Will a 2670qm bottleneck 780m

Hi all,
I'm upgrading my machine with a 2670qm in it and was just wandering the above question.
If yes, how much of a throttle can I expect and would paying the premium of a Haswell CPU
be worth the price?
thanks in advance.

 

Not really, it's more or less comparable to i7 3610qm.

It will be fine, enjoy your new GPU.

 

Agree with Undyingghost. Have a great time with it.

 

You'll need something faster. Even an i7-4700MQ bottlenecks the GTX 780M in some games when AnandTech tested it.

 

Source please?

 

AnandTech | Mythlogic Pollux 1613 / Clevo P157SM Review

AnandTech | Mythlogic Pollux 1613 / Clevo P157SM Review

Keep in mind the 2670QM is quite a bit slower than the CPU's that were tested.

 

That's very sad to read. The CPUs always have a lot of leftover power in those games. Skyrim likes faster cores, but people stream skyrim for a living and Skyrim leaves over more than enough CPU power for them. I rather they set the ability to use low-high CPU power as a setting in those types of games and let you decide if it takes up 80% or something of your CPU (if it can). Most of these games don't even crack 40% of an i7-950, regardless of loving higher clock speeds >_<

 

The list of games that your CPU will bottleneck isn't too extensive.
RTS games, Farcry 3, Crysis 3, Natural Selection 2, Sleeping Dogs are the heavy hitters that spring to mind.
I'd personally place priority upgrading to a 780m first before the CPU. Or I'd get a 680m and use the change to get a 2xxx XM which are pretty good value now.

 

Thinking I might stick with the 2670qm for now, wait a year then go for one of the xm's further down the line. Im thinking of starting a build soon anyway so doubt this will be my main gaming rig for too long anywho... To be honest I was mainly concerned about being able to play the new GTA whenever ita released on its highest configs by so I think the 780m should see me through for another while..

 

Sure. Well if GTA4 is anything to go by, GTA may be more CPU dependant. I can't wait for this game too. We have a lot of time before it comes to PC though

 

BF3 and Skyrim are also very CPU-dependent and your 2670QM will not be enough.

 

It will be enough to push decent framerate. Some really CPU heavy games will show some slowdowns compered to i7 4700qm but it's far from bottleneck.

 

Ive played both on it, skyrims a awfully tempremental game (I've seen it run better a 2720, 6970 than on mine) but bf3 i had running without many issues somehow.Would a 3920xm be worth the upgrade? Would it be more economical than say a 4700qm?

 

Sorry let me be more clear. My question is literally just; would a 3920xm be good enough for these games and do you envision it be good enough, in general for the next couple of years?
I've been receiving advice on a separate thread (from octiceps included) regarding my laptop and I'm just trying to work out the beat course of action for me to take.
Thanks again.

 

Yes it will be, especially if you overclock it.

 

If a 3920xm overclocked isn't enough then I'm afraid we are all out of luck in the world of laptops.

 

There are always the mobile workstations/small suitcases from Clevo with desktop Intel hexa-cores.

 

Lol true. For the truly insane

 

Playing BF3 or modded Skyrim uses 30-40% of my 2670qm, CPU is never the bottleneck and u dont have to worry about it, unless the game is poorly designed of course or some issues with configs or drivers.

 

I've seen BF3 use up to 80% of my i7-3630QM, which is supposedly 25% faster than the 2670QM, so not sure what's going on there. In general, overall CPU usage is 30%-60%, dependent on the map and number of players of course.

Keep in mind that the GTX 780M is more than twice as fast as the 6990M, so it is much more likely to be CPU-limited in certain games.

 

Perhaps he was playing at a lower resolution? From memory I think I played battlefield a step down from 1080p on my 6990.

 

Yes, with the 780M you will experience some serious under utilization. The CPU and GPU work together, if one is much more powerful than the other, you will not have a good experience. If you had the 2820QM, I may not be so certain.

Upgrading to the 3920XM would be worthwhile.

 

Depends on the game, or you can get a 1440p/1600p monitor to compensate.

 

30-40% any map, perhaps 50% spikes, pretty much all ultra no AA 1080. This in AIDA which keeps a graph of utilization. GPU is always the bottleneck and this we can see with test after test like in youtube, a cheap AMD FX-8350 or a Intel dual core doing just as good as a 1000 dollars i7 even with a fast GPU. You would need a TOP desktop GPU or SLI of good GPUs to start to see some bottlenecking. And in the case of notebooks i would say u need at least a SLI 780m to bottleneck.
And either way, a basic 4700MQ is just about 20% better than the 2670QM so it doesnt worth it and the better models are really expensive. Just wait until you can test the 780m with your 2670QM and if it do throtle which i doubt perhaphs a OC could be enough.

