77936867 Proof using power 8
2020-12-22 19:03:22 gfx906-0 77936867 OK       800   0.00% 1579c241dc63eca6  616 us/it + check 0.28s + save 0.09s; ETA 13:20
2020-12-22 19:03:28 gfx906-0 77936867        10000   0.01% fc4f135f7cf4ad29  627 us/it
2020-12-22 19:03:34 gfx906-0 77936867        20000   0.03% 3cd1bd9d5e09cbc5  634 us/it
2020-12-22 19:03:41 gfx906-0 77936867        30000   0.04% c4e0ff35e3290d98  633 us/it
2020-12-22 19:03:47 gfx906-0 77936867        40000   0.05% dffe1b1b0d748128  639 us/it
2020-12-22 19:03:54 gfx906-0 77936867        50000   0.06% 52e286945371ed29  641 us/it
2020-12-22 19:04:00 gfx906-0 77936867        60000   0.08% 0945da4dc08bdd95  637 us/it
2020-12-22 19:04:06 gfx906-0 77936867        70000   0.09% 7131fa4eb77f4bb2  640 us/it
2020-12-22 19:04:13 gfx906-0 77936867        80000   0.10% 8d76071d27ee4221  635 us/it
2020-12-22 19:04:19 gfx906-0 77936867        90000   0.12% 0bacff453b2f470e  624 us/it
2020-12-22 19:04:25 gfx906-0 77936867       100000   0.13% 6d7296b9e2830f50  637 us/it
2020-12-22 19:04:32 gfx906-0 77936867       110000   0.14% 8cbfd4435622bda7  642 us/it
2020-12-22 19:04:38 gfx906-0 77936867       120000   0.15% 79ae5dad855057ad  642 us/it
2020-12-22 19:04:44 gfx906-0 77936867       130000   0.17% 50c97bcbf876231f  638 us/it
2020-12-22 19:04:51 gfx906-0 77936867       140000   0.18% e1db15f897271496  637 us/it
2020-12-22 19:04:57 gfx906-0 77936867       150000   0.19% 127631386c6a9b17  629 us/it
2020-12-22 19:05:03 gfx906-0 77936867       160000   0.21% 25b7b6206fc6f085  633 us/it
2020-12-22 19:05:10 gfx906-0 77936867       170000   0.22% 416816b0d9f4bba8  640 us/it
2020-12-22 19:05:16 gfx906-0 77936867       180000   0.23% 6bee5d054f770861  643 us/it
2020-12-22 19:05:23 gfx906-0 77936867       190000   0.24% f37f068f014b18a0  635 us/it
2020-12-22 19:05:29 gfx906-0 77936867 OK    200000   0.26% f0b04b45b0855bd2  641 us/it + check 0.28s + save 0.09s; ETA 13:50
2020-12-22 19:05:36 gfx906-0 77936867       210000   0.27% 43eb2fc2424d8aac  639 us/it
2020-12-22 19:05:42 gfx906-0 77936867       220000   0.28% a1081c6dc6a7689f  631 us/it
2020-12-22 19:05:49 gfx906-0 77936867       230000   0.30% 2387818d3d3d0d01  640 us/it
2020-12-22 19:05:55 gfx906-0 77936867       240000   0.31% a9deae45055e5216  637 us/it
2020-12-22 19:06:01 gfx906-0 77936867       250000   0.32% 89fcab15218f7cac  641 us/it
2020-12-22 19:06:08 gfx906-0 77936867       260000   0.33% 55da428da4cf928a  641 us/it
2020-12-22 19:06:14 gfx906-0 77936867       270000   0.35% dc349756c5f05abf  641 us/it
2020-12-22 19:06:20 gfx906-0 77936867       280000   0.36% 3564af24488443f4  634 us/it
2020-12-22 19:06:27 gfx906-0 77936867       290000   0.37% 63fb281a06f78198  636 us/it
2020-12-22 19:06:33 gfx906-0 77936867       300000   0.38% 990aa099aad5bf9c  639 us/it
2020-12-22 19:06:40 gfx906-0 77936867       310000   0.40% 61e14297a2cc0096  645 us/it
2020-12-22 19:06:46 gfx906-0 77936867       320000   0.41% 37e630c5f956cf8a  621 us/it
2020-12-22 19:06:52 gfx906-0 77936867       330000   0.42% 66ccde7e28ce2b33  621 us/it
2020-12-22 19:06:58 gfx906-0 77936867       340000   0.44% d4a7cff61adaa84e  622 us/it
2020-12-22 19:07:05 gfx906-0 77936867       350000   0.45% d57b659cc1ca2753  636 us/it
2020-12-22 19:07:11 gfx906-0 77936867       360000   0.46% 992df79b843f90de  641 us/it
2020-12-22 19:07:17 gfx906-0 77936867       370000   0.47% 10b0b99eba490a1e  640 us/it
2020-12-22 19:07:24 gfx906-0 77936867       380000   0.49% 56b1e40cd2666109  634 us/it
2020-12-22 19:07:30 gfx906-0 77936867       390000   0.50% ecccd874a8a0d961  640 us/it
2020-12-22 19:07:37 gfx906-0 77936867 OK    400000   0.51% c03f94396a5aa29e  643 us/it + check 0.28s + save 0.09s; ETA 13:51
2020-12-22 19:07:43 gfx906-0 77936867       410000   0.53% bf44242560060429  637 us/it
2020-12-22 19:07:50 gfx906-0 77936867       420000   0.54% 7f656173fb521927  634 us/it

storm5510 · 2020-12-23, 00:25

Quote:

Originally Posted by moebius

I would also rather do TF on the card, but you can still create a benchmark for me if you have the time..

