Intel vs. AMD - The Road Ahead

[Molman]

Thanks for that unrelated post, now tell me where in the article I was talking about the Server/Workstation Market? To quote myself:

I am only going to be talking about their desktop processors as that is the only concern we have in this article. The server/workstation and mobile segments are totally different issues altogether.

Don't be too hard, I think his post is quiet relevant in that the Multiprocessor market 'is' slowly moving to the desktop. Anyhow I thought I'd make this little post of what the current roadmaps, rumors and releases show for the Athlon and P4 this year. I've only looked at the top desktop CPU's coming out and ignored the other segments.

Current
AMD - AMD Athlon XP (Palomino) 2100+
Intel - Intel Pentium 4 (Northwood A) 2.2Ghz

Q2
AMD - AMD Athlon XP 2000+, 2200+ & 2400+ (Thoroughbred)
Intel - Intel Pentium 4 (Northwood B) 2.26 & 2.4Ghz

Q3
AMD - AMD Athlon (Barton) 2600+
Intel - Intel Pentium 4 2.5Ghz, 2.6Ghz (Northwood A), Intel Pentium 4 2.53Ghz, 2.66Ghz (Northwood B)

Q4
AMD - AMD Clawhammer (Hammer) 3400+
Intel - Intel Pentium 4 2.8Ghz (Northwood B)

AMD
Palomino - 0.18u, 266mhz FSB
Thoroughbred - 0.13u, 266mhz FSB
Barton - 0.13u, 266mhz FSB, 512 KB L2 cache
Hammer - Totally new. See links and discussion above

Intel
Northwood A - 0.13u, 400mhz FSB
Northwood B - 0.13u, 533mhz FSB

I don't know about about everyone else, but at no point do I really see AMD falling far behind Intel (if at all).

[Acid]

In the end it's all a matter of waiting for the benchmarks. Some people will agree with the article and some won't, but until we see some evidence, everything is just speculation. At this point, arguing is pointless.

[viditor]

"Charles, firstly the article was directed at the high-end market, not the value market which is why I focused totally on the AthlonXP and the P4"

Fair enough Asher, I was referring to your comment on page 5
"In the value segment, AMD plan to keep the Duron on the Palomino core till Quarter 3 of 2002 , when they will be releasing their Appaloosa core that not only shrinks the die to 0.13, but also bumps the FSB up to 133MHz (266MHz DDR)."
My point was that by the end of this year (about the time Intel's 533 MHz parts come out) the Athlon XP will BE the value market, and the Clawhammer will be what the P4 is competing against.

"Secondly, I didn't comment on the Hammer because as you said, it IS a whole new architecture and it is impossible to speculate on how it will perform, especially against Intel's 64-bit processor"

Again, fair enough...my point is that this is a common mistake today (IMHO) because while the Clawhammer will indeed be 64-bit capable, it's market is the high-end desktop. Itanium won't be it's major competitor (that's Sledgehammer's domain), the P4 will be...and while a bus and clock speed increase will help the P4 enourmously (as you pointed out), it shouldn't come close to what the Clawhammer's new architecture can achieve.
As an aside, Clawhammer is already sampling and being demonstrated. We should be seeing reviews as soon as the NDA's run out (a little birdie told me...<grin>).

Cheers,
Charles

[Anonymous]

Hammer should have to start off a 533mhz bus, which was reported a long time agoto make Intel's 400 look dated. dual channel ddr for the sledge hammer, not clawhammer.

[Anonymous]

Hammer should have to start off a 533mhz bus, which was reported a long time agoto make Intel's 400 look dated. dual channel ddr for the sledge hammer, not clawhammer.

. . .arrrh I don't believe Hammers bus has be specified (feel free to point me to something that says otherwise) and 533mhz P4's are due out soonish.

[Molman]

. . .arrrh I don't believe Hammers bus has be specified (feel free to point me to something that says otherwise) and 533mhz P4's are due out soonish.

. . .oops forgot to log in

[Anonymous]

true, 533 is not as big as a deal now, it's now even with the p4

where I read it was from long ago, but AMD is not using the old DEC bus anymore with hammer, but Hypertransport. Ok I agree that it doesn't mean a 533fsb, but it will be easier without it.

[Anonymous]

What would AMD have to do to make the Athlon run at a higher FSB frequency? What is involved engineering-wise? Is it possible to do something like the quad pumped bus of the P4?

[Oldbugger]

How can the article purport to look ahead and ignore Hammer......which is a highend desktop chip as much as a server....the Intel Itanium is relegated to servers cause of the cost and and its very architecture......but the Hammer will compete directly with the PIV.....thats what needs to be looked at if you are indeed looking forward......

[Mr.Tweak]

. . .oops forgot to log in

Just a matter of days now before the 533MHz FSB P4 is released... ;)

[Mr.Tweak]

What would AMD have to do to make the Athlon run at a higher FSB frequency? What is involved engineering-wise? Is it possible to do something like the quad pumped bus of the P4?

