Hmmm the new P4s with hyperthreading should make Intel reconsider ...

[MrGrim]

... the way they name their models.

They will be 20% faster than an equally clocked Northwood right? I think they would be suckers if they didn't pull an AMD and market their 3.0GHz with hyperthreading P4 as a 3600+ P4.

(for those that don't get the meaning of my post, rating is bollox.)

 

[DaveSimmons]

Originally posted by: MrGrim
... the way they name their models.

They will be 20% faster than an equally clocked Northwood right? I think they would be suckers if they didn't pull an AMD and market their 3.0GHz with hyperthreading P4 as a 3600+ P4.

(for those that don't get the meaning of my post, rating is bollox.)

You do realize intel is planning to do this with their bannias p3-clone notebook chip don't you?

For Joe Ignorant Consumer who doesn't know or care about work-per-clock-cycle the PR rating is not bollocks it's a helpful pointer to the relative power of an AMD CPU.

<-- typing this on intel system, not a fanboy of intel or AMD.

 

[MrGrim]

You do realize intel is planning to do this with their bannias p3-clone notebook chip don't you?

Then they are are idios too.

For Joe Ignorant Consumer who doesn't know or care about work-per-clock-cycle the PR rating is not bollocks it's a helpful pointer to the relative power of an AMD CPU.

Sure. And perhaps when the Barton chip comes out they should keep the rating and have a 2.0GHz CPU marketed as 9999+.

 

[smadavid]

No, Intel is not planning to do it with banias. They are going to market the chip primarily based on power usage / battery time. I don't remember the exact details (this was a while ago), but they're still going to use MHz to indicate chip speed. Marketing the batter life of a chip is definitely not the same thing as using a performance rating instead of MHz to indicate speed.

 

[Lonyo]

What happens if AMD and Intel both do PR's based on a chip and then that chip stops being produced. IF Intel did it with new P4's then stopped making old ones and had 4000+, with AMD on, say 3800+ or whatever, what would it be compared to? It could really get screwy if AMD and Intel both did PR because then we'd have no base speed. That would confuse Joe Consumer even more.

 

[Utterman]

Originally posted by: MrGrim
... the way they name their models.

They will be 20% faster than an equally clocked Northwood right? I think they would be suckers if they didn't pull an AMD and market their 3.0GHz with hyperthreading P4 as a 3600+ P4.

(for those that don't get the meaning of my post, rating is bollox.)

Hyperthreading is done through software that support it. So a rating of 3600+ or whatever wouldn't be justified.

 

[CZroe]

Intel has CONSISTANTLY improved chip performance within the same generation without resorting to such tactics. Remember the Pentium when it got its MMX refresh? OBVIOUSLY software didn't take advantage of MMX immediately, so the chip was just designated "Pentium XXXMhz MMX"
Then the PII got a Procesor ID, and SSE and a minimum FSB of 100Mhz all of a sudden it was a "Pentium III," not a "Performance-Rated PII." It took some time for software to become optimized for SSE, but it has happened. Then the P4 came along with SSE2 and an entire architechture change which is obviously going to run older X86 code less efficiently (because nothing is optimized for it yet). Thanks to AMD pushing CPUs with old software in mind to astronomical clock speeds, the P4 has remained the CPU with unoptimized software and has remained clock-for-clok slower (There's is no magical "IPC" value that was "lowered" for higher P4 clock speeds). People claim software, such as benchmarks, that utilize SSE2 optimizations are "cheating" AMD just because they are moving to the platform that has a higher ceiling for the future. "MMX" benchmarks wern't cheating. I did read that Intel will "not emphasize" megahertz as the performance-selling factor for Banias (Which is highly optimized for Pentium Pro/PII/PIII code), but that does not mean they will invent a new speed measurement. Hyperthreading slows down many unoptimized applications, even many apps designed for dual CPUs! So it's not even considered the P4's MMX, much less something to "performance rate." It will simply be the Northwood-core "Pentium 4 HT." It will even have "HT" on the logo just like the "M" for the mobile part and "MMX" on the old Pentiums.

