Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
I am almost certain that Hector Ruiz is a US citizen. But anyway...
What does the Xbox2 have to do with AMD?
I finally pulled the trigger.
Just brought home a new Sharp Actius AL27 Athlon64 notebook from Sam's Club. Boought it there mostly for the return policy than from exhaustive price comparison--though it seems to be cheaper there than elsewhere now that I look. Quiet, fast, great looking machine with a glossy extra-bright screen. 6.5 lbs, light for a DTR. This is going to replace my Athlon XP 2000 Shuttle SSF box as my main system.
And so, as I come home to test it out, I find AMD has shot up 1.3 points, and Dell is saying they're going to use AMD chips.
Ah, not a bad day.
"Moving up the sweetspot is the last you want if your invenories are below it."
That is an interesting observation. Are you saying that you think this is a reason why they cancelled 4ghz? More likely a silver lining to a dark cloud, given the damage to Intel's image.
begging for customers forgiveness
Barrett just settled the issue of whether or not they really could have produced a 4ghz chip as the PR stated.
They couldn't.
This was supposed to be the year of reckoning for Intel in
server MPUs.
Yeah, well, last year was supposed to be the Year of Itanium. Whatcha gonna do?
BTW, doesn't it cost a lot more to build a 2M cache part, and increase the die size?
They have to do something with all that fab capacity. The big costs are fixed.
BTW, I seem to be in the minority here, but judging by the overclockability of the P4, and BTX on the way, it seems to me that Intel's explanation is plausible.
Right now, the P4 platform really could be more improved by other things than clock boost.
"...talking as if they just slapped it together"
From the EETimes article:
According to some reports, Intel has been so rushed to make the multicore shift that one of its first dual-core server chips will not even sport arbitration logic. Instead it simply slaps two cores on one die, asking the existing bus-contention protocol to sort out the doubled workload.
That sure confirms what I was thinking.
Idle Speculation RE: multi-cores
I wonder if part of Intel's Xeon->Itanium transition plans might not involve a dual core processor, one of which would be IPF, the other EM64T. Not sure how feasable that is from the hardware or software side, but IPF needs some way of getting better x86 performance if it wants to play in the Xeon space.
"I don't blame them, it's about time. I would think it would be a tactic to lead AMD to follow...once again"
Umm... AMD is leading. How can Intel "lead them to follow" if AMD is not already following? This is true especially since Ruiz has outlined a path of divergence from Intel-imitation and responsiveness to customer demands.
Anyway, I doubt that either company learns all that much from IDF. I suspect they learn alot more from their customers--NDA or not--about what the other is doing. Let's say Intel is going to HP to pimp technology X under an NDA. If HP likes it, they will say "can your next generation chip do [all the things X can do]?" They don't have to specifically mention the Intel technology, but the insinuation would be pretty strong.
I remember a while back Sun described in an interview the Intel technology what would become EM64T--which surely had been disclosed under NDA. What's Intel going to do? Sue its customers?
Oh wait..
"It's not like it was creating new silicon; it's merely putting two processors in one package."
I know its an unnamed source. I know it might be a clueless marketing person. However, this combined with Intel's vagueness about it and lack of a roadmapped codename lead me to believe it was not a single die.
I'm not calling Intel a liar, it was dual-core and in the same package, I just doubt it was on the same die.
AMD desktops outsell Intel desktops 54% to 45%
http://slashdot.org/articles/04/09/13/2241256.shtml?tid=142&tid=118&tid=1
This would seem to suggest the dual-core demo was P4-based (because Dothan doesn't have Hyper-Threading):
"Bill Leszinske, director of digital home marketing at Intel, told reporters Wednesday that he did not know if Intel would phase out Hyper-Threading after the dual-core strategy was implemented. For now, it appears that Intel's dual cores could have both."
http://www.eweek.com/article2/0,1759,1644477,00.asp
I would be really surprised it they had a demonstrable dual-core Dothan right now.
