Toms CPU charts, which is good for MAME comparisons?

[rockin_rick]

Using Tom's Hardware CPU charts-

http://www23.tomshardware.com/cpu.html

Which best compares to MAME performance?

Thanks,
Rick

[Fozzy The Bear]

Using Tom's Hardware CPU charts-

http://www23.tomshardware.com/cpu.html

Which best compares to MAME performance?

Thanks,
Rick

Depends why you're asking.......  If you're on a budget, then the fastest processor you can afford.

If money is no object.... then the fastest processor available at any cost will give you the best Mame performance.

There really is no other way to answer this.

Best Regards,
Julian (Fozzy The Bear)

[SGT]

Using Tom's Hardware CPU charts-

http://www23.tomshardware.com/cpu.html

Which best compares to MAME performance?

Are you asking which benchmark would be similiar to running Mame?

[Tiger-Heli]

If SGT is correct - I would say PCMark is probably the closest.

If you just want general CPU advice -

Something like a Sempron 2200 ($40 on Pricewatch) will play most late 80's games fine, but will be slow on games like cruisn' usa.

Something like an Athlon 64 X2 3800 will give better performance to $90.

If you can swing it, the Core 2 Duo 4300 at $180 is faster than the above at stock clocks and easily overclocks to ridiculous speeds, will play cruisin at full frame rates.

Even that won't play the most demanding CHD games.

[rockin_rick]

Are you asking which benchmark would be similiar to running Mame?

Yeah, that's what I meant.  I think that there are two main issues with the benchmarks.

With Mame not taking (full) advantage of dual cores, then I'd think that a benchmark that only measures the speed of a single core in multi-core devices would provide a more meaningful/accurate comparison of CPU's for use with Mame.  I'm guessing that some or most (all?) of those benchmarks take advantage of dual/multi cores and the score is based on that.

Also, a benchmark that operates the CPU like Mame does would be more accurate.  For instance, a benchmark that includes a lot of floating point math may skew the results if Mame does little FP math.  (not sure what Mame uses...) 

PCMark 2005 - CPU is the best?  Does it only measure one core?

Thanks,
Rick

[ahofle]

My understanding is that the C2D still beats everything else in terms of MAME performance, despite the fact that MAME isn't specifically optimized for multiple cores. 

[dmckean]

The Pentium M's and C2D's are still something like 30% faster than a similar Pentium 4 at the same clock speed when using just one core.  Plus duel cores still help some with MAME because system processes are able to run completely in the backround and let's MAME have slightly more CPU time.

[Tiger-Heli]

PCMark 2005 - CPU is the best?  Does it only measure one core?

That was just a semi-educated guess on my part.

[rockin_rick]

(It is my opinion) that the "Lame 3.98 beta....." benchmark at Tom's Hardware is the closest (not dead on) chart for comparing MAME performance.  I believe that Lame only utilizes one core, like MAME.  I came to this conclusion after reading CPU reviews and their benchmarks, and compared that info with MAME benchmarks, etc. 

Rick

[u_rebelscum]

Also, a benchmark that operates the CPU like Mame does would be more accurate.  For instance, a benchmark that includes a lot of floating point math may skew the results if Mame does little FP math.  (not sure what Mame uses...) 

Mame is almost all integer, AFAIK.

The Pentium M's and C2D's are still something like 30% faster than a similar Pentium 4 at the same clock speed when using just one core.

AFA mame goes, I though PM & Core 1 were about 30%, Core 2 about 70%, and Athlon somewhere around 25-35% faster than P4 at same GHz.  (Athlon is wider range because some games like intel or AMD chips better.)  And that solo, dual, or quad don't make very much a dent.