Intel 5x faster per watt.

[Intelia]

Paul said to day that the new Conroe chips will be 5x per watt faster. conroe will be 65 watts .I am not sure what smithfield is . will say 130 watts. that means Conroe will be 2 1/2 faster than smithfield. Now thats fast

http://www.theinquirer.net/?article=25619

 

[Furen]

Just remember that Intel also says that it outperforms AMD and makes it sound like it invented x86-64 and dual core. I'm not saying that Conroe will suck or that Otellini is lying, I'm just saying that Intel is not the best source for "unbiased" specs on it's CPUs and they're probably overstating the benefit.

 

[Pabster]

Let's just hope this isn't another "Pentium 4 and Netburst to reach 10GHz!!!" debacle

 

[Avalon]

Originally posted by: Intelia
Paul said to day that the new Conroe chips will be 5x per watt faster. conroe will be 65 watts .I am not sure what smithfield is . will say 130 watts. that means Conroe will be 2 1/2 faster than smithfield. Now thats fast

http://www.theinquirer.net/?article=25619

It's not going to be 2.5 times faster than a smithfield. It's just going to have significantly decreased power usage. I'm pretty sure it will be faster than smithfield, but nowhere near that sort of jump.

 

[Intelia]

If you take Pauls words at value ????? Than it will be 2 1/2 times faster. He clearly states Conroe will be 5x faster per watt. And merom would be 3x faster than Dothan per watt.

Now if intel can hold true on this its going to be very large improvement .5x faster per watt thats fantastic. Conroe only 65 watts and 2 1/2 faster than any present Intel uing 130 watts.

Now woodcrest at 80 watts has my attention I want to know way more about whats going on here.

 

[Vegitto]

Originally posted by: Intelia
Paul said to day that the new Conroe chips will be 5x per watt faster. conroe will be 65 watts .I am not sure what smithfield is . will say 130 watts. that means Conroe will be 2 1/2 faster than smithfield. Now thats fast

http://www.theinquirer.net/?article=25619

It just means that:

-If a Smithfield at 3.0 GHz uses 100 Watts, a Conroe at 3.0 GHz will use 20 Watts
OR
-If a Smithfield at 100 Watts can run (at the speed of) 3.0 GHz, a Conroe at 100 Watts could run (at the speed of) 15.0 GHz.

But I think it's the first one.

 

[bunnyfubbles]

5x performance per watt is intel's way of hiding the fact that it isn't that significant of a performance increase, certainly not across the board. They can't market the minimal performance increase, but they can market the power saving.

What sucks is that we'll have to wait for the 2nd half of 2006 before we can get our hands on any of this. Although I expected as much, Apple's keynote hinted at such a timeframe.

 

[Intelia]

Originally posted by: Vegitto

Originally posted by: Intelia
Paul said to day that the new Conroe chips will be 5x per watt faster. conroe will be 65 watts .I am not sure what smithfield is . will say 130 watts. that means Conroe will be 2 1/2 faster than smithfield. Now thats fast

http://www.theinquirer.net/?article=25619

It just means that:

-If a Smithfield at 3.0 GHz uses 100 Watts, a Conroe at 3.0 GHz will use 20 Watts
OR
-If a Smithfield at 100 Watts can run (at the speed of) 3.0 GHz, a Conroe at 100 Watts could run (at the speed of) 15.0 GHz.

But I think it's the first one.

NO clearly states performance perwatt. 5x faster per watt . Thats what it says nothing .Like how your tring to spin it.
Conroe uses 65 watts. not 20 not 100.

 

[Matt2]

I agree with Vegitto.

if it really was 5x faster per watt, then a Conroe at 100watts 3.0GHZ would be equal to a Smithfield at 15GHZ??

I don't think so.

Intelia, you gotta remember that Intel is marketing it's product to sell. You don't think Paul's putting a spin on things to talk up his new architecture?? I would be after the Prescott disaster.

 

[touchmyichi]

It makes sense to the average user to reduce power use, but this still isn't helping their position against AMD. Oh well, Intel can feel free to fall behind a decent amount just as long as AMD doesn't beat them by a huge margin.

 

[Bona Fide]

How about we all just wait and see the production samples before making any hasty judgements?

 

[Ronin]

Originally posted by: Furen
Just remember that Intel also says that it outperforms AMD and makes it sound like it invented x86-64 and dual core. I'm not saying that Conroe will suck or that Otellini is lying, I'm just saying that Intel is not the best source for "unbiased" specs on it's CPUs and they're probably overstating the benefit.

You do recall that Intel tried (tried is better than not doing it to begin with) to tape out dual core long before AMD ever did, and the Itanium was the 64bit corollary to AMD's? MS had the 64bit version of Windows for Intel before the AMD version was ever considered.

Don't get me wrong. I'm an AMD fan through and through, but I don't like it when misinformation is presented.