An entire new architecture I'd say.

[Oldbugger]

I certainly hope that AMD and Intel keep as close as they are....I would hate to see AMD jump ahead as much as I would hate to see Intel jump ahead too.....look at the benefits we have gotten over the last 2 years.....and our current crop of processors has a heap of application redundancy in them....they will last and do the job WE want for quite a while...and all this is because Intel and AMD have been competing.........

I hope it continues to a close contest.......up to the Athlon Tbirds there was a real convergence of technologies...all heading in the same direction and using the same technology.....

With the Athlon Tbirds/XPs and the PIVs the paths have diverged a little making that bit harder to get an apples to apples comparison......but who the F%$# cares.....my Tbirds will do JUST AS GOOD A JOB as a PIV.....labels don't matter...whether it is a PR 990000 ..or a bloody squillion gigahetz .......we NEED AMD and Intel to remain close in the "comparitive" performance stakes but we need an INDEPENDENT assessment of processor capabilities for that....

At the moment Intel is boasting about the revs it does while AMD says it can pull more weight.....true both of them....but its like the speeding driver who has to tell the judge by how much he broke the law.....you can't rely on either......

I was a little disappointed with Asher's article claiming to look forward but focussing on one aspect of processor performance....indeed the article reflect more the support the processor gets from the mobo and chipset that was the tone.....especially when you consider simple little things like the OS and drivers can make more difference to a machines' performance probably than raw grunt.......without much being evident.....

What we need is to look at how the differing architectures will enhance the processing capacities of the chips....like the long pipeline of the PIVs is good for graphics type stuff..especially with SSE2.... and how the onboard memory controller of the Hammer will boost the performance....I think the architectures are too different for a direct comparison without that comparison being on specific aspects of the processors capabilities....and how do you get an overall winner?????who knows..it depends on what you want to do with the processors as to which is most suitable......it really is that different already.......I just hope Intel and AMD don't wake up to that fact and keep trying to outdo each other....I'm gunna luv getting a XP 2gig+ chip for next to nothing.......

[Anonymous]

I think the article is flawed because it doesn't really mention Hammer, it just says that by the end of the year AMD will have 256 KB L2/266 MHZ fsb machines against dual channel DDR 533 MHz fsb P4's. Clawhammer is launching in Q4 and Sledgehammer is launching first half 2003. Clawhammer launches at 3400 +, so I'd say that by the end of the year AMD will be positioned at the front of the desktop market, unless Hammer misses its release (and current indications point to anything but).
In the low end segment AMD will have a cheap but high performance 512KB L2 Barton core getting pumped out at UMC vs. a crippled P4 Celeron. It's easy to see what will be the best in that segment.
In the high-end desktop market AMD should have a 3400+ Clawhammer produced on a 0.13 micron SOI process (with 3 more speed grades getting released within 1 quarter of its introduction) vs something like a 3GHz P4. I don't think anyone can deny that AMD's flagship will at least be competitive.
AMD will be untouchable in the server market. Sledgehammer will provide dual-channel DDR bandwidth for each CPU in configurations of up to 8, with high speed hypertransport connections between the CPUs. AMD will have full compatibility with all x86 software, will support 64 bit for addressing more than 4GB of memory, and should handily outperform Itanium. I'd take a guess that AMD will probably be significantly cheaper also, even if this isn't that important in this market.

[viditor]

"Just a matter of days now before the 533MHz FSB P4 is released... "

Huh? Did I miss a meeting? All news articles and reviews I've seen (including rumours...) put the release at September-ish (if nothing bad happens...). And if the launches of the past 2 years are any indication, Intel seems far more prone to "paper launches". This is why I don't expect to see any 533 MHz FSB parts until Xmas time (though I may be thinking a bit too cynical here...).
From THG, 25/2/02:
"We have tested the processors that Intel won't be bringing you until Fall of this year at the earliest. Most of all, we look at the Pentium 4/2666 and Pentium 4/2533, which are the first CPUs to offer support for 533 MHz Rambus"
That's why I was comparing it to the Clawhammer instead of the Athlon as they will probably both be on the shelves within a few months of one another. One last tidbit (educated guessing...), it is appearing more and more as if Microsoft will be supporting the Hammer series in both 32 and 64 bit modes...

From Amdmb.com
"Microsoft is going to announce on Wednesday a new Windows XP version for 64-bit operating systems and that it will have support for the AMD Hammer line of processors"
This is speculation on their part, but I tend to believe that MS will be making that announcement soon none the less...

http://www.amdmb.com/index.php#News-4284

Cheers,
Charles

[Mr.Tweak]

We've also been testing a P4 533MHz FSB processor, I can't say anymore though due to an NDA agreement. I may be wrong about the release date, time will tell.