Message boards :
Number crunching :
New Benchmark Thread - times wanted for any hardware, CPU or GPU, old or new!
Message board moderation
Previous · 1 . . . 10 · 11 · 12 · 13 · 14 · 15 · 16 . . . 19 · Next
Author | Message |
---|---|
Send message Joined: 8 Mar 15 Posts: 30 Credit: 78,345,300 RAC: 102 |
http://www.geeks3d.com/20140305/amd-radeon-and-nvidia-geforce-fp32-fp64-gflops-table-computing/ Here you can see the FP32 / FP64 GFlops of many GPUs. I think that GTX Titan and HD7970 are the best option for cost/power. ASUS X570 E-Gaming AMD Ryzen 9 3950X, 16 core / 32 thread 4.4 GHz AMD Radeon Sapphire RX 480 4GB Nitro+ Nvidia GTX 1080 Ti Gaming X Trio 4x16 GB Corsair Vengeance RGB 3466 MHz |
Send message Joined: 9 Jul 17 Posts: 100 Credit: 16,967,906 RAC: 0 |
I just started with an RX 570 running at the stock clock of 1244 MHz under Win7 64-bit, and supported by two cores of an i7-4771. And I am using the latest AMD drivers (Radeon 18.9.3) if it matters any more. My last five work units averaged 121 seconds. https://milkyway.cs.rpi.edu/milkyway/results.php?hostid=737912&offset=0&show_names=0&state=4&appid= I was surprised at how well it did compared to my GTX 1070, which ran about twice as long as I recall. But the really significant part is how cool it runs by under-volting to 0.900 volts. I used the Phantom utility, as it is an Asrock card, and it is now at 50 C. Also, GPU-Z shows the GPU power at 50 watts, but that is not the whole card. My UPS shows a difference of 65 watts between running and not running, or considering the 90% efficiency of the power supply, 58.5 watts to the card. That is also lower than the GTX 1070. It has been a while since I had an AMD card, and if you set them right, they run rather well even if the DP performance is not what it used to be. |
Send message Joined: 6 Mar 10 Posts: 4 Credit: 37,798,404 RAC: 0 |
System Specs: AMD Ryzen 1700X @ 3.5GHz (stock) Asus Prime x470-Pro TeamGroup 32GB DDR4-3000 (16GB x 2 dual-channel) Zotac GeForce GTX 1060 3GB (single fan) Xubuntu 18.04 LTS w/ 4.15 kernel nVidia 415.13b Unix drivers BOINC 4.9.3 x64 Average Benchmark Summary (taken from 5 WU's in a row): 250 seconds per 'bundle4_4s' workunits 281 seconds per 'bundle6_2s' workunits Average GPU Usage (as according to output of 'nvidia-sml'): 92% load 54°C temperature 62% fan speed 62W / 120W power consumption 154MB per 'bundle4_4s' WU 384MB per 'bundle6_2s' WU Detailed WU Information (as copied from BOINC upon WU completion): Application: MilkyWay@Home 1.46 (opencl_nvidia_101) Name: de_modfit_sim19fixed_bundle4_4s_NoContraintsWithDisk260_3_1541104502_6822657 State: Ready to report Received: Sun 18 Nov 2018 05:09:00 AM EST Report deadline: Fri 30 Nov 2018 05:09:00 AM EST Resources: 0.5 CPUs + 1 NVIDIA GPU Estimated computation size: 42,151 GFLOPs CPU time: 00:00:52 Elapsed time: 00:04:10 Executable: milkyway_1.46_x86_64-pc-linux-gnu__opencl_nvidia_101 Application: MilkyWay@Home 1.46 (opencl_nvidia_101) Name: de_modfit_sim19fixed_bundle4_4s_NoContraintsWithDisk260_3_1541104502_6576179 State: Ready to report Received: Sun 18 Nov 2018 05:09:00 AM EST Report deadline: Fri 30 Nov 2018 05:09:00 AM EST Resources: 0.5 CPUs + 1 NVIDIA GPU Estimated computation size: 42,151 GFLOPs CPU time: 00:00:50 Elapsed time: 00:04:10 Executable: milkyway_1.46_x86_64-pc-linux-gnu__opencl_nvidia_101 Application: MilkyWay@Home 1.46 (opencl_nvidia_101) Name: de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140_1_1541104502_6652016 State: Ready to report Received: Sun 18 Nov 2018 05:09:00 AM EST Report deadline: Fri 30 Nov 2018 05:09:00 AM EST Resources: 0.5 