X
Business

Intel's technically legitimate but shady benchmarks

When I slapped AMD's hands earlier today for pretending that Core 2 Duo didn't exist, I offered AMD a chance to point out similar marketing shenanigans from Intel.  AMD came back to me with two web pages from Intel.
Written by George Ou, Contributor

When I slapped AMD's hands earlier today for pretending that Core 2 Duo didn't exist, I offered AMD a chance to point out similar marketing shenanigans from Intel.  AMD came back to me with two web pages from Intel.com so I'm going to analyze these complaints in public with you.

Complaint #1 from AMD: AMD says that Intel is pretending that AMD doesn't have an embedded chipset solution when in fact AMD has the ATI-based 690G (which I gave a nice review here).  However upon further inspection I found that this appeared to be simply out of date.  This was not selectively out of date like AMD comparing newer AMD Turion X2 processors from this year to last year's Intel Core Duo processor.  The Intel web page is equally out of date for both Intel and AMD since it talks about the older Intel 945/955 chipsets and not Intel's latest 965 and 3-series chipsets.  So this is just an old web page that compares older Intel products to AMD's situation when AMD didn't have a chipset before the ATI merger.  Intel is looking in to updating their website so this doesn't appear to be a deliberate or major foul on Intel.

Complaint #2 from AMD: Intel is comparing a new Intel X5355 2.66 GHz quad-core CPU to an older Opteron 275 2.2 GHz CPU.  I asked Intel why they did this and they told me that it was the highest available score for this particular benchmark.  Last time my colleague David Berlind called this kind of behavior from Intel "felonious" but AMD admitted to me that every benchmark posted by Intel in their presentation was in fact the best score that AMD or its partners published on SPEC.org.  So Intel seems to have an excuse that they merely posted the best score available from AMD and that it wasn't Intel's responsibility to bring a better AMD player to the game whereas AMD deliberately left out better scores from Intel that were published and available on SPEC.org.

When I asked AMD why they didn't get better scores published, AMD tells me they're just a small company that doesn't have the resources necessary to get the most up to date scores published for the newest AMD servers.  So this might be a valid point from AMD but Intel technically can claim they did nothing wrong.  Perhaps it would be a good idea for AMD to completely skip out of some of the less popular benchmarks rather than doing it half-heartily and only send their second string players to the game.  Intel is exploiting the situation that AMD isn't sending top players to the game so they're taking advantage of the situation.

So my assessment of this complaint is that while Intel technically has a valid excuse, it looks shady because it is so obviously misleading when you compare a new Intel CPU to AMD's older 2.2 GHz Opteron.  It's perfectly valid to compare Intel's best quad-core to AMD's best dual-core processor since it's AMD's fault for not having a quad-core CPU, but using an Opteron 2.2 GHz processor on a somewhat obscure benchmark even if it's the best published score available looks shady.  AMD also bears some responsibility by not getting its best scores for its best processors published.

Editorial standards