Sonnet 7200 PCI programming

[kev]

INFO CENTER

http://www.powerbridge.de/download/m...SICHRA-PG1.pdf
http://www.ustechnologies.com/admin/...s/PrPMC800.pdf
this is for the HARRIER ASIC bridge chips the white motorola 51-W6005B02
Part Number 29L4328 IBM ASIC HARRIER ASIC CBGA720 CBPN 4000A FOR 51-W6005B02

C++
http://www.vxdev.com/docs/vx55man/vx...00/sysLib.html
http://read.pudn.com/downloads132/so...ierPhb.h__.htm
The HARRIER ASIC is a combined FALCON/RAVEN. So the
* HARRIER PHB (PCI Host Bridge) is equivalent to the RAVEN

http://www.mvme.com/manuals/MVME5100-programmers.pdf
this is the HAWK ASIC bridge chips the glob top motorola 51-W6000B02 or full plastic 51-W6000B03

PLX Bridges


ftp://ftp.freecalypso.org/PowerPC/su...100414-00C.pdf
https://www.linux-mips.org/wiki/Marv...em_controllers
http://www.ic72.com/pdf_file/g/415218.pdf
ftp://ftp.freecalypso.org/MIPS/suppo...v95_12-Feb.pdf
ftp://ftp.freecalypso.org/PowerPC/su...60/datasheets/
ftp://ftp.freecalypso.org/PowerPC/su...60/datasheets/
https://www.digchip.com/datasheets/p...-64130-pdf.php
for the Marvell Discovery II MV64340
MV64360/361/362
Marvell Discovery III MV64440
Galileo Discovery GT64260
MARVELL Discovery controllers GT–64260A/B, GT–64261A/B, GT–64262A/B, GT–64130 bridges
Marvell Horizon III MV96360
GT-64241

http://pdf-file.ic37.com/PdfOld/icpd..._datasheet.pdf
http://www.powerbridge.de/download/m...10_UM_0209.pdf
ftp://csiph.com/incoming/timc/PDFS/PowerPC/refbdman.pdf
ftp://ps-2.kev009.com/incoming/timc/...C/CPC700FA.pdf
http://www.chipfind.net/datasheet/html/ibm/cpc700.html
this is the CPC700 IBM bridge its white and looks like the motorola HARRIER ASIC
the IBM AVIGNON CPC710-133 is different looking

BOOT code is online and each one can do PCI AGENT mode
let the compatibility BEGIN!

FYI http://www.powerbridge.de/ has a lot of info
PowerPlus II/III

PPCBug Firmware Package User’s Manual Part 1 and 2
http://npg.dl.ac.uk/documents/edoc245/ppcbuga.pdf

[Hedeon]

Originally Posted by DarrenHD

Ok Update from me:

Now using mmulib/68060.library

Oustanding issues still:

-AmigaAmp 2.25 still has initial hang/freeze for about 10 seconds before the graphics load, but once the graphics for AmigaAMP load then it works fine with WarpUP Engine.
-SysSpeed still reports 2-5Mhz 060.
-PicViewWOS seems to work, but has incredible delay/freezing, it's very slow.
-LowViewWOS the same.
-PPCMemTest still fails to allocate memory.

Thanks
Darren

Don't know about PPCMemTest...works flawlessly here. But I kludged something into the latest build regarding the slowness of the above programs. I'm afraid it needs some manual editing but if you set bit 31 of the hunk size of any 68K code hunk inside the hunk header (mostly the first hunk) then it gets loaded to FAST RAM. So for SysSpeed for example:

000003F3 00000000 00000002 00000000
00000001 00007633 0000010A 000003E9

Setting bit 31 of the size of the first hunk results in 80007633.

I don't think I can make it easier than this at the moment.

[gulliver]

Originally Posted by Hedeon

Don't know about PPCMemTest...works flawlessly here. But I kludged something into the latest build regarding the slowness of the above programs. I'm afraid it needs some manual editing but if you set bit 31 of the hunk size of any 68K code hunk inside the hunk header (mostly the first hunk) then it gets loaded to FAST RAM. So for SysSpeed for example:

000003F3 00000000 00000002 00000000
00000001 00007633 0000010A 000003E9

Setting bit 31 of the size of the first hunk results in 80007633.