CPUs + 1 NVIDIA GPU Estimated computation size: 37,764 GFLOPs CPU time: 00:01:35 Elapsed time: 00:04:41 Executable: milkyway_1.46_x86_64-pc-linux-gnu__opencl_nvidia_101 Application: MilkyWay@Home 1.46 (opencl_nvidia_101) Name: de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140_1_1541104502_6825958 State: Ready to report Received: Sun 18 Nov 2018 05:09:00 AM EST Report deadline: Fri 30 Nov 2018 05:09:00 AM EST Resources: 0.5 CPUs + 1 NVIDIA GPU Estimated computation size: 37,764 GFLOPs CPU time: 00:01:34 Elapsed time: 00:04:41 Executable: milkyway_1.46_x86_64-pc-linux-gnu__opencl_nvidia_101 Application: MilkyWay@Home 1.46 (opencl_nvidia_101) Name: de_modfit_sim19fixed_bundle4_4s_NoContraintsWithDisk260_1_1541104502_6825708 State: Ready to report Received: Sun 18 Nov 2018 05:09:00 AM EST Report deadline: Fri 30 Nov 2018 05:09:00 AM EST Resources: 0.5 CPUs + 1 NVIDIA GPU Estimated computation size: 42,151 GFLOPs CPU time: 00:00:52 Elapsed time: 00:04:07 Executable: milkyway_1.46_x86_64-pc-linux-gnu__opencl_nvidia_101 |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
Hello, Seeing as no else answered you! No special program for benchmarking or seeing 'benchmarks', simply go to the Milkway homepage> your account> tasks. Then pick at least 5 Work Units with 227.23 credits & average the time. (Although I'm wondering how prominent the 227.23 WUs are now....) Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
Ok guys, I've been looking at the range of credited WUs in this Top participant's valid list - http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=780711&offset=0&show_names=0&state=4&appid= And I'm only seeing 227.62 or 203.92 credit WUs in 200 tasks. So it would seem the 'old' benchmark WU is over along with it's benchmark table. Are you guys seeing the same thing in your valid tasks? If so then I need to change the benchmark requirements to the 227.62 WUs (then we may have a rough comparison to the old table, maybe!). If you guys concur on the above I will start a new benchmark thread (seeing as I can't edit the op! grr) I will kick it off with the usual requirements, as below :- Average from at least 5 WU times using 227.62 credit WUs only (not cherry picked please! ;)). A dedicated physical CPU core for each GPU (for optimal MW WU times). If only using BOINC for CPU tasks, & you have an HT capable CPU, then the only way to be certain of this (bar disabling HT) is to set the BOINC computing preferences (in advanced mode>options) so that you have 1 less CPU thread running then you do physical cores. Don't panic too much about lost CPU ppd, it doesn't take long to run MW GPU WUs ;) (see previous table). Please state what speed & type CPU you have, as it now has a significant affect on GPU WU times! Please state GPU clock speeds if overclocked(including factory overclocks) or state 'stock'. Please state whether only crunching 1 WU at a time per GPU, or state how many are run concurrently (I will create a separate table for that). For CPU times please state whether Hyper Threading (or equivalent) is enabled or not, times for both states welcomed :). It would also be useful if you could state your BOINC, driver version, & OS, incase it does make any difference. If you find your WU times are fluctuating more than a couple of % then use GPU-Z or your grx card driver tools to check that your GPU is able to hit near 100% load (although I'm not sure that NVidia cards can hit that for MW), note that even when crunching normally, the GPU load will be on/off on this current MW app, so the GPU load graph should look like a series of blocks. Also