My post was only to write in the post, following instructions...
And thanks for your contribution!!!
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP 1,1-5,1 MacPro5,1: BootROM thread | 144.0.0.0.0
- Thread starter tsialex
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
And thanks for your contribution!!!
@tsialex: It seems I have exactly the same problem with a Mac Pro 4.1 (8-core), BootROM MP41.0081.B07 running OS X El Capitan and Windows 10 (Legacy). From time to time I have used the Boot Camp tool under Windows to change the startup disk. In the last few weeks, the number of memory configurations has increased from 9 to 19 after a deep NVRAM reset and normal usage. This is probably critical if the 2nd VSS store is no longer working. Is there any way to contact you directly as a new user or should I post the full information here for everyone to be documented?
I’ve flashed my 4,1 to 5,1 and installed Sierra. The Boot ROM is MP51.0089.B00. The System reports that the NVIDIA Quadro 4000 is Metal capable. The drive is a Samsung SSD 860 EVO 1TB
When I try to install High Sierra I get the error that I cannot install because I’m missing a Firmware Partition. If I install 144.0.0.0.0 will that take care of it?
When I try to install High Sierra I get the error that I cannot install because I’m missing a Firmware Partition. If I install 144.0.0.0.0 will that take care of it?
For me, with my ADHDI this is like wading through a swamp of alligators. I don't know what to erase and reformat or how to do it. Do you mean to download and reinstall Sierra?
Quadro 4000 isn't metal-capable, but Quadro K4000, M4000 and P4000 are supported.
Attachments
Make a createinstallmedia installer of High Sierra, nuke your disk, re-format it and then install High Sierra. This is not BootROM related, but incorrect disk partitioning. If you don't know how to do it, it's easy to use the search or to google it and find instructions.
Quadro 4000 is a Fermi GPU (GF100-825-A3), Apple does not support METAL with Fermi GPUs, your screenshot shows a known misidentification. METAL is only really supported with NVIDIA Kepler GPUs.
Okay, to start with you can’t imagine how much I appreciate both your computer knowledge and your attitude about helping people. We need more like you in this screwed up world.
So now that I’ve tried to butter you up ...
I do have the High Sierra Installer on a 16GB thumb drive.
My Sierra 10.12.6 and all my files are on a Samsung SSD. Would it work to plug in a hard drive, erase and format it, install High Sierra on it and use CCC to clone it to the SSD? Or would I still have the same firmware problem?
Another question is does High Sierra require METAL?
Oh, another. Am I now able to use 64GB of RAM and what type should I look for? And another. Why wouldn’t it work to just use Disc Utility and repartition to have only the one since it’s a disc partition problem. I did once start to use a second partition that’s still on my desktop. It’s only about 8GB. Could that actually be the reason I’m getting the install error?
So now that I’ve tried to butter you up
...I do have the High Sierra Installer on a 16GB thumb drive.
My Sierra 10.12.6 and all my files are on a Samsung SSD. Would it work to plug in a hard drive, erase and format it, install High Sierra on it and use CCC to clone it to the SSD? Or would I still have the same firmware problem?
Another question is does High Sierra require METAL?
Oh, another. Am I now able to use 64GB of RAM and what type should I look for? And another. Why wouldn’t it work to just use Disc Utility and repartition to have only the one since it’s a disc partition problem. I did once start to use a second partition that’s still on my desktop. It’s only about 8GB. Could that actually be the reason I’m getting the install error?
I've already written that this is not a firmware problem.Okay, to start with you can’t imagine how much I appreciate both your computer knowledge and your attitude about helping people. We need more like you in this screwed up world.
So now that I’ve tried to butter you up
I do have the High Sierra Installer on a 16GB thumb drive.
My Sierra 10.12.6 and all my files are on a Samsung SSD. Would it work to plug in a hard drive, erase and format it, install High Sierra on it and use CCC to clone it to the SSD? Or would I still have the same firmware problem?
Mac Pro from 1,1 to 6,1 stores the firmware inside a FWB flash memory or a SPI flash memory - nothing is stored on a disk. 2019 Mac Pro have a different design and the firmware is stored inside the T2 managed storage.
You have a disk partitioning problem that it's impeding the creation of the Recovery partition - I'm completely failing to identify what let you think that this have anything to do with Mac Pro firmware?!?
No. Mojave on wards.
Read the thread stickies, you will find the memory reference thread linked on one of the stickie threads (the upgrade one). It's possible to use 4x16GB DIMMs with a single CPU tray that have a X56xx Xeon (not a factory config) while for dual CPU trays, even the most basic one already supports 8x16GB DIMMs from factory.
Yes.
Last edited:
Thank you very very very. Being 85 and with ADHDI is a strange experience. Like there's a little electron zooming around my brain trying to escape. So I need all the help I can get.
That's for what we are here for. Don't hesitate to ask anything.
