So are you talking about a 256KB PPro or a 512K one?
Pentium Pros came with 256K, 512K, or 1MB L2 caches. The budget-oriented 60fsb models (150 and 180MHz) came with 256K only, and the 166MHz ones were all 512K and the top-dog 200MHz models came with 256K, 512K, or 1MB.
The core of a P2 and a pentium pro (and P3 for that matter) are essentially the same thing. The performance issues on 16 bit code where adressed in the P2 redesign though. When running 32bit OS's a Pro and a P2 act very much alike.
The biggest difference is in the L2. The Pentium Pro had it's L2 on chip, on a second die beside the core in a dual cavity package. It ran at full core speed. All P2's have the L2 off chip, and run it at half core speed. So if your application is responsive to the L2 cache, a pro can be faster.
For the majority of tasks though I think the clock speed of the P2 will win out. SETI may go faster on the pro, but most applications most users here are familiar with will respond to the clock speed of the P2. Like anything else in PCs, its just a matter of application.
MMX was largely useless, it affects the results in very few applications for which these processors are suitable.