5870; Core 2 Duo vs i5 -- A Comparison

Systems Used:

Intel Core 2 Duo E6750 @ 3.2GHZ
XFX ATi Radeon HD 5870
Abit IP35-E
XF-I Titanium Pro PCI-Express
Corsair 520HX
8GB DDR2-800
Western Digital 1TB Caviar Black
Antec 900 Case
Windows 7 Pro 64
Catalyst 9.11

Intel Core i5 750
Corsair H50 CPU Heatsink
XFX ATi Radeon HD 5870
Asus P7P55D Pro
XF-I Titanium Pro PCI-Express
Corsair 520HX
Corsair XMS3 4GB DDR3-1600 8-8-8-24
2x Western Digital 1TB Caviar Black
Antec 900 Case
Windows 7 Pro 64
Catalyst 9.11



Core 2 Duo 3.2Ghz, 8xAA 16xAF Max in game settings 1920x1080:


Average Framerate: 31.08
Max. Framerate: 51.85
Min. Framerate: 20.98

Core i5 2.67GHz 8xAA 16xAF 1920x1080:



As you can see even with the Core 2 Duo clocked 600MHz higher than the i5, the i5 setup is 40% faster in regards to minimum frame rates, 42% faster in average frame rate, and a whopping 51% faster in maximum frame rates. Lets have some fun with the i5 setup...


Core i5 4GHz NO AA 16xAF 1920x1080



At this setting the i5 is 71% faster in minimum frame rate, 40% faster in average frame rate and 67% faster in maximum frame rate.


Core i5 4Ghz 8xAA 16xAF 1920x1080:



At this setting the i5 is 40% faster in minimum frame rate, 43% faster in average frame rate and 54% faster in maximum frame rate.

Heaven Benchmark DX11​

Core 2 Duo 2.6GHz: 4xAA 16AF: - NOTE: 5870 WAS AT 950 1250 FOR THE CORE 2 DUO



Core i5 2.6GHz 4xAA 16xAF



Clearly this benchmark is totally GPU dependent.

Company of Heroes DirectX 10​

Core 2 Duo 3.2GHz 8xAA 16xAF Max in game settings 1920x1080



Core i5 2.6GHz 8xAA 16xAF Max in game settings 1920x1080




Here we see an increase of 21% in average frame rate, 13% increase in maximum frame rate, and finally a 34% increase in minimum frame rate. Despite the Core 2 Duo's 18.75% clock speed advantage.

Core i5 4.0GHz 8xAA 16xAF Max in game settings 1920x1080



Core i5 4.0Ghz NO AA 16xAF Max in game settings 1920x1080



I tried increasing the clock speed of the cpu to 4ghz and even no AA, it made no difference to the frame rate.

Resident Evil 5 DX9​

Core 2 Duo 3.2Ghz 4xSSAA 16xAF Max in game settings DX9



Core i5 4GHz 4xSSAA 16xAF Max in game settings DX9



Only reason I ran 4xSSAA in this benchmark was it was the only benchmark I had of my core2duo system. Still we see an average of 10fps increase but the real advantage here for the i5 over the core 2 is constant high frame rates.


Random Benchmark:

Thanks for your work and posting this. I'm currently rocking a e6400 and want to jump to an i5 system and keep my 8800gtx for the time being. This should give me a huge boost in performance with team fortress 2.

no problem

im installing cysis and crysis warhead now benchmarks to follow soon, i dont have core 2 duo to compare though

Good comparison bench. C2D to i5 was a more worthwhile jump than I realized.

Im curious how much of that is quad core and how much is I5 architecture improvement.

Cali3350 said:

Im curious how much of that is quad core and how much is I5 architecture improvement.

Click to expand...

Something else to consider is this: the Core i5 750 runs the cores at 3.2 GHz if the load is only on one or two processors.

http://www.tomshardware.com/reviews/intel-core-i5,2410-4.html

This means that you are REALLY comparing a 3.2 GHz i5 versus a 3.2 GHz Core2 Duo. Thus, your "impressive" improvements are not so amazing. But the processor still does kick the crap out of Core2.

And for those games that can actually stress 3 or more cores, the CPU runs at 2.8 GHz...but this doesn't matter, since the extra CPUs more than make up for it.

I do appreciate the comparisons, however - it is nice to know that I should consider upgrading my Core2 Duo.

defaultluser said:

Something else to consider is this: the Core i5 750 runs the cores at 3.2 GHz if the load is only on one or two processors.

http://www.tomshardware.com/reviews/intel-core-i5,2410-4.html

This means that you are REALLY comparing a 3.2 GHz i5 versus a 3.2 GHz Core2 Duo. Thus, your "impressive" improvements are not so amazing. But the processor still does kick the crap out of Core2.

