How does a 500MHz K6 do on this project?

[jasong]

I know it won't exactly scream along, but I'm looking for somewhere I can do well. Clock-for-clock, if not anything else.

[xilman]

Originally Posted by jasong

I know it won't exactly scream along, but I'm looking for somewhere I can do well. Clock-for-clock, if not anything else.

Good question, to which I have no real answer.

Assuming it has enough memory (I wouldn't recommend running on a machine with less than 128M) it will undoubtedly do useful work. I've used 300MHz PII machines productively while at MSR.

The best approach, I suggest, is to suck it and see. Try running for a few days and see what happens. That would certainly be my approach if I had such a system.

Paul

P.S. The performance of NFS sievers tend to be much more correlated with efficiency of the memory architecture than the raw clock speed, so you may find that your machine out-performs on a clock-for-clock basis --- or you may not. Suck it and see remains my advice.

[Wacky]

At one time, cycle for cycle, my 450MHz G4 tower was fastest in our group.
It certainly wasn't all that fast in terms of overall throughput since its clock was less than 1/4th of the 2GHz of (at that time) "bleeding edge" machines.

However, at half the competitive speed, its memory to cpu speed ratio was twice as good.

[geoff]

Originally Posted by jasong

I know it won't exactly scream along, but I'm looking for somewhere I can do well. Clock-for-clock, if not anything else.

I have a similar machine, I posted a few timings for different projects here, at the moment I have it looking for a factor of MM61 which is the only project that I could find where it matches a P3 clock for clock.