Test the First Flashed GTX285 ?

[TheLostBrain]

Thanks for the info

why would you need efi part of rom on pc without efi itself ?

I appreciate the info... realizing I don't have as a good a grasp as I thought I did on how osx86 is working under the hood - time to start reading up again.

One thing did come to mind though. I know with my current 8800gt I had to wind up disabling a few kexts (AppleUpstreamUserClientDisabler, IntelCPUPMDisabler, etc.) in order to stop some random momentary lockups I was experiencing... and then I later learned that a few guys actually resolved the issue permanently by simply flashing their cards with the mac firmware.

This is more what I meant when I said "preventing compatibility issues". I wonder if there are any similar issues w/ the GTX 285? I'll probably just hold off for now and then if issues arise consider doing the flash then.

Thanks to both you and The Rominator for the help.

 

[netkas]

I appreciate the info... realizing I don't have as a good a grasp as I thought I did on how osx86 is working under the hood - time to start reading up again.

One thing did come to mind though. I know with my current 8800gt I had to wind up disabling a few kexts (AppleUpstreamUserClientDisabler, IntelCPUPMDisabler, etc.) in order to stop some random momentary lockups I was experiencing... and then I later learned that a few guys actually resolved the issue permanently by simply flashing their cards with the mac firmware.

This is more what I meant when I said "preventing compatibility issues". I wonder if there are any similar issues w/ the GTX 285? I'll probably just hold off for now and then if issues arise consider doing the flash then.

Thanks to both you and The Rominator for the help.

Guys was flashing only PC part of macpro's 8800gt rom and it worked fine (Cyclonefr is example)

 

[MacVidCards]

Free GTX285 Flash !!!

I am surprised that only one of these cards has been flashed.

I am willing to do a few more for FREE, just so we can all benefit from the knowledge gained.

All you need is:

1. 2008 or 2009 Mac Pro
2. 2 @ Mac PCIE power cables
3. EVGA (or likely ANY brand) of GTX285 1 Gig

I have the EEPROM chip ready, just need some eager participants. You will be asked to post here afterwards regarding how it works.

Anyone? I am in LA, Hollywood area. I will even accept cards mailed to me, provided you include return postage and promise to post about it.

Just checked at Newegg...they have "Recertified" GTX285s for $259. So, nearly $100 LESS than an Apple 4870.

 

[10THzMac]

I am surprised there is not more take up of the flash option. Might be the case that the expenditure on ANY 285 is high enough that people will just then go the extra for the official version, rather than take any perceived risk with a card.

In the UK the web price for the Mac version is about 400GBP, which is 50% more than the PC EVGA card, and a huge premium over what you say you can get a PC card for.

 

[Guiyon]

I just got in the samples from SST so I'm just waiting until I can afford a $300 brick to flash it. I have at least a good half-dozen EEPROMs to test and a healthy supply of ChipQuik

 

[MacVidCards]

Good News !!

I have been PM'd...a second person has stepped up to the plate and we will soon have more data regarding flashing the GTX285.

Keep in mind that once the next card comes out, the PC GTX285 prices will drop like a paralyzed Falcon while the Mac GTX285 prices will likely remain high for some time. (The Max X1900XT is STILL $399 at Apple Store)

So, our pioneering work here will lead to better options in 6 months to a year that will become more mainstream.

Someone more clever than I (like...Netkas perhaps) may even figure out a way around the 256K issue. Meanwhile, I am happy to do a little soldering for the community's benefit.

 

[zebity]

Testing 2GB 285...

Rominator & Co,

I have ordered EVGA GTX 285 Mac (1GB RAM & EFI64) Mac Edition and a EVGA GTX 285 with 2GB RAM (PC/BIOS) card.

The objective is to try to flash the 2GB card with the the Mac EFI64 ROM to see if I can have a 2GB GTX 285 to work in my Early 2008 Mac Pro.

I am not sure if this has been attempted before and will post the results. At this point I am hoping that I do not have to solder ROM, but I will not know this until both cards arrive and I have checked the EEPROM details of the two boards.

