I'm trying to figure out some stuff about the Xbox. We have Chihiro and we have Dev kits. All nice and pretty, but Now i'm so much doing research I cant find specific information about some stuff. Some of you know allot about different subjects or you know some one who maybe knows. So while I was using google to find more I found a 3th version of the MCPX of the Xbox I dint know off. I have collected some info under here, please take a look. So we have in ALL retail xbox a MCPX X3, this is the final version, I'm not sure if secretly there are differences but because all the software needs to run on all retail xboxes I can hopefully assume there are no differences or only minor bugs removed or a smaller footprint. 1: First question, do we have a High res images of the MCPX decapped? the bare DIE ? like Bunnie did with his? cant find it Ok so I asked something different in the title. The DVT3 development kit came around (June 2001?) This was equipped with a MCPX X2, if my information is correct. Now the final one looks like this (its a 1,6 mobo) which as you read is a MCPX X3 Now What is its exact difference hardware wise? is software incompatible? have we example code or documents about this? Especially because this is a picture of a Chihiro Motherboard: As you can see, the Chihiro is very much compatible, like you can read elsewhere, its even uses XBE files and game images are encrypted FATX format files(Representing a hard-drive(?confirm)) Some games run also on DVT3's that are equipped with this mcpx x2 chip. (Also on DVT4 or Retail with 128MB mod? please confirm this) Anyway, now I found this image: Its probably hard to see, but according to the poster of this thread: http://www.assemblergames.com/forums/showthread.php?7221-Prototype-XBOX-Motherboard-HD-Combo its a Prototype XBOX Motherboard with a MCP X-MODE 2. Notice the "Arcade" connector behind/infront of the Case fan. (I now think its a test/debug port for the cpu[2015]) Also do someone has all his pictures in High-res? Please? So whats this last board? Was it used to DEV Xbox games and Pre chihiro? and finally used the MCP X2 or MCPX X2 for final chihiro and X3 for Retail xbox. Now Have you images of different xbox mcp chips, please pm, or reply here. Do you know more about this development. Do you know how MS made this all and/or have some documents about this? Yes, I do know about the Alpha1 and Alpha2. Thats an other topic, but came the dvt3 after this or did MS other research and development between? Im not scared for Schematics, Tech documents or even code examples. I want to know the version history as far as we together can reconstruct and maybe help some people get Chihiro code run on retail hacked stuff, make more games run and maybe fix unkown bugs in the cxbx emulator. Thank you for your reading time and feel free to reply, but please keep it on topic (yea, a large topic ) last bit, Ive seen images of the following "versions" 5 so far: MCP X-mode 2 MCP X2 MCPX X2 MCPX X3 MCPX X3P (not sure, but my latest notes state I found some info on this X2 uses testkeys X3 internal ROM Also there are some revisions of the mcp code (mcpx drivers included in the kernel) A01 DVT1,2,3 (diferent NVA cores?, CPU clock changes at least)init tables 133MHz mem, 155MHz NVCLK? (can we find this in alpha?) A02 Faster clocks (think FSB, NVCLK) (3 or more clock variants) Above is based on "sourcecode", and litle dissasebly done by me of retail code, should do alpha code to compare (still work to do..) below is based on JayFoxRox and my own code disassebly (lost my notes on that, but did remember finding most codes, not the names sadly(D3-D4) B01 (JFR: some debug/dev? B02 (mainly found in code, pre retail?) C02 MCP_REV_C03 (JFR: retail 1.0?, CA: comments point in that direction indeed) If we are building for QT which implies C03 (or higher), the following should be defined For C02, it should not be defined C03 (JFR: Debug+Retail 1.1?) I did find "revision" notes in the comments and code D01 (comment in source: "...This change is required only for MCP revision D01 ...")(JFR: Retail 1.3-1.4? Chihiro (type 3)) D03 (JFR: Retail 1.6?) This is after 4400 Kernel source, so no info (is posible) D04 (JFR: 'Fox' xbox) D05 (JFR: 'Ex-Julian' xbox) JayFoxRox also notes some info about the NV2A, wich is a NV20 updated for the xbox. Building a Frankenalpha (just a AlphaI or II) will require a Nvidia NV20 chip. I did not crossreverence binairies yet with the alpha code, and did not yet looked into the code in the kernel for NV20 and NV2A diferences, but there are multiple revisions of the NV2A (and some NV20) code in the 4400 kernel JayFoxRox found some code indicating these versions, and I dint find them the same way, but appear in code comments https://github.com/JayFoxRox/xqemu/wiki/MCPX-and-bootloader Kernelsource is 1.00.4400.1 probebly build on 21-07-2003. Released in small public newsgroups under the name of "Barnabas-2084". (could be the japanese sega leak, wich I named so, because I did got Symbols for a japanese sdk also close to 4400 and first(?) notions of "Arcade"(?)... I do not claim it was a leak at sega's part, just that it is very weird to have symbols in japanese so close (if not, need to recheck) to the kernel sources) Also, some information claim full source tree must exsist, ive requested before, if you have it, or know more, do so in the right thread or PM me. Thread: https://www.assemblergames.com/forums/showthread.php?44462-Open-Source-Xbox-Live/ Update 22-6-2015: Some small parts are off or seem incorrect atm. I continue my search ofcourse and try to update where I can. My current believe is that DVT4 is the retail first generation, DVT6 middle and XBLADE could be the 1.6v xbox because of the changed chipset (especialy the video and inclusion of the TSOP(kernel). A recent (personal) discovery is that some SDK includes tools to assemble code for the Decoders onboard (DSP56300 24Bit DSP) and im also looking into the Developepment tools used by MS. (source for NT has found his/her way online, and shows some interesting similarities to the xbox, wich is ofcourse very logical for being NT/2000 based xbox) I also edited some details in the main post, because I found some revision info, that confirmed my findings in retail code. Im still not realy sure about the DVTn codes, ive done some matching on the MCPX subcodes (next to the date codes wich seams to follow the normal use yyww (year, week) Lot of retail pictures of D3, and not X2, with the chinese sources "only" selling X2s in the D3 version. currently researching the either "fake" or "real" chinese mcpx-x2 ic. datecodes, next to "confirmed" verions, seem to indicate: B2 mostly 1.0 D1 1.1, 1.2~3, 1.4~5 D3 1.6 But, mcpx is replaceble, and only 20 xboxes with images. (either send me your motherboard picture along with xbox version or ask/wait for poll) B2 2001 (maybe earlier) D1 is atleast 2002 (according to my spreadsheet) D3 could be 2003
Not seeing any "arcade connector" though I do see the USB headers, power/eject button header and an AT power supply connector. All v1/1.1 XBox boards sported an AT connector. EDIT: Nevermind, was looking in the wrong direction.
