The Incomplete Works of Infocom, Inc.

[jindroush]

I don't think Altirra emulates U1MB, and also this sounds like one of the corner cases which nobody 'needs' to be emulated. And, we still don't know if it works now on real hw, have to wait for confirmation

[drpeter]

Altirra does emulate U1MB- there's a check box under memory configuration:

 

[drpeter]

40 minutes ago, jindroush said:

this sounds like one of the corner cases which nobody 'needs' to be emulated

I think @phaeron's view is that anything that works on real hardware should work on Altirra and vice versa - in theory, one should be able to develop with Altirra as a target and be confident that things will work on real hardware

Edited August 12, 2023 by drpeter

[drpeter]

16 hours ago, DrVenkman said:

 

The test system is a PAL system with VBX and U1MB. 

What is your U1MB and VBXE setup- did you match it exactly in Altirra?

Edited August 12, 2023 by drpeter

[+DrVenkman]

3 hours ago, jindroush said:

I don't think Altirra emulates U1MB, and also this sounds like one of the corner cases which nobody 'needs' to be emulated. And, we still don't know if it works now on real hw, have to wait for confirmation

It does, and I use it that way all the time. 

10 minutes ago, drpeter said:

What is your U1MB and VBXE setup- did you match it exactly in Altirra?

Yes, in every way that matters (extended memory settings, OS selection, PBI BIOS, BASIC state, Hard Disk state, VBXE enabled and base address …) This ain’t my first rodeo.  

[jindroush]

@DrVenkman I meant I'm waiting for confirmation that fixed 0.9.2 is really fixed and working on your setup.

[drpeter]

3 hours ago, jindroush said:

I don't think Altirra emulates U1MB, and also this sounds like one of the corner cases which nobody 'needs' to be emulated. And, we still don't know if it works now on real hw, have to wait for confirmation

OK, it seems that enabling PORTB memory sharing for VBXE in Altirra replicates the problem:

unfortunately, with this setting the problem persists in version 0.9.2

 

I assume that there is some way to disable this setting with real VBXE hardware, so that this will work??

[drpeter]

4 minutes ago, drpeter said:

OK, it seems that enabling PORTB memory sharing for VBXE in Altirra replicates the problem:

unfortunately, with this setting the problem persists in version 0.9.2

 

I assume that there is some way to disable this setting with real VBXE hardware, so that this will work??

 

@phaeron will be relieved to learn that this seems to be a settings dependency rather than an emulation failure

[+DrVenkman]

4 minutes ago, jindroush said:

@DrVenkman I meant I'm waiting for confirmation that fixed 0.9.2 is really fixed and working on your setup.

Will test once I’ve had coffee. It’s only 8:13AM here.

4 minutes ago, drpeter said:

I assume that there is some way to disable this setting with real VBXE hardware, so that this will work??

That is not a standard use case for VBXE these days - as I understand it (and I am far from a VBXE expert) the first versions of the VBXE core allowed using the board for PORTB extended memory but since many/most people these days who have VBXE already have U1MB boards in their machines, this core is deprecated. People use U1MB for extended memory and VBXE strictly for video.

[drpeter]

14 minutes ago, DrVenkman said:

Will test once I’ve had coffee. It’s only 8:13AM here.

That is not a standard use case for VBXE these days - as I understand it (and I am far from a VBXE expert) the first versions of the VBXE core allowed using the board for PORTB extended memory but since many/most people these days who have VBXE already have U1MB boards in their machines, this core is deprecated. People use U1MB for extended memory and VBXE strictly for video.

Unless I've misunderstood (I know very little about VBXE) I'm a bit confused again then what the problem is with your setup, if you don't have memory sharing enabled.

It seems odd that enabling memory sharing in Altirra seems to replicate your problem. Two different potential issues with the same symptoms perhaps?

Edited August 12, 2023 by drpeter

[+DrVenkman]

26 minutes ago, drpeter said:

Unless I've misunderstood (I know very little about VBXE) I'm a bit confused again then what the problem is with your setup, if you don't have memory sharing enabled.

It seems odd that enabling memory sharing in Altirra seems to replicate your problem.

The version I was testing last night DOES run with 80 column VBXE text under Altirra, but note that option is un-checked by default:

 

******

 

But in any case, the updated version 0.9.2 seems to be working correctly on real hardware now. I tried a Z3 game with VBXE and Extended Memory (a combination which did not work yesterday), and several Z4 and Z5 games and all load/display correctly with VBXE 80 column text. Thanks!

 

EDIT: For the sake of completeness, I should note that this version also continues to operate correctly under emulation. One odd/interesting fact that is probably a result of my settings somewhere in Altirra: the base loader program and the interpreter/game files operate faster on real hardware loading from my FujiNet device than they do in emulation, despite whatever drive emulation settings I select in Altirra. I gotta have something misconfigured in Altirra somewhere. But the important thing is, your updated version runs on real hardware and emulation now with correct VBXE 80 column text!

[jindroush]

Glad to hear that the simple fix fixed it all. I have some (maybe irrational) problem with VBXE docs and understanding them, so I didn't exactly follow them and used provided example.

