TT VME Bus Fix - Does it really make it faster?

[Atarian Computing]

From my blog: http://www.atariancomputing.com/blog/tt ... keitfaster

So I decided to attempt the VME Bus Fix as per Atari's rework procedure. The fix saves 1 clock cycle per transfer so it just might be faster. Before I did that, I took screenshots of two benchmarks:

IMG_0161.JPG

IMG_0163.JPG

I have a 2MB Mach 64. Only NVDI 5.03 and Idek's Nova drivers are loaded. Reference is the exact same setup and system sans Mach64. The Mach64 @ 1280x1024x256 colors is pretty much as fast as TT High. That says a lot. The ET4000 is faaar behind.

Here is the same setup after the VME Bus Fix. First result is vs the first screenshot above. The rest are with the same TT High reference.

IMG_0198.JPG

IMG_0199.JPG

IMG_0200.JPG

So the results are pretty conclusive and somewhat expected. Wasn't expecting huge differences but discernable gains. Average results are not affected that much but looking at single tests the gains are 1-3% faster. This is with gfx card. I don't know how it saturates the VME bus. Which is why I tried 16bit color as well. I haven't installed Lightning yet, I wonder what speed differences we would see with USB file transfer.

[Steve]

Very interesting. I have a 2mb Mach64 as well, so I might run a comparison bench.

[Cyprian]

what is that "VME Bus Fix"?

[Atarian Computing]

what is that "VME Bus Fix"?

It is an Atari issued rework procedure, one of several over the years to keep older TTs up to date as newer revisions had them incorporated in the motherboard.

Here is Atari's guide: http://dev-docs.exxosforum.co.uk/TT030_Board_Rework.pdf
Here is my visual guide for REV.C: https://www.exxosforum.co.uk/forum/viewt ... f=31&t=887
and REV.D: https://www.exxosforum.co.uk/forum/viewt ... f=31&t=902

[Cyprian]

ok thanks