I have previously had success flashing Quadro FX 5600, but this was relatively easy as the PC boards always had EEPROM chips on them that were big enough, so there was not need to solder. I want the 285 so I can get 30 processors with CUDA 1.3, as FX 5600 only has 16 processors with CUDA 1.0 .

There is delay of nearly 2 weeks for the Mac 285, so I will advise results once all the parts are available.

Cheers,


Zebity.

 

[MacVidCards]

I am nearly 100% certain you will find either a 64K or 128K chip o any PC version of this card.

It only "needs" 64K, so even 128K is a "gift" from luck gods.

I thought I posted about the 2 Gig card I have.

I put the Mac ROM on it but in limited testing it would work briefly and register 2 Gigs, but if I opened a 3D ap in Windows it went zonkers.

I am pretty sure I can move memory strings from proper card over to ROM but it takes time. (See the 8800GTS thread I have....I moved memory timings over to increase power of 8800GTS and it worked)

I need to get a 2008/09 Mac Pro to test on....

 

[netkas]

I am nearly 100% certain you will find either a 64K or 128K chip o any PC version of this card.

It only "needs" 64K, so even 128K is a "gift" from luck gods.

I thought I posted about the 2 Gig card I have.

I put the Mac ROM on it but in limited testing it would work briefly and register 2 Gigs, but if I opened a 3D ap in Windows it went zonkers.

I am pretty sure I can move memory strings from proper card over to ROM but it takes time. (See the 8800GTS thread I have....I moved memory timings over to increase power of 8800GTS and it worked)

I need to get a 2008/09 Mac Pro to test on....

try to create hybrid rom for 2gig card

 

[MacVidCards]

try to create hybrid rom for 2gig card

Brilliant idea...if only I knew how.

I never really understood how Pipolomo42 knew just how to assemble that 4870 ROM.

I tried using that script to "sew" a couple Nvidia bits together but no joy.

PPC was so much easier, 1 ROM and No Checksums. Although they did have those tedious tokenizing strings.

Maybe it's time for a contest.

 

[10THzMac]

Brilliant idea...if only I knew how.

Netkas injector works fine with 2G PC card. I have to wonder now about the effort/reward ratio for the ROM effort.

Netkas: has ANYONE reported to you on using 295 with your injector?

 

[netkas]

Brilliant idea...if only I knew how.

I never really understood how Pipolomo42 knew just how to assemble that 4870 ROM.

I tried using that script to "sew" a couple Nvidia bits together but no joy.

PPC was so much easier, 1 ROM and No Checksums. Although they did have those tedious tokenizing strings.

Maybe it's time for a contest.

the only difference will be different location and size of efi.part in 285rom

rest of script should work fine.

 

[MacVidCards]

2nd Flashed GTX285....with pretty color pictures !

We Americans LOVE our color pictures. So to spice up this thread, I'm including some.

A fellow member from up North sent me an EVGA GTX285 for "the treatment"

Bad news is of the 3 cards I have checked, 2 had 64K chips and one had a 128K.

So even if we could whittle 28K out of the ROM, we would just be back to our hit-and-miss on 128K chips.

I have shown the chip as it appears in NvFlash and as it appears up close.

I am also attaching the output from the card once it had a 256K chip.

The flash takes quite some time on a Mac, not sure why. (couple of minutes perhaps...dots advancing very slowly on bottom of screen)

This card took the flash and worked as before pn my first Gen via Netkas magic helpers. It should be fine in an EFI-64 machine now.

Will ship it back to owner and wait to hear how it works for him. Looking forward to another success.

 

[zebity]

Where is ~CRC32 on ROM and what Region is this calculated on...

Rominator & Co,

do you have any idea where the CRC is stored and what portion of ROM is involved in CRC calculation...

I presume that this must be a subset of ROM as otherwise the "tricks" that people have been doing by directly editing ROM using Hex Editors to tweek clock speeds etc would fail as they would affect CRC.

To confirm this I wonder if anyone has ever done on the fly adjustment via RivaTuner/NiBiTor and then redump ROM and confirmed that ~CRC32 reports the same return as pre "tuning"?

