 |
|||||||||||
|
|
|
|||||
 The AMD Phenom II X4 965 saw a 13% decrease in the average frame rate when disabling two of the four cores. Performance dropped by 71fps to just 62fps when testing with Battlefield Bad Company 2, while the minimum frame rate also dropped by 19%. Then, when disabling another core, turning the Phenom II X4 965 into a single core 3.4GHz processor, performance was slashed by another 40% with the average frame rate reaching just 37fps.
 The Intel Core 2 Quad Q6600 processor saw an even greater reduction in performance when going from four cores to just two. The dual-core configuration was 23% slower when testing with Battlefield Bad Company 2 when measuring average frame rate performance, while the minimum frame rate was 26% lower. Finally, when going from two cores to just one, the performance plummeted by 47% to an average of just 28fps, while the minimum frame rate of 20fps was less than half that of the dual-core configuration.
 Interestingly when testing with the Intel Core i7 920 processor we only saw a 3% reduction in the average frame rate when going from four cores to just two cores. Even the minimum frame rate was only reduced by a 7% margin, dropping from 58fps to 54fps. The Core i7 920 did suffer with just one core as the average performance dropped to 41fps, making it 39% slower than the dual-core configuration.
 Similar margins were seen when testing with the Core i5 750 processor when compared to the Core i7 920, please remember the 920 does have Hyper-Threading disabled. Again the dual-core and quad-core configurations delivered similar performance, while the Core i5 750 really only suffers in Battlefield Bad Company 2 with just one active core.
 Far Cry 2 is thread optimized and as a result we see a 27% performance decrease on the Phenom II X4 965 when going from four cores to just two. Then, when going from two cores to just a single core, performance is reduced by a 42% margin with the average frame rate dropping from 33fps to just 19fps.
 The Core 2 Quad Q6600 saw a similar performance loss, as the 42fps seen with all four cores enabled was reduced by a 36% performance margin, to just 27fps with two cores enabled. Finally the single core configuration was 48% slower when compared to the Core 2 Quad Q6600 working with two cores enabled.
 This time the Core i7 920 did become considerably slower with the 3rd and 4th cores disabled, as the dual-core configuration was 40% slower with an average of just 37fps. Performance was then decreased by another 46% margin when going from two cores to just a single core.
 The performance margins where again much the same for the Core i5 750 processor as they were for the Core i7 920 processor. |
|||||
|
|
Chootia |
Posted on: 07/21/2010 10:06 PM
Thanks. I upgraded to the Core i5 750 for Bad Company 2 with a Radeon HD 5850 and could not be happier. My old Core 2 Duo system with the Radeon HD 4870 was choppy as hell in this game. Seems like it was more the CPU! |
|
PFULMTL |
Posted on: 07/22/2010 01:37 AM
Aww, I wanted to see my CPU in there, AMD AthlonII X4 620 compared to PhenomII X4, since there is only one AMD on the list here. Was pretty surprised on the benchies comparing the i7 and i5. I thought the i7 would have much more FPS. |
|
ProX |
Posted on: 07/22/2010 03:26 AM
@ PFULMTL - with HyperThreading disabled the Core i7 920 and Core i5 750 are the same processor really  If you want to know how the Athlon and Phenom compare just ready a CPU review. |
|
Ash |
Posted on: 07/22/2010 05:32 AM
@ ProX - not quite: different turbo speeds, different memory bandwidth, different PCIe lanes potential. |
|
archange Posts: 9 Joined: 2010-07-22 |
Posted on: 07/22/2010 07:05 AM
@ Ash: all of which prove to have little effect in real-life gaming and in the context of this testing. So I agree with ProX. @ Steve Walton: 1. Why on Earth would you choose to disable HT on the i7?! 2. I agree, disabling cores on the quads still yields better performance than using "real" dual cores; I suspect cache size has quite an impact here. I would have liked to see some real dual core data, especially when nobody in their right mind would buy and then disable cores on purpose. 3. I found the picture in the article title misleading: The two Zalman coolers led me to believe that you were going to test with CPU cores in excess of 8. In another context, great job Steve; I like coming back to Legion. It allows me to get a "down to Earth" feeling again. |
|
Steve Posts: 80 Joined: 2010-02-08 |
Posted on: 07/22/2010 07:19 AM
Thanks for the feedback guys. @ archange: The reason HT was disabled was so that we could see the impact of having just 1,2 and 4 cores. With HT we would be showing the impact of 2, 4, 8 threads and that is not what we were trying to do. As for real dual-core performance that was shown in the article. A Core 2 Quad processor is two Core 2 Duo's stuck together. Furthermore the Phenom II X4 and Phenom II X2 are also the same processor so again you are seeing a real dual-core here. Anyway thank you for the feedback. |
