3930k bottlenecking 2x GTX Titans?

[Cacao]

Hey guys, I recently got my whole setup up and running and right out of the gate I noticed quite a bit lower performance than I expected, so I got my OSD working and I noticed that my GPU usage never went above 60% (in both GPUs). Also I should point out that in GPU benchmarks such as unigine they both went up to almost 100%. At this point everything was still running at stock clocks, so I decided to get my 24/7 CPU overclock working (got it stable at 4.6ghz on 1.4v, kind of crap for a custom loop eh? Guess I'm unlucky..) and sure enough my GPU usage jumped to 80+%, so currently I can only confirm this for BF3 at 1920x1080 at Ultra settings.

Pretty much before the overclock I would get around 70fps at around 40-60% usage and after the overclock I get a consistent 130+fps at 80+% usage. I also got a pretty descent overclock on my titans and the overclocked CPU still managed to handle it. I will test with more games and different resolutions tomorrow but I guess this means that a stock 3930k actually bottlenecks 2 titans.

It's not much of a problem since I would've been running that overclock anyway, but I still found it a bit strange, any thoughts?

[TechSage]

It will not bottleneck, trust me and everyone on this forum. There is no possible way. 3930k is almost as good as it gets.

[<boss2939>]

I blame drivers even a 3570k would not bottleneck those titans

[defunkt]

Game's wont use all of your gpu. Synthetic benchmarks just push shizz to the max.

[Nexus]

Try other games and see if you get a similar pattern.

From what you've said it does sound as though the CPU is holding you back. If it wasn't you would not have noticed such an increase in performance and GPU utilisation.

 

Do more tests and report back to us.

[I'm Batman]

it doesnt bottleneck mines

[Cacao]

Try other games and see if you get a similar pattern.

From what you've said it does sound as though the CPU is holding you back. If it wasn't you would not have noticed such an increase in performance and GPU utilisation.

 

Do more tests and report back to us.

I definitely will tomorrow. I did some research and it turns out that bottlenecking is a real factor when you're running multiple high end cards, even with extreme edition CPUs and especially at lower resolutions like 1080p.

I'll try some more games and in surround on 3 displays and I'll let you guys know tomorrow.

[WoodenMarker]

I blame drivers even a 3570k would not bottleneck those titans

Nor would a 2500k.

[TechSage]

I blame drivers even a 3570k would not bottleneck those titans

Nor would a 2500k.

They wouldn't bottleneck because an i7 is an i5 with hyper threading but core vs core it should be the same so none of them should bottleneck. I agree with you sir.

[Cacao]

Nor would a 2500k.

 

 

Game's wont use all of your gpu. Synthetic benchmarks just push shizz to the max.

 

 

I blame drivers even a 3570k would not bottleneck those titans

 

 

It will not bottleneck, trust me and everyone on this forum. There is no possible way. 3930k is almost as good as it gets.

All of you are saying it *won't* bottleneck, which I would've agreed with a few days ago, but then how do you explain what happened? 

 

Also if you do some googling you'll find that lots of people actually have bottlenecking issues at 1080p with high end SLI cards. I think the whole "as long as you have an i5 nothing will bottleneck" is only valid up until a certain point, bottlenecking is a real problem with high end cards in SLI, particularly at 1080p.

[<boss2939>]

It has to be driver support a 2500k shouldn't bottleneck it. Think about it most games use 50-60% cpu usage unless it minecraft, but if you have seen the ares 2 vid from newegg the z77 platform actually bottlenecked it a lot even the older sandy bridge e 3820 didn't bottleneck it. Which confused me so I can blame is driver's especially since it's on 1080p have you tried 1080p+ resolutions. It still confuses me that a older architect cpu bottlenecks less than a newer ivy bridge cpu both use pcie 3.0 so drivers sry for errors typing on my HTC one with 0.40 mbit internet.

[exyia]

did you enable PCIe 3.0? It's usually on 2.0 by default and easy to miss

 

I don't care about how many people think the difference is neglible - I found a huge improvement on actual Titan SLi

 

<-- Titan SLi and 3930k owner (haven't OC'd the cpu yet)

[Cacao]

did you enable PCIe 3.0? It's usually on 2.0 by default and easy to miss

 

I don't care about how many people think the difference is neglible - I found a huge improvement on actual Titan SLi

 

<-- Titan SLi and 3930k owner (haven't OC'd the cpu yet)

I enabled PCIe 3.0 and i went from 114fps to 115fps in unigine heaven :D 

Do you have BF3? Can you do a test at 1080p at stock speeds and tell me the GPU usage and framerates?

[exyia]

I enabled PCIe 3.0 and i went from 114fps to 115fps in unigine heaven :D

Do you have BF3? Can you do a test at 1080p at stock speeds and tell me the GPU usage and framerates?

 

BF3 is how I measure my performance because it's a game I actually play for more than an hour (unlike Crysis), and I got at least a 15fps bump with much fewer dips in fps. forgot to mention I'm on 5890x1080 res

 

a single 1080p will just outright downclock Titan SLi - nowhere near enough stress to put the GPU's to work at full boost speeds. Had to game on a single 1080P while waiting for one of my monitors on an RMA, the Titans kept running at ~700MHz

[Ryan Leech]

It seems that if I were to buy 2 titans I would have about the same performance as you.

[Cacao]

BF3 is how I measure my performance because it's a game I actually play for more than an hour (unlike Crysis), and I got at least a 15fps bump with much fewer dips in fps. forgot to mention I'm on 5890x1080 res

 

a single 1080p will just outright downclock Titan SLi - nowhere near enough stress to put the GPU's to work at full boost speeds. Had to game on a single 1080P while waiting for one of my monitors on an RMA, the Titans kept running at ~700MHz

Yeah in surround my cards definitely go to work, but what confused me was that on 1 display I only got like 60-70fps and my usage didn't go above 60%, which made me think that my CPU was bottlenecking it and sure enough it fixed it when I overclocked (repeating original post).