http://www.dailytech.com/Article.aspx?newsid=7277

4MB L2 cache (Conroe) => 6MB L2 cache (Penryn)

I think something has changed:
1. The L2 cache associativity may increase from 16-way to 24-way, which will raise the efficiency of the cache system.
2. From the 3MB low-end version, the die size will be lower than 100mm^2, and the manufacturing cost will be lower.

I think AMD should worry about these. :?

Penryn is not a simple die shrink of Conroe...

Just like they doubted they'd get their asses handed to them by C2D....

Conroe is a great chip put I think Penryn will not be a vast improvement, just a +5-10% refinement. Increased cache associativity and increased cache size are also a bit contradictory; if you improve the cache efficiency, you tend more to reduce it's size than increase it, not to mention that even for Intel's cache-hungry architecture, it's incidence on performance decreases a lot above 2M/core. This thing of the from 2M/core to 3M/core looks a lot like that of the 3.4GHz P4 EE with 512K L2/2ML3; it's top performance gain over a standard 3.4GHz P4 was only 5%.
Altogether, I think Penryn has a lot better features to make it perform better than cache increase, which I think is more for marketing boasting than for a beefy performance punch.

From what I have seen I would expect to see a 5 - 15% depending on the app incease in clock for clock performance penryn over conroe. This is a decent refresh they have going on...the kicker is the new smaller scale with less power and higher clocks that will be achieved with the lower heat output. It will be a good push for the arch just before they concentrate on nehlam more. (sry if I misspelled that.)

Best,

3Ball

The purpose of the increased cache size is to support the higher clock frequencies of Penryn. The greater associativity of the cache is probably to maintain the performance of the cache that has been expanded by 50%.

+- 10% per clock improvement + 3.6-4.0 GHZ top bin will make Penryn formidable.

I would be genuinely impressed if K10 beat that by a significant margin.

I would be genuinely impressed if K10 exists.

I would be genuinely impressed if K10 exists.

You have become a fanboi already.

I would expect Intel can retain the performance crown with Penryn family.

Of which brand? I've been on AMD for years and last week bought a new X2!!! 8)

I would be genuinely impressed if K10 exists.
Well, if seeing is believing, we've not had anything to believe,... well, someone might count the task manager demo, the countless fake CPUZ screenshots on fudzilla and all those cool LED fans AMD demoed last week

I would be genuinely impressed if K10 exists.
Well, if seeing is believing, we've not had anything to believe,... well, someone might count the task manager demo, the countless fake CPUZ screenshots on fudzilla and all those cool LED fans AMD demoed last week

m25, you're right. I've been too negative on this. When you clinically add up all the hard evidence that you've outlined there is no doubt whatsoever. I regret being a neighsayer. Please, forum members. Forgive me.

I forgot to add the countless claims from AMD like "the K10 will deliver ultimate performance" or "ours is a true quad core CPU" ... I'm so tired of all this BS...

Money talks and bulls**t walk