check using task manager that your CPU does actually have the spare load to give to MW (& btw, GPU crunching won't show up in the TM). ********************************************************************************************************** Regarding AMD vs NVidia for MW@H, AFAIK it's still the case that even when an NVidia card is matching an AMDs card for raw Double Precision power, the AMD card still does better, although running concurrent WUs on NVidia's closes this gap a lot. Not sure why this is the case, although I have vague memories of MW being written for AMD cards only initially, IIRC! Evans & vseven I see you've already posted WU times for the 227.62 WUs, if you can each provide me with an average from at least 5 WUs I will kick off the table with your times :). Btw Vseven, you mentioned 'new software', what new s/w? Also what CPU was that V100 running on? XArchAngel Interesting info you have their, but you didn't mention once which credit WUs they were! ;), crucial for the benchmark table. Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
Send message Joined: 9 Jul 17 Posts: 100 Credit: 16,967,906 RAC: 0 |
And I'm only seeing 227.62 or 203.92 credit WUs in 200 tasks. Yes. Fire away with your new thread. |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
I just started with an RX 570 running at the stock clock of 1244 MHz under Win7 64-bit, and supported by two cores of an i7-4771. And I am using the latest AMD drivers (Radeon 18.9.3) if it matters any more. A link to valid tasks is no use for my table, I need a calculated average of 5 benchmark WUS :), also you say the last 5 averaged were 121s, but looking at your current page (for 227.62 credit) WUs I'd say the average is about 193-194s, something on your rig change? [edit] Just seen your inbetween reply ;). I've just remembered this thread has been stickified, I'll have to get this one unstuck & the new one stickified once we've received further confirmation on the 227.62 WUs. Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
Send message Joined: 9 Jul 17 Posts: 100 Credit: 16,967,906 RAC: 0 |
A link to valid tasks is no use for my table, I need a calculated average of 5 benchmark WUS :), also you say the last 5 averaged were 121s, but looking at your current page (for 227.62 credit) WUs I'd say the average is about 193-194s, something on your rig change? OK, I think you have it straight now. The 121 seconds was the average of the last 5 (running 1x), but I then increased it to 2x. It seems to be mildly helpful, with a proportionate increase in output for a corresponding increase in power. I am interested to see how the RX 570 does in comparison to the other cards; I am mainly interested in efficiency though, rather than just output. |
Send message Joined: 31 Dec 11 Posts: 17 Credit: 3,172,528,345 RAC: 0 |
Here's some of the new 227.62 tasks for your database. All times are with one task running per GPU. OS: Arch Linux kernel 4.19 BOINC Version: 7.12.1 GPU Driver: Catalyst 15.12 CPU: Intel Xeon E5 ES 10 core @ 2700MHz (ht off) GPU: AMD HD7970 @ 1200/1400 10 WU avg run-time - 38.23s GPU: AMD R9 290 @ 1000/1300 10 WU avg run-time - 70.94s |
Send message Joined: 16 Jun 16 Posts: 9 Credit: 781,147,932 RAC: 16,407 |
Avg (of 5) = 120.33 seconds from the 227.62 WUs AMD A12-9800 stock; ddr4-2400, onboard (iGPU) R7 graphics Boinc 7.9.3, Ubuntu 18.04.1 LTS, OpenCL 1.2, amdgpu-pro 18.50 drivers. Only MW@Home running. No CPU tasks. 1 WU at a time. Still not 100% stable. Will investigate if power delivery issue or driver issue. Locks up when asked to do too much. Stable if left to crunch.