|
pcgamer |
Posted on: 07/22/2010 08:50 PM
good article, i'm glad i've bought q6600 3 years ago (costed only 25% over e6600) despite almost everyone in that time wrote in benchmarks that it's better to buy dualcore, but i knew better, now i dont have to change my cpu in next 2-3 years, only changed gpu, now i'm laughing at my friends that argued about my bad decision, yes - i'm cruel XD, just checked some benchmarks at pcgameshardware, all of those games gets more fps when going 2to4cores and most of them plays better with q6600@2.4ghz than e8400@3.0ghz (and more cache): Dirt 2, Dragon Age Origins, Left 4 Dead 1&2, GTA4, Anno 1404, Ghostbusters, ArmA 2, Prototype and probably some more... |
|
archange Posts: 9 Joined: 2010-07-22 |
Posted on: 07/23/2010 05:31 AM
@ Steve: I see your reasoning, but I still would have liked to see regular i7 benchmark results next to the other graphs - given that I own a 920. With that said, I suppose the results can't be more than a few percent off in either direction. And yes, all we want is to make your life more complicated  Looking forward to the next review. |
|
Offordef |
Posted on: 07/23/2010 03:28 PM
Hi Steve, Good article as we become to expect from you,you must be busy swapping hardware. It would have been interesting to see the resulst with SLI/Crossfire to remove the GPU bottleneck however cannot have it all (and I cannot afford it anyway) Guess that Quad Cores will be sufficent until the new console generation arrives. |
|
Razor |
Posted on: 07/23/2010 08:23 PM
Hey nice article. That was a very interesting read. Anyone building a new PC for gaming should get a quad-core anyway but of course not everyone does |
|
Dakaan |
Posted on: 07/25/2010 06:47 PM
Hey. I did my own testing (work in a hardware store ;-) before building my new machine. And came to the same conclusion you did. the i5 750 and even the i7 930 performs basically equal in a 4core vs 4core situation. I wound up buying the i5 platform after i found out that gaming performance on the i7 actually decreased after activating hyperthreading. Read up and the issue seems to be that cache problems slow the i7 down a little with HT enabled. HT only seems to be a speed benenit when you are truly using 6 or more threads heavily. |
|
Effran |
Posted on: 07/27/2010 02:33 PM
Very nice work there. My next upgrade will be to a quad-core for sure! |
|
Lex Luger |
Posted on: 08/02/2010 05:15 PM
Great article, lots of good games choices for sure, but still, I would of loved to see more quad core game engines tested. -Grand Theft Auto 4 -Empire or Napoleon Total War -Assassins Creed 2 -The Last Remnant, (uses more cpu than any other UE 3 game) -Resident Evil 5 -Mlb or NBA 2k9 or 10 (jaw dropping graphics) -ARMA II (extreme graphics) -Starcraft 2 Elder Scrolls 4 or Fallout 3 (oblivion was first quad core game) |
|
ProX |
Posted on: 08/03/2010 05:00 AM
Strange picks there Lex Luger. StarCraft II for example is hopeless, uses just 2 cores. NBA 2k9 or 10 use only two cores and look very average in my opinion. Resident Evil 5 is old but it does use more cores but a single core can play it fine. The rest are okay but I think we got the idea. |
|
PCMechanik |
Posted on: 08/11/2010 03:05 AM
All this info is great, but with a Dual Core I 5 670, with hyper-threading, isn't it just as good as a quad core? Has anyone tested an I 5 670 against an I 5 750? |
|
ProX |
Posted on: 08/11/2010 12:58 PM
@ PCMechanik - really not sure what that has to do with anything? At least to do with an article that explored the impact 1,2 and 4 cores had on performance. If you want to count the Core i5 670 as a quad core be my guest. That said HT is not nearly efficient as dedicated cores, so the Core i5 750 will be faster in games that utilize all four cores. You also have to question who would spend $300 US on a dual-core processor with Hyper-Threading? Any gamer that does when the Core i5 750 costs $200 US probably needs to have their marbles checked. The Core i5 670 also has half the L3 cache of the Core i5 750, OUCH!!! |
|
travbrad |
Posted on: 11/01/2010 04:58 AM
I've been looking all over for something like this. It's greatly appreciated. I think my E8400 will last me a bit longer with some OCing  |
|
Cheewie |
Posted on: 12/24/2010 04:57 AM
Thanks! Great test |
|
DAG |
Posted on: 08/31/2012 03:52 PM
This is a great article regarding an interesting point - do cores matter and how much. However, I am perhaps more interested in the real world decision - what is the gain based on cpu variants, but considering everything that is included in that purchase (e.g. the extra cores + hypertheading, etc.) Moreover, if the extra bells and whistles of a i7 are disabled, it really doesn't help make a decision on which cpu I should buy. While this comment is very late, perhaps the new articles will also include a comparison of level playing field of processors only, AND normal operating experience with the new cpus. Just a thought. |