My Conroe compared to Penryn..

[warezme]

I ran these benches and you all can tell me if they are a valid ballpark preview:

My comparison

See the data of my E6400 @ 3.33Ghz compared to the data from Beijing of 3.33Ghz Penryns in 3DMark06. I tried to get as close to the speed of their chips but since their multiplier is like 10x 333 for 1666FSB, I had to bump my FSB to 1666 on an 8X multiplier and the mem slightly to balance.

3DMark06 is the only benchmark I have that is the same as theirs but even then I run mine in XP instead of Vista so there is a variance there also. I have the HL2 benchmark but not the DEM's they used for comparisons.

It is not an apples to apples but what is interesting as I have said before, the performance improvement while good is not that great when compared to C2D of the same speed. If I had used a E6600 I believe my scores would have been equal or better than Wolfdale of the same speed because of the larger cache factor.

Here's a quick graph but the data is at the top link for speed of CPU, bus and mem.

 

[warezme]

I meant to say their FSB of 1333 and mine of 1666...

 

[djgandy]

Penryn was never going to stomp on Conroe etc clock for clock. Penryn is C2D like they are.
I'm not sure why people are under the impression that huge clock for clock increases were going to happen. My impression from the offset is that penryn will allow higher clock speeds at lower temps and power comsumption. Higher clocks = more processing power.

It also bundles some new SSE instructions and slightly better IPC performance just to put smiles on our faces.
Sounds like a good die shrink to me.

EDIT:

BTW your chart shows wolfdale on average as good as a 3.6ghz conroe in the CPU bench (overall is unreliable, mainly based on gpu power)

Sounds good to me!

 

[warezme]

Penryn was never going to stomp on Conroe etc clock for clock. Penryn is C2D like they are.

Exactly, I would like to see more perspective on the current hoopla on the reporting of Penryn. If it is no more than a faster C2D than wouldn't it be conceivable that Barcelona still stands a good chance of one upping Intel providing their new architecture is killer.

BTW your chart shows wolfdale on average as good as a 3.6ghz conroe in the CPU bench (overall is unreliable, mainly based on gpu power)

yes, I was afraid it would be GPU skewed since the only comparable bench I have they used is 3DMark06 but as I noted I think a 4MB conroe would have been a more comparable match based on the CPU index of 3DMark06

 

[djgandy]

Exactly, I would like to see more perspective on the current hoopla on the reporting of Penryn. If it is no more than a faster C2D than wouldn't it be conceivable that Barcelona still stands a good chance of one upping Intel providing their new architecture is killer.

It's 5-10% faster clock for clock in general. That means pretty much everywhere.
Higher clock speeds at lower power is where its at. Expect 3.6ghz parts.
AMD are going to struggle to do anything about that.

I don't see how it's not faster than C2D?

 

[RichPLS]

But from the results so far, it appears that it is indeed getting anywhere from a minimum of 5 to 10% better performance per clock (taking out cache size and MHz diff), and upwards of 40% better towards the top end...
On top of this it clocks higher.

 

[warezme]

But from the results so far, it appears that it is indeed getting anywhere from a minimum of 5 to 10% better performance per clock (taking out cache size and MHz diff), and upwards of 40% better towards the top end...
On top of this it clocks higher.

I think the "silver bullet" for Intel is the smaller die and low power. If you think of this way 333x10 = 3.33Ghz for Wolfdale, at 333x12 your already hitting just about 4Ghz. Similarly a small bump in FSB to 400 just at 10x multiplier hits 4Ghz right of the bat.

Scaling up over 4Ghz something Netburst couldn't do (easily) looks to be cake for the 45nm Penryn. Thats the killer part.

 

[RichPLS]

Agree