AMD Triple Core Phenom 8600 Early Results

It looks like someone has managed to snag a B2 revision Toliman processor and run a few benchmarks with it. The Toliman processor is essentially a 4-core Phenom with a core disabled which is due to the stringent testing of processors off the assembly line. Since K10 is a monolithic core, any minor errors effectively limit AMD’s ability to churn out quad-core processors compared to Intel’s MCM design so to counter-act the low yields AMD has decided to simply disable the bum cores and peddle the rejected processors as tri-core units. This will result in a very broad product line with dual, tri, and quad-core processors lining up between the $100 and $300 price-range, making it all that much easier harder for consumers to choose the processor fit for their applications. Due to the pricing between Intel and AMD, these tri-core processors will compete against dual-core Wolfdale processors which should be a very interesting matchup. Wolfdale will have higher clocks but Toliman will technically have a 50% increase in potential through-put. I’m still putting my money on Wolfdale but Toliman may be very competition in the OEM market.

Here are a few screenshots from the Phenom benchmarks along with benchmarks I just ran to compare them to a simulated tri-core Intel processor along with a similarly clocked Yorkfield processor. The results still don’t look good for Toliman but I’ve been told this processor had the TLB patch which typically results in a 10% performance penalty.

AMD 8600 Cinebench

AMD 8600 Cinebench

Yorkfield Tri-core Cinebench

“Tri-core” Yorkfield Cinebench

Yorkfield Cinebench

Yorkfield Cinebench

AMD 8600 Super Pi

AMD 8600 SuperPi 1M

Yorkfield Super Pi

Yorkfield Super Pi

Even though the systems weren’t identical, they were close enough in clockspeeds that the gross differences in performance aren’t within the margin of error. Unfortunately I wasn’t able to run my QX9650 at 11.5 x 200 as the lowest FSB possible with this board was 333MHz. The interesting bit in all of this was the rather attrocious performance scaling of the Toliman processor. A 2.66 ratio compared to 2.79 on the simulated “tri-core” Yorkfield was rather shocking. It sure sounds like all the propaganda flooding from AMD about “true” quad-cores being better might have been fluff.

No Comments Yet

You can be the first to comment!

Speak Your Mind

  • Comment Policy: No flaming unless directed at the author for good reason. All other flaming will be turned into a comment about ponies and rollerskates.