exxos's Atari, Amiga & retro forum

ST Medium resolution, no caches TT Medium resolution, with cache TT Medium resolution, with cache, program in TT-RAM Images by stephen_usher.

Stephen - What spec is your TT?

For example which TT-Ram board do you have etc? Just curious as I want to compare against my TT :)

Also once my TT has finished building NetBSD packages I'll run tests on the FPU overclock.

It's only got Atari memory cards. The Fast-RAM card is a 16MB I bought a "new old stock" a couple of years ago.

stephen_usher wrote: 31 May 2018 08:08 It's only got Atari memory cards. The Fast-RAM card is a 16MB I bought a "new old stock" a couple of years ago.

Thanks Stephen. I ran the same tests on my TT only difference is I have ST-Ram upgraded to 10MB, the Storm/Thunder TOS with Storm TT-Ram card with 256MB and Thunder-IDE card in the TT. I put the FPU 48MHz overclock back to stock 32MHz.

Looking at the ST-Med Res captures - Most of the stats are identical apart from GEM Window is higher by 56% on your and the Storm card is faster (+120%) for the ALT-RAM test

ST-Med Test


TT-Med Test

Hey Icky,
My TT has the same configuration as yours and guess what, it gets exactly the same results lol, what a surprise :) I worked out you guys are using the base STFM as the comparison.
Oddly though there seems to be a difference in our FPU benchmark, Stephen and Icky get the same result but I don't. Is my FPU slower or faster I'm not sure how to read the results.

TT w/ 10mb STRAM / 64MB TTRAM STORM / ST MED vs STFM ST MED


TT w/ 10mb STRAM / 64MB TTRAM STORM / TT MED vs STFM ST MED

FPU test will show 0% because there are no benchmark files using a FPU on a ST. Plus ST FPU is IO mapped which is not supported anyway.. Only proper LINF F FPU supported, IE Falcon.. In which case I think only Falcon as the FPU benchmarks.. This is all explained in the GB6 readme file.

exxos wrote: 09 Aug 2018 16:26 FPU test will show 0% because there are no benchmark files using a FPU on a ST. Plus ST FPU is IO mapped which is not supported anyway.. Only proper LINF F FPU supported, IE Falcon.. In which case I think only Falcon as the FPU benchmarks.. This is all explained in the GB6 readme file.

So you're saying that the TT FPU results mean nothing and could be random? And only the Falcon has proper FPU results?

Steve wrote: 09 Aug 2018 16:57 So you're saying that the TT FPU results mean nothing and could be random? And only the Falcon has proper FPU results?

How can they be random ? all the images above show 0%...

Reference machine even shows "FPU not present" So you can't compare "nothing" to "something", result will always be zero, as it shows in above results.

exxos wrote: 09 Aug 2018 17:00

Steve wrote: 09 Aug 2018 16:57 So you're saying that the TT FPU results mean nothing and could be random? And only the Falcon has proper FPU results?

How can they be random ? all the images above show 0%...

Chris I think you misunderstand me, I am not comparing my FPU bench to the stock STFM in the image, I am comparing it to the other TT results above from Icky and Stephen. Theirs show something like 3.xx where as mine is at around 7.xx. Do you see what I mean?

Steve wrote: 10 Aug 2018 02:32 Chris I think you misunderstand me, I am not comparing my FPU bench to the stock STFM in the image, I am comparing it to the other TT results above from Icky and Stephen. Theirs show something like 3.xx where as mine is at around 7.xx. Do you see what I mean?

I would suggest opening a new topic and posting all relevant and relating information to see what's going on...