MP 1,1-5,1 MacPro5,1: BootROM thread | 144.0.0.0.0

[tsialex]

Anyone here has a mid-2012 Mac Pro built in the second semester of 2013? I only have 3 dumps of mid-2012 Mac Pros made after July 2013 and I need more.

Seems that the last ones have a newer hardware descriptor and I only found one example until now.

Usually mid-2012 Mac Pros have the Base_21 descriptor with this header/version:

01050000 00000809

I have a dump that still has the Base_21 but with this header/version:

01060000 00000809

If anyone have mid-2012 with 2013 build dates and could PM the dump, preferably later build dates, I'll be grateful. No need to send it again if you ever sent it before.

The easiest way to check is running MachineProfile:

View attachment 829574
[doublepost=1554028890][/doublepost]Btw, refurbished 5,1 Mac Pros have different headers/versioning, usually 010A thru 010D. This one is from a mid-2012 refurb:

010D0000 00000809

refurbished 4,1 Mac Pros seems to start with 0107:

01070000 00000810

[doublepost=1554029101][/doublepost]Mid-2010 Mac Pros that had the backplane replaced by Apple after 2013 would be interesting too.

I was trying to get more dumps to confirm the 0106 header version that I only have one dump, what I got? A dump with a new 0107 header…

Seems that every time I try to confirm something I just get new questions

 

[crjackson2134]

I was trying to get more dumps to confirm the 0106 header version that I only have one dump, what I got? A dump with a new 0107 header…

Seems that every time I try to confirm something I just got new questions

Cool, something new learned.

 

[tsialex]

Cool, something new learned.

I really want to know when Apple stopped the MP5,1 production. Initially I thought that it ended around early September, but I already have one dump that shows a backplane that was made in the end of it, so probably Apple produced mid-2012 around October too. Maybe in the future we will find a Base_22 hardware descriptor?!? I don’t doubt it anymore.

 

[bsbeamer]

I really want to know when Apple stopped the MP5,1 production. Initially I thought that it ended around early September, but I already have one that the backplane that was made in the end of it, so probably Apple produced mid-2012 around October too. Maybe in the future we will find a Base_22 hardware descriptor?!? I don’t doubt it anymore.

If it’s at all helpful, I do recall being contacted from Apple Business rep around the Thanksgiving US Holiday that any remaining BTO (custom) orders for the “tower” needed to be placed before December that year. They ended that style Apple Business program with dedicated rep shortly after.

 

[tsialex]

If it’s at all helpful, I do recall being contacted from Apple Business rep around the Thanksgiving US Holiday that any remaining BTO (custom) orders for the “tower” needed to be placed before December that year. They ended that style Apple Business program with dedicated rep shortly after.

This is interesting, so the production continued until beginning of December, at least. A Mac Pro with a SSN from December will probably have a backplane BD of early November, the earliest one that I have is from late September for a complete Mac Pro, not replacement parts.

Another thing, replacement parts continued to be made after 2013, I have a dump from a backplane made in 2014 that don’t have the 17 digit serial of refurbished backplanes.

 

[XNorth]

The manufactured date for my cMp was November 2012. Let me know if that could be helpful.

 

[tsialex]

The manufactured date for my cMp was November 2012. Let me know if that could be helpful.

Thx for your offer, really appreciated.

I’m trying to find MP5,1s made at the end of the production, after October 2013.

Replacement backplanes made in 2014 are interesting too, but those are more complicate to identify, you have to look at the MLB/LBSN sector for the real BuildDate.

 

[Woof Woof]

I was trying to get more dumps to confirm the 0106 header version that I only have one dump, what I got? A dump with a new 0107 header…

Seems that every time I try to confirm something I just get new questions

That's what happens when corporate doesn't give you budget until November, and you didn't want a 2013 Mac Pro because you had an investment in Aja capture cards ;-)

 

[handheldgames]

Thx for your offer, really appreciated.

I’m trying to find MP5,1s made at the end of the production, after October 2013.

Replacement backplanes made in 2014 are interesting too, but those are more complicate to identify, you have to look at the MLB/LBSN sector for the real BuildDate.

Were they manufactured in 2014 or configured in 2014? It's hard to think apple would run the manufacturing line on such a small unit basis.

 

[tsialex]

Were they manufactured in 2014 or configured in 2014? It's hard to think apple would run the manufacturing line on such a small unit basis.

Manufactured, but this is replacement parts, not full systems.
[doublepost=1554337516][/doublepost]I only have one dump, I’ll wait more examples before any conclusions.

