Sorry for the late reply tictoc, I think the email sub got buried :o, or there isn't 1. Anyway, thx for the times, you've posted a new fastest time :).

Btw are they standard 7970 or GHz edition versions? Either way, nice o/cs :)

My 7970's are original reference models. The 7970 in my Windows 7 rig has been crunching and folding at 1200 Mhz since release day. It has spent most of it's life under water, so core and VRM temps have always been kept in check. It is actually stable at 1250Mhz for a few other projects.

My other three 7970's are also standard reference models. The PCB on the original reference 7970s is a very robust design, with none of the short-comings that many of the later model boards had.

Just adding mine.I usually get around 34 seconds. Had some issues with nbody computations the last few days so won't show those days runs as it caused problems.

1083040166 797988285 610227 17 Apr 2015, 13:10:24 UTC 17 Apr 2015, 13:27:07 UTC Completed and validated 34.07 2.08 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1083040169 798034017 610227 17 Apr 2015, 13:10:24 UTC 17 Apr 2015, 13:28:14 UTC Completed and validated 34.21 2.23 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1083039191 798063748 610227 17 Apr 2015, 13:09:17 UTC 17 Apr 2015, 13:27:07 UTC Completed and validated 36.19 2.33 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1083039193 798063750 610227 17 Apr 2015, 13:09:17 UTC 17 Apr 2015, 13:26:00 UTC Completed and validated 34.11 2.09 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1082175457 797417371 610227 16 Apr 2015, 20:57:39 UTC 16 Apr 2015, 21:16:39 UTC Completed and validated 34.18 2.42 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1082174419 797416576 610227 16 Apr 2015, 20:56:32 UTC 16 Apr 2015, 21:14:25 UTC Completed and validated 34.12 2.59 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1082174420 797416577 610227 16 Apr 2015, 20:56:32 UTC 16 Apr 2015, 21:15:32 UTC Completed and validated 34.07 2.61 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)

Sapphire HD 7870 XT Boost
GPU Clock: 925 MHz
Boost Clock: 975 MHz
Memory Clock: 1500 MHz (6000 MHz effective)

Runs at stock which is 975mhz when working, tried overclocking it in AMD CCC but it never seems to work.

Rest of rig:

Intel 4790k at stock
24gb RAM
Z97 motherboard

Runs at stock which is 975mhz when working, tried overclocking it in AMD CCC but it never seems to work.

My Sapphire HD 7950 Boost used to do the same thing, it was hitting the TDP limit. If your card has a bios switch make sure it's set to position 1, that's what fixed it for me.

Thanks, will take a look at the weekend. I don't see it going over 80c most of the time, not sure what the TDP of these cards is but doubt its less than 90c.
I may try the Sapphire Trixx software again and see if that lets me overclock it.

TDP is about power usage rather than temperatures.

Btw you stated your XT's GPU clock is 925 MHz then state it runs at stock of 975 MHz, don't you mean it runs at boost speed of 975 MHz? ;)

Thx for your time anyway, it's good to have a 7870 XT in the list again :), (will add it in the next few days).

tictoc
Thx for the info :)
What are the short comings on latter 7970s? (I'm looking at buying a 2nd hand 7970 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

Ha, yes sorry, 925mhz stock, 975mhz boost, but as I always see it at 975mhz that's stock for me :-)

It's a good card, gets through units quick enough for me.

Sorry can't help you on the 7970 question :-)

Yea the 7870 XT is still a pokey card :)

The 7970 question is for tictoc ;)

---

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

Hi all. Where do I get the "speed" numbers from exactly?
Ive got several rigs, one with dual Radeon 7850's.
Im getting what appears to be crappy performance based on the "credit" im receiving. It's much worse than my CPU rigs, by a factor of 4x or more.
And that's not saying much, because my CPU rigs are also very flakey. I can't figure out how to get all of the cores working. Im always running less than the total number of cores...

Trying to understand how to read the numbers...

Seems like it's saying that the 4700 is baseline and the 7850 is only 61% as fast (slower) than the 4700? Doesn't make sense.

(1.000) ATI Radeon HD 4700/4800 (RV740/RV770)
(0.696) AMD Radeon HD 7870/7950/7970/R9 280X series (Tahiti)
(0.615) AMD Radeon HD 7850/7870 series (Pitcairn)

My Dual 7850's run MUCH slower than my cpu's in terms of the "credit" I am receiving. I still don't know where to get the timing numbers you are refering to in the above posts.

Which is the fastest GPU available currently?

I've no idea where you're getting those numbers from, you need to read the 'run time' table I've posted here, just above (& at AnandTech). That's why I directed you here ;). The 'speed' numbers are in this thread, lol , look at the table in the post dated April 16th.

Unfortunately I have no HD 7850 times here, probably because the their times would be very slow due to their poor double precision power. Although I would think they are still faster than most quad core cpus, for MW.

From an efficiency point of view you'd better off running 7850s on another project, but if you only want to crunch MW@H then I'd be interested to know what times you get for 106.88 credit WUs :).

Oh & the fastest available GPU is not the same as the fastest GPU for MW@H, games rely on single precision GPU power, MW relies on double precision.
Currently the fastest GPU for MW is the AMD based Tahiti, so 7870 XTs, 7950, 7970 & R9 280/280X. The absolute fastest is the HD 7970 GHz edition (not including dual GPU cards or overclocking).

---

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

Trying to understand how to read the numbers...

Seems like it's saying that the 4700 is baseline and the 7850 is only 61% as fast (slower) than the 4700? Doesn't make sense.

