SSE4 / Extra L2 cache - Will it make a big difference? [N/Y]

[Tom]

With the new Penryn processors coming out, I was wondering if the new SSE4 and extra L2 cache will make a big difference in folding speed? Granted they will be faster because of the higher clock speeds, but I wondered if there will be a bigger boost with the added improvements?

[sneakers55]

With the new Penryn processors coming out, I was wondering if the new SSE4 and extra L2 cache will make a big difference in folding speed?

SSE4 won't make a difference until there is a core that takes advantage of it.

As to the extra cache, that depends on the specific WU.
I can see a difference between 2 MB and 4 MB but not consistently.

[Tom]

Thanks!

That makes sense with SSE4. Now there will be something new for the programmers to play with.

I know there is a difference between the C2D's and Conroes when it comes to folding, the extra cache being the boost factor. I am guessing the extra with Penryn will help even more on the cores that use it. I guess we will find out soon enough.

[toTOW]

The Yorkfield is about 5% faster than a Kentsfield at the same frequency. This is mainly due to SSE units improvements. On some WU, the cache will be also usefull but it won't be the case for all WU.

Everything else than SSE and SSE2 is not used by processing cores, so SSE3 or 4 won't change anything until a core taking advantage of them is released ... and I don't think it's currently on PandeGroup's todo list.

[Tom]

Thanks for the info.

I will probably skip the Yorkfield due to the 1600FSB increase and integrated memory controller coming up with the Nehalem processors. Since 8 core processors aren't slated until around 2014, I hope they will come out with a desktop motherboard that will use 2 or more processors very soon. I could use a couple right now!

I'm very temped to try a Wolfdale E8400. They will be cheap and overclock past 4Ghz. Plus I may have a spare motherboard coming soon.