PearPC Quad-IDE build

[redscorp]

Summary of releases:

ppc-redscorp-04062011.rar - Binary package from Tronix.
Read this post for more details.

redscorp-release-20080329.1.rar - Binary package
ppc-src-20080329.1.rar - Source code.
* bug with big packets is fixed. Enjoy!

redscorp-release-20080329.rar
+ add PCAP connectivity support.
* add time stamp in logging.
* improved Ethernet traffic filtering in RTL8139.
Read detailed description in this post.

redscorp-release-20071122.rar
* last CVS source (0.5pre).
+ sparse HDD support patch.
+ experimental 4 IDE channels patch.

redscorp-release-20071203.rar
* quad IDE (as previous)
+ key combinations handling improved.
WARNING: Bugs are found with mouse handling. Not recommended to use!

redscorp-release-20071204.rar
bufix release.

redscorp-release-20071228.rar
* parallelized read/write operations with IDE drives.

Please report bugs in this thread.

--

This is a night build:
* last CVS source (0.5pre).
+ sparse HDD support patch.
+ experimental 4 IDE channels patch.

Screenshot

ppc-quad.png (173.59 KiB) Viewed 44925 times

Package contain 5 win32 binaries for different CPUs. No additional files like video.x are present.

redscorp-release-20071122.rar

Package

(732.48 KiB) Downloaded 2211 times

Limitations:
You can boot OSX only from 1st IDE channel.

UPD: Bug reports are welcome... If something doesn't work post you report here in this topic.
If everything works fine post your results here.
Your screenshots are also welcome.

[ElGuillermo]

Hi!

Didn't have the time to test it well... I'll do it in some hours or (better for me) monday or tuesday. Seems to work OK for the moment

[dodgermuk]

Tested it using ppc-intel-sse2. Had no problems. Sceenshot included

[ElGuillermo]

Hi!

I tested your ppc-athlon-sse3.exe build
on an Ahtlon64 X2 4200+ (Manchester core) running at stock speed
with 2Gb of RAM (1Gb allocated to PearPC)
on Windows XP Pro SP2
Emulating a G3 running Tiger 10.4.11

...and have NO bug to report yet...
I didn't try the sparse HDD feature.
I didn't notice any speed increase/decrease from my usual build.

Well done Seems we now can use 8 disks at the same time with no compatibility problems
I can't wait for some more new options and code improvements

[redscorp]

OK. So good so far.
NEW RELEASE:
* quad IDE (as previous)
+ key combinations handling improved.

redscorp-release-20071203.rar

(762.95 KiB) Downloaded 1785 times

So now its possible ti use key combinations like "Right-Alt+Ctrl" in config file. This will be interesting for those who use to press "alt+Ctrl" to get back cursor from VirtualPC or vmWare... I use this in my config:

Code: Select all

key_toggle_mouse_grab = "Right-Alt+Ctrl"

Interesting to hear you opinion about this feature...

[ElGuillermo]

Hi!

Thanks Redscorp ! Downloading it now... could be interesting for people like me, who use some software that require function keys... I'm glad I now can use my F1-F11-F12 Quark X-Press shortcuts again

EDIT :
There is something wrong (at least with ppc-amd-sse3.exe, ppc-amd-sse2.exe and ppc-generic-mmx.exe)
I've even tried to launch ppc-intel-sse3.exe on my Athlon64 X2, and got the exact same results

- Logs show a large number of [IO/CUDA] <Warning> Event processing timed out. Event dropped.
- mouse is almost frozen in Tiger.
- Tiger's clock seems to act as crazy, jumping several hours forward at each refresh between freezes...
so : Tiger screensaver is entering active mode after some seconds of "inactivity", instead of 10 minutes, as I set it.
- Sometime Tiger's core totally crashes, displaying corrupted console outputs melted with desktop's background.

I've tried to fiddle with core affinities in Windows XP Pro to see if it would fix the timer's strange behaviour : no success

++G

[redscorp]

Hmm.... Tiger really behaves strange... I tested this release with panther only.
I'll look what's happens there..

[ElGuillermo]

Hum... I'd like to help you more, but I don't have any copy of Panther to be run on your builds...

Did you fiddle with timer's settings ?
Did you change your compiler configuration ?

The previous builds were -afaik- bugless... If you didn't change anything in the code but hotkeys handling, there must be some I/O polling code that is wrong... As a matter of fact, Tiger runs well in your latest build UNTIL YOU MOVE THE MOUSE (and I assume it is the same if you try to use keyboard, but didn't get as far because of freezing...) I'm pretty sure there is an I/O polling process and/or buffer management which is not "right" (non coder's advice, though...)

[redscorp]

Try this release.

redscorp-release-20071204.rar

(762.51 KiB) Downloaded 1962 times

[ElGuillermo]

OK, I've tried the new ppc-amd-sse3.exe with success... Well done
Nothing special to report : behaves like the nov 22 build : no bug found yet...

Tiger runs well again I've been running an entire XBench measuring (results are a little higher than those I get with my previous "fastest" build see : viewtopic.php?f=9&t=99&p=45197&hilit=xbench#p45189 ) but we can't totally rely on XBench within an emulated platform... Still, your latest build seems to be the fastest we know actually

Did you find the reason why your previous build wasn't "good" ? Do you believe you can improve emulation speed ?