Its Ok, I was thinking I ask a lott in a heavy full post. Also these images are not mine and not the same xbox. The last image has the arcade con. Maybe not very much people know about this, and ill have to research myself some more. So this thread may be slowly getting old untile someone or me finds new info. or Chihiro emulation or is running completely on a retail/dvt(3/4).
The X3 has a 512-byte x86 boot ROM that overlays FFFFFE00-FFFFFFFF in x86 space. It contains the secret boot code used as the seed of trust for the Xbox boot sequence. The X2 doesn't have that boot ROM. As a result, X2 systems boot directly from the flash's high 512 bytes. I don't know of any other differences. After all, debug kernels work fine on a retail Xbox once you get past the MCPX ROM protection. Retail systems can even be upgraded to 128 MB. I don't know the difference between "MCP X2" and "MCPX X2". Maybe "MCPX X2" actually has a 512-byte boot ROM for debug kits?
Ive been reading some stuff and It makes sense to me now, also thanks Myria. I did find a Die shot of the mcpx3 from bunny in a higher resolution, but its not very helpful (He also stated that its difficult to make sense out of the top layer.) so next to being a bootrom its main purpose is being a south-bridge. with means some functions, Jtag tied to ground and lpc bus access (on with a serialport can be added.) I wonder if bunnie ever finished his boards he talks about (for on the lpc bus.)http://bunniestudios.com/bunnie/proj/anatak/xboxmod.html#lpc is there anyway to detect a version number of the mcpx or working bios detector for debugs ? I know the bootrom is never stored or moved to the cpu/ram but are other parts visible on the memory adresses?
hehe i still have a xbox1 here with a PC-BIOXX lik-sang prototype too bad i lost all the sfotware for it and cant update it anymore... oh well atleast its running the last build of evox firmware...
Thats I asume just a DIP40 or smaller eprom connected to the mobo, and D0 shortent to GND. Not sure, but if its not working you can try burning a Xbox bios to it using an eprom burner (Like the Wilemprogrammer or dirty trick Arduino)
I have one of the boards, Engineering Sample, has the arcade connector behind the CPU but also has a 4/5 pin connector between LAN and Video ports, this one pictured you can see where the USB connector would go. On mine I have a cable that actually goes up to the DVT board that sits on top of the hard drive for the rear-facing USB port. This is in a DVT3 btw.
I love seeing pictures Between lan en video is usb? can youy take pictures of that area following traces (if any highres images are posible ) It would prove JVS kinda wrong unless that connector (4 to 5 pins) is the said arcade connector (JVS, connected to MCPX like IC as RS232 kind? RS454?, dont qoute this) and the larger one... very intresting still. debug?
Post must have been lost in the Server move, http://imgur.com/WavX4dB,K91rrf4,IGjgamj,pZ16zsL,pqOE4JK,krf2YCJ
I remembered ive seen those, Thank you for reposting. They give some information and more new questions. Also I just see those traces from the jvs/usb port(small one, J6A1) go near resisitor R6M1 of to somewhere (it looks like not populated pads) Anyway, probebly working usb, for whatever reasons? Closeup of the xmode2 is nice bit shiny but cool. Nice to have for reverence. Thank you again @Syclopse
Got more info, but is not very trustworty. Image research seem to indicate years of which the subversions got released like the B2 in or arround 2001, D1 is atleast 2002 and the D3 could be in 2003. (no D3 X2 seen, only in China... wonder why?)
Just thought i take a quick shot off an MCP X2 next to an MCPX X2 and a retail MCPX X3. The MCPX X2 and MCPX X3 both are burned out as you can see! The MCP X2 is installed on a prototype DVT4 board with the part number: A38727-006 [GALLERY=media, 1529]DSC03183 by N64 freak posted Sep 8, 2016 at 11:56 AM[/GALLERY]
Had you seen the decapped images at: https://siliconpr0n.org/archive/doku.php?id=azonenberg:nvidia:mcpx
Yeah seen it, sadly not much to see there, not even the secret rom (so no MCPX rom extraction) cant even tell where the DSPs are or the USB stuff. the overview seems to have all the routing still on it (powerlanes?) it needs some demetalization first. Would do myself if I had a proper microscope. (fumes be scary tho)
Too bad someone didn't clear off the 7 layers of metal routing all the signals/power to the silicon die below. Hard to make out anything below through that rat's nest.
