Guild Wars 2

Purely considering the HT technology on the I7 4770K and 780’s in SLI has anyone observed a performance increase?

Personally I have observed enabling SLI (780’s) on this game would actually lower over all performance and have a bigger FPS drop which is odd but not really unexpected.

For HT, I have not noticed any difference at all.

HT doesn’t help much as it only adds threads, this game uses 4 threads heavily, one of which is the rendering thread and when that is capped it doesn’t help to add any more threads, so that’s that.

As for SLI, the performance of HD 7790 is adequate and after that the CPU becomes a bottleneck, GTX 780 is way above that so SLI shouldn’t help. Also the use of multi GPUs requires some optimization and if GW2 is lacking that you could possibly see a decrease. Hard to say since I only have 1 card.

I get better performance, at least slightly better, with crossfire on my 7970s since I changed to Intel i7.

Avelos.6798:

I get better performance, at least slightly better, with crossfire on my 7970s since I changed to Intel i7.

Interesting. Could it be that CrossFire is supported better than SLI?
Also what did you upgrade from? If you had a weaker CPU and the same GPU it could be that your CPU was bottlenecking that much and now it’s released it a bit.

Yes my i7 feed my 7970s much faster than my AMD FX-8350 ever would.

SLI/CF has always added a slight loss of performance due to the higher system overhead of running 2 cards simultaneously. However thats mainly offset by the increase in graphical performance the 2 chips offers. In a game like gw2 where cpu power is at a premium you will see a decrease in frame rates from 1 to 2 780’s assuming you are not gpu capped.

You only run multi card setups becasue at 4k or eyefinity/surround resolutions, a single 780 is not fast enough and will max out before hitting an i7’s cpu cap. In this case you need to use SLI to run the higher res. However the 780/ti is not really capable at running super high resolutions as the current card manufacturers only have a 3gb version which is, depending on the game, barely sufficient for a 4k display and insufficient for a 3xQHD surround system which is what I currently am using (3x Dell U3011@2650x1600).

In SLI your vram is not cumulative, 2 cards with 3gb vram does not equal 6gb vram, In reality the vram data is mirrored across the two cards which for the 780 means that when SLI’ed you still only have 3gb. Thus at those resolutions you will max out the vram and suffer massive performance penalties for using the system ram. In these cases, a 4gb 290/x in CF or a 6gb titan/black SLI is whats i would go for, although the 4gb 770s in 2 or 3 way may also be a viable solution (I recommend against 3 way sli as a lot of games are only coded to deal with at most 2 gpus. 3 way CF/SLI scaling has been historically bad)

locx.6412:

HT doesn’t help much as it only adds threads, this game uses 4 threads heavily, one of which is the rendering thread and when that is capped it doesn’t help to add any more threads, so that’s that.

As for SLI, the performance of HD 7790 is adequate and after that the CPU becomes a bottleneck, GTX 780 is way above that so SLI shouldn’t help. Also the use of multi GPUs requires some optimization and if GW2 is lacking that you could possibly see a decrease. Hard to say since I only have 1 card.

Okay but does it add even though just a little bit more performance or none at all ? Have you done some testing ?

Pandemonium.6453:

Okay but does it add even though just a little bit more performance or none at all ? Have you done some testing ?

http://www.tomshardware.com/reviews/guild-wars-2-performance-benchmark,3268-7.html

This test is old but its the only one iv found. For sandy bridge, the pentium to i3 (only difference was HT) at same clock had a 2 fps difference.

I assume the spread would be similar or smaller going from a i5 to a (non 2011 socket) i7 with HT. In most games HT is really only effective for the i3’s, as many games can utilize 4 threads, but few use 8, which is why in many other benchmarks, the i5 is very comparable to the i7 (at same clock).

As for actual testing I’m currently running a 3750k so i can’t test HT atm.

That’s because all HT does is provide a chance to improve the overall efficiency of the core by an unknown amount. Most of the time it’s the same as a 10% overclock. Sometimes it’s more, occasionally a lot more and occasionally the impact is actually negative.

http://www.anandtech.com/show/7189/choosing-a-gaming-cpu-september-2013

That’s the most recent comprehensive test I know of between HT and not HT in gaming.

Edit: This is part 1

http://www.anandtech.com/show/6934/choosing-a-gaming-cpu-single-multigpu-at-1440p