Duel of Dualcores: Wolfdale vs. Phenom X2 vs. Conroe

The German hardware magazine PC Games Hardware compared the 3 CPU types with single player benchmarks of Crysis. The E8400 which comes with 333 x 9 and 3 GHz was downclocked to 266 x 9, so it ran with the same clockspeed than the E6600 and the Phenom X4 9600 which had 2 cores disabled. The pictures say 9700, but I think it was a 9600 which is what CPU-Z reports too.

The text is again in German, but the pictures speak for themselves I guess.

Source: http://www.pcgameshardware.de/?article_id=622892

Link to the pictures ("nächstes Bild" means next picture, clicking on the picture will enlarge it):
http://www.pcgameshardware.de/?menu=...=741790&page=1

Wolfdale is shaping up nicely:)

Wow... 6 to 10 FPS difference between Wolf and Phenom. That is huge :eek:

the PHENOM in cpu-z is the 9600

Nobody is perfect ;) I already stated that in my first post :)

Quote:

---

The pictures say 9700, but I think it was a 9600 which is what CPU-Z reports too

---

the fun begins when you overclock them

i believe that any chip is only worth it's max overclock no matter the model number in CPUz :D

Good review, have you tested them Oliver?.

Best Regards.

Quote:

---

Originally Posted by dinos22

the fun begins when you overclock them

---

actually the fun ends for phenom :p:

@fr3ak
nice info ftw :up:

The Phenom didn't overclock too well. 2.7GHz was 06 stable and the max I could booth with while overclocking fron the BIOS. AMD OVerdrive always crashed when I tried to oc with it, so there might be some BIOS issues limiting the overclock, but I dont think we will see much more than 3 GHz from the Phenom any time soon.

On the other hand, I don't think it will be a problem hitting 4 GHz on air with the E8400. I havent tested it yet, but I will try it out under my SS over the weekend, doing air cooled OC next week. The QX9650 I tested was able to hit 4,375 MHz on air 06 stable, so I think 4 GHz on the E8400 shouldn't be much of a problem using 24/7 voltages.

How did you disable two cores? Is this option available in BIOS or it is done somehow else?

I'm definitely getting a Penryn, that it's this big FPS difference between Conroe and Penryn at same clock speed even in Crysis, imagine what it will be like when comparing a max clocked Conroe vs max clocked Wolfdale in other CPU demanding tasks too. In average Wolfdale perhaps will clock 400 ~ 500MHz higher than Conroe too so. :)

My plan is getting an E8400 as I expect it to be quite cheap 200 EUR or so at launch and sell my current E6750 for hopefully around 140~150 EUR here in Finland (it's expensive here), so perhaps sth along 50~60 EUR loss and then my goal is 4.2~4.3GHz stable (I think it's very possible judged by what I've seen by the E8500 and E8400 results so far). At 4.3GHz it would be ~15% clock speed advantage and in CPU demanding tasks Penryn seems to be even ~10% faster clock for clock than Conroe so I'd get around 25% performance boost for 50~60 EUR spent, works for me and I just love playing with new hardware that clocks nicely. 4.3GHz C2D 24/7 stable on air would have a "satisfaction value" for me too. :)

Why downclock? Better compare easily achievable 4GHz+ Wolfdale and see the difference grow :)

Quote:

---

Originally Posted by Cooper

How did you disable two cores? Is this option available in BIOS or it is done somehow else?

---

msconfig

Benchmarks of a E6850 vs. the E8400 will follow. I didn't do the benchmarks, I am busy with other stuff right now, but I will try to OC the E8400 on both air and SS/LN2 until next week.

The benchmarks above should only demonstrate the clock for clock performance of the different architectures.

