@scabit

Quote:

wrote a gui config program for UAE called UAEConfig many years ago.

That was you? Oh so it was. Yes I tried that and experimented with it. I'm sorry but I thought it was bad It used to annoy me by overwriting configs IIRC and I don't think volumes were written correctly. It did show promise and I was hoping for an update.

Feel free to tell me how one of my programs went sour.

@ALL

Linux-ppc and MacOSX-ppc built and ready.

http://threader.zapto.org/e-uae-ppc-jit/beta1/

e-uae-ppc-jit-beta1-lnx:

Last edited by mike on 18-Jan-2014 at 04:38 PM.
Last edited by mike on 18-Jan-2014 at 04:03 PM.
Last edited by mike on 18-Jan-2014 at 02:12 AM.
Last edited by mike on 18-Jan-2014 at 02:12 AM.
Last edited by mike on 18-Jan-2014 at 01:32 AM.
Last edited by mike on 18-Jan-2014 at 12:04 AM.
Last edited by mike on 17-Jan-2014 at 11:48 PM.

_________________
C= Amiga addict
,,,
(Oo)
⎛☮ໄ
ﮑὠՀ
Couldn't care less what other people think, seeing that there's concrete evidence they don't.

@Hypex

Yes....the very first version had a few problems. But due to popular user demand, I updated the program twice I think over the years and cleaned up a bunch of stuff, added some things, and never had any complaints or problems with it after that.
Are you sure you tried the newest version of it?

Scott

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

Looks like I used the latest 1.4 from the Depot. I think my main problem was I wanted to just load a config but "starting" UAE overwrote it everytime! So I don't like things saving unless I press a "save" button. I would have liked it to present a list of configs (apart from the default) I could select and run. And use it as a UAE launcher. But its methodology was too dangerous for me.

@Hypex

[OFF] ILLEGAL instruction is not supported by Petunia, that is intentional. None of the exceptions are supported, that needs a switch between the user and the supervisor mode, which is complicated.
But it is in the documentation somewhere, I guess. At least it was included in the readme I had done.

_________________
Álmos Rajnai

@mike

on what machine and cpu you run this ?

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@Rachy

Is it complicated because it just is or because it follows the static emulator by emulatng the instrunctions in the super mode of the host native CPU?

Given it's emulated I don't know why it would do that. But I was told it does!

Anyway, so you can't exactly switch to static emulator if encoutering a trap code? And I take it you only emulate user mode then

So many questions, so that bug report I made about trap code exceptions is moot?

BTW, only docs I see are the old emulators.readme and latest release notes. Your work is going missing.

At this point I feel like complaining like the OS4 USB2 people do with Poseidon. "Gee, I wish OS4 had the MorphOS 68K emulator, it runs MonmAm right out of the box!"

@tlosm

Powerbook 5,6 (1.67ghz ppc g4). But via VNC so should actually be faster, i think, atleast thats my running theory right now.


Btw, Compiled the OSX binary with gcc-4.6, it's ever so slightly faster
http://threader.zapto.org/e-uae-ppc-jit/beta1/E-uae-gcc4.6.app.tar

However , picasso96 had to be compiled with gcc-4.0.1 - anyone got any clues? (gcc-4.4 has the same problem)
gcc-4 -DHAVE_CONFIG_H -I. -I. -I. -I../src/include -I../src -I../src -I.. -I/Library/Frameworks/SDL.framework/Headers -D_REENTRANT -DCPUEMU_0 -DCPUEMU_5 -DCPUEMU_6 -DFPUEMU -DJIT -DAGA -DAUTOCONFIG -DFILESYS -DBSDSOCKET -DSUPPORT_THREADS -DUSE_SDL -DFDI2RAW -DDEBUGGER -DSAVESTATE -DENFORCER -DACTION_REPLAY -DXARCADE -g -O2 -mdynamic-no-pic -fomit-frame-pointer -Wall -Wno-unused -Wno-format -W -Wmissing-prototypes -Wstrict-prototypes -DREGPARAM= -g -O2 -DNO_MAIN_IN_MAIN_C -DUSE_GL -MT picasso96.o -MD -MP -MF ".deps/picasso96.Tpo" -c -o picasso96.o picasso96.c
/var/tmp//ccTceUGS.s:2196:Parameter error: r0 not allowed for parameter 2 (code as 0 not r0)
/var/tmp//ccTceUGS.s:6401:Parameter error: r0 not allowed for parameter 2 (code as 0 not r0)

