Phenom 2 (deneb) vs. Core2Quad vs. Corei7

[Tempered81]

phenom 2 940 @ 3ghz

c2q q9550 @ 2.83ghz

corei7 940 @ 2.93ghz

phenom 9950 @ 2.6ghz

http://www.xtremesystems.org/f...=3464965&postcount=459

all 2x/4xAA w/ single gtx280

STALKER CLEAR SKY, 1680x1050

Phenom II x4 940 3Ghz = 41.7
C2Q Q9550 2.83Ghz = 41.2

CRYSIS WARHEAD, 1680x1050

Phenom II x4 940 3Ghz = 35.7
C2Q Q9550 2.83Ghz = 35.0

DEAD SPACE, 1680x1050

Phenom II x4 940 3Ghz = 219.0
C2Q Q9550 2.83Ghz = 214.4

FAR CRY 2, 1680x1050

Phenom II x4 940 3Ghz = 54.4
C2Q Q9550 2.83Ghz = 46.2

WORLD AT WAR, 1680x1050

Phenom II x4 940 3Ghz = 48.7
C2Q Q9550 2.83Ghz = 57.3

WORLD IN CONFLICT, 1680x1050

Phenom II x4 940 3Ghz = 92.7
C2Q Q9550 2.83Ghz = 95.0

 

[masteryoda34]

Even more interesting is that these benches show Phenom II 940 being competitive with i7 940 in games.

I sense a price war coming.

 

[SickBeast]

When is the Phenom 2 coming out?

 

[OCGuy]

Just like most have been saying, there is no real advantage there, even being clocked slightly higher. I would also like to see how far you can OC the Phenom II on air within AMD voltage spec against a C2Q with a 1.3625V max Vcore. This is getting exciting!


@ masteryoda34: Price war would = :thumbsup:

 

[masteryoda34]

Originally posted by: SickBeast
When is the Phenom 2 coming out?

January 2009

 

[TidusZ]

Are these even reputable? the supposed source doesn't have them anymore. I could have made these benchmarks myself using paint in less than an hour while I was stoned. I sure hope they're real though.

 

[Cogman]

Originally posted by: TidusZ
Are these even reputable? the supposed source doesn't have them anymore. I could have made these benchmarks myself using paint in less than an hour while I was stoned. I sure hope they're real though.

No source, so it might not have much to it. But heres hoping it does this well, if not better.

 

[SickBeast]

Originally posted by: Cogman

Originally posted by: TidusZ
Are these even reputable? the supposed source doesn't have them anymore. I could have made these benchmarks myself using paint in less than an hour while I was stoned. I sure hope they're real though.

No source, so it might not have much to it. But heres hoping it does this well, if not better.

The results looked somewhat erratic to me. Usually you don't see such swings in the benchmark scores when it comes to just the CPU. For the Phenom 2 to dominate in some games, and then lose badly in others, it doesn't make sense to me.

 

[rogue1979]

Well, we don't know for sure yet, but it looks like Phenom 2 and Corei7 are close in gaming performance. Based on this possibly premature opinion, there is something I just don't understand.

When AMD originally moved the memory controller on-die in the Athlon64 architecture gaming performance showed a big improvement, even before dual core X2.

Intel finally beat that with C2D and no on-die memory controller, how is it when they finally do get the memory controller on-die there doesn't seem to be much if any boost for gaming?

 

[Extelleron]

Originally posted by: rogue1979
Well, we don't know for sure yet, but it looks like Phenom 2 and Corei7 are close in gaming performance. Based on this possibly premature opinion, there is something I just don't understand.

When AMD originally moved the memory controller on-die in the Athlon64 architecture gaming performance showed a big improvement, even before dual core X2.

Intel finally beat that with C2D and no on-die memory controller, how is it when they finally do get the memory controller on-die there doesn't seem to be much if any boost for gaming?

It's mostly because the cache structure on Nehalem is much less favorable to games than Penryn, and in many cases that balances out the faster memory access Nehalem gets with the IMC.

With Yorkfield you had 12MB fast L2..... now you have 256KB fast L2 per core, and a much slower 8MB L3. Games really like that large L2 cache.

Meanwhile with A64 AMD actually increased cache sizes, from 512KB in the last Athlon XPs to 1MB in the top of the line A64s.

 

[Concillian]

The benches are close enough that price would be my deciding factor if I were buying a quad today... or early next year, rather.

 

[SickBeast]

Originally posted by: Extelleron

Originally posted by: rogue1979
Well, we don't know for sure yet, but it looks like Phenom 2 and Corei7 are close in gaming performance. Based on this possibly premature opinion, there is something I just don't understand.

When AMD originally moved the memory controller on-die in the Athlon64 architecture gaming performance showed a big improvement, even before dual core X2.

Intel finally beat that with C2D and no on-die memory controller, how is it when they finally do get the memory controller on-die there doesn't seem to be much if any boost for gaming?

It's mostly because the cache structure on Nehalem is much less favorable to games than Penryn, and in many cases that balances out the faster memory access Nehalem gets with the IMC.

With Yorkfield you had 12MB fast L2..... now you have 256KB fast L2 per core, and a much slower 8MB L3. Games really like that large L2 cache.

