Intel Q3 and Q4 Roadmaps: 1066MHz FSB and Stack Protection Nearby Kristopher Kubicki on July 11, 2004 1:38 AM EST
- Posted in
Its been a while since we had a full blown Intel processor lineup, so we took the time to lay out what's new this week. Hopefully you took the chance to read up on our little primer on Intel's XD technology, which is essentially identical to AMD's NX technology. Unfortunately, this now means there will be two versions of Intel's 775 processors floating around (for 580 and below). The new processors are denoted by the "J" suffix, so when you're browsing for your next LGA775 CPU pay particular attention to the suffix to assure you are buying the "E" stepping, XD version of the processor.
|Intel Desktop Lineup LGA775|
Notice that Intel's next generation CPU is labeled as a 7xx series processor, traditionally reserved for the Israeli designed Dothan/Banias cores (Pentium M). Intel seems to have known confusion may abound since we noticed this peculiar disclaimer at the bottom of the roadmap.
Intel processor numbers are not a measure of performance. Processor numbers differentiate features within each processor family, not across different processor families.
We of course believe Intel since it is hard to imagine a 1.0GHz Ultra Low Voltage Dothan outperforming the 720J, but uninformed customers may find themselves in difficult positions to make purchases.
In other desktop CPU news, it looks like the Intel 560 chip is just starting to ship. This is unfortunate since it needs to be immediately phased out by next quarter for the 560J stepping. Typically we do not recommend waiting for new hardware releases for purchase decisions, but if you have your heart set on a LGA775 processor waiting until the next stepping is strongly advised.
|Intel Value Lineup LGA775|
|Intel Celeron M Lineup LGA775|
Since Dothan (Pentium/Celeron M) continues to push its way into the blade market, and already dominates the mobility market, we are keeping an eye on Intel's eventual plans to move the CPU platform to the desktop too. There are no new significant changes to Intel's value roadmap, but just like the desktop portion of the map, there are "E" steppings of the processor with the XD bit. These labeled with a "J" suffix.
|Intel Pentium M Lineup|
|Intel 755||2.00GHz||2MB||400MHz||Already Available|
|Intel 745||1.80GHz||2MB||400MHz||Already Available|
|Intel 735||1.70GHz||2MB||400MHz||Already Available|
|Intel 758J (LV)||1.50GHz||2MB||400MHz||Q1'05|
|Intel 738J (LV)||1.40GHz||2MB||400MHz||Q3'04|
|Intel 753J (ULV)||1.20GHz||2MB||400MHz||Q1'05|
|Intel 733J (ULV)||1.10GHz||2MB||400MHz||Q3'04|
|Intel 723J (ULV)||1.00GHz||2MB||400MHz||Q3'04|
The Pentium M roadmap appears the most convoluted of the batch. Even though we will see processors based on "E" stepping with the XD bit, there are still slated processors on the way which do not incorporate this security feature. It seems logical for Intel to not spend resources updating their older Socket 478 processors, but it does seem odd to not implement the security feature on newer processors. Stay tuned for more CPU and chipset roadmaps coming up this week!
Post Your CommentPlease log in or sign up to comment.
View All Comments
stephenbrooks - Monday, July 12, 2004 - linkSomething I heard elsewhere is that Intel uses two 'parallel' copies of this numbering system, one for the mobile chips and one for the desktops. So you see two copies of the Intel "350J" chip, which are actually completely different chips. I guess the only way to sensibly distinguish them is to call them "Mobile 350J" or "Desktop 350J". Why couldn't they have used the even-numbered centuries for the mobile ones*, or a simple prefix making 350J vs. M350J, etc.?
[* a possible reason for this is that Opteron already uses 2xx and 8xx]
One thing to bear in mind is that mathematical elegance is not high on the list of priorities of people in marketing (7 out of 10 probably can't do long division, but I didn't mean that-) with Intel in a weak position, perhaps having a confusing numbering system actually statistically helps them generate higher revenues than a system that allows transparent comparison of performance across families? Transparency only benefits the performance leader! Think about it...?
The other option is that they've just completely ballsed it up.
Anemone - Monday, July 12, 2004 - linkYep thumbs down on this dumb numbering scheme.
Interested in how the 3.76 @ 1066 fares, as it will probably be the final teller in the end of year system build for me. By then the NV4 and a 90nm A64 should be a fair comparison to this 3.76 chip (sorry I don't even want to USE that number scheme).
We'll just have to see how they do, but Intel is looking rather bleak atm.
mkruer - Monday, July 12, 2004 - linkPerhaps Intel should just use AMD PR rating system, at least that made sense. What they have now is nuts. Oh well I guess AMD will win the PR in the end, because we all know 1600+ is greater then 770J
sprockkets - Monday, July 12, 2004 - linkI guess you are correct, the EE had 512kL2 cache and then 2MB L3, good point but still dumb numbering scheme.
Pumpkinierre - Monday, July 12, 2004 - linkLaunch date on the Pentium-M 750J 1.86Gig is Q4 04 same as the 720J. 2x1.86 equals 3.73 (allowing for 3rd decimal).
KristopherKubicki - Sunday, July 11, 2004 - linkActually the 720J is not a P4EE. Its a completely different chip that hasnt been announced yet.
sprockkets - Sunday, July 11, 2004 - linkOh, yeah, I'm sure lots of people will know the difference between a 720J and a 730J. Wait the 720J is the super duper EE P4 Prescott chip and the 730J is the next speed level of that chip, oh S--- that's a completely different processor running at less than half the speed of the 720J yet I have the higher number.
At least you can get a nice Pentium M now for the desktop, but again, it's going to be so f-ing hillarious when you see a 720J and 730J next to each other for sale, and not even the informed people remembering which is which.
ranger203 - Sunday, July 11, 2004 - linkI think intel is shooting themselves in the foot with this numbering scheme.... some precessors i.e. 770, have such a high number "770" that future chips will start to be numbered funny, and what happens when they break 1000, they might as well go back to naming them by GHz