DriveEuro

In Tosh's post earlier, I think he meant* to put "4ghz" and not 4gbs of ram...... Most people don't even need 1gb of ram..... 4gbs is just a ridiculous number....

__________________
[color=orange]ASUS P4P800 Deluxe [color=black]-[/color] 2.4C --> 3300mhz [color=black]-[/color] Mushkin PC3200 at 220 5-2-2-2 (cpu limited) [color=black]-[/color] BBAti 9800Pro at 430/375 [/color]
[color=black]3[/color][color=gray]D[/color]M[color=black]a[/color][color=gray]r[/color]k[color=yellow]01[/color] [color=black]3[/color][color=gray]D[/color]M[color=black]a[/color][color=gray]r[/color]k[color=yellow]03[/color] 1600*1200 1280*1024

ToshiroOC

Indeed not, I meant 4gbs And WaltC... I said "average consumer". The video editing chaps and desktop publishers are not average consumers, they are high end consumers and thus a different market than what I'm talking about.

Of course, all I'm trying to say is that Intel thinks HT is more important for the consumer than 64 bit processing, not that I agree with them.

Ah, ok, I misinterpreted what you said there then. Still, they need to be recompiled for 64 bit to get that boost, so not all programs will have that option, though hopefully many will.

I agree. Your point? I'm playing devils advocate here to a certain extent, just because I'm trying to say that Intel is going the route of HT, and AMD is going the route of 64bit...

Thanks, but I know what multitasking is I thought it a basic enough concept that I didn't have to differentiate between it and hyperthreading, though I guess I was wrong in assuming that.

Of course its a marketing phrase... but there is real technology behind it, regardless of whether 64 bit tech from AMD trumps that technology.

Quite easily, I don't Where did I say that x86-64 and HT would be mutually exclusive in Prescott?

I think that Intel simply won't go the way of x86-64 in Prescott at all, since enabling it mid-launch would be a big big deal and probably would entail a socket change, meaning it would be more profitable to market it as a separate chip from Prescott... and if its only Prescott+x86-64 not as many people will be interested in it, hence my belief that Yamhill won't surface in Prescott at all, but instead at the earliest in Tejas.

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

WaltC

Toshiro, all I'm trying to get across here is that these are not Itanium customers, but they need the extra ram in their businesses. And so for people like that for whom the price of Itanium is too high, they'll see x86-64 as a godsend for two primary reasons: it costs a lot less initially, and they can maintain the worth and the performance of their present x86-32 software. It's just not a good idea for Intel to try and tell people they don't "need" certain capabilities--because lots of people will think that they do. Not a smart marketing move by Intel, IMO.



And all I've been saying is that I know this is what Intel tells people it thinks--but in reality it's not really what Intel thinks... Intel wants to charge huge premiums for 64-bit computing. What AMD is doing by bringing it to the masses kind of upsets that applecart. Since you and I both agree that HT and 64-bit computing are not mutually exclusive, and that they are not the same thing, what does that mean when Intel answers a question about x86-64 with remarks about HT except that they are dodging the question entirely?

It would be the same as if you asked me about MHz clock rate and I said, "We think fpu performance is more important" or something like that. I would not have answered your question.

But recompiling is easy compared to writing native IA-64 software--easy as pie; but the great thing is that 32-bit aps don't need to be recompiled to run on A64 because they run on it natively! (Unlike with Itanium, which has to run them in emulation, which is very slow comparatively.)

OK, so what does HT do other than run multiple threads more efficiently in a P4 than they run with HT turned off? If you really want to run "two threads at once" without multitasking you need two physical cpus. (Two logical cpus are still the one single physical cpu.) All HT does is improve scheduling and execution of multiple thread multitasking within the single cpu--it's not the same as two cpus. Ergo: HT increases approximate IPC performance within the cpu when running multithreaded software (when the HT overhead doesn't exceed the benefit, that is--which is very unlike the way two physical cpus handle multiple threads simultaneously.) And that's why multithreaded performance on a dual-cpu rig blows a single HT-enabled cpu out of the water. Intel would like for people to believe they have two cpus--of course!... That's the marketing end of it. But the reality simply boils down to how fast it runs per clock on the one cpu people have with HT cpus.

You don't need A64 to do better than Intel HT--is the point I've been making. Even the AXP currently sold processes appreciably more data per clock than an HT-enabled P4 running multithreaded software. Clock the cpus the same MHz speed and run the performance tests and you'll see that without HT AXP is faster than a P4 with HT enabled. A64 just greatly expands that per-clock performance spread, even in 32-bit software.



Heh... Intel thinks nothing of socket changes... But anyway--let's use HT as an example. Many P4's shipped with HT circuitry in the cpu but disabled. It wasn't until Intel got yields where it wanted them that it even enabled the circuitry in the cpu at the high end. As yields got even better Intel began enabling them in slower P4 cpus. They didn't put that disabled HT circuitry in those cpus because they never planned to use HT one day, did they? Intel has stated the "disabled" YamHill circuitry is already present in Prescott. It's disabled, of course, so it won't ever work in the initial cpus, just as the disabled HT ciruictry is forever dead in those older P4s. But I have a feeling that if Intel sees that A64 is too popular, like magic, x86-64 will "come alive" in Prescott just as HT found life in the P4--eventually. And, maybe it won't hit until Tejas after all--but even so, if Intel goes x86-64 it will be as a direct result of the need to compete with A64/Opteron, IMO.

ToshiroOC

*shrug* I agree with you enough I don't feel the need to debate it anymore

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

@/2ct!<

Good, then no one has to move this to the flame forum eh

__________________

WaltC

Heh... It's probably not you I disagree with as much as it's Intel marketing I disagree with. I really don't like it when Intel puts out FUD on a competitor's product. It will probably backfire on them, though, as human nature being what it is the best way to get someone interested in something is to tell them they don't "need" that something... It immediately perks up people's curiosity, if nothing else.

Murdock

When Windows 3.0 came out Microsoft told us 512KB of memory is enough for desktop computers
1MB was too much,they where wrong as we all know now whe use 1GB now the 486 computer was the limit Intel says at the same time whe know better now 3.2 GHZ i have now ,128kb memory on the graphic card was enough , and now they say 64 bitcomputing is not neseserry,( read my lips) over a year or maybe 2
whe forgotten 32bits computers with our current configurations, whe shall laugh about it, thats my opinion about this issue


greetz
Murdock

Murdock

O B.t.w. if the car industry was so intensivly speedy developed as the computer whe all drove in cars with hydrogen, and not with this gas whe dive now ,and the global epolution was never happend.



Murdock

ToshiroOC

Perhaps the idea that there was more room for expansion with current day technology and current day markets for processors than cars would be worth consideraton?

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)