Nice comparison :) And after seeing this maybe I'll preorder an E8400 now for only 170€ :)

Quote:

---

Originally Posted by RPGWiZaRD

I'm definitely getting a Penryn, that it's this big FPS difference between Conroe and Penryn at same clock speed even in Crysis, imagine what it will be like when comparing a max clocked Conroe vs max clocked Wolfdale in other CPU demanding tasks too. In average Wolfdale perhaps will clock 400 ~ 500MHz higher than Conroe too so. :)

---

It's a 800x600 bench intended to show cpu bound differences/strengths, at normal resolutions of 1280x1024, 1680x1050, 1600x1200, 1920x1200 the difference would be very hard to measure I think.

Quote:

---

Originally Posted by Fr3ak

Benchmarks of a E6850 vs. the E8400 will follow. I didn't do the benchmarks, I am busy with other stuff right now, but I will try to OC the E8400 on both air and SS/LN2 until next week.

The benchmarks above should only demonstrate the clock for clock performance of the different architectures.

---

Nice work, my Wallet hates you LOL!

Quote:

---

Originally Posted by dinos22

the fun begins when you overclock them

i believe that any chip is only worth it's max overclock no matter the model number in CPUz :D

---

+1 :up:

Quote:

---

Originally Posted by Levish

It's a 800x600 bench intended to show cpu bound differences/strengths, at normal resolutions of 1280x1024, 1680x1050, 1600x1200, 1920x1200 the difference would be very hard to measure I think.

---

And your point was? I'm interested in CPU performance, not GPU, so high resolutions doesn't interest me but FYI I usually play at 1280x960 so should be a slight difference anyways. And there's better CPU benchmarks than Crysis even at 800x600 too. :)

What he is saying is there those framerate advantages in reality just isn't there, unless you do really game at 800x600 of course.

Quote:

---

Originally Posted by stealthbomber

What he is saying is there those framerate advantages in reality just isn't there, unless you do really game at 800x600 of course.

---

Well I took it for granted that I think every1 at XS forums knows it by now, some1 always has commented on that in like every CPU review with low resolutions.

If you want to bench CPU performance low res is the way to go. There are other applications where CPU performance is more important than GPU so PC games in low res is also an "indication" how the CPUs performs like in the CPU dependant applications. If the only thing you're doing with your comp is to play PC games on your 22" or 24" LCD then I guess it won't be that interesting but then you would be better focusing on graphics cards instead. :p:

Just saying, no need to comment on the obvious, it gets so repetitive.

Crysis is ver GPU limiting. Even the internal CPU benchmark is. With a gaming resolution of 1280 x 1024 with all details and/or AA/AF, the differemce between the CPUs is not that big. To show what the CPUs are capable of, you have to reduce the resolution. It might not be useful if you only play Crysis, but for other games, the faster CPU in Crysis will most likely be a fast CPU there as well.

I also do GPU-benchmarks with a higly overclocked quadcore, even if most people wont run the CPU that fast, but I want to make sure to show video card performance in that benchmark and not what Fps someone would get using a CPU @ stock paired with the specific video card.

Usually there are both variants being tested: First with a CPU and GPU at stock, then depending on what is tested CPU or GPU performance.

Wow thats a difference in speed. Not just a little amount but a decent amount. Besides the fact they surely will beable to OC even higher. But as said thats low res.. is there a higher res they tested this on?

Quote:

---

Originally Posted by Skyline GT-R

Nice comparison :) And after seeing this maybe I'll preorder an E8400 now for only 170€ :)

---

WTF, it's listed for only 170€? Where? (German or dutch site I suppose :p:)

I have a hard time believing Intel would push the prices this low at launch though, I mean E6850 which has same frequency 9x333 costs like 210 EUR on the cheaper sites today and E8400 is better and a new product... ~200 EUR is the lowest I could expect, anything lower would just be a surprise and only reason to do so that I could think of would be to make the life harder for AMD on the desktop market. :p:

Quote:

---

Originally Posted by RPGWiZaRD

WTF, it's listed for only 170€? Where? (German or dutch site I suppose :p:)

I have a hard time believing Intel would push the prices this low at launch though, I mean E6850 which has same frequency 9x333 costs like 210 EUR on the cheaper sites today and E8400 is better and a new product... ~200 EUR is the lowest I could expect, anything lower would just be a surprise and only reason to do so that I could think of would be to make the life harder for AMD on the desktop market. :p:

---

http://www.pimspcshop.nl/page.htm?de...RDER)&id=8981&

cheap and under 200euro.

