Altirra 2.80 released

[jeffpiep]

 

Make sure you have changed the translation mode in Bobterm from Atari to ASCII. Modems expect conventional line endings and won't work with ATASCII EOLs.

Thank-you for the quick response. I also didn't realize I needed to configure the modem within the 850. Now it works!

[ijor]

 

Yes, it currently emulates a 65C802, not a 6502 or 65C02. This is because I bootstrapped the full drive emulators off of the core I already had for Veronica, and I need to strip it down to a proper 6502/65C02 core before ship. The main thing to watch out for is that RMW abs,X will take 6 cycles instead of 7 and single-cycle NOPs aren't available. abs,X wrapping is fine since the bank address is discarded.

 

This means no NMOS "undocumented" instructions, right? That will break, at least, one program. One of the Archiver/Happy utilities use them, just for obfuscation purposes, IIRC.

[Kyle22]

IllOps and those who use them are evil.

[tschak909]

Disagree, sometimes it's the difference between an algorithm being fast enough, or not.

 

-Thom

[Kyle22]

Yes, but if it's that time critical, it's easy to test what type CPU you have and use the correct instructions for it. If you don't have room in RAM for both sets of code, then load an overlay from disk. It is easily done, and it can satisfy all Atari users.

 

65c02, 65c802, 65c816 get their own improved instructions to work with in the time-critical areas, and everyone is happy.

[HiassofT]

Yes, it currently emulates a 65C802, not a 6502 or 65C02. This is because I bootstrapped the full drive emulators off of the core I already had for Veronica, and I need to strip it down to a proper 6502/65C02 core before ship. The main thing to watch out for is that RMW abs,X will take 6 cycles instead of 7 and single-cycle NOPs aren't available. abs,X wrapping is fine since the bank address is discarded.

Thanks a lot for the info! These differences shouldn't be an issue for the speedy, at least the timing-critical serial code isn't affected by that.

 

During my tests I noticed another minor issue, referencing symbols in the debugger doesn't seem to work on the emulated drive - they show up fine in the disassembly though. eg:

Altirra:0> ~1s
Target now set to 1:Disk Drive CPU.
Altirra:1> ya RXSLOW $fd12
Altirra:1> x *
$FD12  manual!RXSLOW
Altirra:1> ? RXSLOW
Unable to parse expression: Unable to resolve symbol "RXSLOW"

so long,

 

Hias

[phaeron]

Update:

http://www.virtualdub.org/beta/Altirra-2.90-test32.zip

http://www.virtualdub.org/beta/Altirra-2.90-test32-src.zip

 

• Added a power-on delay setting to deal with slow disk drives. This gives the disk drives more time to init before the computer powers on. "Auto" automatically applies a delay for the 1050 Duplicator, ATR8000, and Percom drives.
• 65C816 CPU speed can now be changed on the fly. Changing the CPU type still requires a reboot.
• Fixed a bug with drive states getting screwed up when swapping or rotating images with both full and standard disk drive emulators involved.
• Full disk drives are now using a 6502/65C02 emulator as appropriate instead of 65C802.
• Fixed broken STZ zp,X instruction in 65C02 mode. (It worked correctly in 65C816 mode.)
• Formatting over an existing disk image now works -- the full drive emulators no longer force a virtual image, and fixed a bug where auto-save could fail on the new disk because the emulator forgot which file format to use.
• There is now an option to disable auto-rewind on the tape drive on cold reset.
• Added a new confirmation dialog for some cases where the emulator forces a cold reset on a settings change or on exit. The confirmation dialog has a "don't show this again" checkbox, and there is a button in Options to reset them so they show again after being ignored.

Side note, 1050 Turbo appears to have an issue with density detection when turned on with a double density disk already in it. Not sure why yet. It works fine if you "insert" the disk after it powers on, after which it sees the disk change.

 

During my tests I noticed another minor issue, referencing symbols in the debugger doesn't seem to work on the emulated drive - they show up fine in the disassembly though. eg:

 

This is another issue with multi-target support that I need to work out. Looks like some portions of the debugger are filtering by target while others are not. Issue is, there's only one custom symbol module, so I'm not sure how to resolve this just yet.

[Triads]

Even more excellent work.. Thanks Phaeron..

[Keatah]

Back in the day I did various upgrades to my Apple II+ and Apple //e. I got a lowercase chip, a shift-mod for proper use of the <shift> key. Various ROM upgrades, new character set, new MouseText, 65C02, and of course all sorts of interface cards.

 

Everytime a new version of Altirra comes out it's like like the Apple II days all over again!

