sirocyl

sirocyl

@sirocyl

noted computer gremlinizer

working on a @styx-os.

 

laptop.
                                                                                                     

"accidentally-vengeful telco nerd"
—Tom Scott

platform sec researcher, OS dev, systems architect, composer; Other (please specify). vintage computer/electronics nut.

I am open to tag suggestions - if there is something you want me to tag on my posts, leave a comment. <3

log inask
take a look at
this cool bug I found 🪲
discord
@sirocyl
mastodon/whatever
social.treehouse.systems/@sirocyl
bluesky
bsky.app/profile/sirocyl.bsky.social
revolt.chat (occasionally active)
@sirocyl#5128
new blog:
significantb.it/
styx linux OS project
styx-os.org/
sirocyl
@sirocyl
bennyfactor
@bennyfactor
sirocyl
@sirocyl

I'd like to introduce my other favorite devkits:

giant PS2, PS3 Enterprise Edition, and "just a Power Macintosh G5"

bennyfactor
@bennyfactor

if that ps3 had ps2/ps1 support I would be extremely covetous of that industrial sony design language

sirocyl
@sirocyl

Bad news, and good news.

There is not, publicly, any hardware PS2 support on the DECR units, as they do not equip the EE or GS onboard. However, the EE+GS, RDRAM, and a bridge chip, do/did exist on a PCI card that is used for developing and testing the PS2 BC software modules in the PS3 firmware.

That card is rare as hens' teeth. However! Its components are used in actual PS3 hardware, for the backwards compatibility feature.

I am looking to build compatible replacements for that PCI card.

I have a good idea of what it is built from, so here's the deal.

I'm looking for dead, PS2-capable PS3's.

That's CECHA and CECHB models, preferably; but CECHC and CECHE may also be useful.
Scrap units. Landfill potential. I'm looking for systems that are beyond YLOD, and beyond hope. No reball, no 40nm RSX swap, and no amount of any kind of repair, no matter how expensive, would bring the full system back online.
Water damaged or destroyed boards are even acceptable, I'm only looking to harvest chips from these boards.

If you'd like to contribute, toss me a line in my ask inbox.
I'll only need a couple, at first (for reverse-engineering and research), but if this works, I'll need more broken PS3's.
If you're a PS3/game console repair shop with a ton of these, and do component-level repairs and BGA rework, we can get in touch; the other parts on the board like SYSCONs and VRMs are more useful for you, so if you'd rather keep the full boards, I'll tell you what to harvest - it'll likely be EE+GS or GS, RDRAM, and a bridge chip, and potentially some video logic, clock and onboard power supply/regulator modules.
The board models in question are the COOKIE series - COK-001 and COK-002. No VERTIGOs, SEMTEXes or DIAMONDs, no SURTEESes, KTENs, DYNAMOs etc.

DO NOT DESTROY YOUR PS3.

I shouldn't have to say this. I do not need live sacrifices, I need already long-dead bodies.
Don't break PS3's for this, that's stupid.
If you have a working or repairable PS3, it probably would be better to repair it than to send it my way.

I can't promise anything, but I might be able to work on designing some hardware that can:

  • bring PS2 hardware BC back to many newer PS3's, with a hardware mod.
  • add PS2 hardware BC back to PS3 development units such as the Reference Tool units.
  • enhance PCSX2 compatibility and performance on the PC.
  • work as a graphics card, a standalone GScard.

Long story short: CXD9208GP is an SIF to PCI bridge chip, and encapsulates all the peripheral hardware and I/O of the PS2's EE - the BIOS, the IOP, timers and interrupt controllers, the SPU2, SIO/pad and USB, SSBUS and DEV9, and the CDVD are emulated in software on the host machine, over an onboard conventional PCI connection. I'm reverse-engineering this. There's probably a similar bridge chip for the GS by itself, if it's not all-in-one connected over PCI directly; but it's not mentioned among the hardware on the PS3 dev wiki at the moment.

You must log in to comment.

in reply to @QuestForTori's post:

sirocyl
@sirocyl

This box - the Saturn Address Checker, a chassis by SI Electronics - has one (presumably SCSI) interface port on the back. Another devkit, the Sophia Systems Saturn Programming Box, has SCSI proper and an RS-232 port, which probably makes it better suited for server stuff. :)

login to reply
neckspike
@neckspike

The HST-3220 (2nd hardware revision, the model with the round buttons) came in white (Sega), grey (Victor), or black (Hitachi). The original model with the oval buttons came in grey from all three manufacturers.

login to reply

in reply to @sirocyl's post:

in reply to @sirocyl's post:

dog
@dog

The PowerMac G5 was available in a similar CPU configuration to the 360, and also had a very similar GPU available. There weren't very many PowerPC G5 workstations available, and the PowerMac G5 was the most readily available and cheapest. They used it until they had their own dev hardware available.

login to reply
nortti
@nortti

I would guess it's a combination of:

  1. the design is not fully finalized and packed into the smallest possible space yet, so you get larger protype parts
  2. these are more one-off, so you may want to be able to service them
  3. they contain additional hardware not included in the retail version, especially for older consoles
login to reply
modulusshift
@modulusshift

Additional RAM so you don't have to optimize prematurely, diagnostic readout versions of key pieces of hardware, especially networking, extra slots so that if you're developing some kind of peripheral you can talk directly to the CPU, easily removable hard drives so you can unplug, move to your coding/compiling computer, copy in the new version, and slot it back into the dev kit as quickly as possible... lots of stuff. These are usually basically full computers that happen to be able to run console code.

login to reply
sirocyl
@sirocyl

Mostly this. The devkits usually have close-to-finished retail hardware. Early prototype devkits aren't uncommon, and some are very large, like the 3DS target board or the CellBE (PS3) "Cytology"/"Shreck" boxes; but they rarely get used past the console release.

Many development units have an extra single-board computer (Sony for example) which is used for networking (remote debug/program upload), instrumentation (e.g., JTAG, UART, GPIO), media emulation (for discs/cartridges), and activation/licensing (to keep them from being used by pirate groups... usually unsuccessfully.)

Other development units (e.g., Nintendo until the Switch) make use of a specialized board, built from the components of a retail console (CPU, RAM, misc. hardware) but with additional instrumentation built-in (like video capture, network emulation and remote debug facilities, additional RAM, FPGAs to handle the above) and the handhelds in particular usually have a "remote" which is in the chassis of the handheld, and connected to the unit with an umbilical cord carrying video, sound and input.

The PS3 DECR "Reference Tool" is a special case in a way - it has two disk drives (one for the PS3 system, and one for BD emulation; both connected to the main machine though, emulation is handled by a hypervisor driver iirc) and a massive cooling setup for the BE CPU, and the RSX GPU. It looks for all intents and purposes like a 2U server chassis.

login to reply
sirocyl
@sirocyl

Not quite; the hardware differs quite a bit between retail and the G5 protos. Games can be compiled to run on either kit, but an existing XeXDK-ready game would not easily run on the G5 without extensive rework (iirc most of the important libraries, like the DirectX graphics pipeline interfaces and system call library, are statically compiled; and the G5 is not powerful enough to emulate these and run the game program at an appreciable speed. This and, the G5 is dual-processor, the Xenon is tri-core and equipped with eDRAM that the G5 does not have.)

login to reply

in reply to @sirocyl's post:

Pinned Tags