I guess the 45nm make them much cheaper to make and sell

Quote:

---

Originally Posted by RPGWiZaRD

Well I took it for granted that I think every1 at XS forums knows it by now, some1 always has commented on that in like every CPU review with low resolutions.

If you want to bench CPU performance low res is the way to go. There are other applications where CPU performance is more important than GPU so PC games in low res is also an "indication" how the CPUs performs like in the CPU dependant applications. If the only thing you're doing with your comp is to play PC games on your 22" or 24" LCD then I guess it won't be that interesting but then you would be better focusing on graphics cards instead. :p:

Just saying, no need to comment on the obvious, it gets so repetitive.

---

QFT! Should be stickied, Hi Res for GPU, Low Res for CPU.

Quote:

---

Originally Posted by flopper

http://www.pimspcshop.nl/page.htm?de...RDER)&id=8981&

cheap and under 200euro.

I guess the 45nm make them much cheaper to make and sell

---

Yes of course 45nm should probably make it a bit cheaper to produce but 173€ pre-launch for a CPU like this is amazingly low however you put it (in fact it's what I paid for this E6750 if I remember correctly). Too bad this site doesn't ship to other than Netherlands & Belgium, otherwise I might had placed an order right now. :D Hope other sites will follow the example. In US this deal would correspond something like a 220~$230 pricetag on sites such as newegg.

At least I know which CPU will be my next one now, Q9450 was the interesting one for me before but the price and the fact that 8x will cause mobo bottleneck probs when overclocking with a Penryn I'd rather just pick the cheap E8400 with 9x multi and aim for 4GHz+ (well max out the CPU rather than mobo) with barely losing any money selling my current CPU.

Quote:

---

Originally Posted by flopper

http://www.pimspcshop.nl/page.htm?de...RDER)&id=8981&

cheap and under 200euro.

I guess the 45nm make them much cheaper to make and sell

---

WOW!

too bad they didnt bench with an athlon 64 x2 :D
i wonder if it would be faster or slower than phenom

Quote:

---

Originally Posted by saaya

too bad they didnt bench with an athlon 64 x2 :D
i wonder if it would be faster or slower than phenom

---

Do you really want to know, it could be very depressing :eek:

Benches of a Athlon X2 at 2.4GHz will follow :) Most likely tomorrow.

Quote:

---

Originally Posted by saaya

too bad they didnt bench with an athlon 64 x2 :D
i wonder if it would be faster or slower than phenom

---

Since the x2 6400 generally beats out the e6600 the answer is pretty obvious.

Quote:

---

Originally Posted by flopper

http://www.pimspcshop.nl/page.htm?de...RDER)&id=8981&

cheap and under 200euro.

I guess the 45nm make them much cheaper to make and sell

---

nice! I love it :D even cheaper than I expected
I'm going to pick one in January if they can really reach high clocks easily

Quote:

---

Originally Posted by Periander6

Since the x2 6400 generally beats out the e6600 the answer is pretty obvious.

---

It doesn't matter because this was done clock for clock at 2.4ghz.

Quote:

---

Originally Posted by Fr3ak

Benches of a Athlon X2 at 2.4GHz will follow :) Most likely tomorrow.

---

that should be interesting.. although i m not sure if its fair.. are there any comparisons with an e6600 vs a q6600 with 2 cores disabled?(if that is possible) native quad was never meant to be a dual core.

Quote:

---

Originally Posted by villa1n

that should be interesting.. although i m not sure if its fair.. are there any comparisons with an e6600 vs a q6600 with 2 cores disabled?(if that is possible) native quad was never meant to be a dual core.

