Does anyone have any info about the early PSX 'OS' SDK?

what a tease
thanks for the update!


libgpu/libgte, I wonder what it looked like, probably the same thing without the in line macros.
Which as I understand it weren't standard precompiler defines, but instead linker specific code which injected the actual binary code, obfuscating hardware details from devs? or am I getting that mixed up with DMPSX (the Assembler of the op-codes)
Ah, nope I was right first time.. that's crazy!
From: gte7.pdf August 31, 1995

[1.7. ]: When we can use new in-line GTE functions.
In-line GTE functions are built-in type libgte functions.These functions
is fast because no stack access or PC (program counter) change is
required when they are called.
Our in-line GTE functions are different from conventional c++ inline
functions. The conventional in-line functions needs source code of the
functions. but ours does not needs them. The linker attatched the in-
line functions in object code level. Therefore what you need is the new
linker and *.obj code in libgte.
This in-line functions are working in R&D level on some UNIX (especially
Sony NEWS-OS) environment, then now we are coverting to the PC
environment which everyone uses.
So it takes for a month or so. we are trying to release the first sample
in 7/E.

​

I'm guessing NEWS desktop was part of the 20k purchase price back in 94-95


I found this https://winworldpc.com/product/news-os/4x
I wonder if there's a emulator that can run it.
Looks like it's RISC not M68k
http://katsu.watanabe.name/doc/sonynews/newsos.html

​

 

Looks like it was on both the R3000 and MC680x0..

NEWS-OS 4.0C 4.3BSD X11R4 OSF/Motif MC68020/30
NEWS-OS 4.0R 4.3BSD X11R4 OSF/Motif R3000

68k series were used in early macs, Amiga as well as others so plenty of emulation of 68k around, just the rest of the hardware I guess.. There was BSD for the Amiga as well iirc. R series MIPS CPUs were used on SG machines.

 

The machines that the early PS-X development environment ran on were R3000 MIPS based - they don't seem to have been very popular - back when I was looking for one I managed to find quite a few of the 68K/020/030 based machines and some of the later (R4K, R4.4K) MIPS based ones - but I never got my hands on one of the ones from the same series that was used for the PS-X development.

The ones I saw also had an IOP in them that was controlled by a custom Sony blob - which made running the machines under anything but NEWS-OS pointless since the inability to control the IOP made the machine useless.

 

"Looks like it's RISC not M68k" was in regards to the iso: Sony NEWS-OS 4.2.1R (ISO) 4.2.1R
The R in means RISC (R3000)
https://translate.google.com.au/tra...abe.name/doc/sonynews/newsos.html&prev=search

I wonder how close is a PSX debug/test unit (with 8mb RAM) to a diskless NEWS HW?

https://translate.google.com.au/tra...nabe.name/doc/sonynews/model.html&prev=search

 

I don't believe the games ever ran natively on NEWS, I'm not even sure there was a NEWS workstation that included PS1 hardware (like a PS2TOOL but the two sides had a similar CPU).

While NEWS had an R3000, the PS1 uses a VHDL re-implementation by LSI. I don't believe any of the PS1 chipset existed before LSI came on board. It's not like the N64 where SGI effectively shrunk down one of their workstations, so you were able to start development on one of those workstations. Or the dreamcast where you could do development on a PC using a special PowerVR graphics card. Or the Xbox 360 where developers were using old PowerMacs.

AFAICT the target box was just a prototype PS1 that could be attached to a NEWS workstation. Any similarities between the target box and NEWS case being a cost saving exercise.

The original libgte implementation exists in all SDK, but it has quite a large overhead because of the functions prologue and epilogue. Not only do you have to execute all that extra code, but it played havoc with the PS1 direct mapped cache.

DMPSX isn't just for obfuscation, there are restrictions on GTE instruction placement which need to be enforced and without modifying the SDK assembler they had to add another tool which performed another pass on the instruction stream. The alternative GTE instruction representation may have been used to help identify them.

early libgpu is a bit of a mystery though. There are some clues that I have meant to follow, but never got round to it.

 

yes, they save $ on the case, like DTL-H2010 cd-rom , it use retail CD player case.

 

Sony offered NEWS bundled with PlayStation hardware in Japan as DTL-H5000. Let me dig my documents...

 

The photos are here. I hope that everything is visible.

https://beta.photobucket.com/u/psx-collector/a/f05f6f6e-dae5-48d1-9e9d-0006d249af20

 

is terget box = programmer tool ?
any more info ?

 

What do you mean?

 

because early runtime library still link up to prototype target box , the prototype is look like news one ? or like you pager one programmer tool add in news system ?

 

I kissed the monitor when I see these documents.
Twilight, any more photo? May be on the NEWS computer itself?

 

Fantastic documents!

 

Target Box just used very similar case. Nothing else NEWS related. Target Box is just early PlayStation without CD-ROM drive. Just one big board inside.

Would be great if someone could translate them from Japanese. Do I understand correctly that Sony offered NEWS based workstations with pre-installed PlayStation development hardware?...

No pictures. Only these copies of the documents about NEWS hardware. I have some price lists but nothing very special.

 

Machine translation can do reasonably well these days, even from scans/pictures...