Give me the parameters for the test and I will run it.

moebius · 2020-12-23, 00:45

Quote:

Originally Posted by storm5510

Give me the parameters for the test and I will run it.

gpuowl-win -iters 200000 -prp 77936867 or gpuowl-win -iters 200000 -carry short -prp 77936867
depending on which configuration runs faster for you.

tServo · 2020-12-23, 01:54

Quote:

Originally Posted by MattL

I bought a Radeon Pro VII from B&H last month. It's a new system with an AMD 3700X.

Here's a recent DC:

Code:

2020-12-13 09:49:47 gfx906-0 84296071 FFT: 4.50M 1K:9:256 (17.86 bpw)
2020-12-13 09:49:47 gfx906-0 84296071 OpenCL args "-DEXP=84296071u -DWIDTH=1024u -DSMALL_HEIGHT=256u -DMIDDLE=9u -DAMDGPU=1 -DWEIGHT_STEP_MINUS_1=0.098346913950080425 -DIWEIGHT_STEP_MINUS_1=-0.089540847887837977 -DIWEIGHTS={0,-0.089540847887837977,-0.17106413233520301,-0.24528775277054959,-0.31286532729881017,-0.37439194850560675,-0.43040942388187292,-0.48141104701641219,-0.055691883143657521,-0.14024603259434201,-0.21722913181077724,-0.28731909905033931,-0.35113315119484023,-0.40923323900316377,-0.46213099568677224,-0.020584484771131676,}  -cl-std=CL2.0 -cl-finite-math-only "
2020-12-13 09:49:47 gfx906-0 84296071 ASM compilation failed, retrying compilation using NO_ASM
2020-12-13 09:49:49 gfx906-0 84296071 OpenCL compilation in 2.39 s
2020-12-13 09:49:49 gfx906-0 84296071 maxAlloc: 15.0 GB
2020-12-13 09:49:49 gfx906-0 84296071 P1(0) 0 bits
2020-12-13 09:49:49 gfx906-0 84296071 PRP starting from beginning
2020-12-13 09:49:50 gfx906-0 84296071 OK         0 on-load: blockSize 400, 0000000000000003
2020-12-13 09:49:50 gfx906-0 84296071 validating proof residues for power 8
2020-12-13 09:49:50 gfx906-0 84296071 Proof using power 8
2020-12-13 09:49:51 gfx906-0 84296071 OK       800   0.00% 37d10eab5ea9325d  636 us/it + check 0.29s + save 0.10s; ETA 14:53
2020-12-13 09:49:56 gfx906-0 84296071        10000   0.01% 56dfc915b7d9b1a9  637 us/it
2020-12-13 09:50:03 gfx906-0 84296071        20000   0.02% 7ead874b22960f01  639 us/it
2020-12-13 09:50:09 gfx906-0 84296071        30000   0.04% 95a5fa6a25f1938a  639 us/it
2020-12-13 09:50:16 gfx906-0 84296071        40000   0.05% 3dfac0fb9cdc046d  640 us/it
2020-12-13 09:50:22 gfx906-0 84296071        50000   0.06% d74b3a862f8bc1e9  640 us/it
2020-12-13 09:50:28 gfx906-0 84296071        60000   0.07% cae747451d88d5b0  640 us/it
2020-12-13 09:50:35 gfx906-0 84296071        70000   0.08% bce7c61fd4990794  640 us/it
2020-12-13 09:50:41 gfx906-0 84296071        80000   0.09% ddab7fe2666699b7  640 us/it
2020-12-13 09:50:48 gfx906-0 84296071        90000   0.11% c4b82ccbdb8e4c88  640 us/it

What operating system and what driver are you using? If your driver is not current, you might a boost if you upgrade it.

Zenzoma · 2020-12-23, 03:10

Quote:

Originally Posted by tServo

What operating system and what driver are you using? If your driver is not current, you might a boost if you upgrade it.

Windows 10 and the latest driver from AMD.

Xyzzy · 2020-12-23, 03:20

What are the temperatures like? Do you have a GPU-Z screen shot to share?

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
GPUOWL AMD Windows OpenCL issues	xx005fs	GpuOwl	0	2019-07-26 21:37
Testing an expression for primality	1260	Software	17	2015-08-28 01:35
Testing Mersenne cofactors for primality?	CRGreathouse	Computer Science & Computational Number Theory	18	2013-06-08 19:12
Primality-testing program with multiple types of moduli (PFGW-related)	Unregistered	Information & Answers	4	2006-10-04 22:38

2020-12-23, 03:20	#2662
Xyzzy "Mike" Aug 2002 202A₁₆ Posts	What are the temperatures like? Do you have a GPU-Z screen shot to share?