---

Unlike Q6600, X4 got all 4 cores with a shared L3. So no changes from X2, X3 or X4 as such.

With Q6600 vs E6600 you could end up disabling the "wrong" cores. Either with 2 single cores oN MCM with 4MB each, or 2 cores on same with shared 4MB.

Quote:

---

Originally Posted by RPGWiZaRD

Yes of course 45nm should probably make it a bit cheaper to produce but 173€ pre-launch for a CPU like this is amazingly low however you put it (in fact it's what I paid for this E6750 if I remember correctly).

At least I know which CPU will be my next one now, Q9450 was the interesting one for me before but the price and the fact that 8x will cause mobo bottleneck probs when overclocking with a Penryn I'd rather just pick the cheap E8400 with 9x multi and aim for 4GHz+ (well max out the CPU rather than mobo) with barely losing any money selling my current CPU.

---

I seen a few sites that offer pre order.

Its a new price introduction however, we saw it with the conroe update where the 1333 was cheaper than the old ones.
I got my current one with a good deal and i guess I drop it for a good wolfdale at at least 9x since if anything is true they will go beyound 4ghz with ease.

AMD is struggling, there is nothing I can view as a contender for overclockers.
Six pack Joe is another matter if amd can get cold chipsets and cold cpu´s out the door they can do well.

I look forward to push one to 4.5ghz or so.
;)

Quote:

---

Originally Posted by Shintai

Unlike Q6600, X4 got all 4 cores with a shared L3. So no changes from X2, X3 or X4 as such.

With Q6600 vs E6600 you could end up disabling the "wrong" cores. Either with 2 single cores oN MCM with 4MB each, or 2 cores on same with shared 4MB.

---

Yeah that makes sense, didnt think there was a way to target.. hrm, but in that regard then, with the phenom, are any clock cycles wasted checking the dead cores?

y use games to compare cpu's when there are specific programs like cinebench designed for that.
see, u can give me the cinebench results, and i can infer how good they are for gaming based on that. also, i can then compare it to that of my own pc, or other pc's. with game benches, im gonna have to get an identical graphics card, identical graphics driver, run the bench with identical settings just to compare results lol. and there is probably more variation in the game benches too.

Yea Cinebench, console emulators for PC (often graphics is taken care by CPU calculations, making CPU requirement quite high :)) are good CPU benchmarks. But low res PC gaming is also quite fine but Crysis is a bit too GPU demanding, would gladly see a bit older game used if using a pc game.

Quote:

---

Originally Posted by villa1n

Yeah that makes sense, didnt think there was a way to target.. hrm, but in that regard then, with the phenom, are any clock cycles wasted checking the dead cores?

---

No, they wont waste any cycles. They are completely inactive.

Quote:

---

Originally Posted by RPGWiZaRD

Yea Cinebench, console emulators for PC (often graphics is taken care by CPU calculations, making CPU requirement quite high :)) are good CPU benchmarks. But low res PC gaming is also quite fine but Crysis is a bit too GPU demanding, would gladly see a bit older game used if using a pc game.

---

yea, but the cpu benchs like cinebench are more accurate and precise, and far more easy to replicate, cause things like what graphics card is being used and what drivers are being used dont play any real role in the score.

Quote:

---

Originally Posted by grimREEFER

yea, but the cpu benchs like cinebench are more accurate and precise, and far more easy to replicate, cause things like what graphics card is being used and what drivers are being used dont play any real role in the score.

---

Are you familiar with emulation scene? At least I've been for many years and I'm a betatester for PCSX2 PS2 emulator for example and to me this is the perfect CPU benchmark since it produces graphics which is mostly taken care of by CPU, only the graphics plugin utilizes a bit of the graphics card but it's not enough to show any impact at all on the CPU performance, I mean I could downclock my 7900GTO to well below stock clocks and see 0 FPS decrease in most games, some are more gfx demanding than others though so it can vary a bit. It's either CPU or GPU that is the limiting factor, and depending on plugin and by enabling better quality settings the GPU load quickly rises so you can use it to bench GPU too if you like if using high enough quality.

