Shanghai @2,7 to buldozer anything in virtualization

[duploxxx]

Shangai @ 2,7 ghz VMmark scores are out, kills any living Intel system out there.

shangai @2,7ghz 2p 8core score : 11,22
http://www.vmware.com/files/pdf/vmma...11-12-R805.pdf

harpertown @ 3,33ghz 2p 8core score : 9,15
http://www.vmware.com/files/pdf/vmma...0-09-ML370.pdf


shangai @2,7ghz 4p 16core score : 20,35
http://www.vmware.com/files/pdf/vmma...11-12-R905.pdf

tigertown @2,93 4p 16cores score: 14,14
http://www.vmware.com/files/pdf/vmmark/vmmark_hp5.pdf

Dunnington @2,66 4p 24cores score 19,10
http://www.vmware.com/files/pdf/vmmark/vmmark_ibm4.pdf

fools @ intel, why start with a market leading desktop platform why you know you get killed in the server part. It will take 5-6 months to get back in 2p and 9-10 months in 4p and higher.

[donitsi]

You could say the reverse about AMD... from server to desktop

[Macadamia]

Well, Intel can shun Virtualization, Cloud Computing, and Netbo-oops.

Meh, regardless of how the competition performs, Shanghai is extremely solid as a drop in upgrade or even a new purchase to extend F farms.

[zanzabar]

just wait for the intel parts without FB ram the i7 will take all the performance crowns back this month for 1/2/4P, but this is a huge improvement if the am2+ or am3 dose 3.6GHZ i might get one

[Macadamia]

just wait for the intel parts without FB ram the i7 will take all the performance crowns back this month for 1/2/4P, but this is a huge improvement if the am2+ or am3 dose 3.6GHZ i might get one

4P Nehalem (Beckton) is not going to be out until H2 or even late 2009.

And it uses... OH SNAP, FB-DDR3.

[donitsi]

Have to agree with Duploxx here... Shanghai platform is the clear choice if virtualization were your thing

[zanzabar]

4P Nehalem (Beckton) is not going to be out until H2 or even late 2009.

And it uses... OH SNAP, FB-DDR3.

i had read tri channel ECCddr3 with a raid 5 like option, and IBM had said to be sampling them to data centers in q4

[Shintai]

With FB-DIMMs you can mix DDR2 and DDR3. The memory controller only see´s the AMB chip.

And I think you refer to DDDC. Double device data correction. So you can actually have 2 DIMM failures instead of 1. And ofcourse the usual hotplug memory etc.

[Sly Fox]

You could say the reverse about AMD... from server to desktop

The moneys in server-side sales these days.

[JumpingJack]

You could say the reverse about AMD... from server to desktop

I don't think it really matters which way the release goes, it depends on where in the validation process the respective product goes. In 2006, Intel launched Woodcrest first (server) then Desktop, for Barcelona AMD launched Desktop first then (server).... yada yada.

The launch sequence is dependent upon many things.

[Jacky]

I don't think it really matters which way the release goes, it depends on where in the validation process the respective product goes. In 2006, Intel launched Woodcrest first (server) then Desktop, for Barcelona AMD launched Desktop first then (server).... yada yada.

The launch sequence is dependent upon many things.

What do you think?
Is HT that critical or does Intel simply take more time for validation of a new platform to be on the safe side? Or does it also depend on yields, inventories, etc?

[Bellisimo]

What do you think?
Is HT that critical or does Intel simply take more time for validation of a new platform to be on the safe side? Or does it also depend on yields, inventories, etc?

shanghai is easy to validate, just put a nonproven cpu on a proven platform, the cpu is just a K10 with some enhancements
with nehalem you have to validate an entire new architecture, on an entire new platform, more difficult

[duploxxx]

I don't think it really matters which way the release goes, it depends on where in the validation process the respective product goes. In 2006, Intel launched Woodcrest first (server) then Desktop, for Barcelona AMD launched Desktop first then (server).... yada yada.