I don't think I can make it easier than this at the moment.

Hi, is this a general way to force some 68k apps to take advantage of FAST RAM, or it only works under your own hackish PPC enviroment with the mentioned hardware?

Thanks in advance.

[Hedeon]

It is normal Loadseg() behavior. Bit 31 is fast 30 is chip and when both are set the memory attributes are taken from the next long word

[gulliver]

Originally Posted by Hedeon

It is normal Loadseg() behavior. Bit 31 is fast 30 is chip and when both are set the memory attributes are taken from the next long word

Is it too much too ask you for for a simple cli program that can patch its argument with what we discussed, so as not to have to do it by hand and be prone to mistakes?

[jack-3d]

Shogo works now, Hedeon checked my system and found Picasso96/rtg.library in wrong version.

- as always BIG thank you!

jack-3d - vids?

[Hedeon]

Originally Posted by gulliver

Is it too much too ask you for for a simple cli program that can patch its argument with what we discussed, so as not to have to do it by hand and be prone to mistakes?

CLI programs are prone to misuse.

The latest library now patches the hunk by default (in memory, not on disk). So SysSpeed should now correctly give the amount of MIPS of the 68K.

Included an option to turn this feature off.

[Hedeon]

Originally Posted by jack-3d

Shogo works now, Hedeon checked my system and found Picasso96/rtg.library in wrong version.

- as always BIG thank you!

Great to hear! How's FreeSpace now?

[amigoun]

What is the correct version of the rtg.lib?

[grelbfarlk]

I use P96 2.1e so rtg.library should be 40.4029.

[Hedeon]

The one on Aminet does not seem to work. It's from 1998 or something around there. The latest is from like 2014.

Edit: Sorry, mixed up rtg with rtgmaster....

[amigoun]

Originally Posted by DarrenHD

-PicViewWOS seems to work, but has incredible delay/freezing, it's very slow.
-LowViewWOS the same.

-Did anyone get PowerUP apps to work by installing PPCLibemu? I installed PPCLibEmu and PowerUP apps start but then freeze when trying to access the PPC.

Datatype based picture viewers are expected to be quite slow, but I have found ShowPPC being quite good (although still missing some features) and PPC native (thus quite fast) viewer.
speed on Sonnet G3 400/512
XGA (1024x768px) JPEG picture loading time 0,7 sec
7.4MP (3888x1907px) JPEG picture loading time 7,5 sec
As far as I remeber this picture viewer is PUP, but works nicely thru ppclibemu.

[starship]

Hi
I am watchig deathtrial(mankind) and demoscope demos.
Colors are wrong on me. Everything on purple and green.
I am using p96 2.1e rtg.library 40.4029.
Cybergraphics.library 41.4 on radeon9200.
Are colors wrong on you?
On deathtrial colors are normal at windowed and aga mode. But cybergraphics 15bit colors wrong.

[grelbfarlk]

Originally Posted by starship

Hi
I am watchig deathtrial(mankind) and demoscope demos.
Colors are wrong on me. Everything on purple and green.
I am using p96 2.1e rtg.library 40.4029.
Cybergraphics.library 41.4 on radeon9200.
Are colors wrong on you?
On deathtrial colors are normal at windowed and aga mode. But cybergraphics 15bit colors wrong.

Sounds like wrong endian mode. If you can select a different screenmode use non-PC mode if you have one available.

Just tried Deathtrial it doesn't give you a screenmode requester and it fails to use CGX-15-bit mode if one isn't available. Can't seem to use Newmode either. Works for me in Windowed mode, but slow. Colors seem right.

[starship]

I tried 800x600 8 bit and 640x480 8 bit screen modes on my workbench but colors are the same on demos.

[Hedeon]

Tried on CSPPC and a CVision?

It might not work otherwise:

http://eab.abime.net/showthread.php?t=71650

[starship]

Oh yes There are some incompatibility problems without cvision. I thought I had a library error.
Thank you.

[Hedeon]

Demoskope is endian problem. I used ModePro to catch the screen and change from 320x240PC to normal 320x240 and now the colors are correct. Don't forget to get the WarpOS jpeg.library. The one included is 68K i think.

The demo crashes after the opening screen on my machine btw with a PPC data access to a non-existing memory range.

[starship]

It is working on me. Thank you.