Page 51 of 54

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Sun Jan 19, 2020 8:56 pm
by exxos
Have no idea why it won't work for your guys. Probably have to wait until I get my H4 to look into it. Hardly had any issues here, but TOS206 does seem to need a few resests on the reset switch to boot, but i've not really looked into that issue yet, might be just iffy reset on my board.

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Sun Jan 19, 2020 9:10 pm
by PhilC
Ok, will try it in an STFM this week

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Sun Jan 19, 2020 10:39 pm
by exxos
PhilC wrote:
Sun Jan 19, 2020 9:10 pm
Ok, will try it in an STFM this week
I assume you are trying without ultrasatan plugged in as well ?

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Sun Jan 19, 2020 10:53 pm
by PhilC
Just the floppy before I added anything else to complicate matters.

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Mon Jan 20, 2020 12:01 pm
by exxos
Tried a totally different STFM, behaves exactly like the one I had on test for almost 24H...

Blitter is slow on this machine as well.. but I dont think a stock CPU was on this machine, so need to look into that later.

IMG_4822.JPG
IMG_4822.JPG (141.44 KiB) Viewed 1421 times

EDIT:

So this new STFM at 8MHz with SEC or stock 68000 blitter speeds are expected. Though with running 50MHz with blitter , the scores are still slower than a STE.

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Mon Jan 20, 2020 3:22 pm
by exxos
Finally solved the slower than STE problem... And I actually found this problem about a week ago as well and totally forgot about it :roll:
basically I was not issuing DTACK on blitter access.. but as to how this even worked at all is beyond me, I can only assume the blitter times out or something and completes the cycle whether it sees DTACK or not.

So these are the scores below, the STE scores I pasted next to the SEC scores. now as I said ages ago, there is not much speed gain here at all, because over 32 MHz, we are at the point of diminishing returns big time.

ROM access just won't go any faster, its maxed out for a 55ns ROM... I did contemplate a while ago copying ROM into a SRAM chip so we could use maybe 10ns chips for example. The problem is, the SEC bus cycle time is basically 40ns anyway at 50MHz, so increasing ROM speed would not really do anything without increasing the CPU speed, which cannot really be done as the CPU is pretty much maxed out as it is already.

I think we are now at the end of the 68000 road. The CPU is maxed out, the ROM is maxed out.While we can add fast RAM to speed things up, this is not actually part of speeding up the CPU itself.

So aside from getting this working on the H4,I can finally put this puppy to bed. The booster work started October 9, 2013, so getting on for 7 years of work on these boosters!

1.jpg
1.jpg (261.33 KiB) Viewed 1403 times

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Mon Jan 20, 2020 3:29 pm
by JezC
:goodpost: :bravo: :cheers: :yay: :yay2: :dualthumbup: :cheer: :excited:

Great work to finally nail this puppy!

As part of an homage to your earlier boosters, I plan to fit one onto an H4 ( when they finally arrive). I somehow doubt it will threaten your benchmarks though :lol:

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Mon Jan 20, 2020 4:54 pm
by exxos
JezC wrote:
Mon Jan 20, 2020 3:29 pm
As part of an homage to your earlier boosters, I plan to fit one onto an H4 ( when they finally arrive). I somehow doubt it will threaten your benchmarks though :lol:
Passes the time I guess :lol:

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Mon Jan 20, 2020 4:54 pm
by exxos
Just trying out a few random things, all seems happy...

IMG_4824.JPG
IMG_4824.JPG (66.09 KiB) Viewed 1393 times
IMG_4825.JPG
IMG_4825.JPG (108.82 KiB) Viewed 1393 times
IMG_4826.JPG
IMG_4826.JPG (84.63 KiB) Viewed 1393 times
IMG_4827.JPG
IMG_4827.JPG (98.47 KiB) Viewed 1393 times

Re: V2.5 BOOSTER CURRENT PROTOTYPE STATUS (SEC BOOSTER)

Posted: Wed Jan 22, 2020 10:15 am
by exxos
I have been doing some tests with faster oscillators up to 75Mhz. The results are really strange. it does not matter if 64Mhz or 75Mhz is installed (50mhz all works fine).

Firstly it will only boot into TOS104. EMUTOS and TOS206 do not appear to even tried to boot. In fact 206 the monitor is out of sync like it is stuck in reset.

When booting 104 it will load up and generally do every test in GB6, but there are some odd graphic glitches now and then. Kinda like https://www.exxoshost.co.uk/atari/last/MEGAST/mega1.jpg sometimes. Other times text or bit planes seem to get corrupted. Though it does not make any difference if the blitter is enabled or not. I have tried 2 different motherboards, the first one does not appear to give any graphics glitches, but the second one does. Does not matter if I slow down ROM access or not either.

Even more odd, when I boot into GB6 and press "A" for all tests, it will generally run through all the tests just fine. But as soon as I press "L" I get 3 bombs.. it does not matter if I press "L" after doing tests, or as soon as GB6 loads it crashes with 3 bombs every time. Normally GB6 will bring up a YES NO box when pressing "L" but it only seems to get as far as clearing the screen.

Code: Select all

WINDOW CLOSE MAINID
CALL CS

BUT=form_alert(1,"[2][ Do you really want to   | loop all tests indefinatly | until the end of time | or something explodes? ][ YES | NO ]")
CALL CS
So far I have been unable to have find any clue why all this weirdness is happening. I have tried slowing down the ACIA access but I do not think it made any difference.

Even more oddly, it will run YAART for a hour without any errors at all. So it is not like the CPU is simply maxed out ?!