Comparing CPU frequencies
 

I have two computers, a 1 year old laptop and a 3 year old desktop. I'm wondering why the two perform about the same when the clock speed is so different.

For pure floating point such as rendering, the performance is exactly the same, while for memory/cpu intesive tasks the laptop is slightly ahead. The latter is a factorization task, taking up about 600Mb of memory. Here are the specs:

laptop:
Intel Pentium M 738, 1.4GHz, 2Mb L2 cache
FSB 100MHz, Bus speed 400MHz
1GB + 128Mb DDR-SDRAM @ 133MHz
FSB:DRAM 3:4, 2-2-3

desktop:
Intel Pentium 4, 2.8GHz, 512Mb L2 cache
FSB 133MHz, Bus speed 533MHz
512+512Mb DDR-SDRAM @ 133MHz
FSB:DRAM 1:1, 2.5-3-3

The main difference, apart from the cpu frequency, is L2 cache, but I can't see why this would make much difference as the data isn't reused during solving.

The laptop appears to be much slower. Appearances are deceptive here: it's a pentium M, which is much more efficient than a normal pentium 4. Typically, if you want to compare it's performance, you need to multiply the clock speed by 1.5 or more, depending on the type of pentium 4. You appear to have the older variant, with 533 Mhz FSB and relatively slow ram.

If the clock speeds and CPU types were identical, you'd still have a big difference in the amount of cache memory, which will give the laptop a big advantage during memory intensive tasks. It's exact advantage depends on the computer program, compiler, operating system, etc. etc. It could even give you a boost with non-repetetive data.

All in all, I'm not surprised by the good performance of that laptop. :)

I see.. but I still don't understand why the laptop is 1.5 times faster for pure floating point operations. I'm not very up to date on computer hardware these days, but is this related to the trick AMD did some years ago, passing information on rising/falling edges of the bus signal?

The Pentium M was well know for having a very speedy architecture that performed better than faster clocked chips like the Pentium 4. The Pentium M is also a more expensive chip to manufacture and has many power conservation features needed in laptops. In general (warning: very simplified) the fewer clock cycles that you want any given instruction to execute in, the more transistors you have to dedicate to that instruction, and the CPU has a finite number of transistors to utilized. In this case I think that the chip’s efficiency and 2meg cache is having a very pronounced positive effect on the laptops performance.

Measurements like MIPS (Millions of Instructions Per Second) and FLOPS (Floating Point Operations Per Second) do not have a direct correlation to a CPU’s clock speed, as a single operation can take from one clock cycle to over twelve clock cycles to complete. Instructions very from chip set to chipset also. Even within the Pentium family instructions can be different and take different number of cycles to complete. If you want a starter page with info on the Pentium M try this one for more info : http://en.wikipedia.org/wiki/Pentium_m

Intesteting.. I've always used the desktop pc for these tasks, didn't even occur to me to try this tiny 5lb laptop.