Error appears to be triggered here

asm volatile (
"addi %2, %2, -1 \n\
mtctr %2 \n\
lwz %3, 0(%1) \n\
1: stwbrx %3, 0, %0 \n\
addi %0, %0, 4 \n\
lwzu %3, 4(%1) \n\
bdnz 1b \n\
stwbrx %3, 0, %0"
: "+r" (dst), "+r" (src), "+r" (words), "=r" (tmp)
:
: "ctr", "memory");


addi %0, %0, 4 does addi r0,r0,4

Last edited by mike on 21-Jan-2014 at 12:58 AM.
Last edited by mike on 21-Jan-2014 at 12:32 AM.
Last edited by mike on 21-Jan-2014 at 12:11 AM.
Last edited by mike on 20-Jan-2014 at 05:27 PM.

_________________
C= Amiga addict
,,,
(Oo)
⎛☮ໄ
ﮑὠՀ
Couldn't care less what other people think, seeing that there's concrete evidence they don't.

@mike

That's correct the code is using outputs as inputs, in addi, the first “:” in the asm volatile is for outputs, the second “:” is for inputs. And the 3rd is for reserving registers, there for the use of “memory” her does not make sense, what you should do is, replace “memory” whit “r0”; so that “r0” is not used by input or output registers. Also you should move “src, words to the second “:” symbol.

And also because dst is output it should be written “=r” not “+r”, but in this case maybe you do not want to keep the changes to dst address, so maybe its just fine whit just “r”

Unless you need to store tmp, then you should remove "=r" (tmp) and reserve a register instead, remember it is faster to store to a register then to variable, as now the result is going be returned to tmp.

Then it should compile as normal, i hope.

Last edited by NutsAboutAmiga on 21-Jan-2014 at 11:59 AM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 11:59 AM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 11:33 AM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 11:24 AM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 11:23 AM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 11:18 AM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@mike

addi %2, %2, -1

basically an subi, subtract 1 from counter.

mtctr %2

set the ctr register for number of loops.

lwz %3, 0(%1)

load from src into tmp.

stwbrx %3, 0, %0

byteswap the content of TMP and stored into DST.

addi %0, %0, 4
lwzu %3, 4(%1)
bdnz 1b

increment DST whit 4, loads from src and overwrite TMP, whit the content of (src+4), to no need to incrementing src (as lwzu does that automagicly), the loop continus until the CTR is zero.

stwbrx %3, 0, %0

store result of TMP into address (0+dst) as byte swapped.

So basically this is routine that loads from src, byte swaps it, and store it to dst, tmp is only used as temp register, so returning a value into tmp does not make sense.

If you need to reserve R0 to prevent it from being used as argument, you might as well use it as Tmp, that's what I'm thinking.

Last edited by NutsAboutAmiga on 21-Jan-2014 at 06:11 PM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 01:38 PM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 01:37 PM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 01:02 PM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 01:01 PM.
Last edited by NutsAboutAmiga on 21-Jan-2014 at 12:52 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@NutsAboutAmiga

Hi,

Thanks. this worked:

asm volatile (
"addi %2, %2, -1 \n\
mtctr %2 \n\
lwz %3, 0(%1) \n\
1: stwbrx %3, 0, %0 \n\
addi %0, %0, 4 \n\
lwzu %3, 4(%1) \n\
bdnz 1b \n\
stwbrx %3, 0, %0"
: "+r" (dst), "+r" (src), "+r" (words), "=r" (tmp)
:
: "ctr", "r0");

Strangely, without JIT the benchmarks jumped in sysinfo from 16k to 37k, about party with linux now if not faster, since my osx seems to have a serious regression stuck in somewhere, which is strange since sysinfo doesnt use picasso

http://threader.zapto.org/e-uae-ppc-jit/beta1/E-uae-gcc4.6.app-1.tar

Without Jit


With Jit





Btw, i have some code lined up for Amcc 4x0 support, but i need a way of detecting the amcc cpu. This is because icbi is broken on amcc 440 atleast, and iccci needs to be called to flush icache.


