First steps with the ATW800/2

[rubber_jonnie]

@dml

OK, so I copied the config.jsn over in full rather than editing on the MSTE and used the T4 version and it's working :)

IMG20260615143254.jpg

IMG20260615142248.jpg

My config.json now looks like this:

Code: Select all

{
	{
		"output" : [
			{ "width" :						"320" },
			{ "height" :						"240" },
			{ "image_format" :			"tga" },
			{ "image_compress" :		"0" },
			{ "jpg_quality" :				"90" }
		]
	},
	{
		"transputers" : [
			{ "seurat" : 			"on"},
			{ "slot1_device" :	"T4" },
			{ "slot1" : 				"on" },
			{ "slot2_device" :	"T4" },
			{ "slot2" : 				"on" }
		]
	}
}

So aside from the image size that I tweaked, pretty much the same as you suggested.

[rubber_jonnie]

Ah - the tramtool RAM test is currently hardwired to test 2MB (GeekDot's TRAMs) and I think the vintage TRAMs you have are 1MB so that would explain (A) bad RAM test result and (B) same results on each card.

I'll post an updated version soon which figures out the RAM size automatically (or maybe a quick spin of a 1MB version) but have been busy with the other thing meantime.

That would be cool :)

[rubber_jonnie]

The config.jsn should probably look like this, for your T4 setup:

(I'll check this is set up correctly in the .zips next time I do an upload - a new version will be along this week sometime).

Code: Select all

{
	{
		"output" : [
			{ "width" :						"256" },
			{ "height" :						"192" },
			{ "image_format" :			"tga" },
			{ "image_compress" :		"0" },
			{ "jpg_quality" :				"90" }
		]
	},
	{
		"transputers" : [
			{ "seurat" : 			"on"},
			{ "slot1_device" :	"T4" },
			{ "slot1" : 				"on" },
			{ "slot2_device" :	"T4" },
			{ "slot2" : 				"on" }
		]
	}
}

But of a tabby indent mess there but you get the idea. :)

Mine is much the same now :)

[dml]

Great, it seems to be working!

Odd that the first pattern (kishimisu) comes out blue/green - the red channel is missing. Might be worth a reboot and retry with that one. Otherwise I'll check it myself on T4 before uploading a new version.

The second test (pink pattern one) looks correct so whatever that is, does not look like a real problem.

I find occasionally the ATW800/2 needs a power cycle. There is no hardware reset line tying the Atari to the Transputers or the FPGA so a warm reset or resetbutton does not suffice when things get strange ;)

[rubber_jonnie]

Great, it seems to be working!

Odd that the first pattern (kishimisu) comes out blue/green - the red channel is missing. Might be worth a reboot and retry with that one. Otherwise I'll check it myself on T4 before uploading a new version.

The second test (pink pattern one) looks correct so whatever that is, does not look like a real problem.

I find occasionally the ATW800/2 needs a power cycle. There is no hardware reset line tying the Atari to the Transputers or the FPGA so a warm reset or resetbutton does not suffice when things get strange ;)

Perhaps to do with the colour depth I was using? I had it at 32 bit to see how it did, I'll try again in 256 colour mode. I can't recall what bit depth I was in for the pink one.

I have made a mental note about the reset.

[dml]

Perhaps to do with the colour depth I was using? I had it at 32 bit to see how it did, I'll try again in 256 colour mode. I can't recall what bit depth I was in for the pink one.
I have made a mental note about the reset.

Ah that probably is the reason - I did implement 32bit but have never tested it (my MegaBus ATW800 isn't flashed for it yet). So the colour format was a 'guess'. I suppose I got it backwards RGBA-ABGR :)

The best mode to use for it currently is 65kcols. It looks a bit bandy in 256c mode because I didn't add a dither/palette generator to it yet.

So everything is probably working as well as can be expected for now at your side.

[rubber_jonnie]

Ah that probably is the reason - I did implement 32bit but have never tested it (my MegaBus ATW800 isn't flashed for it yet). So the colour format was a 'guess'. I suppose I got it backwards RGBA-ABGR :)

Ahh, ok. BTW mine is VME.

The best mode to use for it currently is 65kcols. It looks a bit bandy in 256c mode because I didn't add a dither/palette generator to it yet.

Noted, I will flip to that when I reboot the machine next.

So everything is probably working as well as can be expected for now at your side.

I think it is :)

[rubber_jonnie]

Perhaps to do with the colour depth I was using? I had it at 32 bit to see how it did, I'll try again in 256 colour mode. I can't recall what bit depth I was in for the pink one.
I have made a mental note about the reset.

Ah that probably is the reason - I did implement 32bit but have never tested it (my MegaBus ATW800 isn't flashed for it yet). So the colour format was a 'guess'. I suppose I got it backwards RGBA-ABGR :)

The best mode to use for it currently is 65kcols. It looks a bit bandy in 256c mode because I didn't add a dither/palette generator to it yet.

So everything is probably working as well as can be expected for now at your side.

Running in 65K colours on the base machine and I guess this is what it should look like?

IMG20260615151227.jpg

[dml]

Yes that looks correct colourwise - although I notice something else very weird now....

The scene parameters did not load - the params dialog has only a few buttons, missing all of the sliders and stuff to control the scene. :?


It maybe was unable to read the info.jsn from the scene directory....

Did you unpack the zip with STZip by any chance? or did you unpack it first then transfer it?


(yep looking more closely, I can see the error message in the log window - can't find the file. so it could be missing, or damaged - or maybe the current working directory for the PRG is configured to be somewhere other than the PRG itself? this is something you can change via the desktop 'install application' to set the default path for loading files)

[dml]

Actually I guess maybe you replaced the scenes directory with the T4 zip, instead of merging them? The base zip has the info.jsn files so that would explain it?