Now I have Mac 285 ROM and PCI Blocks Addresses I should be able to do a binary compare of EVGA 2G x86 BIOS with equivalent x86 BIOS portion of Mac 285 ROM. If the two x86 BIOS portions are essentially same then this would validate idea that EFI64 ROM is created by getting EFI portion and BIOS ROM portions and then concatentating them together (sewing) with addition of some "extra" headers and CRC check.

Is there anyone out there who understands the details of EFI Boot loader and so can provide guidance of the ROM file structure?

Cheers,

Zebity
(MacBook Pro, Mac Pro & Mini Mac)

 

[gugucom]

This thread contains the sum total of what is known regarding flashing these. So, so far we have found 2 with 64K chips and one with a 128K chip. Not promising, but it is possible that there are cards out there with 256K chips.

Meanwhile, I am still open to making a SECOND flashed GTX285 for someone who brings their 3,1 or 4,1 Mac Pro and a 1 Gig EVGA card over. I have a 2 Gig version I need to test........

When I was looking into differences between PC 3870 and 3870 Mac&PC I found that one difference was the size of the serial EEPROM. Of course the Mac&PC had the bigger 128kB version. I decided to replace an 64kB with a 128kB in a PC card and found a specialist in Germany www.bios-fix.de who fit a SST25VF010A-33-4I-SAE for 29€. I thought that wasn't such a bad price considering that the flashing and mailing of the card was included. Perhaps that address is usefull for other European experimentors. I eventually sold that card as a PC card because there were obviously other modifications which I couldn't identify. I also try to attach the spec sheet of the EEPROM.

EDIT: Of course you can also put a 256 kB EEPROM on and they will load your ROM file before soldering the EEPROM to the card which is handy.

 

[netkas]

Rominator & Co,

do you have any idea where the CRC is stored and what portion of ROM is involved in CRC calculation...

I presume that this must be a subset of ROM as otherwise the "tricks" that people have been doing by directly editing ROM using Hex Editors to tweek clock speeds etc would fail as they would affect CRC.

To confirm this I wonder if anyone has ever done on the fly adjustment via RivaTuner/NiBiTor and then redump ROM and confirmed that ~CRC32 reports the same return as pre "tuning"?

Now I have Mac 285 ROM and PCI Blocks Addresses I should be able to do a binary compare of EVGA 2G x86 BIOS with equivalent x86 BIOS portion of Mac 285 ROM. If the two x86 BIOS portions are essentially same then this would validate idea that EFI64 ROM is created by getting EFI portion and BIOS ROM portions and then concatentating them together (sewing) with addition of some "extra" headers and CRC check.

Is there anyone out there who understands the details of EFI Boot loader and so can provide guidance of the ROM file structure?

Cheers,

Zebity
(MacBook Pro, Mac Pro & Mini Mac)

there is no CRC, bios part of rom must have some predefined checksum (16-bit)

 

[zebity]

Right you are...

Netkas,

ahh... I thought that the nvflash tool did a check of embedded CRC id against actual ROM prior to allowing that ROM to be loaded into card as protection against loading corrupted ROM.

But it appears you are right. I just mangled up ROM via HexEditor and then ran nvflash over both the original and mangled ROMs and it happily treated both as valid, even though the mangled ones CRC was different.

#nvflash --version evga285m.rom /* clean rom */
NVIDIA Firmware Update Utility (Version 5.80)

Image Size            : 152064 bytes
Version               : 62.00.58.00.06
~CRC32                : A36A53A8
OEM String            : NVIDIA
Vendor Name           : NVIDIA Corporation
Product Name          : NVIDIA GeForce GTX 285

Product Revision      : Chip Rev   
Device Name(s)        : GeForce GTX 285
Board ID              : C826
PCI ID                : 10DE-05E3
Subsystem ID          : 3842-1080
Hierarchy ID          : Normal Board
Build Date            : 03/06/09
Modification Date     : 04/28/09
Sign-On Message       : NVIDIA GT200 Bringup

