cMP {classicMacPro}{4,1 & 5,1} - Upgrade Guide Sticky Discussion

Discussion in 'Mac Pro' started by n8ballz, Jan 4, 2018.

Thread Status:
The first post in this thread is a WikiPost, and can be edited by anyone with the appropriate permissions.
  1. Mumford macrumors regular

    Mumford

    Joined:
    Oct 8, 2006
    Location:
    Altadena, CA
    #26
    I just took an SSD I had from an old MBP that still had the OS on it from mid 2016 and installed it in a Mac Pro 5,1. It booted and ran just fine.
     
  2. justin1911 macrumors newbie

    Joined:
    Dec 4, 2017
    #27
    I attempted to upgrade to two X5690s. I was unsuccessful. In fact, at the moment I'm worse off than I started because now the RAM in slot 1 doesn't show up, no matter which chip I put in the slot. And the red light is on. I've heard this could be related to the tightness of the heat sink, but I suspect I fried the slot when I didn't have a RAM stick fully seated. So I put the original CPUs back in it, and its running fine, minus RAM slot 1.

    I couldn't get the machine to boot, and when I finally did, I got the same kernel error black screen over and over, even when trying to boot to USB. I got the white apple screen and the chime. I'll try and dig up the picture I took of the error and report back.

    I am going to try it once more tonight. If I can't get it, I'm going to send it off to an upgrade service on eBay. They are pricey, but the time investment involved is worth it alone.

    My questions are these:

    My CPUs were from eBay from two different vendors. They appear to be the same, and have the same writing on the IHS, but I have no way of 100% confirming that they are a "matched pair". Any tips?

    Also, since these were used one of them could be bad. I haven't found a decent way to to confirm that though, other than to check for damage to the gold plates. I can't see any. Does anyone know a way to test a CPU?

    Should I upgrade one CPU with the 2nd slot empty, and then upgrade the 2nd CPU?

    And lastly, should all of this fail, which I'm expecting it will, what should I look for in a good CPU upgrade service? Or is there a recommendation or preference?


    Thanks so much. You all on this forum have been the only thing to keep me sane at times. :)
     
  3. Squuiid macrumors 65816

    Squuiid

    Joined:
    Oct 31, 2006
    #28
    2009 cMP or 2010+?
    If 2009 did you delid or use washers?
     
  4. ededoad macrumors newbie

    Joined:
    Dec 18, 2015
    Location:
    Corpus Christi Tx
    #29
    I had a 4,1 that was running as a 5,1 with sierra. I put in my SSD with El Capitan and tried to boot it. On the 3rd try it booted but the firm ware version reverted to 4,1. Repeted a
    --- Post Merged, Feb 9, 2018 ---
    I hit the wrong button by accident, anyway repeated attempts to flash the firmware failed. My boot rom ware ends in B08 and others in this thread say El Capitan and B08 are not compatible. I have ordered a Yosemite Install disk and try from there.
     
  5. justin1911 macrumors newbie

    Joined:
    Dec 4, 2017
    #30
    Early 2009 4,1. I did not delid, although I am considering doing it. But I also didn't use washers. The videos I watched that didn't delid, said not to use washers, or didn't mention them.
     
  6. Squuiid macrumors 65816

    Squuiid

    Joined:
    Oct 31, 2006
    #31
    Oops. You’ve quite likely permanently damaged your cMP.

    Delidding or washers are a must.
    My preference is delidding without a doubt.
     
  7. owbp macrumors 6502a

    owbp

    Joined:
    Jan 28, 2016
    Location:
    Belgrade, Serbia
    #32
    Especially X5690s being about 100-120 USD and dual board still around 400 USD.

    @justin1911 , there are many YT videos, or guides, that are using stock CPUs in their 4,1 dual upgrades, and is a big problem for people that are doing it for the first time. It is the worst way you could go.

    Even if you do it right the first time, what will happened when you (or the guy that buys it from you) wants to disassemble it for the proper cleaning or replace the thermal paste...?
    Check the CPU pins on the motherboard's socket for the bent ones.
    Maybe you could try and straighten them if there are some that are bent.

    https://www.anandtech.com/show/2800/upgrading-and-analyzing-apple-s-nehalem-mac-pro/11

    borkedsocket2.jpg borkedsocket.jpg
     
  8. justin1911 macrumors newbie

    Joined:
    Dec 4, 2017
    #33
    Can you be more specific regarding the permanent damage? I assume you're referring to the RAM slot?

    To be clear, I'm posting to you all now from the Mac Pro 4,1 and have been using it as my daily driver for the past month since the upgrade.
     
  9. Squuiid macrumors 65816

    Squuiid

    Joined:
    Oct 31, 2006
    #34
    See owbp’s post above.
    Your RAM issue is indicative of the above.
     