(1.000) ATI Radeon HD 4700/4800 (RV740/RV770)
(0.696) AMD Radeon HD 7870/7950/7970/R9 280X series (Tahiti)
(0.615) AMD Radeon HD 7850/7870 series (Pitcairn)

My Dual 7850's run MUCH slower than my cpu's in terms of the "credit" I am receiving. I still don't know where to get the timing numbers you are refering to in the above posts.

Which is the fastest GPU available currently?

Milkyway uses Double Precision (Aka FP64) for it's calculations so if you want to compare go by that.
The FP64 performance of the 7850 is only 1/16 of it's FP32 (Single Precision) performance which comes out to 110 GFlop/s. The HD 4770 does around 192 GFlop/s in FP64.

Fastest gpu would probably be the GTX Titan Black (around 1700GFlop/s FP64) but the R9 280X is much cheaper and does around 1000GFlop/s FP64.

Ah yea the GTX Titan has greater DP power, but last I heard their are problems getting to fully utilise it's power for MW.

*******************************************************************************

Regarding your 7850 times I get an average of 139.4s from these 9 106.88 credit WUs.
Are both your 7850s at the same stock clock speeds?

1088745178 802272651 21 Apr 2015, 16:25:38 UTC 21 Apr 2015, 17:25:33 UTC Completed and validated 140.03 5.57 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088745182 802272655 21 Apr 2015, 16:25:38 UTC 21 Apr 2015, 17:25:33 UTC Completed and validated 136.56 5.40 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088741893 802270217 21 Apr 2015, 16:22:19 UTC 21 Apr 2015, 17:22:18 UTC Completed and validated 141.32 5.46 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088741899 802270223 21 Apr 2015, 16:22:19 UTC 21 Apr 2015, 17:22:18 UTC Completed and validated 136.34 5.91 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088740869 802269513 21 Apr 2015, 16:21:13 UTC 21 Apr 2015, 17:20:05 UTC Completed and validated 140.43 5.30 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088738915 802268022 21 Apr 2015, 16:19:01 UTC 21 Apr 2015, 17:18:58 UTC Completed and validated 136.22 5.52 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088737761 802267206 21 Apr 2015, 16:17:55 UTC 21 Apr 2015, 17:17:51 UTC Completed and validated 136.19 5.76 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088736603 802266366 21 Apr 2015, 16:16:53 UTC 21 Apr 2015, 17:16:45 UTC Completed and validated 145.41 5.38 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)
1088736608 802266371 21 Apr 2015, 16:16:53 UTC 21 Apr 2015, 17:14:33 UTC Completed and validated 142.12 5.68 106.88 MilkyWay@Home v1.02 (opencl_amd_ati)

To compare, your X6650 Xeons WU times (which are fluctuating significantly btw!) are roughly from 5300-6400s per 106.88 credit WU.
Which even accounting for 12 cores (unless you're using HT?) that still wouldn't be enough to outproduce 1 7850 (5300/12 = ~442s).
Are you looking at the average credit column for your info? That can take a few weeks before it reaches actual output.

---

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

I've no idea where you're getting those numbers from.

http://milkyway.cs.rpi.edu/milkyway/gpu_list.php

Saw this on the anandtech website, Seems to reflect Boinc with double-precision calculations. I know this isn't MilkyWay@Home, but should be close...

While im on the topic of GPU performance, does anyone know anything about these?

The Tesla and Quadro are the professional workstation products from Nvidia. Mainly for graphics and GPGPU financial processing. Those are based on the old Kepler family of chips. I don't think they have come out with any professional cards based on the new Maxwell chips yet. Those products always have had full math performance and are not hamstrung like the consumer video cards.

Cheers, Keith

---

I have a Tesla K20c, Quadro K4000 and K4200.
Quadro's are not good for crunching, since they are mostly low-profile, thus with not adequate cooling.
The Tesla runs much better. I gave it a run on MW@H some time ago and the times were somewhat comparable to Radeon 280/280X. Besides having a much better FP64 performance, they offer ECC, which is a feature that's better to turn off for crunching. One also needs to point out that there are only OpenCL apps on MW available, and the Tesla would perform much better if there would be CUDA FP64 apps.
I'm currently running Einstein on the Tesla, but can try MW again if you need more precise numbers. Anyway, it's a too expensive card for such purposes.

No need, unless you feel like it. I was more interested in the comparison with other cards. I had thought since this was a higher-end professional card, it might offer better performance, but if it's similar to a 280x, then that's good enough for me to make a choice.

HP offers a 4U, half-width blade which supposedly supports up to 8x GPU's. That would allow for 16 GPU's in a 4U rackspace. I was wondering how this might perform compared to other HW.
http://www8.hp.com/us/en/products/proliant-servers/product-detail.html?oid=5177945

That's most probably for single-slot GPUs only. For crunching you need GPUs with better cooling (dual-slot at least), when the airflow in case is not good, you'll need to keep an additional spare slot between them too. So 3 slots per GPU.
From my experience the Tahiti tends to run about 8-10 C hotter when doing FP64. Power consumption can go even twice as high.
So in the winter I'm doing MW@H and in summer I switch to lighter tasks (like Einstein) ;-)
I haven't observed such a big difference between SP and FP64 tasks on the Tesla, but I'll check again.

I've no idea where you're getting those numbers from.

http://milkyway.cs.rpi.edu/milkyway/gpu_list.php

Ah ok, yea their far to vague those lists, that's why I created this thread :) (I think I mentioned that

Good point on the AT F@H benchmarks, odd that the 280X was slightly faster than the 7970 GE though.

So are your 7850s at default clocks then? I'd like to post that time for your 7850 as it would be the 1st 1 on the list :)

Mumak

If you don't mind I'd like to see what time that card gets :), & any other cards that aren't listed in the table, you haven't put any times in this thread yet ;). (I'll post a new table 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