Yorkfield Q9450 vs Kentsfield Q6600 at 3.6Ghz

[Johnny Bravo]

Just a few numbers from my ES Q9450 that I've been playing with for awhile


http://valid.x86-secret.com/show_oc.php?id=297476

Running at the 1333MHz bus it features 2x6MB L2, 6-8 multiplier range and some lovely 45nm technology squeezed into it.

First some bad news .....

Seems the FSBwall it pretty low on these ES chips - 478MHz


http://valid.x86-secret.com/show_oc.php?id=297477

The x8 multiplier really limits the maximum GHz overclock on these cpus. HEre's hoping that the retail ones scale better!

...Let's see how it scales up to its predecessor the Q6600 ?

System setup

abit IX38 QuadGT
2x1024MB Team Group 450@4-3-3-8
Vvikoo 8600GT
Swiftech Compact H2O cooler
Silverstone 850W


I've decided to run both at 3.6Ghz (8x450Mhz) due to the limitation on the Q9450 FSB. Both cpus run on the same system running the same specs. The only difference was the the Q9450 vcore was 1.3volts whereas the Q6600 was 1.45volts:

Lets see some comparisons:

1M SuperPi
Q9450


Q6600


32M SuperPi
Q9450


Q6600


32M/1024M WPrime
Q9450


Q6600


3DMark06 CPU Score
Q9450


Q6600


Power consumption and Heat
Q9450


Q6600


Check out that current rating with all four cores loaded, the Q9450 is drawing nearly half that of the Q6600. Add to this the fact that the Q9450 requires less voltage and it's safe to say that the Q9450 is a very efficient cpu. As a result the temperatures are alot lower on the Q9450 when running all four cores - so less complicated cooling is required.

[Leeghoofd]

Great stuff too bad they don't have something at a reasonable price with a 9 multi... my QX9650 has been delayed till 23rd of January pffffffff

[WoZZeR999]

I was thinking about getting a 9450, but with a lower FSB wall, and low multi, its making me rethink about which of the new cpu's I want. This or E8500.

[CERO]

looks good, but like mentioned pretty much limited by 8x multiplier....

pita leeghoofd, hopes it comes soon m8!

[KTE]

The 2k6 and power section pics aren't loading for 15 minutes now.

Firstly, yep, all the Yorkies I've seen apart from the QX have been low FSB, C2 achievable MHz. Though they are ES.
Secondly, do you have any other benchies? Systools 32M, EVEREST, Pov-Ray, WinRAR, AM3, NuclearusMC, SuperPrime, CPUZ latency, some encoding/decoding?

Thanks.

[Blacklash]

If you want 4.0GHz on air you best go Q9550. Thing about the Q6600 is it will often be easier to get 3.8 with it due to FSB walls and the lower multi the Q9450 is dealing with. I'd rather a 3.8GHz Q6600 than a 3.6GHz Penryn and to each their own.

[Johnny Bravo]

The 2k6 and power section pics aren't loading for 15 minutes now.

Firstly, yep, all the Yorkies I've seen apart from the QX have been low FSB, C2 achievable MHz. Though they are ES.
Secondly, do you have any other benchies? Systools 32M, EVEREST, Pov-Ray, WinRAR, AM3, NuclearusMC, SuperPrime, CPUZ latency, some encoding/decoding?

Thanks.

Thread fixed - cheers

Will run other benchies if there are others interested? Don't have the programs atm

[GripS]

My god over a full second faster in Super pi 1m at the same clock speeds/ram speeds. Very impressive.

[JoeBar]

Thanks for the tests.


The 8 multi plus the fsb wall makes a 9450 not a viable solution for enthusiasts. Only 9550 with 8.5 should be considered just to fight for the 4ghz barrier.


If E8500 turns out to be a great oc'er there will be many dual core reverts...




NVidia must have something to do with this

[Ender17]

just when I thought I would be satisfied with a Q6600
the power and heat is what is really making me second guess

[Supertim0r]

I highly doubt the Q9550 will do 4ghz stable daily. In my testing, it's good for some quick benching (1-16m, aquamark) 3dmark was max 3.95ghz

[Ender17]

I just want 3.6GHz, but with only 8x multiplier I would need 450 FSB and I don't know if the D975XBX2 can do that

[Supertim0r]

max quad fsb on XBX2 is 390-395ish AND is it supported on this 975x mobo ?

