I obviously didnt make my point very well. Of course if you took a conroe and bumped up its cache and fsb etc it would perform very similarly to a penryn. they are the same core (mostly) after all. this is mostly a die shrink / tick phase. with that you can get extra stuff for the same die size and power...which is what intel is doing here. gesher will be the tock phase - new architecture on the same process.
so intel is using that extra stuff to improve performance. they are not jumping back into the ghz race out of pure fear of barcelona. btw one day intel will ship a penryn with the same fsb and cache as a conroes you can find today. and they will perform very similarly - at probably lower power (and cheaper for intel to fab).
how barcelona and penryn will match up remains to be seen. i've heard all kinds of % increase numbers but of course have not seen any real benchmark results. and if barcelona does put amd back out in front, then intel has headroom for more cache, higher clock, more cores etc. that it can leverage to help out.
i know i've been saying "intel" and "they" but should probably say "we"...but i'm in the labs doing network software and nobody ever tells me anything. i learn about things from slashdot like everyone else.
Penryn is mostly just a die shrink. All things equal (clock, FSB, cache) it should not be any faster or slower than a Conroe.
Moving to 45nm gives you extra headroom for clock speed, extra transistor budget, etc. So they might just be demoing systems with similar power envelopes/cost/whatever.
Throw some SSE4 enabled apps in the mix and the Penryn would outperform an equalized Conroe by a fair margin.
Actually it looks like 36.5%, but that is exactly my point. If you want to boast, pick the best metric, not the worst.
The data I'm interested but I can't seem to find is the percentage of total processors. Is that 36.5% of performance produced by 10% of the total processors in the top 500 systems, 20%, 99.9%???
What? If you are going to use number of systems as your metric you might want to check it a little closer : 5% for PowerPC vs. 66% for Intel makes it look like a great decision.
Smithfield is two completely separate cores, one die, with some arbitration logic thrown in to get to the shared front side bus. No shared cache.
Presler is two cores, two dies, with something to get to the shared bus, i dont know what. Obviously separate cores and cache.
Yonah, two cores, one die, shared L2.
The reason the Intel dual cores dont perform as well as AMD on benchmarks is that all of Smithfield's IPC goes through that FSB. Will be the same for Presler. For Yonah, i dont know. Since the cores on Yonah share the L2, they will probably have some faster interconnect. maybe just the shared cache alone will automagically take care of it.
8 of which are special purpose. and it isnt shipping yet.
intel's ixp2800 has 17 cores, 16 of which are special purpose, and have 8 hardware threads each. and has been shipping for nearly 2 years.
now, there is no way an ixp will ever be used in a pc, but intel (and many other companys) have been shipping special purpose multicore chips for a long time
some posters above give the speculative fiction answer, but i read a long time ago (and i wish i could remember who wrote it) that SF was used by (serious?) science fiction writers to distance themselves from SciFi movies about giant brains attacking people.
one of the coolest things I read about banias/centrino was this IDF demo, where it was drawing only 7 watts while doing 30fps mpeg4 encoding, then dropping down to 1 watt when it was done.
Ok, I read everyone else's replies to your post, and nobody bothered to mention the best reason I can think of. I run my notebook at 1600x1200x24, but the lame projectors here at work only support 1024x768x16. So, on a really bad week, I'd be switching my resolution, colour depth several times a day. Of course, my notebook is stuck in win2k until i can figure out how to get away from outlook (etc.) so its not really a problem... yet.
Star Trek TNG used phase shifting, and sub space to fix almost every problem.
I wouldn't doubt it if the engineers behind QBM weren't sitting around bullshitting about that when one of them said (jokingly) maybe we could use subspace to get better memory bandwidth, or phase shifting, or...
