Friday, May 26, 2006

Intel Woodcrest Benchmarks Claimed WRONG!

I spent the entire half of today in the Intel seminar launching the Bensley platform, a new platform with allegedly great improvements over their old stuff. The Platform Manager himself showed us some `recently available' benchmarks showing that Intel had finally caught up with AMD and will NO LONGER PLAY CATCH UP.

I thought, Wow. Opterons had so much lead over the Intels, and nothing much has changed except for the CPU, what must have changed, I thought, was that the new 0.65nm process gave the Woodcrest processors so much more lead. I even asked the platform manager WHEN he was about to integrate their memory controller unit into their CPU, and he told me that this is not necessarily yet (in view of the benchmarks) and in a few years they'd do it.

Imagine my surprise, when I came back, I found this thread in my Instant Messaging:

Intel Woodcrest performance claim a fraud

In summary, the blogpost content includes:
1) Intel based their results on the tpc.org results, but Intel claimed that the AMD Opteron machine was using 64bit software like the Intel machine, but at the tpc.org website, the Opteron machine was loaded with 32bit software.
2) The AMD machine was using different, smaller hard disk storage compared to the Intel machine


Anyway, let these guys clarify this first. Intel Server Platform Group has been informed of this blogpost, and he'll respond.

3 comments:

Sharikou, Ph. D. said...

More information on Intel's Woodcrest benchmarks.

Intel's guerilla benchmarketing

Unknown said...

LOL an honor for Sharikou to post comments here! I'll take a look.

Unknown said...

I read the links. I am not partisan to any camp, and I think my for my super small subscriber base it suffices to have a highlight on your page, unless Intel comments or something. So far, my contacts cannot comment as clearly, this game is played on a much higher level than them.