Impressions of GTX660 single and in SLI

[John Tauwhare]

^^^ Nice testing John. Looks like we got the next IDC over here

Lol, er, let’s not get carried away. But a pale and feeble imitation. I am not worthy…

The numbers on the X axis of the graph the CPU speed in GHZ right?

Yes, chart now updated with axis labels.

I hate to say this, but Heaven is not the right benchmark to test CPU bottlenecks. It is among the least CPU-dependent benchmarks available. I'd suggest you try a 3dMark11, or perhaps a popular in-game benchmark.

Yep, as you say and the chart shows. BUT in several posts people blame a low score in Heaven on their CPU. In fact it doesn’t take a lot of CPU to run it full speed and a Vishera would not be disadvantaged against a 2700K. Between 3.0 and 5.4 I get only +10 points. In testing on a SINGLE GTX 660 there is only a 2% drop off from 4.8 (696) to 1.6 (682).

If I get time I’ll run 3dMark11.

 

[willomz]

On certain games the architecture can make a bigger difference than the clockspeed.

e.g. on Hitman
A 3770k @ 2.5ghz is equal to a 8350 @ 4ghz
The difference between 4.5ghz and 3ghz on a 3770k is only 3fps.

Remind me, why do we bother to overclock again?
An extra 1ghz gives less than a 2% gain :/

 

[John Tauwhare]

I hate to say this, but Heaven is not the right benchmark to test CPU bottlenecks. I'd suggest you try a 3dMark11, or perhaps a popular in-game benchmark.

CPU dependent:



Best run at 5.0GHz:



Crashed at 5.4 and 5.2 but I'm sure I can squeeze a little more.

 

[guskline]

John thanks for your work. Don't destroy your cpu or gpus just for a few extra points!

 

[John Tauwhare]

John thanks for your work. Don't destroy your cpu or gpus just for a few extra points!

Gotta push the envelope..nothing ventured nothing gained!



At 50x100, 1319/1650.