#nvflash --version evga285x.rom /* edited rom */

NVIDIA Firmware Update Utility (Version 5.80)

Image Size            : 152064 bytes
Version               : 62.00.58.00.06
~CRC32                : 0C3A1C65
OEM String            : NVIDIA
Vendor Name           : NVIDIA Corporation
Product Name          : NVIDIA GeForce GTX 285

Product Revision      : Chip Rev   
Device Name(s)        : GeForce GTX 285
Board ID              : C826
PCI ID                : 10DE-05E3
Subsystem ID          : 3842-1080
Hierarchy ID          : Normal Board
Build Date            : 03/06/09
Modification Date     : 04/28/09
Sign-On Message       : NVIDIA GT200 Bringup

So CRCs are different but it does not report any problem... I had obviously falsely assumed that CRC was one of problems with being able to "sew" ROMs.

Speculatively it would seem that most of the EFI code in ROM is to "display" the Apple Logo and Spinning Wheel on boot, with primary data vendor id, clock, pci id etc sitting in x86 BIOS portion of the image and extracted from there by EFI image. As what is missed between EFI ROM vs. plain PC ROM and your injectors is boot behavior (ie Logo & Spinning Wheel). This might explain why EFI code section is so large relative to x86 BIOS section, as it is it doing a lot of extra "feel good" stuff, pending receiving signal from the Machine to activate the display.

Cheers,

Zebity
(MacBook Pro, Mac Pro & Mini Mac)

 

[MacVidCards]

Speculatively it would seem that most of the EFI code in ROM is to "display" the Apple Logo and Spinning Wheel on boot, with primary data vendor id, clock, pci id etc sitting in x86 BIOS portion of the image and extracted from there by EFI image. As what is missed between EFI ROM vs. plain PC ROM and your injectors is boot behavior (ie Logo & Spinning Wheel). This might explain why EFI code section is so large relative to x86 BIOS section, as it is it doing a lot of extra "feel good" stuff, pending receiving signal from the Machine to activate the display.

Cheers,

Zebity

My fiddling with G92 cards backs this up.

Cards are "run" from PC BIOS.

This is exactly the sort of progress I was hoping to make in my "EFI Renaissance" thread. Once we crack this sewing thing we will be able to add many cards.

 

[GrnEydDvl]

Great Success

Got the card back from The Rominator, installed it, and so far everything is working great.

I've tested it in OS X and Vista 32-bit. Runs like a champ. As expected, my flashed card seems to behave just like the OEM Mac gtx285's.

I was slightly nervous shipping the card to Rominator, but he got it done and back to me ASAP. Card shipped out on a Tuesday and I received it back by the following Monday. Thanks Rom!

A fellow member from up North sent me an EVGA GTX285 for "the treatment"...



... Will ship it back to owner and wait to hear how it works for him. Looking forward to another success.

 

[zebity]

Getting a 2GB GTX 285 Up and Running

Rominator & others,

ok I got my EVGA GeForce GTX 285 2GB FTW Edition (overclocked) version of card and here are ROM details:

#nvflash --version
NVIDIA Firmware Update Utility (Version 5.80)

Adapter: GeForce GTX 285      (10DE,05E3,3842,1187) H:--:NRM B:01,PCI,D:00,F:00

The display may go *BLANK* on and off for up to 10 seconds during access to the EEPROM depending on your display adapter and output device.

Identifying EEPROM...
EEPROM ID (9D,007B) : PMC Pm25LV512 2.7-3.6V 512Kx1S, page
Reading adapter firmware image...
Image Size            : 64512 bytes
Version               : 62.00.5E.00.80
~CRC32                : E52B9551
OEM String            : NVIDIA
Vendor Name           : NVIDIA Corporation
Product Name          : GT200 Board - 08920053
Product Revision      : Chip Rev   
Device Name(s)        : GeForce GTX 285
Board ID              : C82E
PCI ID                : 10DE-05E3
Subsystem ID          : 3842-1187
Hierarchy ID          : Normal Board
Build Date            : 03/26/09
Modification Date     : 05/12/09
Sign-On Message       : GT200 P892 SKU 0053 VGA BIOS

#nvflash --pciblocks
NVIDIA Firmware Update Utility (Version 5.80)

Adapter: GeForce GTX 285      (10DE,05E3,3842,1187) H:--:NRM B:01,PCI,D:00,F:00

The display may go *BLANK* on and off for up to 10 seconds during access to the EEPROM depending on your display adapter and output device.

