exxos's Atari, Amiga & retro forum

frank.lukas wrote: 03 Oct 2023 20:53 GB for a PAK030-50Mhz and ET4000

https://forum.atari-home.de/index.php/t ... #msg229653

ref TT16MED

exxos wrote: 03 Oct 2023 17:55

Darklord wrote: 03 Oct 2023 17:53 Just for accuracy's sake, the Pak 68/3 results I turned in with
my STacy are at 40mhz (both '030 CPU and 68882 FPU) and
no Alt-RAM...

I think at the time I requested all tests be run from ST-RAM to keep it getting confusing. Though at the time, there was pretty much no alt-ram solutions to test anyway.

I guess I really need some new benchmark files with alt-ram involved.

It's good to have a benchmark, but if you are to show the advantages of having the add-on/modification(s), then the results should show that. For example, the PiStorm-Atari results all reflect the use of Alt-RAM. That's how the add-on was designed to be used. I'll send you some PiStorm-Atari GB6 files. Maybe you can create a database of results we can pick up if needed?

I think you might be tainting results running in 256 colours as it's not like for like comparing against 4 colours in medium res.

exxos wrote: 04 Oct 2023 10:24 I think you might be tainting results running in 256 colours as it's not like for like comparing against 4 colours in medium res.

Concur.

@dad664npc , try your CT60 comparison in shifter mode medium res with NVDI (apples to apples) and you might see those VDI figures get a bump.

BW

Badwolf wrote: 04 Oct 2023 10:50

exxos wrote: 04 Oct 2023 10:24 I think you might be tainting results running in 256 colours as it's not like for like comparing against 4 colours in medium res.

Concur.

@dad664npc , try your CT60 comparison in shifter mode medium res with NVDI (apples to apples) and you might see those VDI figures get a bump.

BW

I tend to (cautiously) disagree.

I wouldn't expect to increase (maybe even decrease) that figures by much with a shifter resolution. To my understanding, the 256 colour mode emulates an ET4000 - in a completely separate (from ST memory) frame buffer (comparable to Fastram). For the Pi, that means it can speed up transfer to and from there natively while it would be limited to ST bus transfer speeds in Shifter mode? Of course it depends on the nature of the video driver (if it really uses that to its advantage), but it appears to me the Pi's GPU has a native 8-bit palette mode (don't know about its bit pattern organisation, tbh) it could use with near-native performance (while it had to fiddle around manipulating single bits with a native Atari screen organisation). NVDI could make a difference, however.

mfro wrote: 04 Oct 2023 13:37 I tend to (cautiously) disagree.

I wouldn't expect to increase (maybe even decrease) that figures by much with a shifter resolution.

Ah yes, quite possibly true. So medium res running on the ET4000 emulator, anyone? :twisted:

BW

I guess the way is to run a stock system with ET4000 and save that as the GB6 defaults. Then you get the stock benchmarks and can compare to that.

The question is, however, how good is it running in ST-Low or ST-Medium resolution as that's what games will use.

The ET-4000 is all very well for DTP or other such desktop use but let's face it, 99.999% or the time STs will be used for playing games.

I'm trying to show off the capabilities of the PiStorm on the ST(e) platform. I don't see any reason to report benchmark results if they are being restricted.
If you want to play games using the PiStorm in low or medium resolutions then it aint gonna be slower than a real 68000.

stephen_usher wrote: 04 Oct 2023 14:39 The question is, however, how good is it running in ST-Low or ST-Medium resolution as that's what games will use.

The ET-4000 is all very well for DTP or other such desktop use but let's face it, 99.999% or the time STs will be used for playing games.

Then a PiStorm isn't for you