 

[tsialex]

The manufactured date for my cMp was November 2012. Let me know if that could be helpful.

I tried to send the instructions for you by PM on how to dump and other info needed to validate the dump, but seems you only receive PMs by people you follow, I imagine.

 

[tsialex]

I've been noticing one weird pattern lately. Some Mac Pros are not erasing old memory config parameters from the NVRAM after you change RAM.

Today I've found a dump that has 16 memory configs for just 4 16GB DIMMS:

1. 0x120076
2. 0x12281D
3. 0x12301D
4. 0x12381D
5. 0x1244D9
6. 0x124cd9
7. 0x1254D9
8. 0x1262fb
9. 0x126afb
10. 0x127acc
11. 0x1282cc
12. 0x128ACC
13. 0x130076
14. 0x13281D
15. 0x13301D
16. 0x13381D

I'd expect just 8, 4 in the 1st stream/store and 4 in the 2nd stream/store.

I'm going to keep an eye on this, seems a problem/bug to me.
[doublepost=1554462869][/doublepost]First I was thinking that this was a collateral problem of Netkas upgrade, since some parameters are stored differently in MP4,1 NVRAM from MP5,1, like the blobs that make TITAN GPUs not work, but this is a mid-2012.

Weird, no?

 

[bsbeamer]

Do RAM modules that do not register manufacturer code, part number and serial number play a factor into this at all?
Does upgrading from single CPU to dual CPU tray have any impacts?

 

[tsialex]

Do RAM modules that do not register manufacturer code, part number and serial number play a factor into this at all?

Another curiosity, some firmware along the way changed the way the memory config data are stored, some dumps have big binary blobs with nothing identifiable (maybe a SPD dump?!?), but some recent ones have little blocks, almost an ASCII way to store the config, with the serial and some data clearly identifiable.

Does upgrading from single CPU to dual CPU tray have any impacts?

I don’t know if this have implications yet, still early days. It’s something that I thought that was a fluke, then become an observable pattern lately.

It’s not easy to check this here, my Macs never had this behavior, I’ll probably need to find someone that have the problem and have dual and single trays plus time to do tests.

 

[h9826790]

I've been noticing one weird pattern lately. Some Mac Pros are not erasing old memory config parameters from the NVRAM after you change RAM.

Today I've found a dump that has 16 memory configs for just 4 16GB DIMMS:

1. 0x120076
2. 0x12281D
3. 0x12301D
4. 0x12381D
5. 0x1244D9
6. 0x124cd9
7. 0x1254D9
8. 0x1262fb
9. 0x126afb
10. 0x127acc
11. 0x1282cc
12. 0x128ACC
13. 0x130076
14. 0x13281D
15. 0x13301D
16. 0x13381D

View attachment 830490

I'd expect just 8, 4 in the 1st stream/store and 4 in the 2nd stream/store.

I'm going to keep an eye on this, seems a problem/bug to me.
[doublepost=1554462869][/doublepost]First I was thinking that this was a collateral problem of Netkas upgrade, since some parameters are stored differently in MP4,1 NVRAM from MP5,1, like the blobs that make TITAN GPUs not work, but this is a mid-2012.

Weird, no?

I wonder if this is the reason why some people can use 32GB DIMM, but some others cannot (even for the exact same part number).

 

[tsialex]

I wonder if this is the reason why some people can use 32GB DIMM, but some others cannot (even for the exact same part number).

I commented about this with you last year, maybe the multiple memory configs has implications with this problem.

 

[Spacedust]

Windows 8.1 Pro installed twice in UEFI mode then upgraded to 141.0.0.1 and I got this, but all works perfectly fine:

DECIMAL       HEXADECIMAL     DESCRIPTION
--------------------------------------------------------------------------------
0             0x0             UEFI PI firmware volume
16524         0x408C          UEFI PI firmware volume
24972         0x618C          CRC32 polynomial table, little endian
35787         0x8BCB          mcrypt 2.2 encrypted data, algorithm: blowfish-448, mode: CBC, keymode: 8bit
49948         0xC31C          UEFI PI firmware volume
524288        0x80000         UEFI PI firmware volume
540812        0x8408C         UEFI PI firmware volume
549260        0x8618C         CRC32 polynomial table, little endian
560075        0x88BCB         mcrypt 2.2 encrypted data, algorithm: blowfish-448, mode: CBC, keymode: 8bit
574236        0x8C31C         UEFI PI firmware volume
1048576       0x100000        UEFI PI firmware volume
1114112       0x110000        UEFI PI firmware volume
1213557       0x128475        XML document, version: "1.0"
1343511       0x148017        bzip2 compressed data, block size = 100k
1376256       0x150000        UEFI PI firmware volume

 

[tsialex]

Not erasing previous memory configs parameters is definitively a thing.

