Minimig (Amiga) core discussion

https://github.com/MiSTer-devel/Main_MiSTer/wiki

Moderators: Mug UK, Zorro 2, Greenious, spiny, Sorgelig, Moderator Team

zomgugoff
Atari freak
Atari freak
Posts: 66
Joined: Tue Jan 29, 2019 10:12 am

Re: Minimig (Amiga) core discussion

Postby zomgugoff » Tue Sep 24, 2019 11:12 am

Is direct video supposed to be working on Minimig? I saw an update that said some changes were made for it, but I can't get a signal out of it at all. OSSC and a CRT monitor both don't seem to detect anything, even with scandoubling enabled.

EDIT: Nevermind. Forgot to update.

apolkosnik
Atari freak
Atari freak
Posts: 69
Joined: Sat May 18, 2019 3:20 pm

Re: Minimig (Amiga) core discussion

Postby apolkosnik » Tue Sep 24, 2019 1:53 pm

I was able to successfully run the cputest for RTD instruction, but I didn't have the time to test any further.
Technically, the change only entails the execution change to RTD, so this state is better than the mess that was happening before implementation.
I'll take a look later today at your attachment, and try to set up some more test cases.
-A

kolla wrote:
Sorgelig wrote:here is the test version with RTD:
Minimig.zip
Need to test the build against stability


I can confirm that the glitches are gone.

Also, with this core, LoadModule causes a freeze and flashing "power led", but after a reset (from OSD or ctrl-a-a) it reboots and continues, with kickstart module loaded.

The test binaries I got from Oliver still do not work with the test core with RTD.
(meh, silly forum doesn't accept lha extension, so here is a "pdf")

User avatar
remowilliams
Captain Atari
Captain Atari
Posts: 246
Joined: Mon Apr 02, 2007 1:49 am

Re: Minimig (Amiga) core discussion

Postby remowilliams » Tue Sep 24, 2019 2:59 pm

Sorgelig wrote:here is the test version with RTD:
Minimig.zip
Need to test the build against stability


Seeing image corruption that did not exist with previous test version.

Image

Milongero
Atari User
Atari User
Posts: 31
Joined: Sun Nov 25, 2018 2:24 pm
Location: Germany

Re: Minimig (Amiga) core discussion

Postby Milongero » Tue Sep 24, 2019 3:06 pm

Hello,

interesting problem.

I tested Lotus 2 and Lotus III on the new Amiga core and get a Guru meditation at Lotus 2 and the core hangs completely at Lotus IIII.
By the way, I have both - a MIST and a MISTer. And both (MIST & MISTer) have exactly the same problem.
The games ROMs work on the old MIST core from 2017.

Greetings Lutz

Sorgelig
Ultimate Atarian
Ultimate Atarian
Posts: 5358
Joined: Mon Dec 14, 2015 10:51 am
Location: Russia/Taiwan

Re: Minimig (Amiga) core discussion

Postby Sorgelig » Tue Sep 24, 2019 3:42 pm

remowilliams wrote:
Sorgelig wrote:here is the test version with RTD:
Minimig.zip
Need to test the build against stability


Seeing image corruption that did not exist with previous test version.

Image

corruption starts from the beginnig of play?
I found that Gloom behaves weird if you stay on one place for some time like 10 minutes which is not related to instability issue.

User avatar
remowilliams
Captain Atari
Captain Atari
Posts: 246
Joined: Mon Apr 02, 2007 1:49 am

Re: Minimig (Amiga) core discussion

Postby remowilliams » Tue Sep 24, 2019 3:52 pm

Sorgelig wrote:corruption starts from the beginnig of play?
I found that Gloom behaves weird if you stay on one place for some time like 10 minutes which is not related to instability issue.


Yes right from start, three cold boots with same result. BTW - last test core ran Gloom for several hours (while player firing even).

Sorgelig
Ultimate Atarian
Ultimate Atarian
Posts: 5358
Joined: Mon Dec 14, 2015 10:51 am
Location: Russia/Taiwan

Re: Minimig (Amiga) core discussion

Postby Sorgelig » Tue Sep 24, 2019 5:10 pm

Ne build to try:
Minimig2.zip
You do not have the required permissions to view the files attached to this post.

