Red Dawn gets the door prize

.
Sure, they're slower than their equivalent P3 parts (even oc'd a Celery II 566 to 850
roughly performs ~P3 600-650), but for the price they can't be beat.
Celery IIs give current BX owners a chance to let the trusty old BX board run one maybe two more upgrade cycles. By that time, higher clocked P3 Coppermine procs would've gone down some and Duron/T-Bird prices would've gone down some, too. Also, it would give a chance for mobo makers to whip out more Duron/T-Bird boards and/or come out with later BIOS revisions to those that are already out.
For me, its nice to use the old BX mobos one more cycle (6-10 mos). After that, I'll check the market on current prices, improvements on either Intel or Amd-based chipset and go one from there.
I'll probably continue onto the P3 Coppermine on one rig and start off from scratch to probably an Amd Duron on the other. I tend to go slightly behind the current tech b/c it's easier on the wallet and I'm in no rush to get 100+ fps in Q3A

.
To answer Zathras question:
I think it's in either Sharky Extreme or over @ FiringSquad.
I remember what you are inquiring about. They did benches on both the Celery II 800 and the Coppermine 800. They then disabled the L2 (don't know about the L1) on both procs and did benches on them. Frames were identical w/on board cache disabled on both procs. They showed that the major difference between the two was type of L2 cache on board each chip.