So i evaluated an NVIDIA reference GTX 570 as well as an EVGA 570 SC; compared them with a GTX 580 in a relatively simple, 6 game/3 synthetic benchmark performance review... I get done with it an publish it; and I take one of the 570s out an plug the 580 back in.
... And I'm sitting there looking at these two nice GPUs. Then I look at my PSU. Antec Earthwatts 650 (EA650). Now, I know that's a darn good PSU; I've had it for a while and it's powered everything from an A8V Deluxe 939 rig to a litany of AM2+/AM3 setups. I've kept it clean and dust-free on the inside, and I inspect the caps every few months. The lines/rails are pretty good on it, but not as steady as today's $100-$150 PSUs. In other words, it's likely a good PSU for something like a single HD 5830/5850/5870 or GTX 470/480/570/580... On the SLI side it would seem it could put up with maybe dual GTX 465s.
I thought "heck, I'll try these two 570s... they're supposed to be more power efficient; maybe this PSU can handle them". So I plug them in and look at the first SLI setup I've ever seen in person (though I've been a staffer/reviewer since 2003; I don't know anyone in my area that's a PC gamer; maybe I should drive up to Rochester once in a while and hang out at FrozenCPU).
Before I plugged them in, however, I de-overclocked. I figured there would be no way these cards would run in this rig with the x6 overclocked to 3.96ghz @ ~1.47 vcore.
... Smooth startup, good idle temps (42)... so I figured it's time to benchmark a bit and compare my single card results (albeit with CPU @ 3.96) with some SLI figures.
AVP Benchmark went fine:
Single card (with CPU @ 3.96 ghz):
SLI with x6 @ stock 2.8 with turbo mode on:
Resolution: 1920 x 1080
Texture Quality: 3
Shadow Quality: 3
Anisotropic Filtering: 16
SSAO: ON
Vertical Sync: OFF
DX11 Tessellation: ON
DX11 Advanced Shadows: ON
DX11 MSAA Samples: 4
Benchmark Summary:
Number of frames: 7816
Average Frame Time: 13.4ms
Average FPS: 74.5
hm, ok, it works. So then I tried Stone Giant:
Single card (with CPU @ 3.96):
SLI (CPU @ 2.8):
Min: 85
Avg: 136.0
Max: 176
... 90% scaling; nice...
So I was convinced at that point to play a game for an extended period of time. I think amongst all of my games (not really that many), Bad Company 2 seems to be the best all-around system-stresser.
So I played for 2 hours on the new Harvest map... full server, 19x10, all settings highest, HBAO on, and played with all the AA modes. It ran great; no issues all the way through.
It's reasonable to say that I have two GTX 570s running @ EVGA SC speeds, in SLI mode using my x6 system with a good (now I'd call it great) 650W PSU.
I wish I had a Kill-a-Watt. Come next review I'll have one though; I've waited too long to get one.
I'm willing to bet that I'm right at the threshold. Well, I think I'll enjoy this for a while... and then I'll get a better PSU as insurance.
... And I'm sitting there looking at these two nice GPUs. Then I look at my PSU. Antec Earthwatts 650 (EA650). Now, I know that's a darn good PSU; I've had it for a while and it's powered everything from an A8V Deluxe 939 rig to a litany of AM2+/AM3 setups. I've kept it clean and dust-free on the inside, and I inspect the caps every few months. The lines/rails are pretty good on it, but not as steady as today's $100-$150 PSUs. In other words, it's likely a good PSU for something like a single HD 5830/5850/5870 or GTX 470/480/570/580... On the SLI side it would seem it could put up with maybe dual GTX 465s.
I thought "heck, I'll try these two 570s... they're supposed to be more power efficient; maybe this PSU can handle them". So I plug them in and look at the first SLI setup I've ever seen in person (though I've been a staffer/reviewer since 2003; I don't know anyone in my area that's a PC gamer; maybe I should drive up to Rochester once in a while and hang out at FrozenCPU).
Before I plugged them in, however, I de-overclocked. I figured there would be no way these cards would run in this rig with the x6 overclocked to 3.96ghz @ ~1.47 vcore.
... Smooth startup, good idle temps (42)... so I figured it's time to benchmark a bit and compare my single card results (albeit with CPU @ 3.96) with some SLI figures.
AVP Benchmark went fine:
Single card (with CPU @ 3.96 ghz):
SLI with x6 @ stock 2.8 with turbo mode on:
Resolution: 1920 x 1080
Texture Quality: 3
Shadow Quality: 3
Anisotropic Filtering: 16
SSAO: ON
Vertical Sync: OFF
DX11 Tessellation: ON
DX11 Advanced Shadows: ON
DX11 MSAA Samples: 4
Benchmark Summary:
Number of frames: 7816
Average Frame Time: 13.4ms
Average FPS: 74.5
hm, ok, it works. So then I tried Stone Giant:
Single card (with CPU @ 3.96):
SLI (CPU @ 2.8):
Min: 85
Avg: 136.0
Max: 176
... 90% scaling; nice...
So I was convinced at that point to play a game for an extended period of time. I think amongst all of my games (not really that many), Bad Company 2 seems to be the best all-around system-stresser.
So I played for 2 hours on the new Harvest map... full server, 19x10, all settings highest, HBAO on, and played with all the AA modes. It ran great; no issues all the way through.
It's reasonable to say that I have two GTX 570s running @ EVGA SC speeds, in SLI mode using my x6 system with a good (now I'd call it great) 650W PSU.
I wish I had a Kill-a-Watt. Come next review I'll have one though; I've waited too long to get one.
I'm willing to bet that I'm right at the threshold. Well, I think I'll enjoy this for a while... and then I'll get a better PSU as insurance.