I've compared my AMD Opteron 165 @ 2.8GHz and Intel setup both at stock vs stock, both at 2.8GHz and also 2.8GHz AMD vs 3.75GHz C2D and then at AMD @ 2.8GHz and heavily tweaked (as usual) with as tight timings as possible vs Intel setup at completely stock settings in bios so 2.66GHz with slacky timings etc. there was to my suprise even ~35% clock for clock advantage to Intel (I had expected maybe ~20% lol). Whit tweaked and maxed out settings on the Intel setup the difference rises to about ~40% clock for clock advantage as at C2D 3.75GHz vs AMD 2.8GHz there was a pretty consistent 72~75% advantage in most games to the Intel setup. The coders had AMD dual core setups at the time they've done huge improvements (such as adding dual core support) and there's nothing in the code that would make it more "Intel" optimized by any means, it's just shows the raw CPU performance with zero impact by GPU performance. :)

It's easy to make exact comparisions with PCSX2, I simply make a save state, load it at the different configs or setups and it loads the exact same spot and I've even matched the same polygon count to be as accurate as possible. I usually test different games in different spots to see how it behaves like and it's been a very consistent FPS advantage for the Intel setup. Same thing with overclocking, the FPS increase is pretty linear with the overclocked CPU amount. :D

I recommend these articles if you want to read more about it. They're old by now and and concerning a bit outdated PCSX2 version but yea it still explains quite good what I'm saying here.
http://forums.ngemu.com/pcsx2-offici...es-matter.html
http://forums.ngemu.com/pcsx2-offici...so-matter.html

Will of course do a comparision between this E6750 and E8400 Wolfdale later on, I'm guessing 10~15% clock for clock FPS boost for Wolfdale in this app, will be interesting to see what it actually will be like though.

Quote:

---

Originally Posted by RPGWiZaRD

Are you familiar with emulation scene? At least I've been for many years and I'm a betatester for PCSX2 PS2 emulator for example and to me this is the perfect CPU benchmark since it produces graphics which is mostly taken care of by CPU, only the graphics plugin utilizes a bit of the graphics card but it's not enough to show any impact at all on the CPU performance, I mean I could downclock my 7900GTO to well below stock clocks and see 0 FPS decrease in most games, some are more gfx demanding than others though so it can vary a bit. It's either CPU or GPU that is the limiting factor, and depending on plugin and by enabling better quality settings the GPU load quickly rises so you can use it to bench GPU too if you like if using high enough quality.

I've compared my AMD Opteron 165 @ 2.8GHz and Intel setup both at stock vs stock, both at 2.8GHz and also 2.8GHz AMD vs 3.75GHz C2D and then at AMD @ 2.8GHz and heavily tweaked (as usual) with as tight timings as possible vs Intel setup at completely stock settings in bios so 2.66GHz with slacky timings etc. there was to my suprise even ~35% clock for clock advantage to Intel (I had expected maybe ~20% lol). Whit tweaked and maxed out settings on the Intel setup the difference rises to about ~40% clock for clock advantage as at C2D 3.75GHz vs AMD 2.8GHz there was a pretty consistent 72~75% advantage in most games to the Intel setup. The coders had AMD dual core setups at the time they've done huge improvements (such as adding dual core support) and there's nothing in the code that would make it more "Intel" optimized by any means, it's just shows the raw CPU performance with zero impact by GPU performance. :)

It's easy to make exact comparisions with PCSX2, I simply make a save state, load it at the different configs or setups and it loads the exact same spot and I've even matched the same polygon count to be as accurate as possible. I usually test different games in different spots to see how it behaves like and it's been a very consistent FPS advantage for the Intel setup. Same thing with overclocking, the FPS increase is pretty linear with the overclocked CPU amount. :D

I recommend these articles if you want to read more about it. They're old by now and and concerning a bit outdated PCSX2 version but yea it still explains quite good what I'm saying here.
http://forums.ngemu.com/pcsx2-offici...es-matter.html
http://forums.ngemu.com/pcsx2-offici...so-matter.html

Will of course do a comparision between this E6750 and E8400 Wolfdale later on, I'm guessing 10~15% clock for clock FPS boost for Wolfdale in this app, will be interesting to see what it actually will be like though.