I_Banker--Great analogies. EOM
I'd argue that AMD64 is exactly the kind of inflection point you're talking about. (The fact that to you, it is "seemingly inconsequential" only strengthens the case.)
No AMD64, then no EM64T. No EM64T, then everyone would have to follow Barret's yellow brick 64bit roadmap to Itanium, first servers and then everyone else.
Among other things, AMD singlehandedly saved Xeon from extinction. Intel fans should rejoice.
The least you can say is that it has had a much greater effect on Intel's plans than the original Athlon. Whether you want to qualify that as an "inflection point" is a matter of degree. Suffice it to say, things would be different without AMD64.
"...A64 represents some sort of inflection point and that Intel is very likely to be damaged severly in just the next few months to a year."
That's it exactly. The debatable question is how much Intel will be damaged and how effectively they can/will come back.
Judging by Intel's practice of late, a product introduction is not logically inconsistent with a one quarter delay in availability. Both rumors can be true...
And, BTW Sgolds, best of luck in your new business. I've gotten alot from your participation in these forums.
Dual-core cache for Opteron and Itanium being discussed on Slashdot:
http://slashdot.org/articles/04/08/26/2150207.shtml?tid=126&tid=142&tid=118&tid=1
A few interesting posts here, but I'm no techie.
WBMW, Interesting, thanks. EOM
Understood, but wouldn't leakage increase the harder the processor is worked? You seemed to imply that it was a constant.
Hmm... wouldn't leakage be a function of how much power is running through the chip (and maybe other variables as well)?
I remember reading that the early Hammer chips were not very overclockable despite the low temp/power figures. It supposedly had something to do with the SOI process. Don't know if that was bogus or not, but it would make it more difficult to infer headroom from a given SOI chip.
Where did you hear the rumors about the k9 design?
I remember reading an interview with Dirk Meyer around the time that p4 got HyperThreading, and he flat-out said that SMT isn't worth the trouble.
bit vs. byte
"A gigabit, which should not be confused with gigabyte, is 1/8th of a gigabyte. . ."
http://en.wikipedia.org/wiki/Gigabyte
"I'm wondering who in tha HELL buys all these A64?"
Skimming your post very quickly, I thought you had written "DELL buys all these A64?".
After years of following AMD, I guess my nervous system is now finely honed to pick up on Dell rumors.
RE: HP AMD Promotion
I was pleasantly surprised by an HP back-to-school insert in the Sunday paper--it featured all AMD computer systems of various configurations. 8 pages total.
I know AMD dollars are at least partially behind it, but it is nice after hearing of the Intel-exclusive HP back to school brochure.
This was in the Los Angeles area, not sure if it was run nationwide.
"There are two flavors of mobile A64s. One has 1 meg. of L2 cache and is rated at a TDP of 57-65 watts. The other is a low power version that has 512k and a TDP of 35 watts."
Hmm... I hadn't given this a whole lot of thought, but, given two chips with similar performance, shouldn't a big cache chip use less power power than a small cache chip? (i.e. because the clock on the big cache chip would be lower, and the cache doesn't take much juice)
RE: Ed Stroglio
That boy ain't right.
This might provide a little desktop pc demand boost...
Doom 3 ready for production.
http://games.slashdot.org/games/04/07/14/1746228.shtml?tid=127&tid=186&tid=204&tid=206
http://www.shacknews.com/finger/?fid=toddh@idsoftware.com
...ASP are going to head south in dramatic fashion.
And that's another reason that the decision to make 32bit K8 is so perplexing. That is, unless AMD is about to announce a foundry deal.
I would think that fab capacity is still a big issue for Dell. Maybe that changes if/when IBM starts fabbing for AMD
I thought this was interesting from the A64 cooler comparison article:
We're still checking with AMD to see why a .2v voltage jump produced such high temperatures. While voltage increases do increase temperatures they normally don't do so by such large amounts. Early information indicates this is a side-effect of building CPU's using SOI.
http://www.sudhian.com/showdocs.cfm?aid=536&pid=2013
Don't worry about price. If Intel starts churning these out on 65 with all that capacity, the price will come down to levels the market is accustomed to paying. Dothan prices will drop as production increases and after they've juiced all they can out of the early adopters and fanboys.
