AMD launches its Quad-Core Opteron; Phenom expected in December

[pjbliverpool]

Less than 10% of your DNA has to change for you to pop out as any of the hundreds of thousands of mammals on this planet; you know this right?

Interesting but I guess you can see the sense in that when you consider how much we have in common, i.e. materials, organs, energy processing etc...

I wonder how much would need to be altered for us to become an insect!

 

[3dilettante]

Non-rate SPEC scores were posted for Barcelona.

I'll reference the first place I saw the numbers:
http://realworldtech.com/forums/index.cfm?action=detail&id=83478&threadid=83478&roomid=2

In 2-socket systems

Int base/peak
AMD 1.9GHz - 9.97/11.3
Intel 2GHz - 14.2/15.6

FP base/peak
AMD 1.9GHz - 10.7/11.2
Intel 2GHz - 14.5/16.9

There was some uncertainty about autoparallelization being enabled for the Intel platform versus the AMD one, so a comparison was made for a single-threaded Intel submission:

11.9 Specfp base 2.0 GHz


Where this puts AMD:

Barcelona, at this low clock, possibly still iffy stepping, and hodgepodge compiler state, is significantly behind on integer performance and slighly lagging in floating point.

The per-clock performance, if my math is right, for integer is lagging by over 25%.
I don't see any combination of last-minute silicon bug fixes, better compilers, and faster desktop memory that is going to erase this deficit.

Floating point performance is less conclusive, as the slight difference between the new platform and an established chip allows for more improvement for Barcelona.

This is, of course, per clock.
All such examples given for AMD have a 1 GHz deficit from the high-end Intel chips.


The faster Phenoms won't be out for quite a while.
2.5 is the next ceiling, and it doesn't seem that we will be seeing a 3 GHz Phenom until at least 2Q 2008.

Penryn will be out in force by then, and likely will keep a lead of hundreds of MHz over Phenom.

The 2-core and 3-core Phenoms are also going to be found in 2008.
What is especially troubling is that dual core models especially are pushed out to 2Q 2008.
Since the expectation is that they should be clocking faster than the quad and tri cores, their late arrival is a huge question mark. The clock speeds are not definite, but it looks possible that AMD's dual-core clocks might lag or barely match Intel's quad core solutions at that time.

Instead, AMD is putting in a few last speed grades of their 65nm Brisbanes, which will still lag in clock from the top-bin 90nm chips.

 

[Unknown Soldier]

Read the full news item

Weird! Anyone else getting weird characters to the news content posted above?

I get the following:

processor, and first quad-core architecture, the Quad-Core Opteron, a processor most know by its codename “Barcelona.â€￾

“We’ve worked closely with our customers and partners to design a new generation of processing solutions embodied by today’s Quad-Core AMD Opteron processor – a four-way winner in performance, energy efficiency, virtualization and investment protection. Early customer response has been extremely positive,â€￾ said Hector Ruiz, AMD chief executive officer and chairman about Quad-Core Opteron, a quite crucial product for AMD considering its current state of affairs.

AMD pricing for Quad-Core Opteron processors has also been made available. The 2300 Series (up-to 2 sockets) starts at $209 (€152/£103) for the 1.7GHz Opteron 2344 HE and goes up to $389 (€282/£192) for the 2GHz Opteron 2350. While the 4+ sockets compatible 8300 series debut at $698 (€506/£344) for the 1.8GHz Opteron 8346 HE and goes up to $1,019 (€739/£502) for the 2GHz Opteron 8350.

Using Mozilla Firefox for FreeBSD.

US

 

[ShaidarHaran]

I see the gobbledy-gook characters as well.

 

[pascal]

No, I still think AMD should be able to improve along the evolutionary path as well. Intel has already proven that it's possible with Core 2, which is more of an evolution from the Pentium 3.

Yes, the Core 2 is an evolution of P3. It follows the sequence P1-P2-P3-P3S-PM-CORE-CORE2. Each step incorporated some level of new engineering.

I still miss my old Tually (P3S with TUSL2-C mobo).

 

[mczak]

Yes, the Core 2 is an evolution of P3. It follows the sequence P1-P2-P3-P3S-PM-CORE-CORE2. Each step incorporated some level of new engineering.

I wouldn't put the P1 in there - conceptually quite different, if you'd want to put that there you can just put in all earlier chips (486, 386,...) as well. However, the PPro clearly belongs there before the P2. And I'd say that the step from core to core2 is the one with the most significant changes in any one step (you could of course argue the evolution of PM to Core had more benefits due to adding the second core, but the basic architecture didn't really change much).

 

[pascal]

Agree mczak :smile:
Correcting the sequence PPro-P2-P3-P3S-PM-CORE-CORE2.
And agree that Core to Core 2 was very important and significant.