[JoeBar]

I highly doubt the Q9550 will do 4ghz stable daily. In my testing, it's good for some quick benching (1-16m, aquamark) 3dmark was max 3.95ghz

I agree (from what i've seen). That why i stated that u have to fight for 4ghz on a 9550.

[Spanki]

...I'd rather a 3.8GHz Q6600 than a 3.6GHz Penryn and to each their own.

Actually, due to the performance differences seen above, I'd be pretty interested in seeing a Q6600 @3.8GHz vs Q9450 @3.6GHz comparison. Any chance of seeing that JB?

[ziddey]

3.6ghz penryn would whoop a 3.8ghz conroe at divx

[barnacle]

Actually, due to the performance differences seen above, I'd be pretty interested in seeing a Q6600 @3.8GHz vs Q9450 @3.6GHz comparison. Any chance of seeing that JB?

If the OP doesn't mind i would be very interested in that test too,
It would be very interesteing to see where the performance intersects....

and of course a penryn woulod would a conroe at such close speeds as one has twice the cores..... what we are intereseted in is the diffefernce between 2 quad core processors just with different cache, due to the newer yorkies being multi limited it may not be time to upgrade that q6600 yet especially if you can get a 3.8ghts 24/7 overclock on it.

[Dostoyevsky77]

Great post! I'm pre-ordered for a Q9450 myself. I hope a BIOS revision on X38 will help ameliorate the Yorkfield FSB wall.

I wonder how the production chips will differ - if at all - from the seed samples...

[disruptfam]

so what will be better perf wise??

q6600 g0 @ 3.8ghz or q9450 @ 3.6ghz?

when is the q9450 coming out retail?

[rge]

Interesting test. Comparing the two 32M times, you get 6.66% performance improvement at same clockspeed (similar to the 6.5% the xbit lab article quoted as average performance gain). 6.66% at 3600 is ~240 mhz, so theoretically a 3.84 Q6600 should equal a 3.6 QX9450, but would be nice to see tested. (1M superpi showed 8.6% improvement, but thats more prone to variability error)

[Hazzan]

Nice to Compare .....I still waiting my Q9450 ES

[Blasar]

Great comparison that brings some light to the differences between the Q6600 and the Q9450.

I´m waiting the Q9450 launch, to buy a Q6600 for 150€.

[Scubar]

Ill wait until i see what the prices of the Q9450,Q9550 and QX9650 will be when they are all available together.

[KTE]

Yeah those were exactly my reservations and I was hoping for easy high FSBs which they've not been so far, neither have the E8400/E8500.

Chip looks good -> for cold/power.

24/7? I'm not and most people cannot run high FSB, most people do not have an X38 either. And so far over 8 I've seen have topped out ~3.8GHz Pi with 1.4-1.5V/sometimes even water and less than that was stable. So in effect it would be 3.6GHz 24/7 again as with Q6600 G0 and that's considering the top two multi models.
So the good comparison is 3.6GHz compared to each other ideally because 450FSB sounds a very good balance to me and achievable by many boards/chips. People can make up their stats looking at just these figures.

The best one I'd like to see is the 2.5GHz stock Yorkfield compared to Q6600 oc'd a little. It'll show what difference there is in the core without the extra cache these Yorkies have.

[mascaras]

First some bad news .....

Seems the FSBwall it pretty low on these ES chips - 478MHz

CPU FSB wall or motherboard FSB Limit???


btw: nice tests Q9450 vs Q6600

i thought the difference in 3dmark 2006 CPU score between both CPUs would be more than 240 marks .....

its seems only in superpi the Q9450 its way better ( cache) , also in Power consumption and Low voltage , but for me thats secundary



Temps difference :

Q6600 @ 3600Mhz@ 1.40v = 60º
Q9450 @ 3600Mhz@ 1.34v = 50º


i can bench 3dmark 2006 with my Q6600 @ 4Ghz and Prime stable @ 3.8ghz with ~1.48v (all Aircooled) also Q6600 have multi x9 , so in my case i think beside Power comsumption and Low voltages its not worth changing !



regards