Observation of power usage of 2 generations of CPU's/GPU's

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
All power measurements are what they draw from the wall per kill-a-watt.
The CPU's are at stock. The video cards are at stock except Evga precision is overclocking them at default settings. Both systems have a DVD drive, and SSD, but the newer system has 2 older hard drives, 750 gig and 1 tb. The older one has 3 sticks of memory (6 gb total) and the newer one 2 sticks (16 gig total). At these usages, I thing only the CPU and GPU usages really make the difference, and the PSU's should be comparable.

So, first box (sort of -1 generation, video cards are, cpu is 2 or 3 back)
PSU Corsair 1200AXi (about 90% efficient)
Intel 58DXSO motherboard with Xeon X5570@3.0 (stock)
2 x 980TI SC video cards running F@H@ about 1300 mhz (varies)

Second box: current generation
Corsair AX850 (about 90% efficient also)
ASUS B150 Pro gaming motherboard with 6700k @ 4.0 mhz (stock)
2 x 1080 FTW video cards running F@H@ about 2000 mhz (varies also)

At 100% CPU and GPU what do you think the loads are ? (Will post results tomorrow)

I am really curious at the guesses. The PPD for the boxes (approx) are 1.2 million ppd for both, but spikes to 1.6 million or even a little more on the newer box when I get lucky on the units. Both can drop down to 900k when both are really bad units (rare)
 
Last edited:

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
OK, 21 views, and nobody is going to guess, so I guess I will just post it (was going to have a little fun)

Old system 575 watts.

New system 425 watts.

Again, almost tied in ppd, but boy is there a power difference. 150 watts.guessing 50 watts cpu, and 50 watts each GPU ???
 
  • Like
Reactions: Ken g6

Orange Kid

Elite Member
Oct 9, 1999
4,328
2,112
146
I was going to guess 500 and 430 figuring 200 on the 980"s and a100 on the rest for #1 and 180 on the 1080's and 70 for the rest. That Xeon must pull a lot more than I would think.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
Well, I had a single 980 (not ti) that was drawing 320 with a 5570, and if you add 200 for another 980, but a ti, you are already at 520, to the 575 seems pretty in line.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
should retry my new config that has a ti on a X5570.

stay tuned

edit: X5570 with a 980ti is 360 watts, so 215 more for a second one ? motherboard might make a 15 watt difference, this is a cheap dell mobo,

I also have a 4790 with a ti, and a 3930k with a ti, as time permits I may hook those up.

The SR-2 with 2 5639's overclocked, and 2 regular 980's is 750 watts. yikes. Must be the motherboard/CPU's
 
Last edited:

Orange Kid

Elite Member
Oct 9, 1999
4,328
2,112
146
OK CPUBoss reports 95 watts on the Xeon and 88 on the I7 with an average of 77 and 71 watts respectively.

GPUBoss reports 250 on the 980ti's and 180 on the 1080's

So 577 vs 431 and your killawatt is darn close to that o_O:)
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
But the interesting thing is the PPD. I think its due to the cuda cores. the 1080 is only 2560, and the 980ti's are 2816. Its advanced architecture might help sometimes, but there is no replacement for cuda cores in F@H.

edit: and the watts change depending on the unit !!!! The more ppd, the harder it works, the more watts they take.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
Oh, and I am now close to 5 million ppd ! watch the averages in the next week.
 

TennesseeTony

Elite Member
Aug 2, 2003
4,209
3,634
136
www.google.com
Mark, you should redo the test/figures, while running FurMark if you have the cards connected via SLI. It will only stress one card if not in SLI (also has to be fullscreen to enable SLI). To load up the CPUs as well, FurMark has a 'CPUburner' that you can configure to run x number of threads.

Of course that means you would have to pause folding for a minute or two...... :D

That would give a 'worst case' power usage scenario though, not necessarily real life power draw.

When accounting for the CUDA cores vs performance comparisons, don't forget the huge clock speed advantage of the 1080.

Here is an old comparison pic, using R9-280X as the GPU. Dual CPUs are also 95w TDP, but 6 cores each. Those old Dell motherboards/PSU's/fans/DDR3 sticks are hungry.... :oops:

am75eo.png
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
F@H will not run in SLI, and I don't even have the SLI cables ! And the 6700k is using DDR3 ( I did not want to ignore 16 gig of DDR3 I already had)
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
Slightly OT, but my 290 is doing 200-250k ppd, and using 400 watts. Same as my X5570 Xeon and 980ti, but they are doing over 600k ppd for the same usage.

If I turn on the 6234 Opterons on that box, they only get 20k ppd, but use an extra 150 watts or more. ( I shut it then down when I saw that they hit 150 and were not done ramping up yet)
 

theAnimal

Diamond Member
Mar 18, 2003
3,828
23
76
F@H will not run in SLI, and I don't even have the SLI cables ! And the 6700k is using DDR3 ( I did not want to ignore 16 gig of DDR3 I already had)
F@H will run fine with SLI, it's been quite a while since it did not work.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,515
136
F@H will run fine with SLI, it's been quite a while since it did not work.

I stand corrected ! However, there does appear to be a few qualifications on that, and I have never tried it. See below (quoted from F@H site)

  • The current multi-GPU installation guide requires more than a basic knowledge of Microsoft Windows.
  • SLI or Crossfire does not have to be disabled to run the GPU client. However, SLI and CF cannot be used to make a dual GPU card or cards perform like one super GPU. The SLI and CF link does not have enough bandwidth to support that functionality. Run one FAH GPU client for each GPU chip.
  • If you are running multiple GPU cards with SLI or Crossfire, you may need to extend the desktop for the client to recognize multiple GPUs.
  • If Windows prompts for a new driver, hit CANCEL. The default Windows driver may not be the best driver to use. Allowing Windows to install a driver automatically runs the risk of having sub-par performances and/or client installation issues.
  • If you run Windows Vista or Windows 7, you may need to have each GPU connected to a monitor (a KVM can also work but not all models of KVM will provide a signal when it is not the active monitor). Another alternative is to use VGA dummy plug

Edit: this entire page looks to be very informative !!

https://folding.stanford.edu/home/guide/wingpuguide/

However, I have never needed to do most of what is there. I install the client I linked to in the 11th Annual F@H race thread and answer the setup questions, and run FAHControl, and it takes care of everything else.

At least for me.
 
Last edited: