Considering SLI? My first experience here...

[JesseLactin]

I was planning to get an excellent GPU later, once I get my watercooling under way.(Dec). I'll probably be getting a GTX 580 when they drop in price. Before then, I'll order a block (or get an EVGA HydroCopper 2 card, but they cost ~ $700 CDN). I got couple questions: how much power (getting a TX650, looking for something like HX / AX 850+) and is it worth it, performance wise (ie. good scaling)

 

[GrahamC]

Interesting read thanks

 

[SmoulderingCircuits]

Cool, it's good to hear a Cross-fire user post their experience with the latest gen of cards. I assume you've experienced no micro-stutter issues that appeared to be common on the older hardware?

Some excellent temperatures there too, that's very cool indeed. The better coolers with the twin large fans really do work well if they get a good air feed.

Cheers,

Scoob.

CrossFire only micro-stutters if you have a CPU bottleneck - probably the same for SLI

 

[Scoob]

CrossFire only micro-stutters if you have a CPU bottleneck - probably the same for SLI

Really? That would make sense. That would explain why my friends dual 4870x2 was so very very bad, he only had a Q9650 at the time, though it was at 4.2ghz or there abouts. Technically with a pair of 570's I do have a slight CPU bottleneck, but it's not much as I'm at 4.6. At stock 3.3 I'd likely notice it more...might do some tests actually as it'd be an interesting thing to demonstrate.

Cheers,

Scoob.

 

[JesseLactin]

Interesting. Deffo something I'll remember.

 

[SmoulderingCircuits]

Really? That would make sense. That would explain why my friends dual 4870x2 was so very very bad, he only had a Q9650 at the time, though it was at 4.2ghz or there abouts. Technically with a pair of 570's I do have a slight CPU bottleneck, but it's not much as I'm at 4.6. At stock 3.3 I'd likely notice it more...might do some tests actually as it'd be an interesting thing to demonstrate.

Cheers,

Scoob.

I get entire seconds or even two second micro-stutters in Crysis, though admittedly with an autoexe.cfg - this is at stock 3.2Ghz. 4.2Ghz was buttery smooth

 

[Scoob]

I get entire seconds or even two second micro-stutters in Crysis, though admittedly with an autoexe.cfg - this is at stock 3.2Ghz. 4.2Ghz was buttery smooth

Hmm, I didn't think a 2 second stutter was considered micro-stutter. Micro-stutter in my understanding is fractions of a second of stuttering. For example, in my friends case, his cards would stutter all the time making it look like the frame rate was low (i.e. <30 fps for a shooter) however the reported FPS would always be respectable and in the 60's.

Still, very interesting that your CPU boost eliminated it. I've not tried Crysis on my current build, though likely even at stock 3.3 the 2500k would likely feed the cards adequately. I might do some tests of SLI with stock CPU vs. overclocked and see if I experience anything similar to you.

Cheers,

Scoob.

 

[SmoulderingCircuits]

Hmm, I didn't think a 2 second stutter was considered micro-stutter. Micro-stutter in my understanding is fractions of a second of stuttering. For example, in my friends case, his cards would stutter all the time making it look like the frame rate was low (i.e. <30 fps for a shooter) however the reported FPS would always be respectable and in the 60's.

Still, very interesting that your CPU boost eliminated it. I've not tried Crysis on my current build, though likely even at stock 3.3 the 2500k would likely feed the cards adequately. I might do some tests of SLI with stock CPU vs. overclocked and see if I experience anything similar to you.

Cheers,

Scoob.

I had the lower fps look of a higher fps occurance for the very short time I ran 5770 crossfire. 5870 delays are huge, like I said. If I'm up early enough this weekend I'll try for 3.6 ~ 3.8Ghz again and see if the maxed out PSU will cope

 

[Lollipop]

Hmm, I didn't think a 2 second stutter was considered micro-stutter. Micro-stutter in my understanding is fractions of a second of stuttering. For example, in my friends case, his cards would stutter all the time making it look like the frame rate was low (i.e. <30 fps for a shooter) however the reported FPS would always be respectable and in the 60's.

Still, very interesting that your CPU boost eliminated it. I've not tried Crysis on my current build, though likely even at stock 3.3 the 2500k would likely feed the cards adequately. I might do some tests of SLI with stock CPU vs. overclocked and see if I experience anything similar to you.

Cheers,

Scoob.

You could try underclocking it

 

[SmoulderingCircuits]

You could try underclocking it

That can be unstable if you leave the voltage at stock and I doubt many people will have underclocked their cards to give you an idea of values to set

 

[Lollipop]

That can be unstable if you leave the voltage at stock and I doubt many people will have underclocked their cards to give you an idea of values to set

I meant the CPU

 

[SmoulderingCircuits]

I meant the CPU

Even so, could still be overvolted into instability