exxos's Atari, Amiga & retro forum

exxos wrote: 14 Jul 2024 21:53 Currently I don't have anything in the auto folder..

How is the blitter not causing an unholy mess if it's enabled yet you've nothing in the auto folder to correct its operation with TT-RAM?

Have you a ROM patch? Is that the issue?

BW

Badwolf wrote: 15 Jul 2024 10:46 How is the blitter not causing an unholy mess if it's enabled yet you've nothing in the auto folder to correct its operation with TT-RAM?

:shrug: Oddly it doesn't always cause a total mess, no idea why. But I have done that many tests and I've not really been paying much attention to the blitter stuff..

Have you a ROM patch? Is that the issue?

Its my build of TOS206.

The score changed again when running BLT_FIX from TTram. It might be some odd cache thing going on skewing the results. I do need to look into it more but it is probably not worth the effort.

Though I got a bit distracted because the whole thing only seems to run when a fan is blowing on the booster. It seems to be the PLD isn't happy for some reason. :shrug:

Some tests:

Changing the timings to "slow" makes it lock up sooner. Noticeably during hard drive booting.

It mostly seems to lock up after loading the hard disk driver just before MAPROM text is displayed, BUT disabling MAPROM, and even disabling TTram has no effect. With nothing in the AUTO folder, it just looks up just as it's getting to desktop, it just has the busy bee and a white screen.

Running the CPU at only 8MHz makes no odds either.

EMUTOS isn't happy either. After the splash screen it just goes to a totally blank screen. Presumably it is not a blitter issue if EMUTOS isn't using it anyway. There is no precedent for anything either just a totally white screen. So I don't think it is operating system as TOS can't get to desktop either.

Looking at the CPU clock, it gets stuck in 50MHz after the crash. So its locked up for some reason.

Putting freezer spray on the PLD and it all works fine. Spray on the CPU has no effect.

The voltage is a little low on the booster. It starts off for the power supply at 5.05V and ends up 4.85V on the booster. Though even hardwiring power (5V & GND) direct from the PSU to booster makes no odds.

There's either some timing problem somewhere where the PLD warming up is enough to break something.. But I don't know what.. Or its simply a matter of the 10ns PLDs simply don't work...

The only possible idea is just to give the PLD a bit more voltage and see if that helps.

Just a random thought... The CPLD isn't trying to supply current on a line which should be open collector is it? That could both be causing the CPLD to heat up more and the voltage be pulled down.

Probably not, but just a thought.

If a cold spray helps the CPLD, maybe a heatsink?

stephen_usher wrote: 15 Jul 2024 11:51 Just a random thought... The CPLD isn't trying to supply current on a line which should be open collector is it? That could both be causing the CPLD to heat up more and the voltage be pulled down.

Certainly an option I guess.

If a cold spray helps the CPLD, maybe a heatsink?

Possibly though I don't really want to "ship" something with some odd issue like that.

I think it may be relating to /AS syncing currently. Something must be "off" on the STE compared to the STFM. I did a small change and its booted up and running BG6 now, but the scores dropped a bit, mostly seems to be relating to ROM speeds though.

Well, things are certainly different given that the PiStorm firmware which works on the STE fails dismally on the ST. Timings must be very different.

These are before and after the /AS "Fix".. Oddly it seems be running marginally faster..

exxos wrote: 15 Jul 2024 11:30 Some tests:

Changing the timings to "slow" makes it lock up sooner. Noticeably during hard drive booting.

That's not a good sign. You probably want 'slow' for the most part to reducing ringing.

Scope the BR line? Too many DMAish things involved in the descriptions not to try to rule that out first.

BW

Badwolf wrote: 15 Jul 2024 12:49 That's not a good sign. You probably want 'slow' for the most part to reducing ringing.

Scope the BR line? Too many DMAish things involved in the descriptions not to try to rule that out first.

I just tried slow again with my firmware tweak and it is also still working fine..

I did scope out a lot of signals previously and they are just so bad I gave up looking at them :lol: I will check out the bus grant signals again though anyway...

EDIT:

Bus grant signals looks pretty good considering how rubbish signals generally are anyway. BR could do with going high sooner, but it's got 1.2K on it from the MB anyway.

Seems a bit slower than the ST536 benchmark. Though I don't remember if that used a older version of MAPROM in those tests :shrug: