Intel testing 45nm.

[party animal]

According to Intel's website, they are demonstrating working 45nm chips. I wonder what kind of ocs we can get out of these :santa:

[proth]

According to Intel's website, they are demonstrating working 45nm chips. I wonder what kind of ocs we can get out of these :santa:

By the time they release that, they'll be down to 16 stage pipeline running the core @ 2.0GHz

Party is over for ultra high clock speeds after Cedar Mills and Pressler

IMO

[party animal]

But if they are running 2.0 stock, i wonder how high they will go. Comparing apples to apples here.

[wingman99]

But if they are running 2.0 stock, i wonder how high they will go. Comparing apples to apples here.

That is a good point, i hope there better than $245 3.2 p4 Northwood

[d94]

But if they are running 2.0 stock, i wonder how high they will go. Comparing apples to apples here.
if they hit 3, that's rock
if they hit 4, that would be un-realistic

'Wolfdale' and 'Ridgefield', for example, two dual-core, single-die desktop chips due 2008, will have 3MB and 6MB of cache apiece.

These should appear alongside 'Perryville' and 'Penryn', respectively single-core and dual-core mobile parts, the former with 2MB of cache, according to the report, the latter with 3MB and 6MB of L2.

On the server side, we have 'Hapertown', an eight-core Xeon with 12MB of cache shared between the cores. It doesn't look like we're going to see a desktop CPU with eight cores until 2008/2009, with the debut of 'Yorkfield', also sporting 12MB of cache.

The same timeframe will see the debut of the quad-core 'Bloomfield', but how much cache it will contain isn't known. The same question hangs over the 'Silverthorne' mobile processor, for which we don't even have a core-count yet. ®
http://www.theregister.com/2005/12/05/intel_45nm_roadmap/
JESUS 12mb cache LOL 8x core WTH!

[proth]

if they hit 3, that's rock
if they hit 4, that would be un-realistic


LOL

[3DFlyer]

I saw a register rumor that stated the Conroe would have a model at 3.333GHz. On and 18 stage pipeline at that. I wouldn't count out high clocks yet. People have stated that you can only go so high with shorter pipes, but that is with AMD CPU's...not Intels. Intel has alot more money to develop technology with, and they got some pretty smart cookies there too. I think we better wait and see what an actual Intel will do with the shorter pipes. It may very well change rewrite the way CPU's work with these shorter pipes. :)

[party animal]

i agree, im actually pretty excited about what Conroe can actually do. Intel has done pretty well with shorts pipelines considering the Pentium M and Core Duo.

[Sjaak]

Keep in mind that what they are showing here are mere samples of a couple of DRAM chips...it may take another year or two before they get actual 45nm processors working.

[>HyperlogiK<]

Take all these things with a pinch of salt, Sony/Toshiba/IBM had claimed that they would have cell being fabbed on 45nm by the second half of this year, that then became 65nm.
It's certainly promising that Intel are making such advancements, but these things take times as well as R&D money.

[hUMANbEATbOX]

well intel has no shortage of r&d money, that's for sure.

[bobad]

By the time they release that, they'll be down to 16 stage pipeline running the core @ 2.0GHz Party is over for ultra high clock speeds after Cedar Mills and Pressler IMO

I've been wondering for 2-3 years when the extreme frequencies would back off and they would go in the direction of smaller die sizes and multiple cores.

Now I bet we'll see "Core Wars" instead of those insane frequencies.

Hmmmm... wonder how fast a 45nm chip at 2ghz with 16 cores would be? :) Even a mild overclock on such a chip would gain a lot of speed. We just won't talk about multi-core heat right now. We can cross that bridge when we get to it. :D