 

[CZroe]

By the way, I LOVE AMD regardless

 

[Mrburns2007]

Who gives a crap what they call it, it's all about performance. Since 99% of people look a CPU clock speed and think that's all there is to it. Then it's really the P4 that's full of it.

AMD should have done what cyrix did and just equal there clock speed to Intels.

 

[MrGrim]

Originally posted by: Utterman

Originally posted by: MrGrim
... the way they name their models.

They will be 20% faster than an equally clocked Northwood right? I think they would be suckers if they didn't pull an AMD and market their 3.0GHz with hyperthreading P4 as a 3600+ P4.

(for those that don't get the meaning of my post, rating is bollox.)

Hyperthreading is done through software that support it. So a rating of 3600+ or whatever wouldn't be justified.

Read the parenthesis.

 

[Macro2]

RE:"No, Intel is not planning to do it with banias. They are going to market the chip primarily based on power usage / battery time. I don't remember the exact details (this was a while ago), but they're still going to use MHz to indicate chip speed. Marketing the batter life of a chip is definitely not the same thing as using a performance rating instead of MHz to indicate speed."

If Intel tries to sell Banias without a PR rating it will go flopsky in the woods. The public only know MHz or equivalent numbers.
Intels own processors will eat Banias alive despite battery life. MH clearly sells, in a lot of ways. Traansmeta tried the low power routw and it's a penny stock now.

Save this post if you don't think I will be right.

 

[CZroe]

Transmeta Crusoe was also not X86, and on a totally different level of instrustions-per-clock ratios with built-in X86 emulation software. Banias will blow it away (Except for perhaps, the 1GHz OQO PC; I WANT IT!).

 

[XMan]

What's the big deal with hyperthreading? Don't current Xeons have it? I seem to remember reading an article that said it didn't offer much, if any, performance improvement.

 

[CZroe]

It doesn't. Just like real dual (SMP) CPUs, many applications actually run slower. In this case, it's a single CPU and it is a little more noticeable, but when applications USE the new abilities, this will change.

 

[bfonnes]

Originally posted by: MrGrim
... the way they name their models.

They will be 20% faster than an equally clocked Northwood right? I think they would be suckers if they didn't pull an AMD and market their 3.0GHz with hyperthreading P4 as a 3600+ P4.

(for those that don't get the meaning of my post, rating is bollox.)

This is a really bad idea... Encouraging Intel to lie? naughty, naughty

But, seriously. I like to keep up on current processors, and I really hate having to memorize 2 different speed ratings for the same processor. I mean it is cool that AMD processors are faster at a given MHz than Intel's, but why don't they just say that on their ad or on the box instead of making us memorize 2 numbers... Geez...

Bfonnes

 

[bfonnes]

People claim software, such as benchmarks, that utilize SSE2 optimizations are "cheating" AMD just because they are moving to the platform that has a higher ceiling for the future. "MMX" benchmarks wern't cheating.

I don't know anyone that has claimed this... If Intel really wants to cheat, why don't they just release a 1 Thz processor

Bfonnes

 

[mchammer187]

what about people that dont run win XP ? than no gain at all

 

[SexyK]

Originally posted by: CZroe
It doesn't. Just like real dual (SMP) CPUs, many applications actually run slower. In this case, it's a single CPU and it is a little more noticeable, but when applications USE the new abilities, this will change.

News flash :HardOCP review of P4 wilth HT

Hyperthreading gives performance gains accross the board. Read up.

Kramer

 

[Wingznut]

Originally posted by: CZroe
Hyperthreading slows down many unoptimized applications, even many apps designed for dual CPUs!

The HyperThreading on the P4-3.0ghz+ will have new instructions. I am not aware of any application that shows a negative performance impact with this "improved" set.

 

[cmdrdredd]

As was stated before...Mhz sells and Intel sells by name alone so Intel with 3Ghz CPU whether it's faster clock for clock than an AMD which is barely at 2.5Ghz is cash in Intel's pockets.