If you are having problems with your CPU tray, read the diagnostic steps of the Apple Technician Guide (what is usually called Mac Pro Service Manual) for your year model and find what's your problem. Google "Apple Technician Guide early-2009 Mac Pro" for example.
Btw, there are several threads on this forum specifically about CPU tray related problems, here this is completely off-topic since CPU trays problems with processors are not BootROM related in any way, please use the search and post on a more appropriate thread.
Btw, there are several threads on this forum specifically about CPU tray related problems, here this is completely off-topic since CPU trays problems with processors are not BootROM related in any way, please use the search and post on a more appropriate thread.
Last edited by a moderator:
I am working with a cMP 5,1 (mid-2010) that was used as a soft-RAID fileserver which I got for free from my work. It still has the dual 2.4GHz quad-core processors, four 4 GB DIMMS, and ATI Radeon 5770 in it, and my plan was to install a old GTX 760 card I have so I can run Mojave.
While I was waiting for a second auxiliary PCI-e power cable for the "new" video card to be delivered, I was sifting through all the information on OpenCore here because I am interested in having access to a boot menu that works with my non-flashed GTX 760.
Then I encountered the information on this thread about the potential issues production cMPs may have with a low amount of VSS Store space, corrupted NVRAM, and non-working garbage collection. So I started to check on the state of my cMPs NVRAM / VSS Store.
Here are things I did before the first dump of my eMPs BootROM that could have affected its NVRAM VSS store:
1. The VSS Store's free space on the first ROM dump was 20093, and the second store was fully empty (65448). The only repeated entries I see in that dump are MemoryConfig, but those seem to just be updated with a sequence letter at the end (ie., "MemoryConfig", MemoryConfih", "MemoryConfii", etc.) There were a bunch of entries marked Invalid.
2. On the second ROM dump (after 5-chime reset and a couple reboots to disable SIP) I found the VSS Store free space was much lower, at 6610. But the second store was still completely empty, and the entry list appears to be the same as before (no repeats, and quite a few Invalid entries.
3. After three more reboots, things changed a lot. There were only 3 entries in the main VSS Store marked Invalid and the free space has jumped up to 52695. But the second VSS store's free space was down to 55222, because it now has entries, all the same as what was in the main VSS Store except no Invalid entries and one was repeated (AAPL,PathProperties).
So if I'm understanding what I've been reading correctly, it appears garbage NVRAM collection was still working on my cMP 5.1. But then things got worrying.
4&5. I rebooted two more times before I got the GTX 760 card installed. On each reboot, the main VSS Store free space went down, to 46446, and then to 40197. There were more entries marked Invalid, but with the free space still well above 50% of the maximum, I decided to move forward with installing the NVIDIA card and beginning the Mojave upgrade with the Boot ROM update to 144.0.0.0.0.
6. After successfully updating to 144.0.0.0.0, I dumped the ROM again, and now the second VSS Store is gone! The main store free space is down to 24513, and I see a new message in the bottom pane output of the UEFITool: "findNextStore: VSS store candidate at offset 10048h skipped, has invalid size FFFFFFFFh".
I haven't started the Mojave install, because it appears the NVRAM has become corrupted. I'm not sure what the best course of action is at this point. Could this have happened because I jumped directly from BootROM MP51.0089.B00 to 144.0.0.0.0?
As an aside, I notice that the link speed for my GTX 760 card is 2.5GT/s, not 5.0GT/s, even though it's a PCI-E 3.0 card.
While I was waiting for a second auxiliary PCI-e power cable for the "new" video card to be delivered, I was sifting through all the information on OpenCore here because I am interested in having access to a boot menu that works with my non-flashed GTX 760.
Then I encountered the information on this thread about the potential issues production cMPs may have with a low amount of VSS Store space, corrupted NVRAM, and non-working garbage collection. So I started to check on the state of my cMPs NVRAM / VSS Store.
Here are things I did before the first dump of my eMPs BootROM that could have affected its NVRAM VSS store:
- Updated Boot ROM to MP51.007F.B03
- Reset NVRAM (2 chimes) and erased 3 HDDs to remove "broken" software RAID config
- Removed one HDD after SMART status showed imminent failure
- Updated Boot ROM to MP51.0089.B00 as part of High Sierra install
- Installed fresh copy of 10.13.6 High Sierra
- Rearranged and reseated four 4 GB DIMMs because at first one DIMM was not being recognized. Now all four are recognized and tested without any errors
- Had to reset NVRAM (2 chimes) again while trying to figure out how to convert High Sierra boot volume from HFS+ to APFS. It took some CLI hoop-jumping, but I got it converted.
1. The VSS Store's free space on the first ROM dump was 20093, and the second store was fully empty (65448). The only repeated entries I see in that dump are MemoryConfig, but those seem to just be updated with a sequence letter at the end (ie., "MemoryConfig", MemoryConfih", "MemoryConfii", etc.) There were a bunch of entries marked Invalid.