---

Although it is nice to isolate components from each other, and that has its merits, extreme isolation does not really have much pragmatic usefulness to us, aside generating a "stat". Computers are not single tasked machines... well unless you are just building a rig for folding ^^, but that aside, how well it works with the components at hand is just as important.
As well, the cross section of stats generated are what give an overall picture, i dont think we should be looking to find one be all end all cpu test, as that will never exist, by virtue of the fact, architecture's are not the same, and have advantages, and disadvantes. I do feel this is a good test for cpu's though, reading through those posts you linked, and would be a nice addition to create a more comprehensive crossection.

sadly my q6600 wouldnt clock worth beans :( (wouldnt even hold 3.6 stable) not to mention the fact that it was running well below ambient on my 24/7 water/tec setup(about 5*c idle and 15*c load), hell my e6600 clock to 4.0 on regular water no problem, stupid things a dud, i cant wait for a new drop in upgrade hopefully ill get some decent clocks this time

3 GHz E6850 vs. E8400 using Crysis benchmarks again and the CPU test of 3D Mark 06:

http://www.pcgameshardware.de/?menu=...=742471&page=1

Quote:

---

Originally Posted by Fr3ak

3 GHz E6850 vs. E8400 using Crysis benchmarks again and the CPU test of 3D Mark 06:

http://www.pcgameshardware.de/?menu=...=742471&page=1

---

I think that's a E6700 OC to 3.0 vs E8400

Yes, I am sorry, but it doesn't make any difference ;) 333 x 9 is 333 x 9 with a Conroe, independant of the model name :)

I just hope there isnt a fsb wall at 450 or such since that would truly suck.
:D

I can assure you that there is no FSB wall up to 540 MHz using a single stage ;)

That was tested with a multi of 9.

Quote:

---

Originally Posted by Fr3ak

Yes, I am sorry, but it doesn't make any difference ;) 333 x 9 is 333 x 9 with a Conroe, independant of the model name :)

---

yeah but what about the ram?
evenhow this testbed is at least faster than 6850.

Around 6-8% perf diff, add the price diff, around 35% cost /perfomance..not bad at all :up:

Doing some draft calculations, in crysis, conroe has a bit better scalability

Quote:

---

Originally Posted by Fr3ak

I can assure you that there is no FSB wall up to 540 MHz using a single stage ;)

That was tested with a multi of 9.

---

air or water?
I am a watercooling fan and freak ;)

Didn't try it on air or water yet :) I am very busy, so my time is limited, which is why I went straight to single stage. I will test it under air and watercooling next week.

Look forward to your results

Quote:

---

Originally Posted by turtletrax

Wow... 6 to 10 FPS difference between Wolf and Phenom. That is huge :eek:

---

Don't troll.

Quote:

---

Originally Posted by Omastar

Don't troll.

---

:) Actually, probably ironically, 6-10 FPS for a low framerate game is huge, % wise and in playability. Averaging 30 to 40 or 45 FPS would suggest the period of time of min frame rates will be much greater at the higher end.

Quote:

---

Originally Posted by JumpingJack

:) Actually, probably ironically, 6-10 FPS for a low framerate game is huge, % wise and in playability. Averaging 30 to 40 or 45 FPS would suggest the period of time of min frame rates will be much greater at the higher end.

---

Klocking 24/7 isnt taken into account.
If I can klock to 4.5ghz on water using wolfdale, I bet the fps will be wider than my 3800mhz conroe.
;)
Wolfdale is faster, not much but we are talking to people who shift radiators if the difference is 0.5 degree.

Faster=better:ROTF:

Not great for the Phenom