Identifying EEPROM...
EEPROM ID (9D,007B) : PMC Pm25LV512 2.7-3.6V 512Kx1S, page
Reading adapter firmware image...
Location        Size Type           
000000:00FBFF  64512 x86

So first problem is that as predicted board has 512K Bit EEPROM, so I will need to get 2048K chip and do some soldering... I see that the "MX25L2005" 2M Bit chips are available on eBay so I have ordered a few.

Note that both the 2GB FTW board and Mac Edition boards have identical x86 (PC BIOS) ROM sizes (64512 Bytes) .... so it looks like doing a replacement of Mac Edition 1GB x86 portion of ROM with 2GB FTW portion and then putting new EFI'ed ROM back onto 2GB FTW might work. This will mean I get an overclocked 2GB GTX 285 on my Mac Pro...

I have done hex dumps of both ROMs and compared them and the differences appear to be mostly related to version id differences/compile date stamps etc, towards start of ROMs, with bigger differences in main part of ROM. I am assuming that larger differences are due to PhysX code on FTW... I have new ROM image ready to go, now just need to get 2M Bit EEPROMs via post and do some soldering.

I presume that when you flashed "blank" GTX 285 after you soldered on new ROM, you did so via "autoexec.bat" sequence, as GTX would not be bootable as it has no ROM in it...

Cheers,

Zebity
(MacBook Pro, Mac Pro, Mini Mac)

 

[MacVidCards]

With blank ROM on it, I used an ATI card to flash from.

Specifically, a 2600XT.

BTW...if you have any other Nvidia cards around with EFI, you can test the 2 Gig via the Injector.

I am having a weird issue related to the "-" that OSX is adding before the "2048" in VRAM.

It appears that many 3D apps take this literally, believing there to be no RAM to run. I have tried a few different things and have been unable to figure out an easy solution.

I imagine Apple will fix in future OS update, but I am not so patient.

My 2Gig is the FTW too...nice card but the "2 Gig Bug" is bugging me.

 

[netkas]

I am assuming that larger differences are due to PhysX code on FTW...

:ROFL:

Since size of bios part is same just replace it.

 

[zebity]

Disassembling EVGA 285

Rominator and others,

I am having a problem opening up the EGVA 285 board. I have removed all the screws on back of circuit board (x8) and the PCI bracket screws (x 2). But cover does not come off.

I loosened the 3 screws on fan, but they do not appear to be what is holding on cover...

Are there other screws under the "GeForce GTX 285" sticker on the fan enclosure? Or do I need to remove the 2 screws on the top inner edge near the SDI connector tab?

Thanks for guidance.

Cheers,

Zebity
(MacBook Pro, Mac Pro, Mini Mac)

 

[MacVidCards]

14 total screws

12 on back (the bigger ones, not little ones around GPU)

And 2 near the DVI connectors.

 

[zebity]

Card is dead as a dodo...

Rominator,

I got 2M Bit EEPROM chip installed into GeForce GTX 285 FTW board, but when I try to flash I get a "No NVIDIA display adaptors found" message.
I am using the inbuilt VGA board on PC to boot to and cannot put second PCI-e board into machine as it only has a single slot...
Is this why I need to boot from machine with Nvidia card in it to get an Nvidia ROM loaded to allow me to bootstrap the Blank 285?
Is is possible to load the new blank 285 ROM without having to have second Nvidia board in the machine?

I do not think the card is totally toasted as it does appear as a PCI device when put into Mac Pro...

see attachment.

Thanks,

Zebity
(MacBook Pro, Mac Pro (Early 2008), Mini Mac)