Look at this one, from the BootROM @Spacedust sent me to check:

1. 0x120076
2. 0x1227ad
3. 0x122fd9
4. 0x1237d9
5. 0x123fd9
6. 0x124892
7. 0x125092
8. 0x125892
9. 0x126b7a
10. 0x12737a
11. 0x127b7a
12. 0x1290d3
13. 0x1298d3
14. 0x12a0d3
15. 0x12b1ef
16. 0x12b9ef
17. 0x12c1ef
18. 0x12ca9d
19. 0x12d29d
20. 0x12da9d
21. 0x130076
22. 0x1327ad
23. 0x132fd9

 

[bsbeamer]

Could this “fill up” and eventually run out of space if lots of RAM swapping/reconfiguration was happening?

 

[flehman]

Is this an issue that can be addressed by your BootROM reconstruction process?

 

[tsialex]

Could this “fill up” and eventually run out of space if lots of RAM swapping/reconfiguration was happening?

I have various examples of totally full NVRAM.

Multiple memory configs + SecureBoot certificates (they are big) + IASInstallPhaseList (can be very long sometimes) can make a NVRAM full and a Mac practically non functional.

Btw, the first Mac that had the NVRAM totally full was a Mac mini late-2012 from @highvoltage12v. After I cleaned it up, I started to check MP5,1 NVRAM for the same problems too.
[doublepost=1554553883][/doublepost]

Is this an issue that can be addressed by your BootROM reconstruction process?

Yes, the intermediate NVRAM has the 1st and 2nd streams empty, one of my early decisions that proved to be totally correct, and this is totally resolved after flashing the reconstructed BootROM.
[doublepost=1554555285][/doublepost]This is probably a thing that Apple noticed very early:

• MP3,1 has 32KB NVRAM for everything on the NVRAM, with just 2MB FWB flash memory
• MP4,1/MP5,1 SPI flash is 4MB and changed the NVRAM to 192KB, 64KB for the 1st stream, 64KB for the 2nd.
• MP6,1 doubled the size of the SPI flash.

 

[Macschrauber]

Does the memory configurations get deleted with zapping pram or doing a nvram -c ?

Explains why it is sometimes necessary to zap when upgrading.

 

[tsialex]

Does the memory configurations get deleted with zapping pram or doing a nvram -c ?

Explains why it is sometimes necessary to zap when upgrading.

With a normal NVRAM, yes. It’s common to need to force the Mac Pro firmware re-do the memory config. It’s an habit of mine every time I change RAM or CPU-tray to clear the NVRAM 3 times.

When you already have multiple configs/full NVRAM, no. That's what intrigued me with the Mac mini dump that I talked about.

At least it's a relatively easy thing to prevent and repair. Btw, this is not something widespread, just some dumps have it. Until more evidence is found, I’m inclined to think this problem as an unfortunate confluence of events, more like the worst case and not the norm.

 

[Spacedust]

I've never changed CPU's in this machine. It came upgraded with 2x5675's, then upgraded from stock RAM (12 GB) to 64 GB RAM and then to 96 GB RAM. Never touched them after this time (probably about 4-5 years).

However I keep changing GPUs, PCI cards and SSDs.

 

[bplein]

I have a 2009 4,1 that I upgraded to 5,1 via the usual firmware flash. I normally run Ubuntu on it, but I have a High Sierra disk that I can drop in for firmware updates.

I finally bought a Mojave compatible PC GPU (Nvidia GT 630 2 GB) so I could get on the NVMe train to perfomance!

I booted to HS and downloaded and ran the Mojave updater. It correctly says that I need to shutdown, hold the button until I see the lights flash or a long beep.

When I folllow these instructions, I wait and after a while (15-20s?) the power lights flash quickly, then more slowly, then stop and the CD-ROM opens. The beep never happens. But I've tried releasing as soon as the fast lights flash, or after.

In either case, the system sits with no video and no other apparent activity. I can wait 30-60 minutes and nothing happens. If I power it off and back on at this point, it boots twice and then I am back in the old OS and it still wants me to update the firmware.

Any ideas what might be happening? Thanks in advance.