|
Send message Joined: 9 Jul 17 Posts: 100 Credit: 16,967,906 RAC: 0 |
Avg (of 5) = 120.33 seconds from the 227.62 WUs That seems very fast to me for an APU, since you are getting times comparable to a lot of GPUs (both AMD and Nvidia) that must have much more graphics power, I would think. So maybe the APU offloads the dual-precision workload to the CPU more efficiently? I really don't know anything about APUs, so maybe someone who does can explain it. |
Send message Joined: 16 Jun 16 Posts: 9 Credit: 781,147,932 RAC: 16,407 |
It is trending slower today as I turned off motherboard core boost (3.8 -> 4.2) to see if I am hitting a wall with what the motherboard is able to deliver. 10WU avg is now 135.06. I think my powersupply isn't up to the task of boosting the cpu while the iGPU units are being used. Or it could just be driver or mobo related. I am going to try other hardware as it was frozen this morning (locked up). Yes, agreed, I think the proximity of the gpu and cpu units is advantageous in an otherwise mediocre chip. Or maybe they share the cache? Avg (of 5) = 120.33 seconds from the 227.62 WUs |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
A link to valid tasks is no use for my table, I need a calculated average of 5 benchmark WUS :), also you say the last 5 averaged were 121s, but looking at your current page (for 227.62 credit) WUs I'd say the average is about 193-194s, something on your rig change? Hi Jim I am (mainly) after singly run WUs, so your 1st time was the right one :), I just wanted to clarify what changed the times. But yea running 2 (& more on some cards)at once certainly can improve throughput. (Btw, CPUs are much slower at MW than GPUs, so that's not what's giving hoppisaur APU a good score IMO). Will add yours, hoppisaur & tictocs time when I post the new thread (at some point! ;)). hoppisaur I see the R7 is a 3rd gen GCN architecture https://en.wikipedia.org/wiki/Graphics_Core_Next#third, which is what the Radeon R9 285 is based on, that's why it's got fairly decent MW times :). Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
Here's some of the new 227.62 tasks for your database. All times are with one task running per GPU. Hi, added times to the AnandTech benchmark thread, but are both those GPUs in the same machine? If not what CPUs are with which GPUs? It is trending slower today as I turned off motherboard core boost (3.8 -> 4.2) to see if I am hitting a wall with what the motherboard is able to deliver. 10WU avg is now 135.06. I think my powersupply isn't up to the task of boosting the cpu while the iGPU units are being used. Or it could just be driver or mobo related. I am going to try other hardware as it was frozen this morning (locked up). Is the 135s time with the CPU at 3.8 GHz with 227.62 WUs? (I'll add that time too if so). Will link the AT thread again in a moment. Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
New benchmark table posted in the AnandTech thread here http://forums.anandtech.com/threads/milkyway-h-benchmark-thread-winter-2016-on-updated-1-2019-gpu-cpu-times-wanted-for-new-wus.2495905/ (more up to date, as I can actually update posts there beyond 1hr!). Mods Could you de-stickfy this thread please? I will post a new thread soon.... Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
Send message Joined: 31 Dec 11 Posts: 17 Credit: 3,172,528,345 RAC: 0 |
Here's some of the new 227.62 tasks for your database. All times are with one task running per GPU. Those GPUs were in the same machine when I posted those times. The CPU is more or less an E5-2680v2, with an all-core turbo of 2700MHz. |
Send message Joined: 4 Mar 10 Posts: 65 Credit: 639,958,626 RAC: 0 |
where is link to your task times..? why you hide computer/s ? or screenshot ..)) this benchmark thread is pure joke, one man show. first tasks is always fast.. after it going slower.. I have also first 100 task about 20-25 sec. and after settle in 40-50, depend on cpu speed. |
Send message Joined: 16 Jun 16 Posts: 9 Credit: 781,147,932 RAC: 16,407 |
Of latest 14 validated 227.62 WUs, avg is 100.65 secs. Chip is at 3.8GHz. I've tweaked BIOS on another board as much as I can. First board died complete with magic smoke. A12-9800 with dual channel DDR4-2400.