Funny thing regarding the speed of loading - I'm using only 'unreal drive', ie. no disk device in Peripheals/Devices, in Peripheals/Disk 'accurate sector timing off' and in Computer/Accelearation: D patch on, D burst on, Both SIOV+PBI, SIO Override on. That should be fastest possible AFAIK.

BTW: How readable are the three 'videomodes' on your setup? I've decided to leave 'standard' colors in 40 cols and VBXE, and BW for 80 cols. Can you share some 'screenshots' of the monitor?

Edited August 12, 2023 by jindroush

[drpeter]

15 minutes ago, DrVenkman said:

the updated version 0.9.2 seems to be working correctly on real hardware now.

That's excellent!

 

Without knowing exactly how the @jindroush's VBXE screen driver is implemented, I'm still left unsure (and curious) exactly what the extended memory problem was that led to a difference in how Altirra and your real hardware setup were behaving (as I noted before, both 0.9.0 and 0.9.2 don't work when Altirra has VBXE PORTB memory-sharing enabled)

[jindroush]

Just now, drpeter said:

Without knowing exactly how the @jindroush's VBXE screen driver is implemented, I'm still left unsure (and curious) exactly what the extended memory problem was that led to a difference in how Altirra and your real hardware setup were behaving (as I noted before, both 0.9.0 and 0.9.2 don't work when Altirra has VBXE PORTB memory-sharing enabled)

It's not a problem of the driver itself, but the problem of cooperation between two devices. Both VBXE and extended memory share $4000-7FFF memory window, so in some cases the writes don't write to VBXE memory, but to extended memory. The solution is to turn off access to extended memory before 'opening' window to VBXE memory. The fix was five pairs of LDA #$FF STA PORTB instructions before touching VBXE memory.

[drpeter]

1 minute ago, jindroush said:

I've decided to leave 'standard' colors in 40 cols and VBXE, and BW for 80 cols.

I suppose in an ideal world (memory permitting) there would be a setting to select the colour(s) (for VBXE modes I guess foreground and background could be fully independent)  

[jindroush]

1 minute ago, drpeter said:

I suppose in an ideal world (memory permitting) there would be a setting to select the colour(s) (for VBXE modes I guess foreground and background could be fully independent)  

I can do that in settings, but that would need some kind of settings storage. Will think about it.

[drpeter]

11 minutes ago, jindroush said:

It's not a problem of the driver itself, but the problem of cooperation between two devices. Both VBXE and extended memory share $4000-7FFF memory window, so in some cases the writes don't write to VBXE memory, but to extended memory.

Any idea what the difference was between the base Altirra VBXE + U1MB setup and the base real hardware VBXE + U1MB setup that led to 0.9.0 working on Altirra but not on real hardware?

[jindroush]

Most probably Altirra does not emulate this behavior (quoting from VBXE docs): "some memory extensions that utilises EXTSEL signal have priority above VBXE".

[+DrVenkman]

1 hour ago, jindroush said:

BTW: How readable are the three 'videomodes' on your setup? I've decided to leave 'standard' colors in 40 cols and VBXE, and BW for 80 cols. Can you share some 'screenshots' of the monitor?

Can you remind me what the three different modes are? I was testing with VBXE text for all three game types (Z3, Z4, Z5). I know there's also a sofware 80 column mode. What is the third?

[jindroush]

Number 1: 40 col mode probably does not need screenshot, as it's default and well known, number 2: 80 col VBXE and number 3: software 80 columns.

[+DrVenkman]

3 minutes ago, jindroush said:

Number 1: 40 col mode probably does not need screenshot, as it's default and well known, number 2: 80 col VBXE and number 3: software 80 columns.

Okay, here you go. Note that it is EXTREMELY difficult to get good, representative screenshots of a CRT. 

My signal chain here is acceptable but not ideal: 15Khz RGB output from the VBXE is going into a cheap GBS8220 VGA scan-doubler and the display is an early 00's 17" Dell SVGA screen. 

[jindroush]

Thanks, the screenshots look great. I think that even the software 80 columns are quite acceptable. VBXE is great, as expected.

What's strange - can you compare running A Mind Forever Voyaging (any VBXE game) in Altirra and on real HW? Because when you check my screenshot of AMFV in the very first post of this thread, you can see, that the 'blue' is also in background, so the letters don't begin on leftmost edge of the blue rectangle. On your screenshot it seems different, the letters begin immediately on the left.

[+MrFish]

10 minutes ago, DrVenkman said:

My signal chain here is acceptable but not ideal: 15Khz RGB output from the VBXE is going into a cheap GBS8220 VGA scan-doubler and the display is an early 00's 17" Dell SVGA screen. 

Looks good to me. That 80-column software mode is some pretty small text (for 320 pixel width).

 

[drpeter]

1 hour ago, jindroush said:

Most probably Altirra does not emulate this behavior (quoting from VBXE docs): "some memory extensions that utilises EXTSEL signal have priority above VBXE".

@phaeron any thoughts?

[drpeter]

13 minutes ago, jindroush said:

the letters begin immediately on the left.

I think (based on my experience with Altirra) that this may be VBXE in 'attract mode' with a cycling background colour surrounding the display window