 

I think it's pretty common knowledge that to run 64-player BF3 smoothly without major FPS drops (<30 FPS) one needs at least a quad-core, and not just an anemic one at that but a recent-generation Intel or AMD one. BF3 64-player multiplayer is one of the most if not the most CPU-heavy game on the market. If you're referring to benchmarks showing little to no difference between an i3 and an i7 Extreme Edition, well that's single player. It doesn't care whether you have a $100 or $1000 CPU but multiplayer is a completely different story. Even I get CPU-limited at times, especially on B2K maps with their enhanced destruction, and I sure as hell don't have a high-end GPU.

Not sure why you're referring to YouTube videos to prove your point since smoke and mirrors are employed all the time. Perhaps you can disable two of your cores, play some 64-player matches on B2K maps, and then get back to us?

The 4700MQ is 30-35% faster than the 2670QM, not just 20%. It's already got a 15% higher clock speed not to mention being 15-20% faster clock-for-clock due to two generations of IPC improvements. Also, the 2670QM is not overclockable at all but the 4700MQ has two unlocked turbo bins.

 

Haswell is mainly UP TO 13% higher in IPC compared to Sandy Bridge.
Real world tests revealed that Ivy Bridge was actually closer to 3-5% IPC increase compared to SB, and HW saw mainly 5-8% increase compared to IB.

The difference in clock speeds between 2670QM and 4700qm base and turbo clocks is actually 9.1% and 9.7% (2.2Ghz vs 2.4Ghz for base clocks, and 3.1Ghz vs 3.4Ghz).

Taking all those factors into account, the 4700qm is roughly 22.7% faster than 2670qm all round - not 30-35%

Are people intentionally coming up with larger numbers for Intel simply because it dominates the market, or what?

I don't think the 2670QM would bottleneck the 780M. Sure, the 4700QM would feed the gpu by 22.7% faster, but its not a big difference that would be 'earth shattering'.

 

I said any cheap CPU but from recent generations of course, common knowledge or not i do run it using 30-40% on multiplayer. I mentioned youtube cuz you can look up for tests and benches, like LinusTechTips himself have this kinda of tests.
And finally, as Deks agreed with me, the diference 2670QM to 4700QM is about 20% as you can see at notebookcheck.
I saw no game using over 40-50% here

 

I'm not making these numbers up, and I'm not an Intel fanboy.

2670QM runs at 2.8 GHz and 4700MQ runs at 3.2 GHz. Those are their true Turbo speeds when all cores are loaded, such as in games. It's a 400 MHz or around 15% increase.

Also, let's look at these results:

3DMark06 CPU: 6882 vs. 5400.2 - 27% increase
Cinebench R10 32-bit single-threaded: 4847 vs. 3899.5 - 24% increase
Cinebench R10 32-bit multi-threaded: 18894 vs. 14842.5 - 27% increase
Cinebench R11.5 64-bit: 6.89 vs. 5.19 - 33% increase

Average: 28% increase

My 30-35% figure was a bit generous, I admit. Frankly, I was expecting the speed difference between 4700MQ and my 3630QM to be more than it actuallly was, but in reality it's only like 5% faster. Haswell sure is disappointing.

But still, an almost 30% increase in CPU performance going from a 2670QM to a 4700MQ is not insignificant when we're discussing games that will be primarily CPU-limited with a GTX 780M. Games many people play such as PS2, BF3, Skyrim, StarCraft II, Grid 2, and others. Plus 4700MQ is overclockable by 200 MHz whereas 2670QM is not, which will allow it to pull away even further.

Frankly, if you look at the AnandTech results I linked to in my previous post and see the supposedly 30% slower 680M beating the 780M by as much 10-15% just by virtue of having a stronger Ivy Bridge CPU vs. a weaker Haswell one, why would you spend hundreds more on the 780M when it can't reach its true potential?

 

These numbers can be a mistake due to Hyper-threading and Intel TurboBoost. With hyperthreading when game utilizes only 4 threads you will never see anything above 50%. In fact, even if it does that numbers you are looking at still are not quiet true.
Plus TurboBoost can increase 2-cores shutting down other 2 (couple miliseconds per second) but that still will make influence on those utilization numbers when it works at 100% while monitor will believe it actually slows down to 25%.
Really, disable 2 cores and look at the difference. or disable 1 core.

 

I'm gonna repeat myself: Most of those YouTube videos you refer to are benching single player. The one Linus did on NCIX Tech Tips was no different and I even watched the video on his channel of the exact single player sequence they used for the timedemo. BF3 singleplayer doesn't care what CPU you have but a dual-core will get slaughtered in a huge, chaotic 64-player multiplayer match with tons of destruction and things exploding left and right. Overall CPU usage won't show you bottlenecks; you have to monitor individual core and thread usage. All it takes is one core or thread being maxed out to introduce a CPU bottleneck. This is why I am extremely CPU-bound in PlanetSide 2 Zergs despite the fact that overall CPU usage during such times is less than 30%. The main gameplay thread runs on only one core and it's very easy for it to get saturated during large battles.

If you still have a hard time wrapping your head around this notion I suggest you disable two CPU cores through Task Manager or MSConfig and then hop into a full BF3 server on a large open map. See if you can't feel the difference.