Bug: turns out gcc-4.0.1 and cctools as its called on osx instead of binutils, weren't up to scratch, with gcc-4.2.1-5666.3 from apple e-uae runs at decent speed without the above bug, it might be worth adding a workaround for people with gcc-4.0 on apple ppc - gcc-4.2-5666.3, etc. for darwin-8 if anyone feels the need http://threader.zapto.org/experimental/darwin8/ppc/

Last edited by mike on 25-Jan-2014 at 02:32 AM.
Last edited by mike on 25-Jan-2014 at 02:30 AM.
Last edited by mike on 21-Jan-2014 at 04:35 PM.
Last edited by mike on 21-Jan-2014 at 03:23 PM.
Last edited by mike on 21-Jan-2014 at 03:05 PM.
Last edited by mike on 21-Jan-2014 at 03:02 PM.
Last edited by mike on 21-Jan-2014 at 03:02 PM.
Last edited by mike on 21-Jan-2014 at 02:47 PM.
Last edited by mike on 21-Jan-2014 at 02:47 PM.

_________________
C= Amiga addict
,,,
(Oo)
⎛☮ໄ
ﮑὠՀ
Couldn't care less what other people think, seeing that there's concrete evidence they don't.

@mike

What machine/speed is that from?

E-UAE ppc jit beta pre 2:

Appears only 68ec020 works. ?

Osx:
http://threader.zapto.org/e-uae-ppc-jit/beta2/E-UAE.app.tar.xz


@Hypex

1.6 ghz g4 powerbook 5,6

Last edited by mike on 09-Feb-2014 at 08:30 AM.
Last edited by mike on 09-Feb-2014 at 08:26 AM.

_________________
C= Amiga addict
,,,
(Oo)
⎛☮ໄ
ﮑὠՀ
Couldn't care less what other people think, seeing that there's concrete evidence they don't.

Which OS (Kickstart/WB) version do you use with E-UAE JIT? For my AOS3.9 installation it is like described within the README: Shortly after booting into WB EUAE crashes resp. the loaded program crashes (Guru).

Another question: Did someone manage to get GFX-Card resolutions (i.e. 1600x1200 etc.) within EUAE for AOS4 (the only version where I managed to do so is the SDL-Version from OS4Depot)?

Many thanks in advance!

@Reth

Quote:

Did someone manage to get GFX-Card resolutions (i.e. 1600x1200 etc.) within EUAE for AOS4 (the only version where I managed to do so is the SDL-Version from OS4Depot)?

I once was able to set up E-UAE on OS4 using the picasso96 driver to have 800x600, and would have been able to go much higher if I wanted to...I think you had to get a special version of the picasso96 driver, and I can't recall how to install it.

If anyone knows, I am interested in finding out as well!

Thanks,

Scott

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

As said: I've got such an installation but only with EUAE SDL version.

Anybody got a hint regarding my other question? What kind of Kickstart/WB are you using with EUAE-JIT (AOS4 version)? Here with AOS3.9 it does not work (like mentioned in the readme). Either it crashes directly after booting into WB or the started program causes a GURU.

Recompiled libsdl, altered the configure scripts to be more like libsdl/test/configure, included the right paths, does this still work on other systems?

Again, pre beta 2
http://threader.zapto.org/e-uae-ppc-jit/beta2/E-UAE-beta-2-osx-bin-1.app.tar.xz

Last edited by mike on 12-Feb-2014 at 05:25 AM.
Last edited by mike on 12-Feb-2014 at 05:01 AM.

_________________
C= Amiga addict
,,,
(Oo)
⎛☮ໄ
ﮑὠՀ
Couldn't care less what other people think, seeing that there's concrete evidence they don't.

@mike

Yesterday i had the opportunity finally for test Euae jit sdl last build from my great friend on my Love machine the pegasos2 ...
My config is Pegasos2 1266mhz 2Gb Crucial DDR 400 ram and the great Radeon 9800 pro
i tested it without the soundblaster live 5.1 plus the 9800pro isnt accelerated No 3d no Composition and...
With Soundblaster i will have small better performaces in eg quake i have 5fps more thanks of this audiocard,

The euae Jit sdl Os4 result is : 29. 70 mips without Jit enabled
and repeat without any video acceleration

I have to setup the jit and i will report how fast is it :) with the sounblaster too.

Last edited by tlosm on 12-Feb-2014 at 08:38 AM.
Last edited by tlosm on 12-Feb-2014 at 08:37 AM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

Quote:

mike wrote: 1.6 ghz g4 powerbook 5,6

but why ECS Agnus in Sysinfo?
How it works Jit in AGA emulation?

_________________

@Seiya

This is not graphic chip emulation it does not matter what EUAE is configured to.

The only thing you need is set the CPU to 68020, no other CPU is supported atm.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS