Xeon X5570 "Nehalem" benchmarked.

[BadTrip]

http://it.anandtech.com/IT/showdoc.aspx?i=3536

Very impressive numbers.

 

[werxen]

yeah but can it run cry... oh screw it.

 

[BadTrip]

lol

 

[jimmysmitty]

Holy jebus. It beats or keeps up with a quad Xeon/Opteron setup. Thats pretty damn impressive.

Now lets see how AMD responds....

 

[BadTrip]

yeah, it looks like an amazing chip.

 

[yomamafor1]

AMD's response...

http://breakfree.amd.com/en-us/antitrust.aspx

 

[BadTrip]

^LMAO

 

[BadTrip]

you are aware these are server benchmarks, and everything you just said has ZERO relevance.

 

[cjl]

We already know Intel plays dirty, lets leave it at that.....

Well considering the Phenom 2 AM3 does pretty well against the i7, I'd say they will be pretty competetive, but at a lower price of course.

Its a shame Intels first 32nm chips will be budget dual cores in socket 1156. i7 owners will have to wait till early next year for the hexa-cores to see 32nm efficiency.

Just so you know, the 2.7-2.9GHz AMD chips in that review were the 45nm parts. AMD doesn't have any better, and won't unless they have a new architecture or process that was somehow not on the roadmap that nobody knows about.

Basically, AMD 45nm is utterly flattened by the Nehalem xeon. In almost every way. By huge margins.

 

[fazers_on_stun]

Just so you know, the 2.7-2.9GHz AMD chips in that review were the 45nm parts. AMD doesn't have any better, and won't unless they have a new architecture or process that was somehow not on the roadmap that nobody knows about.

Basically, AMD 45nm is utterly flattened by the Nehalem xeon. In almost every way. By huge margins.

LOL...

The 6-core version of Shanghai is supposed to launch later this year, right? That'll help close the gap to maybe Grand Canyon size. In the meantime, it looks like Nehalem is doing for server what C2 did for desktop.

 

[werxen]

well AMD does have a point. i mean the i7 is basically all prior AMD technology is it not? its easy to take someone elses stuff and beef it up but credit and respect should be given to the innovator... but then again we do live in a capitalistic society so to hell with it.

 

[spud]

well AMD does have a point. i mean the i7 is basically all prior AMD technology is it not? its easy to take someone elses stuff and beef it up but credit and respect should be given to the innovator... but then again we do live in a capitalistic society so to hell with it.

So you are saying that Intel reverse engineered the P2 into the I7 arch then redid it in their own image??? ???

Word, Playa.

 

[spud]

2x post ;<

 

[jimmysmitty]

You mean XEON is a server chip?????? OMG Thanks for letting us know big fella.
Its based on the i7 architecture, so yes it is completely relevent, just not what you wanted to hear huh fanboy?

....and no, Gulftown (hexa-core @ 32nm) is not planned till early next year, however Clarksdale (Westmere) will debut at 32nm this year in socket 1156, not 1366.
In between Gulftown and Clarksdale, there will be Lynnfield which will be a 45nm quad in 1156.
Makes no sense to me why Intel would release s1156 duals at 32nm and s1156 quads in 45nm afterwards AND have s1366 in 32nm as well.

Actually what happens on the DT market in performance by no means is relevant to server market performance. And vice versa. AMDs Phenom and Phenom II always owned Intel in Server for the most part. But in desktop Phenom did not own. It would keep up in some situations and fall behind in most.

What this shows is that a DUAL CPU (note that part) Nehalem server is able to keep up with in some and beat in most quad CPU Intel/AMD setups. That means that admins can save money by buying dual socket servers instead or replace their current ones with Nehalem CPUs and get even better performance gains since this only used DUAL CPU Nehalem based Xeon servers.

Basically nehamel Xeons are the king of the server ring. There is no fighting that. It was easy to see considering the bandwidth given by a Core i7 alone.

 

[werxen]

So you are saying that Intel reverse engineered the P2 into the I7 arch then redid it in their own image??? ???

Word, Playa.

built in memory controller, 64 bit, actual quad core.... these are all pre existing AMD technologies bro... everything that is in Intel cpus these days were created on an AMD lab table.

 

[dattimr]

built in memory controller, 64 bit, actual quad core.... these are all pre existing AMD technologies bro... everything that is in Intel cpus these days were created on an AMD lab table.

... and far better implemented by Intel. What's the excuse for AMD losing by those margins with their own super "inventions"?

 

[BadTrip]

I think we all knew nehalem was going to be the king of the server market, but what surprises me is by how large of a margin.

 

[BadTrip]

oh, if you need a good laugh, go over to AMDZone and see the comments there.

 

[keithlm]

oh, if you need a good laugh, go over to AMDZone and see the comments there.

Why when we can be just as amused here?

 

[cjl]

built in memory controller, 64 bit, actual quad core.... these are all pre existing AMD technologies bro... everything that is in Intel cpus these days were created on an AMD lab table.

Out of those, the only actual AMD technology is the x86-64. Some processors as old as a Dec Alpha 21066 have an IMC, so it is neither a new concept nor is it really specific to either company. Also, Intel has done a native 80 core before, so a native quad is not exactly AMD specific.

 

[BadTrip]

here is more

http://www.dailytech.com/article.aspx?newsid=14701

 

[yomamafor1]

built in memory controller, 64 bit, actual quad core.... these are all pre existing AMD technologies bro... everything that is in Intel cpus these days were created on an AMD lab table.

Actually no. Intel already built an IMC way back in the 90s when they tried to create a SoC for the masses. Both HP and DEC also created IMC for the same reason back in the 90s.

64bit was actually first developed by Intel as IA64.

Intel came up with the first quad core x86 processor.

 

[cjl]

Actually no. Intel already built an IMC way back in the 90s when they tried to create a SoC for the masses. Both HP and DEC also created IMC for the same reason back in the 90s.

64bit was actually first developed by Intel as IA64.

Intel came up with the first quad core x86 processor.

IA64 has nothing to do with x86-64, so I would say that all modern 64 bit X86 compatible chips are using AMD technology for that. Everything else isn't AMD specific though, as you said.

 

[yomamafor1]

Of course, I agree. I'm just simply pointing out that 64bit wasn't AMD's pioneering, but rather x86-64 extension. That was one of the many few true innovations (not "native" quad core BS) from AMD.

 

[Wisecracker]

.....

What this shows is that a DUAL CPU (note that part) Nehalem server is able to keep up with in some and beat in most quad CPU Intel/AMD setups. That means that admins can save money by buying dual socket servers instead or replace their current ones with Nehalem CPUs and get even better performance gains since this only used DUAL CPU Nehalem based Xeon servers .....

They didn't run against 4p (or 8p) Opty's - not that it would have mattered. They ran 2p 8384s at 2.7GHz. 4p Opty 8384s score 22k in SAP ERP 6.0 two-tier.

IIRC 4p 'Barcelonas' at 2.3GHz were in the 17-18k range. (I'm guessin') 4p Shanghai Opty's at 2.9GHz will be in the 23.5-24k range.