So basically Intel won't have a competitive XeonMP for a while?

D

Duke3d87

Limp Gawd
Joined
Jul 6, 2004
Messages
430
From the way it seems, Intel has been talking a lot about dual core processors and the whole Merom lines as if it they are Intel's saviour. But what about XeonMP? If i am not mistaken the only strong product that Intel will have to fight back with the Tulsa and then Tigerton. Am i mistaken? AMD is aiming to caputre 50% of the 4P server market and imo, the integrated memory controller makes it easier for AMD to do so over Intel...Thoughts?
 
Woodcrest?

As I understand it there haven't been that much difference between Xeons and Pentiums up 'till now, but maybe I've been spoiled by the way Opterons are just basically Athlons that have more OC potential...
 
Duke3d87 said:
From the way it seems, Intel has been talking a lot about dual core processors and the whole Merom lines as if it they are Intel's saviour. But what about XeonMP? If i am not mistaken the only strong product that Intel will have to fight back with the Tulsa and then Tigerton. Am i mistaken? AMD is aiming to caputre 50% of the 4P server market and imo, the integrated memory controller makes it easier for AMD to do so over Intel...Thoughts?
Click to expand...

I suggest you look at IBM 366 , Unisys ES7000 , DELL 6650 benchmarks versus comparable Opteron and you are in for a big surprise.

There is a fair number of benchmarks were Xeon MP Paxville stomps Opteron and it can scale up to 32 way ( 64 cores) while Opteron is limited to 4 way ( 8 cores ) - 8 way is possible but scaling is poor.

So Intel has a competitive Xeon MP now ....only problem is power burned...:)
 
I know we've talked about this and whatnot, but if they are so good, then why is Intel always looked down upon for not having the memory controller and it's apparent lack of scability?

Also why don't they make a Merom based XeonMP? I mean if the whole point of using Cloverton in the future is for permance and low power consumptium would not a 40 watt XeonMP be good? If you can cool a paxville in a blade, quad Meroms would be relatively easy.
 
Duke3d87 said:
I know we've talked about this and whatnot, but if they are so good, then why is Intel always looked down upon for not having the memory controller and it's apparent lack of scability?

Also why don't they make a Merom based XeonMP? I mean if the whole point of using Cloverton in the future is for permance and low power consumptium would not a 40 watt XeonMP be good? If you can cool a paxville in a blade, quad Meroms would be relatively easy.
Click to expand...
I think they need more time to validate with the Xeon MP line, this line seems typically behind the newest stuff, Tigerton is behind Tulsa which itself has yet to be released.

Tigerton is Core Architecture based, it will jsut take sometime to get here is all, look how slow it is taking Intel in it's enterprise line. Montencito is based on a 90nm process.

The Xeon DP line can refresh faster hence you have Woodcrest not too far likely from the desktop launch of Conroe.
 
coldpower27 said:
I think they need more time to validate with the Xeon MP line, this line seems typically behind the newest stuff, Tigerton is behind Tulsa which itself has yet to be released.

Tigerton is Core Architecture based, it will jsut take sometime to get here is all, look how slow it is taking Intel in it's enterprise line. Montencito is based on a 90nm process.

The Xeon DP line can refresh faster hence you have Woodcrest not too far likely from the desktop launch of Conroe.
Click to expand...

Actually , last I've heard is that Woodcrest will be 1st and Conroe will appear after it.

Kinda makes sense , the biggest problems are in the 2S/4S space...not desktops.
 
coldpower27 said:
I think they need more time to validate with the Xeon MP line, this line seems typically behind the newest stuff, Tigerton is behind Tulsa which itself has yet to be released.

Tigerton is Core Architecture based, it will jsut take sometime to get here is all, look how slow it is taking Intel in it's enterprise line. Montencito is based on a 90nm process.

The Xeon DP line can refresh faster hence you have Woodcrest not too far likely from the desktop launch of Conroe.
Click to expand...

What do they do when they validate? I've been wonderng what they do that makes the validtation take so long. And with Montecito, they've had a lot of problems to deal with other then the die size.

savantu said:
Actually , last I've heard is that Woodcrest will be 1st and Conroe will appear after it.

Kinda makes sense , the biggest problems are in the 2S/4S space...not desktops.
Click to expand...
Woodcrest will only be 2P so that still leaves 4P for Tulsa and Tigerton.
 
Duke3d87 said:
What do they do when they validate? I've been wonderng what they do that makes the validtation take so long. And with Montecito, they've had a lot of problems to deal with other then the die size.
Click to expand...

Validation means testing to make sure EVERYTHING works in the highest possible stressed environments. It is entirely timebound and takes much longer than desktop validation, and hence its expense.
 
Yes AMD has made up some huge ground on Intel in the desktop and a huge step in the server market. But in the server market, everything has to work. Stress levels etc... are better than there desktop counterparts. Testing and vaildation takes longer. When a desktop dies, 1 user is affected. When 1 server dies who knows how many user are or could be affected.
 
You must log in or register to reply here.
Back
Top