|
Send message Joined: 26 Mar 18 Posts: 24 Credit: 102,912,937 RAC: 0 |
Got to play with a Tesla T4 for a little bit. Same machine as my Tesla V100 test a couple months ago. CUDA 10.0 and nVidia 410.28 drivers, 8 CPU cores. The T4, kinda as expected, is slower in double precision then its predecessor V100: 119256158 1709623437 15 Jan 2019, 15:26:28 UTC 16 Jan 2019, 13:43:18 UTC Completed and validated 151.29 150.00 227.62 MilkyWay@Home v1.46 (opencl_nvidia_101) 119256165 1709847295 15 Jan 2019, 15:26:28 UTC 16 Jan 2019, 14:53:50 UTC Completed and validated 151.34 149.90 227.62 MilkyWay@Home v1.46 (opencl_nvidia_101) 119256166 1709847329 15 Jan 2019, 15:26:28 UTC 16 Jan 2019, 13:35:04 UTC Completed and validated 151.17 149.10 227.62 MilkyWay@Home v1.46 (opencl_nvidia_101) 119256168 1709847342 15 Jan 2019, 15:26:28 UTC 16 Jan 2019, 14:37:58 UTC Completed and validated 151.21 150.09 227.62 MilkyWay@Home v1.46 (opencl_nvidia_101) 119255936 1709840508 15 Jan 2019, 15:26:28 UTC 16 Jan 2019, 14:19:00 UTC Completed and validated 151.35 149.79 227.62 MilkyWay@Home v1.46 (opencl_nvidia_101) 119256225 1709852404 15 Jan 2019, 15:26:28 UTC 16 Jan 2019, 14:25:53 UTC Completed and validated 151.43 150.00 227.62 MilkyWay@Home v1.46 (opencl_nvidia_101) One WU maxed out the GPU and takes about 151 seconds, overall about 30x slower then the v100 when you consider its taking almost 6 times as long per WU and can only do one at a time. Tesla V100 is still the king for the foreseeable future. |
Send message Joined: 22 Jan 11 Posts: 375 Credit: 64,667,747 RAC: 902 |
where is link to your task times..? why you hide computer/s ? Beg your pardon!!?? If you think you can do a better job then go ahead & do it! Yes it's a 1 man show because I started this off after seeing a benchmark thread in the Folding@home forums years ago, (& a MW one in the AnandTech DC forum). So instead of spouting empty BS insults why don't you explain why this thread is a joke? You're taking the piss after the amount of time I've put into this thread! And I assume you're talking to tictoc about hiding computers, he's entirely entitled to do so! Some people have genuine security reasons for doing so, e.g a person on my team lived in a very sparsely populated area of a small population country & he has many computers, getting his IP to get a rough location of where he is could be enough to find his house. Links to tasks times aren't very useful as the results move on, links to individual tasks may be more useful but it would be a pain to look at every single task for every single time posted by every user. Yes a screenshot would be useful, but not required as this is done on trust, if a time looks dodgy then we can inquire more about it, it could be an innocent mistake, like wrong credit WUs, faulty machine, maths error etc. Theirs no need to take the accusatory tone that you are taking! And no, the 1st task is not always the fastest, something odd about your machine it seems! Or you're badly explaining yourself, that sentence is badly written. If you're getting task times varying from 20-25 to 40-50 then theirs something wrong with your setup, that is not a normal variation, even when running multiple WUs! The average of 5 result tasks smooths out any slight variations that typically occur. Team AnandTech - SETI@H, DPAD, F@H, MW@H, A@H, LHC, POGS, R@H, Einstein@H, DHEP, WCG Main rig - Ryzen 5 3600, MSI B450 G.Pro C. AC, RTX 3060Ti 8GB, 32GB DDR4 3200, Win 10 64bit 2nd rig - i7 4930k @4.1 GHz, HD 7870 XT 3GB(DS), 16GB DDR3 1866, Win7 |
©2024 Astroinformatics Group