  10. ededoad macrumors newbie

    Joined:
    Dec 18, 2015
    Location:
    Corpus Christi Tx
    #35
    After spending a few days reviewing as many of these threads as I could find it seems that many flashed 4,1 to 5,1 eventually have some problems. Mostly will not boot. I have decided to take the cowards way out and bought a 5,1 back board and tray. This is about what I did when I went to a 3,1 from my 2,1. I just bought a mother board and 2 memory boards and crossed some wires cut some holes and its been running fine for the last couple of years.

    If anyone is interested I will let you know how it comes out.
     
  11. Squuiid macrumors 65816

    Squuiid

    Joined:
    Oct 31, 2006
    #36
    Sorry, I don't mean to sound rude here but your research is flawed and/or you have been misinformed.
    Your statement is categorically incorrect.
     
  12. h9826790 macrumors G4

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #37
    I agree that quite a few flashed 4,1 case here. However, IMO, none of them are really due to the 5,1 firmware.

    Most of them are hardware issues (PSU, Logicboard, NB...), which are pretty much firmware independent.

    Some of them are user’s fault. E.g. damage the socket, try to boot from NVMe, installed a unsupported GPU, etc. Again, not really firmware related.

    So far, there is no evidence that a flashed 4,1 has higher failure rate then unflashed 4,1. The 5,1 has less issue is expected because it’s a newer machine. The electronics are younger and still has longer life span to go.

    Also, less thread about unflashed 4,1 may be because those users are simply never upgrade their Mac, so, less issue overall. But this is not directly firmware version related.

    There are plenty of happy flashed 4,1 users here, including me. If the firmware version is an issue, it should pop up on day 1, but not randomly select some users and randomly kill some hardware.
     
  13. kohlson macrumors 65816

    Joined:
    Apr 23, 2010
    #38
    I updated the firmware on my 4,1 about 6 months ago, after doing at least that much time in research in the MP forum. De-lidding, washers, maximum RAM, boot-screen GPUs (and 2.2 slot CPUs), PCIe-based SSDs - the list goes on. It's daunting, but eventually it made sense to me. My goal is to make this a usable 4K FCPX system. But the RX580 I have decided is "just the thing," well, mining has put a temporary halt to my plans - pricing these cards out of reach. And while the system does have its quirks (works great with Apple wireless keyboard and Magic Pad - not really at all with Magic Mouse) it has been a good performer for the several FCPX project so far. And it saved me when the 10.13.1 release screwed up the FCPX performance on my 2013 15-MBP.
     
  14. Theophany macrumors 6502a

    Theophany

    Joined:
    Nov 16, 2008
    Location:
    NW London.
    #39
    Here's an interesting theoretical for you all; have we hit the limit of marginal returns for graphics cards on these systems now or have we still got a way to go?

    I've got a 980Ti in my system and feels pretty punchy, but driving a 3840x1600 display is still pretty taxing for games. Given the age of the machine and the limitations of PCI-e 2.0, do the gains from newer hardware become more and more marginal? Would the performance jump from a 980Ti to a 1080Ti justify the cost of the upgrade?

    Admittedly, I'm looking more to the next generation of nVidia cards which will hopefully normalise GPU prices again, but I wonder if it isn't worth the extra spend because an old CPU, old RAM and old architecture will end up bottlenecking the latest and greatest hardware.
     
  15. h9826790 macrumors G4

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #40
    It really depends on how you use the GPU. If the software is very GPU bounded, PCIe 2.0 is not that limiting indeed. Here are some free benchmarks that you can run on your 980Ti, and than you can compare it to my 1080Ti's result, then decide if it's worth.

    Octanebench (macOS)
    Screen Shot 2018-02-13 at 19.33.25.jpg
    Luxmark 3 (macOS)
    Ball.jpg
    Unigine Heaven (macOS and Windows)
    Screen Shot 2018-02-13 at 19.38.23.jpg Screen Shot 2018-02-13 at 19.39.09.jpg
    Unigine Valley (macOS and Windows)
    Screen Shot 2018-02-13 at 19.38.42.jpg Screen Shot 2018-02-13 at 19.38.48.jpg
    Final Fantasy XV Windows Edition Benchmark (Windows)
    FF XV 4k High.JPG
     
  16. Theophany macrumors 6502a

    Theophany

    Joined:
    Nov 16, 2008
    Location:
    NW London.
    #41
    Interesting! We have the same CPU, so hopefully that should eliminate any variance and make it a straight comparison of GPUs.

    I scored 60.1 FPS and 1515 in Unigine and 51.0 FPS and 2135 in Valley using the same presets as you. That makes the 1080 Ti 27.6% faster in Heaven and 12.9% faster in Valley.

    I guess the caveats there are that the Unigine benchmarks run on OpenGL which are always going to look bad compared to Metal or DirectX in Windows.

