exxos's Atari, Amiga & retro forum

I think I know what the confusion is here.

Where we are comparing as Chris rightly says against an ST we are getting a 0% for Float Math as there is nothing to compare against.

However Steve you are seeing difference between the other TT's in the time it is taking to complete that function and not the output.

Yours is taking 7.x in time to reach the conclusion of 0% where as ours is taking 3.x in time.

To confirm the Float Math when compared against the base TT16MED.GB6 test file actually comes back around 100% mark on my TT in stock. I also did a test against this with my FPU overclocked to 40MHz and got 245% for Float Math.

I never even looked at the "time" figures.. Never give it a thought it would even output "time" when result was 0%.. but makes more sense now!

But yeah, The FPU routine will run say 10,000 loops and output a time in seconds on how long it took to complete.

If the benchmark file has a FPU benchmark done (like Falcon TT etc) then it will show a speed comparison between the file and system and thus show a % figure.

FWIW, its the same "issue" with alt-ram tests...

I've just fitted my TT with a Storm memory upgrade and 2x32MB EDO DIMMs and got an interesting benchmark result relative to the TT-Med benchmark..

Other than the addition of the Storm (and Lightning, which I can't get to work yet) the machine is stock.

stephen_usher wrote: ↑Thu Nov 08, 2018 11:05 pm I've just fitted my TT with a Storm memory upgrade and 2x32MB EDO DIMMs and got an interesting benchmark result relative to the TT-Med benchmark..

Other than the addition of the Storm (and Lightning, which I can't get to work yet) the machine is stock.

IMG_1221.jpg

Welcome to the ranks of the Storm TT’s all you need now is the FPU overclock.

https://www.exxosforum.co.uk/forum/viewt ... f=18&t=744

Why is his float math already 195% though?

Steve wrote: ↑Fri Nov 09, 2018 2:40 am Why is his float math already 195% though?

That's what I was wondering about.

stephen_usher wrote: ↑Fri Nov 09, 2018 8:11 am

Steve wrote: ↑Fri Nov 09, 2018 2:40 am Why is his float math already 195% though?

That's what I was wondering about.

Yes - very interesting on yours which I assume is stock FPU its 195% and on mine which is overclocked its 245%.

What is its baseline its measuring against?

I know the jumper for the FPU clock speed has a 16MHz and 32MHz setting. Did the baseline benchmark have the FPU clock set to 16MHz?

Icky wrote: ↑Fri Nov 09, 2018 8:38 am

stephen_usher wrote: ↑Fri Nov 09, 2018 8:11 am

That's what I was wondering about.

Yes - very interesting on yours which I assume is stock FPU its 195% and on mine which is overclocked its 245%.

What is its baseline its measuring against?

I know the jumper for the FPU clock speed has a 16MHz and 32MHz setting. Did the baseline benchmark have the FPU clock set to 16MHz?

If I get time. I may put my TT to stock FPU at 32MHz take out TT-Ram and take the ST-Ram card out so it’s basic no frills.

Then do the benchmark to see it’s it’s all 100%.

If only I could afford a TT of my own.

There are no jumpers changed on the main board, it's as it was supplied by Atari in 1990. As for the RAM, it's definitely EDO RAM (2x 32MB Compaq branded DIMMs) and the Storm jumpers are set for that.