2. On the second ROM dump (after 5-chime reset and a couple reboots to disable SIP) I found the VSS Store free space was much lower, at 6610. But the second store was still completely empty, and the entry list appears to be the same as before (no repeats, and quite a few Invalid entries.
3. After three more reboots, things changed a lot. There were only 3 entries in the main VSS Store marked Invalid and the free space has jumped up to 52695. But the second VSS store's free space was down to 55222, because it now has entries, all the same as what was in the main VSS Store except no Invalid entries and one was repeated (AAPL,PathProperties).
So if I'm understanding what I've been reading correctly, it appears garbage NVRAM collection was still working on my cMP 5.1. But then things got worrying.
4&5. I rebooted two more times before I got the GTX 760 card installed. On each reboot, the main VSS Store free space went down, to 46446, and then to 40197. There were more entries marked Invalid, but with the free space still well above 50% of the maximum, I decided to move forward with installing the NVIDIA card and beginning the Mojave upgrade with the Boot ROM update to 144.0.0.0.0.
6. After successfully updating to 144.0.0.0.0, I dumped the ROM again, and now the second VSS Store is gone! The main store free space is down to 24513, and I see a new message in the bottom pane output of the UEFITool: "findNextStore: VSS store candidate at offset 10048h skipped, has invalid size FFFFFFFFh".
I haven't started the Mojave install, because it appears the NVRAM has become corrupted. I'm not sure what the best course of action is at this point. Could this have happened because I jumped directly from BootROM MP51.0089.B00 to 144.0.0.0.0?
As an aside, I notice that the link speed for my GTX 760 card is 2.5GT/s, not 5.0GT/s, even though it's a PCI-E 3.0 card.
Yep, now the NVRAM volume is corrupt.
No. The problem is with the design of the NVRAM volume.
This happens a lot, it’s a 10+ years old Mac that was never intended to be running after all this time.
You will need a BootROM reconstruction. I’ll send you a PM later.
NVIDIA GPUs change the PCIe connection speed on the fly, while AMD don’t. NVIDIA default is PCIe v1.0
I want to stay (for now) on 10.13 HS but upgrade my firmware to 144. Can I do this with my non-Metal 5770?
Because, when I download Mac OS Mojave, open the installer, it tells me I need a metal card to run Mojave and doesn't let me select "shut down" like I have seen in the guides. Just to repeat, running Mojave is not my intent. I wasn't aware this was a requirement for the 144 update. How do I get around it?
Because, when I download Mac OS Mojave, open the installer, it tells me I need a metal card to run Mojave and doesn't let me select "shut down" like I have seen in the guides. Just to repeat, running Mojave is not my intent. I wasn't aware this was a requirement for the 144 update. How do I get around it?
It's a requirement, no way get around it using Apple tools. The only way to bypass is by a firmware reconstruction - borrow a METAL supported GPU and do it the normal way.
It’s unfortunate since after doing the 5-chime NVRAM reset, things appeared to be in working order.
Thank you.
For now, I’ve flashed the BootROM with the “best” dump from today (main VSS store free space at 52695, and the second store intact and empty, but with the BootROM version downgraded back to MP51.0089.B00). After 1 reboot the main store free space went down a little, but since the second store was still okay, I just shut down the machine until I can figure out what to do.
I’m just glad I encountered this thread when I did and had a look at my cMP’s NVRAM situation before possibly bricking the machine.
@tsialex quick question: for those of us with (your) reconstructed BootROM, is there any benefit in having a recurring schedule to flash the BootROM "as brand new"..? Say like flashing it once a year or maybe once every 6 months?
I can imagine that if our cMPs managed to survive many years without BootROM tweaks that we might probably be ok waiting longer before flashing it again, but thought to ask if there is any suggested way to go about this. Thanks!
I can imagine that if our cMPs managed to survive many years without BootROM tweaks that we might probably be ok waiting longer before flashing it again, but thought to ask if there is any suggested way to go about this. Thanks!
The Mac Pro early-2009 to mid-2012 platform BootROM design was made back in 2008, when the NVRAM usage was something eventual, now is multiple times per day with a ridiculously tiny space compared to the last iMac without the T-2 architecture, 2019 iMac, that have more than 4x times the space.
It's a completely different scenario today and the MP51 BootROM is now the Achilles heel, you don't want the lack of NVRAM main store space to pressure the NVRAM volume and you should make a recurrent Calendar appointment to flash your never booted image from time to time, something like 90 or 120 days and re-flash it whenever you do significant hardware upgrades, like when changing the CPU tray or the DIMMs.
Yep! Something I’ve been doing ever since the discovery of errant memory configurations that keep getting added in for unknown reasons. In my case, I’ve noticed it seems to happen mostly after an OS update combined with a reboot, or when I boot to a different disk with a different OS a install.