    I did a run of GFXBench Metal too, which is free from the App Store and the results are below - ignore the 'onscreen' tests as these are run at native resolution, which for me is 3840x1600.

    [​IMG]

    For clarity, my cMP is a W3690, 32GB RAM, GTX 980 Ti 6GB FE, 525GB 2.5" SSD.
     
  17. h9826790, Feb 13, 2018
    Last edited: Feb 13, 2018

    h9826790 macrumors G4

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #42
    Here are the 1080Ti's results.
    Screen Shot 2018-02-13 at 22.22.47.jpg Screen Shot 2018-02-13 at 22.22.55.jpg

    And I make a table for easy comparison (only consider the offscreen results)
    Book1.jpg

    So, it seems the 1080Ti still quite a bit faster than the 980Ti with the W3690.

    It seems you are using a 75Hz screen, all onscreen results max at 75FPS. In my case, I am with a 144Hz screen, so all results are limited to 144FPS.

    P.S. Computer spec as per my signature.
     
  18. Theophany macrumors 6502a

    Theophany

    Joined:
    Nov 16, 2008
    Location:
    NW London.
    #43
    Yikes, that's a pretty massive difference! I'll keep an eye out for a 1080 Ti on eBay I think!
     
  19. Squuiid macrumors 65816

    Squuiid

    Joined:
    Oct 31, 2006
    #44
    A reference 1080 FE might be a nice compromise. Cheaper and easy to power in a cMP than a Ti.
     
  20. Theophany macrumors 6502a

    Theophany

    Joined:
    Nov 16, 2008
    Location:
    NW London.
    #45
  21. h9826790 macrumors G4

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #46
    It's quite true indeed. I am sure you know the powering issue. So, the following info is mainly for others to understand why 1080Ti can be a problem.

    I rerun the GFXBench Metal again to capture the power consumption (only run the 7 offscreen tests to maximise the power draw). And the result is like this.
    Screen Shot 2018-02-13 at 23.11.22.jpg
    97.1W is the monitoring limit. The actual power draw should be a bit above 100W (from each mini 6pins) for the 1st and 3rd test. And most other tests were drawing around 90W average.

    This is definitely beyond the 75W official limit. And if I fail to balance the load between the two mini 6pins. The 1080Ti can easily shutdown the Mac by drawing too much power from one of the mini 6pin.

    It seems you already done the Pixlas mod, so powering a highly customised OC 1080Ti should not be an issue.

    But I personally still prefer a reference card. Simply to decrease the chance of having incompatibility issue. It's still unknown at this moment. However, from memory, I have't seen a case complaining about a reference 10xx card is buggy on the cMP. Almost all the "buggy reports" are with non reference card.

    Anyway, a £98 new 1080Ti including shipping, I will buy it without thinking twice (if this kind of deal really exist) :D
     
  22. gjarold macrumors regular

    Joined:
    Nov 14, 2007
    #47
    I have a early 2009 octo mac pro 4,1 ...

    I am short on PCI card slots so I attempted to buy a single card that has *both* a 6gbps SATA m.2 slot *and* some USB3 ports. The card I ordered was this one:

    https://www.amazon.com/gp/product/B06XWXJT4C/

    (SIIG Legacy and Beyond Series USB 3.0 Type-C and Type-A 3 Port PCIe Card With SATA SSD Adapter)

    But this is not what I was looking for - this little card carries the m.2 SSD, but you connect it with a sata cable to the motherboard which means I'm getting the same old, slow, on-board SATA.

    Basically I am looking for a pcie SSD that *also* has a few usb3 ports on it.

    Basically I want a predator hyperX m.2 pcie SSD card that *also* has a few usb3 ports on it.

    Does this combination exist ?

    Thank you.
     
  23. Theophany macrumors 6502a

    Theophany

    Joined:
    Nov 16, 2008
    Location:
    NW London.
    #48
    I bought one and had my money refunded within an hour - it was (quite obviously) a scam on a hacked account. Worth a try just in case!! :p
     
  24. justin1911, Feb 19, 2018
    Last edited: Feb 19, 2018

    justin1911 macrumors newbie

    Joined:
    Dec 4, 2017
    #49
    I disagree. I suspect that if I delid the X5690s, then the RAM issue will go away. Or if I just tighten the heatsink just the right amount. I will have a closer to definitive answer once I finish de-lidding the X5690s.

    Which delid method(s) have you tried and had success with?
     
  25. Squuiid macrumors 65816

    Squuiid

    Joined:
    Oct 31, 2006
    #50
    Vice method without a doubt. Don’t use heat.

    As for the damage, take the CPUs out and see for yourself. If no bent pins then your theory above is correct. But until you check it’s only really a theory.
     
Thread Status:
The first post in this thread is a WikiPost, and can be edited by anyone with the appropriate permissions.

Share This Page