Meanwhile with A64 AMD actually increased cache sizes, from 512KB in the last Athlon XPs to 1MB in the top of the line A64s.

There's that, plus the fact that the memory controller for the C2D paired with its cache was so efficient, that it actually compared favorably to the Athlon 64 in terms of cache latency. In many ways intel didn't even need to move to an on-die memory controller.

 

[myocardia]

Originally posted by: SickBeast
There's that, plus the fact that the memory controller for the C2D paired with its cache was so efficient, that it actually compared favorably to the Athlon 64 in terms of cache latency. In many ways intel didn't even need to move to an on-die memory controller.

No, they did, it just wasn't needed on desktop systems yet.

 

[dguy6789]

I hope these are true. A fast Phenom 2 would be great for competition.

 

[eternalone]

WOW this is really good news for AMD if its true. Im amazed, that the phenom 2 can compete so well with the i7, if true of course, this would be why Intel has been shaken of late with all this phenom 2 talk. WOW. :thumbsup:

 

[evilbix]

I am curious if this is another shanghai vs whatever comparison. I've seen a bunch of benches compare an i7 system vs a shanghai using low-speed ECC memory and throwing up game benches. I'd see some 100% legit deneb benches and know whether to wait or not.

If it is a shanghai it would be amazing to see what deneb can really do. If not, it's still pretty good numbers.

 

[thilanliyan]

Originally posted by: evilbix
I am curious if this is another shanghai vs whatever comparison. I've seen a bunch of benches compare an i7 system vs a shanghai using low-speed ECC memory and throwing up game benches. I'd see some 100% legit deneb benches and know whether to wait or not.

It's using DDR3-1333...doesn't say ECC but there aren't many details.

 

[Kuzi]

These are some great gains. Did you guys notice on World in Conflict, Phenom 2 is 43% faster than the Phenom 9950, while only having a 15% advantage in clock speed. In Far Cry 2 it's 35% faster, World at War 29% faster, Dead Space 25% faster. I guess some games really like the extra L3 cache (and tweaks) on Deneb.

 

[JMapleton]

I'm not an AMD fanboy, I actually lean towards Intel. But I do remember everyone being as skeptical early Conroe benches as people are skeptical about early Phenom II benches, so I will not make any assumptions and say Phenom II's benches are overly optimistic.

 

[error8]

Looks like I'm going to buy myself a cheap quad after all, in the beginning of 2009. Price war is the only war I like.

 

[BTRY B 529th FA BN]

Originally posted by: error8
Looks like I'm going to buy myself a cheap quad after all, in the beginning of 2009. Price war is the only war I like.

+1

 

[JackyP]

There's only one benchmark where the "deneb" beats penryn (eye-balled & adjusted for clock). So Intel's IPC is still unmatched it seems. If the highest end phenom competes with the q9550 price-wise it could sell well and even enable some profit for AMD in contrast to k10. I'm really not overly optimistic due to the huge die size of deneb, though. Still it's certainly an improvement over K10.

 

[Viditor]

Originally posted by: JackyP
There's only one benchmark where the "deneb" beats penryn (eye-balled & adjusted for clock). So Intel's IPC is still unmatched it seems. If the highest end phenom competes with the q9550 price-wise it could sell well and even enable some profit for AMD in contrast to k10. I'm really not overly optimistic due to the huge die size of deneb, though. Still it's certainly an improvement over K10.

Ummm...the Deneb is reputed to be the same die size as the i7.
Text

 

[BlueBlazer]

Looking further into the XtremeSystems thread...

The original source http://forums.vr-zone.com/showthread.php?t=359342

Notice the poster statistics:

Silentjack07
Registered User

Join Date: Nov 2008
Posts: 2

And those 2 posts are in that thread!

Quote:
Originally Posted by geckolizard View Post
no source ??

http://www.expreview.com

the benchmarks had been there for several hours tonight. now i cant find them anymore, do you?

That's what I call fishy...

Furthermore.... http://www.xtremesystems.org/f...=3466299&postcount=504

This one http://www.bit-tech.net/hardwa...ng-intel-core-i7-920/7 contradicts with the so-called (fishy/dodgy) "benchmarks" http://www.freeimagehosting.ne...age.php?a4e1257410.jpg

Note its using the same graphics card also, GTX280, and same resolution 1680x1050.

 

[JackyP]

Originally posted by: Viditor

Originally posted by: JackyP
There's only one benchmark where the "deneb" beats penryn (eye-balled & adjusted for clock). So Intel's IPC is still unmatched it seems. If the highest end phenom competes with the q9550 price-wise it could sell well and even enable some profit for AMD in contrast to k10. I'm really not overly optimistic due to the huge die size of deneb, though. Still it's certainly an improvement over K10.

Ummm...the Deneb is reputed to be the same die size as the i7.
Text

Umm.. that was my point actually =)
K10.5 doesn't beat penryn, lower die size, in IPC for desktop applications (as predicted). Nehalem is faster than penryn in games ~5-10% as far as I have seen.
However, Nehalem is out of their league (price- and performancewise) and not competing for a place in your new gaming anyway.
A single Neha can equal 2 high end harpers or mid-end shanghais for several workstation/ server tasks as shown by TR.