is it possible to run speedsys without hdd test? maybe its taking cpu overhead during IO into consideration. Better bios will allow for faster hdd transfers at the cost of cpu.

Also run some other benchmarks. Speedsys is buggy sometimes.

can this promise bios be used with a NIC and a random onboard or standalone controller?
if yes then we can test with more systems

rasz_pl wrote on 2024-02-12, 09:59:

is it possible to run speedsys without hdd test? maybe its taking cpu overhead during IO into consideration. Better bios will allow for faster hdd transfers at the cost of cpu.

I skipped the HDD tests on the benchmarks, yes.

kixs wrote on 2024-02-12, 10:28:

Also run some other benchmarks. Speedsys is buggy sometimes.

I ran Doom with and without and got 7490 realticks / 74.69 fps each time, so it mush be some strange artifact in speedsys. I'll try some other bench marks after work today and see if anything interesting shows up.

douglar wrote on 2024-02-12, 12:34:

I skipped the HDD tests on the benchmarks, yes.

Maybe it did include it anyway 😀 running from floppy with HDD disconnected would solve this mystery.

douglar wrote on 2024-02-12, 02:45:

I have this motherboard: https://theretroweb.com/motherboards/s/edom-wintech-mv008 […]
Show full quote

I have this motherboard: https://theretroweb.com/motherboards/s/edom-wintech-mv008

When I use a controller with this Option ROM on it: https://www.vogonsdrivers.com/getfile.php?fileid=2018 I get these results in speedsys:

• CPU speed index : 50.12

When I use a controller without an INT13 Option ROM or I use XTide Universal BIOS, I get this result:

• CPU speed index : 47.85

How could that be? Nothing else changes and it doesn't look like the cache or memory timings or WB/WT changed.

Most likely a caching artifact. That Option ROM might cause some changes in memory layout (e.g. by reserving some space at the top of the 640KB conventional memory), which could change what data in main memory competes for the same cache lines.