 

[Nanobaud]

That current Intel processors are designed for xx(?) watts and use xx watts in normal operation is an unfortunate and probably unintended 'feature'. Presumably they will avoid that next time, and a 65-Watt Conroe won't actually use 65 Watts under normal operating conditions. 5x probably comes from something like projecting a 25% improvement in execution and two cores running at half the total power. Just guessing though.

Edit: Note from the graphs shown, Conroe Performance/Watt is 5x over Northwood, not Smithfield.

nBd

 

[Furen]

Yes, you're correct that Intel was the first to look into dual core, but they didnt follow through and then they delivered Smithfield a week before AMD had scheduled the Opteron dual-core release, just to be able to say they're the first company that shipped x86 dual cores. And about microsoft's 64bit OS, did you mean the itanium version? I was talking about x86-64, which AMD most certainly developed (though I'm pretty sure microsoft had a hand in it as well).

 

[Pabster]

Originally posted by: Ronin
You do recall that Intel tried (tried is better than not doing it to begin with) to tape out dual core long before AMD ever did, and the Itanium was the 64bit corollary to AMD's? MS had the 64bit version of Windows for Intel before the AMD version was ever considered.

To be fair, they (MS) did release Windows 64 for Intel first. Thing is, it was for Itanic. That isn't exactly a desktop-class platform now, is it?

x86-64 brought 64-bit to the desktop and AMD was there first.

 

[Lonyo]

No one noticed that Intel said the Prescott core gave better performance per watt that the Northwood core?
Link
Now since at equal clocks, NW and PS are fairly much even, and the Prescott runs hotter, I don't honestly see the PS giving more performance per watt than a NW.

 

[Intelia]

Originally posted by: Matt2
I agree with Vegitto.

if it really was 5x faster per watt, then a Conroe at 100watts 3.0GHZ would be equal to a Smithfield at 15GHZ??

I don't think so.

Intelia, you gotta remember that Intel is marketing it's product to sell. You don't think Paul's putting a spin on things to talk up his new architecture?? I would be after the Prescott disaster.

You agree with vegitto this is what he said.
if it really was 5x faster per watt, then a Conroe at 100watts 3.0GHZ would be equal to a Smithfield at 15GHZ??

If you would read what it says 5x faster per what . So you guys take 5x 3 and come up with 15GHz LOL

 

[Intelia]

Now lets just go for the very smallest gain . Conroe SSE3 64 bit 65 watts if it is 2x faster than a dothan 2.26 I am going to be thrilled to death. Merom is stated to be 3x faster than dothan so Iam being really concervative

 

[Furen]

Well, merom does have TWO cores and is a 65nm die shrink...

 

[Intelia]

Originally posted by: Nanobaud
That current Intel processors are designed for xx(?) watts and use xx watts in normal operation is an unfortunate and probably unintended 'feature'. Presumably they will avoid that next time, and a 65-Watt Conroe won't actually use 65 Watts under normal operating conditions. 5x probably comes from something like projecting a 25% improvement in execution and two cores running at half the total power. Just guessing though.

Edit: Note from the graphs shown, Conroe Performance/Watt is 5x over Northwood, not Smithfield.

nBd

Thats why I went 2 1/2 to account for the dual core .Northwood better than P4P (single )
Springfield P4P dual core. 5x P4C =+ 2 1/2 P4P

 

[Pabster]

...As is Yonah.

 

[Lonyo]

Originally posted by: Intelia
Now lets just go for the very smallest gain . Conroe SSE3 64 bit 65 watts if it is 2x faster than a dothan 2.26 I am going to be thrilled to death. Merom is stated to be 3x faster than dothan so Iam being really concervative

No, you're being an idiot.
Mermon is NOT stated to be 3x faster than Dothan ANYWHERE except your head.
It's stated to have 3 times the performance per watt that Dothna does, and with a die shrink and dual cores, as Furen said, that's fairly reasonable.But it's not necessarily going to be 3x as fast.
Hell, it might be the exact same speed, but consume 1/3rd of the power, and Intel's numbers would still be true.

 

[Intelia]

Originally posted by: Furen
Well, merom does have TWO cores and is a 65nm die shrink...

Thats right so that would mean the Merom single core is about 50% faster than Dothan

 

[Intelia]

Originally posted by: Lonyo
No one noticed that Intel said the Prescott core gave better performance per watt that the Northwood core?
Link
Now since at equal clocks, NW and PS are fairly much even, and the Prescott runs hotter, I don't honestly see the PS giving more performance per watt than a NW.

Hay why don't you just burythe prescott everone else has

 

[FirNaTine]

Intel also may be cherry picking two points on performance per watt. One being the least efficient on the old core and one the most efficient on the new core. Remember how the heat and power requirements did not scale linearly with the processor speed? I think it is way too early to read into numbers like this until we know exactly what cores and what speed, on what bencmark, etc they are comparing. They may also be taking into account the benefits of increased cache on a very cache dependent (not real world) benchmark, the perfromance of newer ram, etc to get the full 5x.

edit: spelling