Discussion in 'General Hardware' started by Mussels, Dec 27, 2013.
Ahem what? Is there an echo? That link in my previous post was to the SS 300W PSU.
Your link points to an 80+ Bronze one. Mine points to the 80+ Gold version. And since we're looking for a high-efficiency 300W Seasonic, I think I'm a bit closer to the mark there
Splitting hairs, but still.
Touche, sir.. Touche! LOL!
Later than expected but another to add. And when I get this 290 switched out for the 7770, I'll return with that info too
manofthem | XFX Core 550Pro | i3 2100 @ 3.1GHz + R9 290 1050/1350 | 60w idle | 94w | P8988 360W | 24.97points per watt | BOINC 92W / 3DMark13 341W / F@H 267W
We need MOAR POWA!
less power, preferably
yes, that's what I meant
Come on, fellas, post some numbers. I know some of us would love to see some details on the power comsumption of gaming pcs, crunching pcs, folding pcs, and the like. It's all for the better of us
Swapped the 7770 back in and reran some numbers. Much lower wattage now
manofthem | XFX Core 550Pro | i3 2100 @ 3.1GHz +7770 | 60w idle | 94w | P3997 140W | 28.55points per watt | BOINC 92W / 3DMark13 142W / F@H 100W
i might not reply to every score posted, but they are being added to the first post.
over months and years, its going to be interesting - with any luck, we should get more power efficient as time goes by.
upgraded both my PC's, so i'll be throwing in a batch of new results for myself within the next 24 hours.
(saving them in this post, so i dont lose them if i reboot. editing as tests come in)
Mussels | Corsair HX1000 | i5 2400 @ 3.85GHz + 7970 | 84w idle | 139w | P.... W.... | points per watt | new mobo is good
Mussels | Corsair TX750 | i3 2130 @ 3.4GHz 1.025v + 4890 1GB | 119w idle | 134w | DX10 card, furmark 720p test 239W - 1423 points | 5.95 *furmark* points per watt | 4890... so.... HUNGRY
Mussels | Corsair TX750 | i3 2130 @ 3.4GHz 1.05v + 1GB Geforce 550 Ti OC | 89w idle | 134w | P2766 184W | 15.03 points per watt | so quiet (218W furmark load, 1632 furmark points, 7.48 FPW)
After using furmark on the 7890 and 550 Ti, i've realised it provides the 'maximum' wattages compared to 3dmarks 'average' wattages.
At this time is when i'm considering a V2 of this thread - hiding the old table in a spoiler and archiving it
so the changes would be from 3dmark to furmark (easier to download, run, and test with - and higher, more accurate wattages) and maybe change the CPU benchmark to something that gives a score in points instead of time, so we can get points per watt there too.
It would be better in the long run. I would think it through how you will do it so you don't have to do a v3.
Dont think furmark would be a good idea, it often gets artificially limited by drivers, and if not, it doesn't really give a good example of gaming load poweruse. Maybe heaven instead?
For now then, could you do a run on your system and compare the three.
i like furmark because its got a simple 30 second test, and anyone can run it very fast. it gave higher wattages than 3dmark did, so to me that means its LESS limited.
heaven is a bit more complex since you can change off the default settings, and it takes longer.
Well what I ment with limited by drivers is that for an instance Nvidia drivers can detect furmark, and will downscale the GPU clock to prevent the card from using too much juice and potentially damaging itself. This isn't workload related but purely furmark related as that program specifically gets recognized.
I would also argue that even if programs like furmark can coax the maximum out of your card,; the result isn't really all that relevant, much like the maximum amount of GFLOPS isn't that relevant. Sure your card might have a higher theoretical peak, but more often than not this only relates to real world workloads in only a very rough way. Edit: There are no games that only let you play around with a furry ball/symbol/couple of letters.
I agree with though though, it really is simple to run, but even that simpleness has its downsides, as in 30 seconds your card probably won't really heat up all that much and as a result might give different results compared to a longer run of some other(or the same) benchmark.
Heaven, like 3dmark also has a default preset anyone can run(I think for free aswell?) It wont take nearly as long however.
I'm also open to other suggestions, but I stand by my advice against furmark
bumpin this thread
Would you take a submission from a new member?
I've become fascinated by the subject of power efficiency, especially with gaming PCs, and felt I needed to make an account and join the discussion.
I'm in agreement with Mathragh about using furmark. It is definitely very easy to test, but I feel it isn't a good indication of typical load power draw.
This is with my PC configured as I use it daily (I've done a fair bit of undervolting on it). Peak power draws recorded during wPrime and 3DMark11P.
NateDawg | SeaSonic SSP-450RT | i5 3350p @ 3.3GHz + 660 OEM | 48w idle | 74.4w | P6097 141W | 43.24 points per watt | Daily use
I do not have a wall meter but I can disconnect everything and let only the computer powered on and measure its consumption by getting data from my electricity meter.
May I do it that way? If yes, I'll be posting my results tomorrow (too late right now).
That wont be accurate at all.
As for a V2 of this thread, Cinebench looks like a good replacement to Wprime - gives a 'points' score, free to download, heavily multi threaded.
So as the new template based on Cinebench for CPU and GPU - if it works for crossfire and SLI, it'd make a very convenient all in one tester here.
CPU + GPU
PPW (Cinebench CPU / W)
PPF GPU (FPS / W)
i7 2600 @ 4.23GHz + 7970 3GB
3.88CB pw (696CB - 179W)
0.30 PPF (70.22FPS / 227W)
New system, new test!
Copy this example
User | PSU | CPU + GPU | Windows idle | PPW (Cinebench CPU / W) | PPF GPU (FPS / W) | Comment
and ofc, we can show screenies for verification, but this is mostly based on trust.
Let me know if theres any issues with using this for a GPU test, if not i'll update the OP with the new setup and spoiler tag the old ones.
Sewje | Ocz 750 | Pentium 3258 3200 @ 4.2GHz + 970gtx | 52w idle | 80w | P8532 225W | 38.78 points per watt | Some serious cpu bottlenecking.
Curious, What's your Fire Strike score?
Aquinus | Seasonic 1000W Platinum | Intel i7 3820 @ 4.2Ghz & 2xAMD Radeon HD 6870s @ stock | 195 watts | 705 cb / 320 watts = 2.20 cb/Watt | 97.18 FPS / 325 watts max (305 avg) = 0.299 FPS/Watt | CFX does not work in Cinebench and the OpenGL benchmark doesn't fully load even just one of my 6870s very well, forget two if it even worked.
Don't you find it the least bit strange that my 6870 is getting a better frame rate than your 7970? That doesn't sound right... I was watching afterburner while it was running it so CFX definitely wasn't active.
well thats easy then, we ignore the GPU aspect of this benchmark and use another one.
recommendations? something easily downloaded (small file size!) and installed that gives a score preferably, that works with multi GPU
Separate names with a comma.