"Since the L3 cache is clocked at the same speed as the memory controller, raising that memory controller's clock speed should be a priority for AMD. This particular issue may be more of a concern in desktops and workstations than in servers, however, given the usage models involved."

Aah... that kinda explains why it's scaling so well to 2.5.
OR- why 2.0 is so handicapped.

Totally depending on your perspective.

thanks for the link , nice review ,
k10 rock ! but opteron price

Emmm YUmmyyy... I wanna have one as well
But the price put off me now...wait for another 6 months

Better wait for phenom.. or better still go for wolfdale or yorkfield

on average, based on the benchmark from anandtech, it seems K10 offers only 15% clock for clock improvement over K8. As we know, Intel's Core arch offering is around 20% faster than K8 clock for clock. keep in mind that Penryn will be around 5% average faster than Core. Where is the 25% / 75%(FPU) improvement that AMD claim since a year ago?

*it could be the buffered RAMs and the slow DDR2 667MHz which was used in the benchmark.

Not all reviews paint the same picture.

http://www.techwarelabs.com/reviews/proces...ona/index.shtml
Inconclusive but a different way of comparison lol.

And benchmarking on a beta BIOS (all reviews) = WTF.

Guys im kinda confused here since i never really follow this tread.

Phenom is the name of its new desktop PC processor right? N the core code name is Agena?

Wut about Barcelona? The code name of the "new" Opteron's core?

Exactly what u think. Simply said Barcelona isnt actually for us. Phenom later will have little improvement which is Hyper Transport 3.

I reach the second page and i already lol. These guys apparently haven't seen a mainboard before.
I also can't seem to find the specs of the intel system.

This really takes the cake:

LOL.

Anyway I'm waiting for an official teardown/benchmark on Barcelona by the boys in blue.

Based on data I've seen so far my conclusion is: relatively unimpressive debut given all the hype (40% + improvements /the 2nd coming of Jesus etc...)
Weaker than Core on INT, about even on FP (more data needed). Power consumption is improved significantly.
HT is the only reason it still rules on 4P systems but Tigerton looks pretty good and CSI/IMC is coming on Nehalem in 2008.
It still is a killer HPC box but not so great on desktops/typical servers. A pretty uneven arch geared for servers rather than desktop/mobile.

Bigger concern for me is how much AMD is making off Barcelona.
Large die size on 65nm yet they can't sell at higher margins due to unimpressive performances vs Core.

This will certainly drop when Intel launches theirs

http://youtube.com/watch?v=R7EZmYth6TM

Despite performance against current core based xeon, the barcelona seems doing well on power consumption and price scaling, which I think might be much ideal choice as compare to xeon in multi processor base environment.

What they must hardly do now is ramp up the clock speed and production. The Phenom should win some and loosing some. The price will determine whether it can sell well or not. But it will be tough for AMD when intel going to 45nm afterward.

those are what.... B1 stepping?

B2 has much-much better improvement, which is the final retail.

---

Added on September 11, 2007, 11:45 amthe other guys also seems to be confused.

Phenom & Opteron are the product names.

Agena and Barcelona are the codenames only.

This post has been edited by soulfly: Sep 11 2007, 11:45 AM

Edited
product name ......corename
Opteron ..............barcelona
Phenom X4 --------agena
Phenom X2 --------kuma

all r k10

This post has been edited by likito: Sep 11 2007, 02:07 PM

Bro. It is Phenom
No hard feeling

-pWs-

ops !! sorry , edited already
phonem need more 60-90 day

This post has been edited by likito: Sep 11 2007, 02:09 PM

The Sep 10 CPUs are "BA" or "B1 Fixed", whichever way you like them.

NOT those lame crippled ones from the Coolaler review.

Though I agree with B2 bringing a shinier outlook and propose that B3 saving their asses from eternal damnation (it's all in the yields).

Stepping usually has no co-relation to yield rate/health.
And don't expect miracles going from B(n) to B(n + x) stepping; these are just minor fixes/tweaks using pretty much the same mask.