Is X2 GTX 670m better then 1 680m?

as title says, is two gtx 670m better then one gtx 680m?

 

No. (ten chars)

 

wahh!! how? i thought 2 would be o.o...

 

The 670m is just a re-branded 570m Fermi card whereas the 680m is a newer Kepler card based off new architecture. There has been a pretty great performance/watt increase with this new architecture.

 

Good topic by the OP.

I've always been confused about this. Isn't the primary benefit of SLI or Crossfire to maintain FPS at higher resolutions rather than being used to gain a massive increase in FPS? :hi2:

 

No not really. It can be used for both. By reducing the workload of the GPU roughly in half, it will definitely improve FPS. The 680m is a beast pure and simple, the 675mx in sli might exceed performance of the 680m by a little bit at stock speeds. But considering you can easily exceed 3DMark11 over 8000 with a simple OC it will be tough to beat.

 

Thanks for clearing that up. :thumbsup:

 

GTX 675MX SLI completely eradicates the 680M. There's no competition there. It's like comparing SLI GTX 560 Tis against a single GTX 570. There's literally no comparison. Heck, one 680M MIGHT come close via overclocking, but the MX-labeled GPUs are based on the same architecture and very very fast.

But for 670Ms... no the victory goes to the 680M. The 670M is equal to my 6970M, maybe a little quicker actually, but two in SLI would just match a single 680M if I recall from what Vantage shows. @_@ Kinda scary.

 

I would say, on paper yes, but in real life, it would depend on the game. Metro 2033 when it first came out, all ATi cards ran horrible compared to Nvidia.

 

SLI is never 2x the performance. Usually 50-60% depending on the game, sometimes a little more, sometimes a little less. So I say 675m or mx will be slighly faster than 680m, not eradicate it. Check out my benches on the NP9370 with 680M as Single and SLI. In general 40-60% performance increase was the result.

 

The top end GPUs never did scale superbly well. The GPUs just below that point generally did better as far as scaling goes.

 

I'm not sure about that. Perhaps part of it is there is a CPU bottleneck at that point. In any SLI config desktop or laptop I've seen 50-60% on average.

 

I had SLI GTX 470s back in the day and they scaled almost 100% in games I played, even Crysis. Most of it is driver related.

I'd rule out the CPU bottleneck to be completely point-blank. An i5-3570k equivalent processor shouldn't bottleneck two GTX 570s in SLI at all. (That's the desktop equivalent of what you've got there.)

 

Here's a good comparison. They don't have 570 but 580 and 580 SLI with a handful of games. Depends on the game, and more like 70-75% improvement: AnandTech - Bench - GPU12

GPU heavy games like Crysis will obviously scale better. Otherwise it's ~ 75% at best.

 

You have to remember we are talking mobile here as well. With the 675m or MX at 100w tsp per card in ski the are main board power concerns along with heat and possibly other consideration like optimization. The single GTX 680m is probably the best simple solution in most cases...........

 

The more I look at it though, the creepy clown guy is kinda right. 675MX SLI will be better than 680m by a bit. Would be curious to see if anyone has that config and by how much, although it seems the option isn't readily available for any laptop, probably because the cost isn't much less than SLI 680m's. Original question was 670M/MX in SLI better than 680m, and that's a solid no. Otherwise 675MX SLI is a moot point I think since it's not really available.