[toddtmw]

• Added a new confirmation dialog for some cases where the emulator forces a cold reset on a settings change or on exit. The confirmation dialog has a "don't show this again" checkbox, and there is a button in Options to reset them so they show again after being ignored.

 

@phaeron,

 

Thank you!

 

-Todd

[solo/ng]

Hi Phaeron, is there a way to export data from Altirra Profiler - Detailed View? Currently I don't even see a way to select and copy the text from that window. It would be great if I could save the collected data to a file (e.g. CSV format) for further analysis.

If there is no such feature, it would be hard to add it? We're profiling our new game at the moment that we want to show in few days on the party in Warsaw and such feature could help us improve it's performace.

 

Thanks in advance.

Edited May 4, 2017 by solo/ng

[Mclaneinc]

New version crashes if you don't have the needed drive firmware rom setup, tried with US Doubler.

 

Not a bug, just a bit unfriendly

[phaeron]

http://www.virtualdub.org/beta/Altirra-2.90-test33.zip

http://www.virtualdub.org/beta/Altirra-2.90-test33-src.zip

 

• Fixed XEP80 tab stops not being properly reinitialized after master reset.
• Fixed some timing issues with SIO command line emulation for Percom and XF551 emulators.
• Fixed crash in 6502 coprocessor core with missing drive firmware.
• Implemented config latch for VBXE (but not actual config upload).
• Fixed an issue in ATBasic with PMBASE getting overwritten even when PMGRAPHICS was not used. 1.54 attached.
• Added copy-as-CSV for profiler list views.

 

atbasic.bin

atbasicx.xex

[_The Doctor__]

Coool, now that's a good pile of fixes.... hmmm this is taking shape. Emulation is improving!

[+MrFish]

• Fixed XEP80 tab stops not being properly reinitialized after master reset.

 

Thanks, the menu looks better now.

[Keatah]

Coool, now that's a good pile of fixes.... hmmm this is taking shape. Emulation is improving!

 

It's a force of nature.

[pseudografx]

Hi phaeron, did you read my private message?

[+MrFish]

• Fixed XEP80 tab stops not being properly reinitialized after master reset.

 

Thanks, the menu looks better now.

 

I should clarify for those who weren't following the thread related to this fix.

 

Thanks, the menu for AtariWriter 80 is now correctly centered.

[solo/ng]

Thanx!

[Swami]

Hello,

 

We noticed the 5200 controller and 5200 trackball in the canned controllers for Altirra 400/800 do not work as real or emulated hardware for the 400/800 and, so, perhaps should be removed from the Input controller options.

 

Thank you.

[antrykot]

Hello,

 

Let me report a couple of issues with POKEY chip in Altirra that I've found

1. Setting POKEY registers as follows
AUDF1=100
AUDC1=175
AUDF3=0
AUDC3=0
AUDCTL=5
SKCTL=3
should cause it to generate a narrow impulse train. It seems to work reliably
on a real hardware, however on Altirra it sometimes works correctly and
sometimes it doesn't. Sometimes the impulse train is generated but it is very
quiet. Most often no sound is produced at all.

2. Behavior of a low timer in a linked timers pair is incorrect when timers
are driven by 64 kHz or 15 kHz clock (it's OK with 1.79 MHz clock).
For example
AUDF1=33
AUDC1=170
AUDF2=1
AUDC2=0
AUDCTL=16
SKCTL=3
the generated sound is wrong and is disturbed by keystrokes. This happens only
when timer emulation is in deferred mode. Forcing it to go out of deferred
mode by enabling interrupts for timer 1 fixes the problem.

3. The pitch of a square wave generated with the following settings
AUDF1=99
AUDC1=170
AUDF2=199
AUDC2=0
AUDCTL=0
SKCTL=139
is one octave lower than it should be.

4. Altirra doesn't implement additional two-cycle delay when timers 1 and 2
are linked and two-tone mode is enabled at the same time.
For example, with the following settings:
AUDF1=10
AUDC1=0
AUDF2=0
AUDC2=136
AUDCTL=208
SKCTL=139
the pitch is too high by almost 2 semitones.

[tschak909]

now THAT'S a bug report.

 

-Thom

[Mclaneinc]

now THAT'S a bug report.

 

-Thom

 

Damn, I'm now suffering from bug report 'inadequacy'....I may need bug report Viagra..

[Mclaneinc]

@Antrykot....What version of Altirra are you using, just for reference and are you using system roms or built in Altirra bios...I doubt if it actually matters but just in case it does..

[antrykot]

I'm using Altirra 2.80, roms don't matter here. There were no significant changes in pokey implementation between 2.80 and 2.90-test33, so I assume those bugs are still present.

Edited May 12, 2017 by antrykot