Everyone keeps reporting how the new C2 processors can overclock like mad (533 to 900, 633 to 1GHz, etc). The C2 cpus do not perform clock-for-clock as their Coppermine similar-core-sibling and I haven't found anyone that takes a look at why. Sure the C2 has 1/2 the cache but if that is truly the only difference than the Cu, then adjusting bus speeds and matching C2 MHz to Cu MHz, they should perform very closely (except in cache intensive apps). So what else has changed with the C2? I smell a story here waiting to be discovered: (1) What is holding the C2 back and (2) can it be undone by the community who loves to turn the tables on Intel? (Of course it only becomes a real story if there is something that can be adjusted to the C2 to make it run like a Cu)To which I responded:
Sam:And the stage was set. We mailed back and forth a bit about the question, with Sam pointing out this Celeron II review at GamePC. The CII is overclocked to 1GHz on a 105MHz bus in this case, and it just ties a 700MHz Pentium III. Sam writes:
I've gotten the impression that the Celeron II and PIII Coppermine perform _very_ close to one another at the same bus and clock speed. Perhaps you read Kyle's review of the Celeron II at the HardOCP and saw the PIII perform quite a bit better in his tests? This is the thing about Kyle's tests: the clock speeds were similar, but the Celeron II was running on a slower system bus.
Check this link...
..for a look at how bus speeds affect the Celeron II.
What else have you read that gives you the impression the Celeron II is crippled beyond the bus speed and cache handicaps?
I would think a C2 and a Cu running at the same bus speed would produce almost equivalent scores. But a 633/66 C2 using a 100MHz bus becomes a 1GHz C2. But, this 1GHz/100 C2 performs like a 700MHz/100 Cu (the 3rd line in the chart). So with bus speeds being equal, the C2 has a 43% advantage in clock speed. Is 1/2 cache really causing a 43% disadvantage when bus speeds are equal? I thought the cpu core would perform almost the same in at least benchmarks that are not cache intensive.I think perhaps the system was bottlenecking on the video card in the GamePC review, but I looked at the numbers a bit, and I think Sam may have a point. The PIII and Celeron II are very close together, and I'd expect the CII's MHz advantage to mitigate the cache size differences, especially in higher-latency FPU ops. I'd love to test it, but I don't have a CII here right now.
So the question is, has Intel limited the performance of the Celeron IIs in some way beyond saddling 'em with a default bus speed of 66MHz and disabling half of the processor's L2 cache? What do you guys think?