Slashdot Mirror
I obviously didnt make my point very well. Of course if you took a conroe and bumped up its cache and fsb etc it would perform very similarly to a penryn. they are the same core (mostly) after all. this is mostly a die shrink / tick phase. with that you can get extra stuff for the same die size and power...which is what intel is doing here. gesher will be the tock phase - new architecture on the same process.
so intel is using that extra stuff to improve performance. they are not jumping back into the ghz race out of pure fear of barcelona. btw one day intel will ship a penryn with the same fsb and cache as a conroes you can find today. and they will perform very similarly - at probably lower power (and cheaper for intel to fab).
how barcelona and penryn will match up remains to be seen. i've heard all kinds of % increase numbers but of course have not seen any real benchmark results. and if barcelona does put amd back out in front, then intel has headroom for more cache, higher clock, more cores etc. that it can leverage to help out.
i know i've been saying "intel" and "they" but should probably say "we"...but i'm in the labs doing network software and nobody ever tells me anything. i learn about things from slashdot like everyone else.
Penryn is mostly just a die shrink. All things equal (clock, FSB, cache) it should not be any faster or slower than a Conroe.
Moving to 45nm gives you extra headroom for clock speed, extra transistor budget, etc. So they might just be demoing systems with similar power envelopes/cost/whatever.
Throw some SSE4 enabled apps in the mix and the Penryn would outperform an equalized Conroe by a fair margin.
as a ...whatever in Ratchet & Clank
Lombax
> Not a great story, but some people areally are gullable.
wow! the word gullable really isn't in the dictionary!
WHAT?
Actually it looks like 36.5%, but that is exactly my point. If you want to boast, pick the best metric, not the worst.
The data I'm interested but I can't seem to find is the percentage of total processors. Is that 36.5% of performance produced by 10% of the total processors in the top 500 systems, 20%, 99.9%???
What? If you are going to use number of systems as your metric you might want to check it a little closer : 5% for PowerPC vs. 66% for Intel makes it look like a great decision.
Unless things changed drastically since IDF:
Smithfield is two completely separate cores, one die, with some arbitration logic thrown in to get to the shared front side bus. No shared cache.
Presler is two cores, two dies, with something to get to the shared bus, i dont know what. Obviously separate cores and cache.
Yonah, two cores, one die, shared L2.
The reason the Intel dual cores dont perform as well as AMD on benchmarks is that all of Smithfield's IPC goes through that FSB. Will be the same for Presler. For Yonah, i dont know. Since the cores on Yonah share the L2, they will probably have some faster interconnect. maybe just the shared cache alone will automagically take care of it.
really?
8 of which are special purpose. and it isnt shipping yet.
intel's ixp2800 has 17 cores, 16 of which are special purpose, and have 8 hardware threads each. and has been shipping for nearly 2 years.
now, there is no way an ixp will ever be used in a pc, but intel (and many other companys) have been shipping special purpose multicore chips for a long time
some posters above give the speculative fiction answer, but i read a long time ago (and i wish i could remember who wrote it) that SF was used by (serious?) science fiction writers to distance themselves from SciFi movies about giant brains attacking people.
Everything else is just fanboy wankery
dont you mean wanklery?
Perhaps they plan to move that logic into a future mobile CPU chip.
yes they do. and into everything
We were trying to prove the possiblity of new course for the collage.
you know, cutting out pictures of chips and gluing them onto a piece of paper really isnt the same thing.
I typed all of this using ALT and the keypad!
one of the coolest things I read about banias/centrino was this IDF demo, where it was drawing only 7 watts while doing 30fps mpeg4 encoding, then dropping down to 1 watt when it was done.
I'm planning on getting my wife some hardware for valentine's....a stripper pole for the bedroom.
Ok, I read everyone else's replies to your post, and nobody bothered to mention the best reason I can think of. ... yet.
I run my notebook at 1600x1200x24, but the lame projectors here at work only support 1024x768x16. So, on a really bad week, I'd be switching my resolution, colour depth several times a day.
Of course, my notebook is stuck in win2k until i can figure out how to get away from outlook (etc.) so its not really a problem
where's your spoiler alert?!!?!?
not to mention that the ALU is running at 5.6GHz
Remember the good old days when we used to laugh at the people that thought they could get viruses (viri, whatever) from email?
I am pretty sure that stuffed penguins are not limited to any one local.
you might want to try ultraedit. not free, but supposedly well worth the price.
Star Trek TNG used phase shifting, and sub space to fix almost every problem.
...
I wouldn't doubt it if the engineers behind QBM weren't sitting around bullshitting about that when one of them said (jokingly) maybe we could use subspace to get better memory bandwidth, or phase shifting, or
Explorers