Have today's desktop PCs surpassed IBM's Deep Blue of 1997?

Well if wikipedia is accurate here are the specs for deep blue:

30 x RS/6000 SP Thin 120MHz P2SC-based system in a cluster. Each contained a special purpose VLSI chess chip. Running AIX.

Processing performance was 11.38 GFLOPS & at the time was the 259th most powerful supercomputer.

Ok, lets take a stab at it. It's pretty hard to tell what the VLSI chess chips were doing, but a reasonable guess is they improved performance of the chess game by doing certain heavy calculations that were slow on the CPU.

I can say for sure that the 120MHz RS/6000's are Dinosaurs by todays standards and an average desktop PC would outperform a bunch of them tied together without even getting hot. If you also count the GPU in a gamer machine e.g. the ATI Radion R800 can achieve 3.04 TFLOPS (I think single precision) and this is not the fastest out there.

Even on an average CPU I'm sure it'll outperform deep blue. Throw in the GPU and utilise CUDA and you'll probably outperform deep blue by over a hundred times.

Computing power is just way way quicker than it was 13 or so years ago.

I saw your question and became curious myself. I found this quote from an interview with a Deep Blue coder in 2007:

Wired News: What is the state of supercomputer-versus-human matchups? How are we humans doing?

Murray Campbell: Not so well! The current world champion, Vladimir Kramnik from Russia, lost a match to a PC program in November, 4-2. If you look at the supercomputer that Deep Blue ran on, I think a present-day Cell processor has as much processing power as that entire system did in 1997.

Source: Wired's interview with Murray Campbell.

Deep blue was also "enhanced with 480 special purpose VLSI chess chips".

Dedicated silicon for a specific task running in a massively parallel configuration can be extremely quick.

Deep Blue could evaluate 200 million positions a second. Two core Duo chips running Fritz could manage 8 million positions a second.