My point is that it is a good idea that will find an eager market when it ramps up and prices drop.
I know AMD plans dual-core Opteron, but without contracting out for fab space, they cannot dual-core their desktop chips for some time to come. But I don't think that will matter in the next 2-3 years...
I'll go on the record to say that it's absurd to believe Intel and analyst assertions that these dual-core chips will be generally available in 2005. 2006 or 7 more like it. Unless this has been a secret project for a long time (which I doubt), then there's a massive amount of work to be done, as current iAMD64 implementation is for NetBurst architecture--that will have to get designed from scratch--all dual core talk was about NetBurst--that will also have to get designed from scratch. Really, much more work than going from say k7 to k8. I wouldn't be surprised to see a single-core version of Dothan64 come out first to satisfy the market until they get all the kinks out of dual-core. And even that is more than a year away at best (worst?).
I hope so.
It's clear that a dual-core dothan-based 64bit chip would be a very compelling product. The question is how long will it be before we see it, and will it still be so compelling then. If all this was just cooked up as a reaction to the failure of Precott--and that's what it looks like--then it will be quite a while before we see volume product. Expect the interim performance increases to come from increasing the cache size and enabling 64 bit on Prescott chips along with the odd speed bump. AMD should be able to compete very well in that scenario.
On the other hand, Barrett has been good at following through on bold new initiatives, it's just that most of them have been collossal mistakes. (I'm chuckling now as I imagine what an unweildly monster the Itanium-compatible Tejas would have been.) But this new push to dual-core is definitely the right way to go for Intel, and it hits AMD right where they can't fight back--fab space.
Not that I think it's wise to read too much into these kind of comments, but good thermal and power characteristics sort of imply good binsplits, don't they? And I don't recall AMD ever commenting directly on binsplits in public statements.
IMO, An industry-wide TPI will never happen for alot of reasons, mostly that it isn't in the interest of whichever company has fallen behind. Intel, especially, likes to control benchmark tests and would not officially endorse a benchmark unless it could have *at least* the sort of control it (is speculated to) have had on some others. For them, it would almost be like outsourcing its PR to AMD and vice versa.
Hell, I don't even think AMD and IBM could agree on a TPI scheme and they're not direct competitors.
The one that strikes me as off is that there's a 64-bit NT kernel that runs on something other than x86 or IPF.
Well, if it's true that it's going to be Power based, then they'd better port a kernel for it! The weirder thing is that they've chosen Power at all, given that it breaks backwards compatibility with Xbox1 and makes the software side more complicated. IBM's deal must have been really sweet, or else MS has something else up its sleeve. The least you can say is that this, along with their support of AMD64, shows that they want to ensure a diverse CPU landscape.
Lately the p4 is starting to look like the souped-up Hondas that I see alot on the road. You know, instead of buying a nice sports car to begin with, you can pour thousands into your Honda Accord for lowering, rims, nitrous, etc. in a futile effort to make it do and look like more than it is. E.g. HyperThreading was the equivalent of a large decorative spoiler--lots of money, lots of hype, negligable improvement. Prescott is the end result of this "design philosophy."
Pentium M is a winner, though.
Of course they could start stacking the chips on top of each other... remember that rumor?
You say that like it's not an obvious truism. Are you hinting at something?
Having it in hardware does no good without support from the software and OS.
You keep saying this, like some kind of self-help mantra. Never mind the 386. Never mind that people buy things all the time because of features they will never use. And, among the more saavy and sober buyers, assurances of future support from leading software companies is enough to give value to the 64bit capability. It's silly to say that 64bit has zero value until the day Win64 comes out, then suddenly it will worth something. For alot of people, it's enough to know that Win64 will come out, many don't even need that.
And, has been noted many times, the server side has had good 64 support for a while now...