Milongero
Atari User
Atari User
Posts: 31
Joined: Sun Nov 25, 2018 2:24 pm
Location: Germany

Re: Minimig (Amiga) core discussion

Postby Milongero » Tue Sep 24, 2019 5:53 pm

Sorry
but
Lotus Turbo Challenge 2 and Lotus III: The Ultimate Challenge still not running.

User avatar
remowilliams
Captain Atari
Captain Atari
Posts: 246
Joined: Mon Apr 02, 2007 1:49 am

Re: Minimig (Amiga) core discussion

Postby remowilliams » Tue Sep 24, 2019 6:49 pm

Sorgelig wrote:Ne build to try:
Minimig2.zip


Not seeing previous issue, or other issues. Gloom ran for 1+ hours.

Milongero wrote:Lotus Turbo Challenge 2 and Lotus III: The Ultimate Challenge still not running.


WHDLoad or floppy? I just tried WHDLoad and they seem to be working.

Milongero
Atari User
Atari User
Posts: 31
Joined: Sun Nov 25, 2018 2:24 pm
Location: Germany

Re: Minimig (Amiga) core discussion

Postby Milongero » Wed Sep 25, 2019 5:14 am

Floppy

With ans without turbo

GoingDown
Atariator
Atariator
Posts: 28
Joined: Sun Jan 20, 2019 7:20 am

Re: Minimig (Amiga) core discussion

Postby GoingDown » Wed Sep 25, 2019 9:52 am

Milongero wrote:Lotus Turbo Challenge 2 and Lotus III: The Ultimate Challenge still not running.


I tested both WHDLoad and floppy. In PAL mode, both work. In NTSC mode, only WHDload version works. Floppy version hangs already on crack/trainer screen

Sorgelig
Ultimate Atarian
Ultimate Atarian
Posts: 5358
Joined: Mon Dec 14, 2015 10:51 am
Location: Russia/Taiwan

Re: Minimig (Amiga) core discussion

Postby Sorgelig » Wed Sep 25, 2019 10:01 am

WHDLoad is not only just for convenient load from HDD. It also adds a lot of tweaks for games against CPU speed, memory config and so on.

apolkosnik
Atari freak
Atari freak
Posts: 69
Joined: Sat May 18, 2019 3:20 pm

Re: Minimig (Amiga) core discussion

Postby apolkosnik » Wed Sep 25, 2019 1:28 pm

Kolla,
I've looked through the testkey4, and it shows no sign of "4e 74" sequence in it, so it's likely not RTD that is causing the problem.
I ran the cputest against a much larger set of instructions, and it produced quite a big text file. It'll take me a bit to go through it and validate. Maybe I'm just running it wrong, since I think I'm seeing some errors resulting from comparing 32bit vs 24bit addresses.

kolla wrote:
Sorgelig wrote:here is the test version with RTD:
Minimig.zip
Need to test the build against stability


I can confirm that the glitches are gone.

Also, with this core, LoadModule causes a freeze and flashing "power led", but after a reset (from OSD or ctrl-a-a) it reboots and continues, with kickstart module loaded.

The test binaries I got from Oliver still do not work with the test core with RTD.
(meh, silly forum doesn't accept lha extension, so here is a "pdf")

harryd91
Retro freak
Retro freak
Posts: 10
Joined: Wed Sep 25, 2019 1:52 pm

Re: Minimig (Amiga) core discussion

Postby harryd91 » Thu Sep 26, 2019 12:35 pm

I'm getting tons of errors on every version after 20190717 (that version works flawlessly). Any ADF I try comes up with checksum or read/write errors as do most hdfs. Loads of random restarts/meditation errors as well. Have tried numerous configurations too and still no luck. Most kickstarts do not even work at all.

Sorgelig
Ultimate Atarian
Ultimate Atarian
Posts: 5358
Joined: Mon Dec 14, 2015 10:51 am
Location: Russia/Taiwan

Re: Minimig (Amiga) core discussion

Postby Sorgelig » Thu Sep 26, 2019 1:02 pm

harryd91 wrote:I'm getting tons of errors on every version after 20190717 (that version works flawlessly). Any ADF I try comes up with checksum or read/write errors as do most hdfs. Loads of random restarts/meditation errors as well. Have tried numerous configurations too and still no luck. Most kickstarts do not even work at all.

looks like SDRAM problem.
20190717 uses much lower SDRAM frequency - that's why it works better for you.