And for those games that can actually stress 3 or more cores, the CPU runs at 2.8 GHz...but this doesn't matter, since the extra CPUs more than make up for it.

I do appreciate the comparisons, however - it is nice to know that I should consider upgrading my Core2 Duo.

Click to expand...

it wont run the cores at 3.2ghz if turbo boost is disabled

pandora's box said:

it wont run the cores at 3.2ghz if turbo boost is disabled

Click to expand...

But DID YOU? And did you verify it with CPU-Z?

defaultluser said:

But DID YOU? And did you verify it with CPU-Z?

Click to expand...

yes and yes

pandora's box said:

yes and yes

Click to expand...

Cool then, here's an additional request: can you re-run some of these benchmarks with half the i5's cores disabled?

Here's how you do it:

Start->Run-> msconfig (you can also just type it into search on Vista/7)

Boot tab, Advanced Options, enable the "number of processors" check box and select 2 processors and confirm. Restart, and you should only have two cores visible to software.
(you can undo this any time by un-checking the box and restarting)


I'm really curious how many of these benchmarks utilize more than two cores, and how many are just seeing that much more performance from the i5's increase in IPC

sure, i'll do far cry 2 and crysis when its finished downloading the patch

Nice of you to run those comparisons. However, I wonder how much those farcry improvements were due to better memory bandwidth. ddr3 1600 vs ddr2 800? The i5 is one hell of a cpu, and will probably be my next upgrade, but there is still alot of life left in these core duo's.

[email protected]
ddr2 1150
asus 5850

highest in game settings on the ranch small test. couldnt figure out how to do that line graph you got, sorry.

http://img413.imageshack.us/img413/3641/farcry2coreduo6850378xa.jpg

very cool. for anyone wondering how much you're leaving on the card with a 5870, 4870x2, or gtx295 level product. same goes for users on gtx260 sli/4870 cf or better setups. as to the question of memory bandwidth, ddr800 with 4-4-4-12 timings roughly equivalent to ddr1600 with 8-8-8-20 timings.

Hey thanks alot. I knew that the Core i5/i7 series was felt faster than the C2D and C2Q.

Core i5 4GHz All 4 Cores 1920x1080 0xAA 16xAF Very High 64bit Dx10
Average FPS: 38.04, Min FPS: 19.09, Max FPS: 44.44

Core i5 4GHz 2 Cores Disabled 1920x1080 0xAA 16xAF Very High 64bit Dx10
Average FPS: 35.12, Min FPS: 19.33, Max FPS: 43.73

Far Cry 2 with 2 Cores Disabled, CPU at 2.6GHz 1920x1080 8xAA 16xAF Max in game settings DX10:



Company of Heroes with 2 Cores Disabled, CPU at 2.6GHz 1920x1080 8xAA 16xAF Max in game settings DX10:

Very nice post. How are you doing those graphs? I know myself with an i7 my framerates got a nice bump from a Q6600, BUT in certain cases only with HT off. I don't have the Q6600 anymore to test but I could throw some HT on/off benches in here for people thinking of moving to an i7.

yea quadcore is great, been enjoying my i7 for a year and it just rocks.

thx for the benches.

Throw in some unrar and par checking benchies and you will have my full attention.

Damn you. I was going to buy a second s775 board in 2 weeks, (I have 2 C2D CPUs. 1 active.) now I just want to put the money into an I7 set up and sell the e6850. (I'd give my friend my q9550/8800GTX, though I was going to give him the e6850/8800GTX.)

Thanks for taking the time to do this, I have some stuff to think about now. *shakes fist* /facetiousness (But seriously thanks. )

I am curious to see if the On Die PCIe controller makes a difference... the i5 has them, and the new i7's..

I guess It would be neat to see a comparison between the old and new i7's.

But w/e, thanks for the great thread.

I just saw this article over at tomshardware (i know i know) regarding video card vs cpu performance:

http://www.tomshardware.com/reviews/build-balanced-platform,2469.html

It doesnt have the 5870 in the tests but it does test the gtx295 and 4870x2 so you can have some idea where the 5870 will land.

Thanks for that. It settled my mind, a bit.

For now I think the P45 chipset mobo/RAM upgrade should do me for awhile before moving to the I7. ($400+ vs $1000+)

I have a E6600 + 5970 and while its a vast improvement to my 260 its being held back by the cpu of course.

But the nice thing is Valve split up gpu vs cpu effects (shader details vs effect details) So Left 4 Dead 2 averages 60 fps at 2560x1600 No AA 4xAF Everything on high except effect details is low.

