Prescott somewhat revealed...

[Luminescent]

This article here, at chip architect, contains very interesting and informative material on the potential prescott processor: its capabilities and enhancements over northwood, with evidence!

It seems Prescott includes the abilities of a 64-bit processor with x86-64 extensions, but won't put them to use (what a shame) until another revision of the core is made; hyperthreading in p4 was the same way.

 

[Saem]

x86-64 in prescott seems VERY unlikely. I don't believe x86-64 was finalised soon enough for this to happen, the one after Prescott (Teja) is another matter.

A large instruction TLB could be for more instructions in flight with a larger reorder window, the large trace cache could be just that, larger. We don't know the full extent of the hyper threading functionality in the current P4s and how much was exposed.

I'm still rather sceptical.

 

[Luminescent]

The article shows, in the features table, that 64-bit extensions will not be available in prescott. Through the block diagram, however, it is revealed that the required integer logic (extra set of simple alu's), register (extra set of RHAT registers, with double the amount of entries as Northwood), and L1 cache additions are in there. It also seems Intel will separate SSE2/3 units and legacy floating point parts.

 

[Saem]

Well currently SSE2 units are half speed, IIRC you can only issue on SSE2 instruction every other clock. The seperation was likely to increase the throughput -- this was predicted by Mr. DeMone of RWT fame.

 

[Luminescent]

With the SSE fmul and fmad in operation (each kicking in every two cycles), 4 32-bit packed results are available per clock.

 

[Tbred2GHz]

The immediate successor to Prescott after it tops out at 5.20GHz will be the "Tejas" core, also produced on a 90 nanometer process and delivering 5.60GHz using a 1066MHz system bus. That's slated to start appearing towards the end of 2004. Tejas will increase in steady increments which appear to be 6GHz, 6.40GHz, 6.80GHz, 7.20GHz, 7.60GHz, 7GHz, 8.40GHz, 8.80GHz and topping out at 9.20GHz. The first Nehalem is supposed to appear at 9.60GHz before Intel succeeds in its goal to produce a 10GHz+ chip, the Nehalem, and using a 1200MHz front side bus. "


The prescott pentium5 at 90nm and 3.4GHz with 1mb cache is looking great on paper. also at 250fsb itll be oced to 4.25GHz and being 90nm this might even be possible on air. If people are getting the 2.8c to around 3.5GHz on 130nm, imagine 90nm. I dont see why 4+ GHz would be a problem even with the first steppings. Later steppings will have people ocing 4GHz to 5GHz

 

[Saem]

The Netburst is DEFINATELY hitting it's stride.

 

[Tagrineth]

You know what's REALLY scary about that?

According to that, Tejas will appear at the end of 2004 at 5.6GHz.

That means we'll be going up TWO MORE GHZ by the end of 2004 with Prescott.

 

[MfA]

Doesnt really make much sense to speed up SSE mul&adds without going superscalar.

 

[Tbred2GHz]

You know what's REALLY scary about that?

According to that, Tejas will appear at the end of 2004 at 5.6GHz.

That means we'll be going up TWO MORE GHZ by the end of 2004 with Prescott.

well thats more than a year from now so it may not be unreasonable if Intel releases 200MHz faster cpus every 6 weeks or something. heck they might just jump by 400MHz now, 200MHz is whimpy by now

 

[Luminescent]

Very true MFA.

 

[Natoma]

Intel won't support x86-64 outside of the Yamhill core, and that's only if the Athlon64 takes off.

 

[Saem]

AFAIK, Yamhill wasn't x86-64, it was Intel's own take on how to do it.

Irrespective the Opteron hasn't taken off -- which is what every sane minded individual thought in the first place.

We'll have to wait a bit more till major OEMs start using it.

 

[Blade]

All I have to say is.. I hope that Canterwood boards will support Prescott like they were advertised to..

 

[Natoma]

AFAIK, Yamhill wasn't x86-64, it was Intel's own take on how to do it.

Irrespective the Opteron hasn't taken off -- which is what every sane minded individual thought in the first place.

We'll have to wait a bit more till major OEMs start using it.

Considering x86-64 was AMD/MS creation, there really is no "Intel's own take" on it. Yamhill is their core that supports x86-64 as specified by AMD/MS.

Opteron has only hit the channels in the last few weeks. Hell, motherboards aren't even really available at the moment. To say it hasn't taken off yet is an erroneous statement to make at this time. It's just come out. Give it a year. Also, that's just the server variant. We'll see how the Athlon 64 stacks up against the Northwood P4's when it's launched next month at consumer speeds of 2.0 - 2.4 Ghz.

 

[Natoma]

All I have to say is.. I hope that Canterwood boards will support Prescott like they were advertised to..

The only board that currently supports the voltage requirements of the Prescott are the 8KNXP and the 8KNXP Ultra, and that is because of the DPS-II, aka Dual Power System. DPS-II supports VRM 2.0 aka VRM 10.0. The Prescott CPUs require VRM 1.5, which means the 8KNXP and Ultra variants will be fine. However, the rest of the Canterwood boards only support VRM 1.0. So basically you're SOL if you've got one of those other boards.

Luckily I purchased the 8KNXP when I built my system in June.

 

[keegdsb]

All I have to say is.. I hope that Canterwood boards will support Prescott like they were advertised to..

The only board that currently supports the voltage requirements of the Prescott are the 8KNXP and the 8KNXP Ultra, and that is because of the DPS-II, aka Dual Power System. DPS-II supports VRM 2.0 aka VRM 10.0. The Prescott CPUs require VRM 1.5, which means the 8KNXP and Ultra variants will be fine. However, the rest of the Canterwood boards only support VRM 1.0. So basically you're SOL if you've got one of those other boards.

Luckily I purchased the 8KNXP when I built my system in June.

Not entirely true, I believe that Asus is making a couple of boards with supposed Prescott support.

 

[Blade]

Keeg: He said 'currently'.

So yeah, I'm screwed. I made pretty much the worst upgrade possible when I got my Asus P4C800 (875p) because it's not going to really support Prescott like it says in the freaking manual. Stupid Intel.

 

[keegdsb]

Keeg: He said 'currently'.

So yeah, I'm screwed. I made pretty much the worst upgrade possible when I got my Asus P4C800 (875p) because it's not going to really support Prescott like it says in the freaking manual. Stupid Intel.

So they're lying in their manuals and product pages ?

 

[Dave H]

We'll see how the Athlon 64 stacks up against the Northwood P4's when it's launched next month at consumer speeds of 2.0 - 2.4 Ghz.

Are you sure about this? I haven't been following the CPU market closely the last few months, but last I heard (and as best I can tell following a spot check at Xbit), Athlon 64 is said to be launching at 1.8 and 2.0 GHz, with few or no speed bumps planned until Q1 of next year. Which, incidentally, jibes very well with Opteron--a much lower volume chip--having just recently hit 2.0 Ghz (a month or two late). And further comports with AMD's difficulties getting the K7 to those sorts of clock speeds.

Frankly, if AMD had the capability to be fabbing chips in the 2.4 GHz range, they'd be doing it already. IMO. But who knows?--maybe they'll do a paper launch of a 2.4 GHz part so that they can claim the performance lead.