User avatar
remowilliams
Captain Atari
Captain Atari
Posts: 246
Joined: Mon Apr 02, 2007 1:49 am

Re: Minimig (Amiga) core discussion

Postby remowilliams » Thu Sep 26, 2019 2:40 pm

I usually do not use floppy images, but I decided to try some known good ones (with a very vanilla config - 68K/512MB Chip/KS12/13/IDE off). There is definitely some inconsistently odd behavior with the core and floppy images.

Sometimes it will just refuse to even boot them (repeatedly showing the Workbench Hand) and sometimes they will mysteriously crash while loading. It usually takes several tries of reloading a config/rebooting the core to get it to work properly.

Sorgelig
Ultimate Atarian
Ultimate Atarian
Posts: 5358
Joined: Mon Dec 14, 2015 10:51 am
Location: Russia/Taiwan

Re: Minimig (Amiga) core discussion

Postby Sorgelig » Thu Sep 26, 2019 2:52 pm

I didn't try any config combinations. I use standard AGA/2M/KS3.1 config and time after time load some demos from FDD. Didn't find any oddities..

harryd91
Retro freak
Retro freak
Posts: 10
Joined: Wed Sep 25, 2019 1:52 pm

Re: Minimig (Amiga) core discussion

Postby harryd91 » Thu Sep 26, 2019 4:25 pm

Sorgelig wrote:
harryd91 wrote:I'm getting tons of errors on every version after 20190717 (that version works flawlessly). Any ADF I try comes up with checksum or read/write errors as do most hdfs. Loads of random restarts/meditation errors as well. Have tried numerous configurations too and still no luck. Most kickstarts do not even work at all.

looks like SDRAM problem.
20190717 uses much lower SDRAM frequency - that's why it works better for you.


Ok thanks think this might be the case as memtest is showing errors above 140Mhz - what frequency is it running at now btw? The wiki says 86Mhz? If so then shouldnt it be fine if testing at 140Mhz is ok?

Sorgelig
Ultimate Atarian
Ultimate Atarian
Posts: 5358
Joined: Mon Dec 14, 2015 10:51 am
Location: Russia/Taiwan

Re: Minimig (Amiga) core discussion

Postby Sorgelig » Thu Sep 26, 2019 5:01 pm

harryd91 wrote:Ok thanks think this might be the case as memtest is showing errors above 140Mhz - what frequency is it running at now btw? The wiki says 86Mhz? If so then shouldnt it be fine if testing at 140Mhz is ok?

Unfortunately memtest cannot test it fully as work depends on core as well.
20190717 uses 86MHz. Newer version uses 114MHz.

kolla
Captain Atari
Captain Atari
Posts: 267
Joined: Thu Sep 17, 2015 11:39 pm
Contact:

Re: Minimig (Amiga) core discussion

Postby kolla » Thu Sep 26, 2019 8:53 pm

So newest core is 16x while old was 12x, to use terminology from a certain other project :) (or x8 and x6 if we use plain A1200 020 as base). Would it make sense to build a few cores with different clock for each release (like that certain other project does)?
-- kolla

User avatar
retrofun
Atari freak
Atari freak
Posts: 52
Joined: Sat Jan 12, 2019 3:12 pm

Re: Minimig (Amiga) core discussion

Postby retrofun » Thu Sep 26, 2019 10:03 pm

Fixed the RTR instruction (for MiST)!

Does NOT fix any issue I currently know (testkey4, Pinball Illusions).

Latest cputest 20190924 from Toni Wilen:

Before:

Code: Select all

CPUlvl=2, Mask=ffffffff Code=00740000
RTR:
data/68020/rtr/0000.dat. 0...