And Shattered Horizon runs fine with 2560x1600 No AA 4xAF max everything. Although the in game fps counter says 40 fps, it feels smooth

pandora's box said:

Core i5 4GHz All 4 Cores 1920x1080 0xAA 16xAF Very High 64bit Dx10
Average FPS: 38.04, Min FPS: 19.09, Max FPS: 44.44

Core i5 4GHz 2 Cores Disabled 1920x1080 0xAA 16xAF Very High 64bit Dx10
Average FPS: 35.12, Min FPS: 19.33, Max FPS: 43.73

Far Cry 2 with 2 Cores Disabled, CPU at 2.6GHz 1920x1080 8xAA 16xAF Max in game settings DX10:

Company of Heroes with 2 Cores Disabled, CPU at 2.6GHz 1920x1080 8xAA 16xAF Max in game settings DX10:

Click to expand...

Hey, would you look at that! The quad cores, they do nothing!

I originally thought that Intel was nuts to push Clarksdale as their first 32nm processor, thinking that quad cores should get the go-ahead. But the fact is most consumer-level processing loads are STILL dual-thread or less, and with that new process, can you say 3.5-4.0 GHz dual-core Nahalem?

Next year will mark the resurgence of the dual core

No, its the software that's not ready for the 4 cores. Keep pushing hardware, and software will catch up. Don't go back and gimp the hardware

Pyrolistical said:

No, its the software that's not ready for the 4 cores. Keep pushing hardware, and software will catch up. Don't go back and gimp the hardware

Click to expand...

You'll still have the choice of purchasing an i5 instead of the i3 - for a lot of professional users, the performance difference is staggering. I was simply stating that gaming still benefits more from two fast cores than four slower cores, and probably will until the next round of consoles comes along.

I'm glad for the 32nm dual-core because right now there's a distinct lack of choice: you either get 4 cores, or 4 cores with 8 threads. Clarksdale will let everyone choose exactly how much performance they need.

defaultluser said:

You'll still have the choice of purchasing an i5 instead of the i3 - for a lot of professional users, the performance difference is staggering. I was simply stating that gaming still benefits more from two fast cores than four slower cores, and probably will until the next round of consoles comes along.

I'm glad for the 32nm dual-core because right now there's a distinct lack of choice: you either get 4 cores, or 4 cores with 8 threads. Clarksdale will let everyone choose exactly how much performance they need.

Click to expand...

but 4 cores for $150 is hard to pass up. i dont see the point in clarksdale beyond 32nm. granted not everyone has a microcenter nearby

pandora's box said:

but 4 cores for $150 is hard to pass up. i dont see the point in clarksdale beyond 32nm. granted not everyone has a microcenter nearby

Click to expand...

Agreed, 2 cores is on the way out on the desktop, but it's definitely not dead yet. The smaller die means the price point can continue to fall well below the $100 mark while still making a profit, which is GREAT news for people building low-end to mid-range gaming systems. Quad core i5/i7 will not be doing that anytime soon, considering that the $150 price you mentioned is a closeout for clearing stock (lowest-end i5 is $200 street).

Clarksdale will probably be the last dual-core desktop part from Intel, but it will not be the last dual-core: low-power portables won't be ready to move up to quad core for another couple process generations, as it just uses too much power.

Good thread...useful, lad, useful.

defaultluser said:

Clarksdale will probably be the last dual-core desktop part from Intel, but it will not be the last dual-core: low-power portables won't be ready to move up to quad core for another couple process generations, as it just uses too much power.

Click to expand...

Last dual core? Intel still makes a few single core models.

I had a similar experience moving from e8400 to i7 920. My numbers in Far Cry 2 more than doubled I believe, and people tried to say I was flame or doing something wrong. FC2 was the one game with a truly dramatic improvement though...the other gains were much less.

Thanks for the effort put in, a nice balanced review.

It's a bit strange to benchmark such different CPUs against each other though, a dual core vs a quad core in modern games which make use of quad cores is going to heavily favour quad core CPUs irrelevent of clock speed.

PrincessFrosty said:

Thanks for the effort put in, a nice balanced review.

It's a bit strange to benchmark such different CPUs against each other though, a dual core vs a quad core in modern games which make use of quad cores is going to heavily favour quad core CPUs irrelevent of clock speed.

Click to expand...

I mainly did it because when i was thinking about upgrading i had a hard time finding comparisons between the older (e6xx) dual core processors and the i5/i7 series, most compare the e8400 to the i5/i7 and it was usually at stock speeds too.