The launch sequence is dependent upon many things.

intel woodcrest was indeed launched before conroe, but conroe was available in volume before woodcrest

sure about those amd parts?

barcelona launch sept 07
agena launch nov 07

[donitsi]

I don't think it really matters which way the release goes, it depends on where in the validation process the respective product goes. In 2006, Intel launched Woodcrest first (server) then Desktop, for Barcelona AMD launched Desktop first then (server).... yada yada.

The launch sequence is dependent upon many things.

I wasn't talking about release orders.

I meant that for example original opterons and athlon 64's were the same processor... They designed it to be server proc and with minor modifications made the desktop proc
Likewise current phenoms are based on server architecture

It simplifies things but I would like to see complitely different architectures for server and dekstop

So Duploxxx first said that Core2 was based on desktop design and then with modifications they made the server version.
That's why countered it with my post

[Helmore]

I wasn't talking about release orders.

I meant that for example original opterons and athlon 64's were the same processor... They designed it to be server proc and with minor modifications made the desktop proc
Likewise current phenoms are based on server architecture

It simplifies things but I would like to see complitely different architectures for server and dekstop

So Duploxxx first said that Core2 was based on Laptop/notebook design and then with modifications they made the server version.
That's why countered it with my post

It needed a small correction.

[donitsi]

Heh, that will do

[JumpingJack]

intel woodcrest was indeed launched before conroe, but conroe was available in volume before woodcrest

sure about those amd parts?

barcelona launch sept 07
agena launch nov 07

Barcelona did not appear to system vendors until Q2 08 ... AMD indeed typically launches into server first, but the Barcelona launch is an illustration as to why DT will precede server as necessary.

Server requires much more rigorous validation due to the mission cirtical nature of the product. DT users can live with a lockup or two over time, they will simply curse MS for a cruddy OS ... enterprise solutions, lockups (downtime) means money ... so server has to be cut to a much higher standard.

Technically, Barcelona launched in Sept. true, but the TLB errata pulled it back for the very reason I mention above. Agena, the TLB errata is no issue for DT, for server it is not.

[JumpingJack]

What do you think?
Is HT that critical or does Intel simply take more time for validation of a new platform to be on the safe side? Or does it also depend on yields, inventories, etc?

See me response to Duploxx above as one line of logic, but if I were simply thinking this through...

- Yield: yes, this is important, and if initial yield or ramp is low in volume you maximize the revenue by launching into low volume/high margin markets first (that way you don't leave customers wanting).

- Competitive pressure: fight the fights you can win, Shanghai will improve the areas where Barcelona is already competitive and close the gap or take a lead in areas that it doesn't.

- Time to Market - AMD has a window of opportunity to maximize market penetration with server since they are not facing Nehalem derivatives in server just yet. This is AMD's marketing/PR claims anyway.

Many factors come to mind.

Jack

[malice85]

Was there any reason to start yet another "fanboy-war"?


@fragmaster, what exactly was the point of your post? Sorry, I didn't get it, maybe you can explain

[BlazingArrow]

I know this sounds stupid, but what does it mean by "virtualization"?

[informal]

I know this sounds stupid, but what does it mean by "virtualization"?

http://en.wikipedia.org/wiki/Virtualization

[BlazingArrow]

http://en.wikipedia.org/wiki/Virtualization

So is "virtualization" the same thing as "benching"?

[gallag]

AMD could be on to a winner with shanghai and virtualization, Cloud computing could be the future and could have some big backers like google and yahoo

[Calmatory]

So is "virtualization" the same thing as "benching"?

Can be used to measure performance in virtualization tasks. Basically it means that single computer can run (multiple) e.g. "4 computers(from software viewpoint) inside it".

[BlazingArrow]

Can be used to measure performance in virtualization tasks. Basically it means that single computer can run (multiple) e.g. "4 computers(from software viewpoint) inside it".

lol oh man...i feel stupid. I'm a biology major, study philosophy and science, but that sentence confuses me.

So, virtulization tasks means testing the performance of a computer, based on how many 'virtual' computers the single computer can simulate?