0:      4e77    rtr

A7: expected 00777ffe but got 00778004
Registers before:
D0: 00000010 D1: 00000000 D2: ffffffff D3: ffffff00
D4: ffff0000 D5: 80008080 D6: 7fff7fff D7: aaaaaaaa
A0: 00000000 A1: 00000080 A2: 00008000 A3: 00007fff
A4: fffffffe A5: ffffff00 A6: 0073ff00 A7:*00777ff8
SR: 0400   PC: 00740000 ISP: 0077ff80 MSP: 00780000
T1=0 T0=0 S=0 M=0 X=0 N=0 Z=0 V=0 C*0
Registers after:
D0: 00000010 D1: 00000000 D2: ffffffff D3: ffffff00
D4: ffff0000 D5: 80008080 D6: 7fff7fff D7: aaaaaaaa
A0: 00000000 A1: 00000080 A2: 00008000 A3: 00007fff
A4: fffffffe A5: ffffff00 A6: 0073ff00 A7:*00778004
SR: 0401   PC: 00741c3a ISP: 0077ff80 MSP: 00000000
T1=0 T0=0 S=0 M=0 X=0 N=0 Z=0 V=0 C*1
OK: No exception generated


With https://github.com/retrofun/minimig-mis ... 429d2c4287

Code: Select all

CPUlvl=2, Mask=ffffffff Code=00740000
RTR:
data/68020/rtr/0000.dat. 0...
data/68020/rtr/0001.dat. 226...
All tests complete (total 226).

apolkosnik
Atari freak
Atari freak
Posts: 69
Joined: Sat May 18, 2019 3:20 pm

Re: Minimig (Amiga) core discussion

Postby apolkosnik » Thu Sep 26, 2019 10:34 pm

The weird thing is that we have instances where the 32-bit addresses get truncated to 24-bits, but when I check with HRTmon, it looks like the 32-bit addresses are not truncated. It could be either something in the ini file or a bug in Tony's program.

4524: 5c39 4508 dbce addq.b #6,$4508dbce
Memory byte write: address 0008dbce, expected 2f but got 29
You do not have the required permissions to view the files attached to this post.
Last edited by apolkosnik on Thu Sep 26, 2019 11:20 pm, edited 1 time in total.

User avatar
retrofun
Atari freak
Atari freak
Posts: 52
Joined: Sat Jan 12, 2019 3:12 pm

Re: Minimig (Amiga) core discussion

Postby retrofun » Thu Sep 26, 2019 10:58 pm

apolkosnik wrote:...
Maybe I'm just running it wrong, since I think I'm seeing some errors resulting from comparing 32bit vs 24bit addresses.

Did you create the 68020 test data with 24bit address space (instead of 32bit address space)?

cputestgen.ini:

Code: Select all

cpu_address_space=24


apolkosnik wrote:I ran the whole suite, and there's a bunch of things that need fixing... Attached.


With 68020 32bit address space test data there are less failures:
cputest-20190927.txt
You do not have the required permissions to view the files attached to this post.

apolkosnik
Atari freak
Atari freak
Posts: 69
Joined: Sat May 18, 2019 3:20 pm

Re: Minimig (Amiga) core discussion

Postby apolkosnik » Thu Sep 26, 2019 11:21 pm

lol, makes sense.

retrofun wrote:
apolkosnik wrote:...
Maybe I'm just running it wrong, since I think I'm seeing some errors resulting from comparing 32bit vs 24bit addresses.

Did you create the 68020 test data with 24bit address space (instead of 32bit address space)?

cputestgen.ini:

Code: Select all

cpu_address_space=24


apolkosnik wrote:I ran the whole suite, and there's a bunch of things that need fixing... Attached.


With 68020 32bit address space test data there are less failures:
cputest-20190927.txt

apolkosnik
Atari freak
Atari freak
Posts: 69
Joined: Sat May 18, 2019 3:20 pm

Re: Minimig (Amiga) core discussion

Postby apolkosnik » Fri Sep 27, 2019 12:26 am

ok, so the first thing that jumps at me is issue with a whole addressing mode, namely Address register Indirect with predecrement, when it happens to be used for both source and dest.

retrofun wrote:
apolkosnik wrote:...
With 68020 32bit address space test data there are less failures:
cputest-20190927.txt


Return to “MiSTer”

Who is online

Users browsing this forum: Dacide, sigman and 19 guests