Someone over at rage3d pointed out that vsync needs to be forced off with a launch command for company of heroes. I figured turning it off in catalyst control center would do the trick, apparently not:


2.67GHz Core i5 8xAA 16xAF Max in game settings 1920x1080



3.8GHz Core i5 8xAA 16xAF Max in game settings 1920x1080




I have the i5 at 3.8GHz only because it took 1.4 volts to do 4ghz while it takes 1.29 volts for 3.8ghz, not worth .1 volts for 200 mhz imo.

Had a bunch of time this Thanskgiving, so I ran a bunch of benchies on my rig. Of all the games that the Op tested, I'm only seeing one where the quad core i5 has an advantage over a dual core, and that's Farcry 2. Of all the other games, it's almost the same performance as a dualie.
Kind of Interesting to see what an upcoming Clarksdale @5ghz will do. If anyone has more benchies of different games, please post.

All benchmarks ran at 1920*1080 with all available in game settings on max. AA settings are set in game with the exception of Batman (set using CCC).

Some images had to be cropped so I could upload them easier

Here's my rig: http://img209.imageshack.us/img209/6695/systemv.jpg

Farcry 2 8xAA 16xAF

http://img413.imageshack.us/img413/3641/farcry2coreduo6850378xa.jpg

Company of Heroes DX9 8xAdaptive AA 16xAF (couldnt figure out how to run it in DX10 so maybe Op can do a DX9 run for comparison?)

http://img267.imageshack.us/img267/8940/cohw.jpg

Company of Heroes DX9 8x SuperS AA 16xAF

http://img260.imageshack.us/img260/8360/coh.jpg

Resident Evil 5 DX9 4xSuper SampleAA 16xAF

http://img121.imageshack.us/img121/7158/re5dx920091126163200444.png

Street Fighter 8xAA 16XAF

http://img204.imageshack.us/img204/71/sf4i.jpg

Batman 4xAA 16xAF No Physics

http://img260.imageshack.us/img260/8654/batmany.jpg

World in Conflict 4xAA 16xAF

http://img338.imageshack.us/img338/6303/wicg.jpg

Hawx DX 10.1 8xAA 16x AF

http://img301.imageshack.us/img301/4379/hawxu.jpg

Oscar Meyer said:

Had a bunch of time this Thanskgiving, so I ran a bunch of benchies on my rig. Of all the games that the Op tested, I'm only seeing one where the quad core i5 has an advantage over a dual core, and that's Farcry 2. Of all the other games, it's almost the same performance as a dualie.
Kind of Interesting to see what an upcoming Clarksdale @5ghz will do. If anyone has more benchies of different games, please post.

Click to expand...

Far Cry 2 is the only one of those games that is CPU limited. Then you went and made the GPU bottlenecking worse by throwing on tons of AA.

JCDenton said:

Far Cry 2 is the only one of those games that is CPU limited. Then you went and made the GPU bottlenecking worse by throwing on tons of AA.

Click to expand...

Huh? Isnt that the whole point of buying these high end graphics card..to play with increased eyecandy? So you're suggesting that we should run games with no AA? Anyway, the whole point of my post is there are only a handful of games that will run much slower with a dualie rather than a quad. There are still alot of folks on dual core platforms such as myself. And we're wondering if it is a necessary to upgrade platforms if buying a 5800 or 5900 series card. Or if one can get by without upgrading and wait for the next best thing. Even with a three year old dual core, it is more than sufficient for 90% of games if you get one of these new vid cards. And as for the more cores vs more clocks, it will be interesting to see how the upcoming clarksdale will perform.
If a 5ghz overclocked Clarksdale pushes more frames than a 4ghz overclocked i5 in, lets say, Crysis 2 or Rage...I see Clarksdale as a viable upgrade.

in company of heroes to get dx10 just put all the sliders in the graphics settings to max, i believe its the shader detail that puts the game into dx10 mode. and the only reason your getting higher frames than me is because you overclocked your video card and your using adaptive aa which is easier on the video card than regular aa.

pandora's box said:

in company of heroes to get dx10 just put all the sliders in the graphics settings to max, i believe its the shader detail that puts the game into dx10 mode. and the only reason your getting higher frames than me is because you overclocked your video card and your using adaptive aa which is easier on the video card than regular aa.

Click to expand...

If I'm not mistaken, MultiSampleAA > Adaptive AA > Super Sample AA in terms of fps/performance. And as for the COH benchmarks, I think I am getting faster fps because I am running in DX 9. Looked around the net and it seems you have to patch the game up to v1.7 to enable dx10 effects. An another thing is despite my vid card overclock, I am running a 5850 and you are running a 5870.