I've found topbench's measurements to be on par with speed perceived during actual use

konc wrote on 2020-04-26, 15:09:

I've found topbench's measurements to be on par with speed perceived during actual use

Interesting - Topbench does not give a direct comparison to other CPUs though, does it? Is there a database of CPU Topbench scores somewhere?

appiah4 wrote on 2020-04-27, 07:11:

Interesting - Topbench does not give a direct comparison to other CPUs though, does it?

You're right, it doesn't

appiah4 wrote on 2020-04-27, 07:11:

Is there a database of CPU Topbench scores somewhere?

Have you looked at the site?
https://dosbenchmark.wordpress.com/downloads/

OK I downloaded TOPBENCH and the database; I get 171 points with SETMUL.EXE L1D which puts me spot on in the DX2-66 neighborhood, which is pretty much exactly what SpeedSys says as well. However, when I launch DOOM it is significantly slower than it runs on a real DX33, this PC behaves more like a fast 386 or slow 486 (Think DX25). Interestingly Norton SystemInfo also says this system scores slower than a DX33. To be honest, I find TOPBENCH to not be a very good indicator...

Any other suggestions?

There might not be one good indicator, but if what you care about is Doom performance then why not use Doom as your benchmark?

BinaryDemon wrote on 2020-04-27, 13:25:

There might not be one good indicator, but if what you care about is Doom performance then why not use Doom as your benchmark?

It's not, I am looking for an indicator that is good for judging overall gaming performance. I am trying to judge performance sweetspots for games like Wing Commander and Ultima VII..

A wild stab, but what about computing Pi and comparing the results?

I would guess a game would be a better benchmark than a synthetic benchmark in most cases because it tests all the subsystems (video/audio/ect).