Pentium 4 2GHz up for sale

[Guru]

http://www.theinquirer.net/03080101.htm

[rylan]

Oh cool.. so that means it performs a the speed of an Athlon 1.4GHz while only costing 2x as much? :P

[Kosh Naranek]

RYLAN
Why did you post that ?
Is it because you don't like Intel or is it becase you don't want people to choose on their own .... some like Intel and some like AMD ?

So why did you post that ?

Is what you posted true ?

If it's not true, then why did you post it ?

[Marshmallowman]

sounds accurate,...its a question of dollars and "sense"

[Belwarrior]

Unfortunately what Rylan says is true, P4 is crippled by Intel (I am not a very technical person) but the original design had P4 with 4 pipelines (or something like that) but what is rolling out of the factory only has 2.
I think this was a ploy by Intel to make AMD come out with their Athlon4 earlier so that the P4 with 4 piplines will perform better. It has backfired as now the current P4 cant even compete with the AMD Thunderbird (case in point : AMD 1.4GHz performing better than P4 1.7GHz)
I was using Intel processors before and have no complaints on them (my P1 100MHz did 24/7), However in the light of what Intel is doing nowadays (they must realise sooner that they must fight to keep their market share, they are not the only lion in the arena) I am using AMD for the simple reasons of $$$ and performance. AMD processors give the best performance/cost ratio.
The Rambus tie-in is a also major flaw by Intel.

So for now my vote goes to AMD.
Hope Intel doesnt cripple its processors anymore, thought they would change after they restricted celerons.

[Greebe]

I would think that the "Day of the OCer" has come into it's own and therefore would be the smartest thing for Intel to unlock the multiplier on all new chips.

Hopefully that would breath a little life back into them

[Wombat]

What rylan said is true. And I try to tell it to everyone I know. The P4 exists because to the uneducated computer buyer:

<b>1.</B> Pentium4 is a bigger number than Pentium3

<b>2.</B> More MHz is faster, right?

So intel is making money from the buyers' naivity. The P4, as it was released, is a suck processor. It can be helped somewhat if the program is compiled specifically for the P4, but that won't help the programs you already have, or most of what you will have.

Maybe when you can get P4's with DDR SDRAM that will help too.

We have a saying here in the lab: "You can always get bandwidth if you're willing to pay for it, but not even God can give you better latency." RDRAM's latency sucks, and it will lose the bandwidth advantage it has when you start to see more multi-channel DDR memory buses.

[Kosh Naranek]

MultiChannel DDR ..... doesn't that give more bandwidth in itself ?

What's that got to do with latency ? If latency is your GOD then you should love the P4 ... becuase it's L1 and L2 cache have the lowest latency int the x86 CPU market, by far

Don't you think there will be multichannel RDRAM as well ?

I can't wait to see what happens when 2 Ghz + P4 rev. 2 ~ 0.13um and DDR boards hit the market !

If the current stories are true ( about the 2Ghz P4 being reduced 50 % in price in late August ) then suddenly the $$$$ advantage is gone and then what.

Read my previous post in another thread regarding the differences between the P3 and P4 and you will see that the P4 is a VERY advanced CPU with a lot of room for improvement.

From my point of view, *ntel is in a better position than AMD to realease faster CPU's, why ? because it is a lot easier to change the multiplier on a CPU than to redesign the Core of your existing CPU's.

And don't forget .. the current P4 is only rev 1.0, meaning no one knows the true potential of the PIV other than *ntel.

[Wombat]

<blockquote>What's that got to do with latency ? If latency is your GOD then you should love the P4 ... becuase it's L1 and L2 cache have the lowest latency int the x86 CPU market, by far </blockquote>

Yes, MultiChannel DDR will give more bandwidth. That's my point. RDRAM currently has more bandwidth, but the latency is absolutely HORRENDOUS. The serial design it uses takes an eternity to start sending data to the CPU. Adding more bandwidth won't help that. Maybe the P4 will get better when it doesn't have a suck architecture to run on, but right now...ick.


<blockquote> and you will see that the P4 is a VERY advanced CPU with a lot of room for improvement. </blockquote>

Yep, the P4 has tons of room for improvement. And tons and tons is needed. It's a poor design (or at least a poor implementation, the original design had potential, but Intel's a fab, not a design house).

<blockquote>From my point of view, *ntel is in a better position than AMD to realease faster CPU's, why ? because it is a lot easier to change the multiplier on a CPU than to redesign the Core of your existing CPU's. </blockquote>

Okay, that's not necessarily true. The P4 has a LOT of latches, and a huge pipeline (Athlons have quite the pipeline too, but <I>really</I> impressive branch prediction. You get to a point where you do have to redo the CPU to get it to go faster. Besides, the K7 is a really fast core already. It's too bad it has to have that clunky x86 front end, or it would probably outrun a G4 (there was a good article on ARS a while back). But I have friends who worked on K8, and some of that tech is really cool.
And I bet BOTH companies are redesigning. That's the nature of the business. It takes years to do a process port, and a couple times longer to do a CPU from white paper.

<blockquote>And don't forget .. the current P4 is only rev 1.0, meaning no one knows the true potential of the PIV other than Intel.</blockquote>

How do you know it's rev 1.0? Maybe 1.0 <I>really</I> sucked. And I design processors for what company? And they're in a partnership with who on IA-64? Gee, I wonder what processors Wombat is working on or knows about?

[Kosh Naranek]

A huge pipeline yes .... but the PIV is NOT dependant on the numbers of x86 decoders meaning ..... the instructions stored in the trace cache are stored as how the expected program flow will be ... loops ~ extrem speed ... and they can be executed over and over again without the need to be traced every time.
Besides, the contents of the trace cache is not storing instructions like normal x85 CPU's but addresses on instructions ..low latency !

I remember when the Pentium Pro came out .... what a shocker !
It sucked in running 16-bit apps... but when the PII came out ... WOW ,total surprise !

How do you know it's rev 1.0? Maybe 1.0 really sucked. And I design processors for what company? And they're in a partnership with who on IA-64? Gee, I wonder what processors Wombat is working on or knows about?

Well ...it's the First official PIV .... so to me it's rev. 1.0

Stop with the attitude .........................................no need for that !

Yes, MultiChannel DDR will give more bandwidth. That's my point. RDRAM currently has more bandwidth, but the latency is absolutely HORRENDOUS. The serial design it uses takes an eternity to start sending data to the CPU. Adding more bandwidth won't help that. Maybe the P4 will get better when it doesn't have a suck architecture to run on, but right now...ick.

Work is being done to reduce the Latency of RDRAM by a factor of 0.5 in which case it will look good !
Remember that a ''simple'' process shift down to 0.13um will give RDRAM running at 1+ Ghz while it takes a lot more work to get MultiChannel DDR-Ram