Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Mac Pro 4.1 flashed to 5.1 issues... RAM ?

H

HackMacPro

macrumors newbie
Original poster
May 30, 2018
7
0
Hello all

I have a dual CPU 2009 MP flashed to 5.1 with these 2 main changes

-CPUs changed to 2 x Xeon W5690-6 core at 3.4Ghz - total 12 core
-GPU changed to GTX 980 TI - 6 gb RAM

Did the refurbish myself and all went well for some time until total system crash and collapse, when I gave up and sent the machine to a specialist.

When it came back repaired, their feed-back was over heating issues and OS - SSD disk failure. I installed OSX El Capitan on a brand new SSD drive and made further add-ons. Made sure also to feed the GTX GPU from both MP AGP connectors and from a spare SATA -using both the 6 and the 8 pin plugs.

It worked fine for some time, then it started -hard crashing frequently, and I detected in fact that the RAM was failing. This is the total 40Gb RAM config occupying all 8 slots in the optimal way according to Apple

- 4 x 2Gb original Apple - 2009
- 4 x 8Gb added OWC - 2010 / 2011 ?

Swapped failing chips and moved them around and the system seems more stable, using 3 out of the mentioned OWC chips, in banks 1 to 3 / CPU A.

The funny thing is that the system shows them running at 1333 MHz, when they're supposedly made for 1033... If that read is true and the CPUs are running these chips 'over-clocked', wouldn't that be unstable, risking to cook them ?

I'd like just the system to work as it used to do and, hoping that my problems are only about old RAM chips, ask you guys out there for advice around

What -reliable and affordable new RAM chips are available out there ?

Will faster than 1333 Mhz chips still work at this lower speed, all the other specs. unchanged ?

Are Apple 's thermal specs an issue with chips that don't comply ?

Thanks in advance

-Update:

System crashed again... aborted boot-ups.. system freezes.. etc
Brief red light near CPU A memory slots, and another chip down.. I think they're being fried at 1333Mhz
 
Last edited:
I'm seeing many vendors selling 'Mac Pro RAM' and checking the chip's specs, many are labelled as -L *low voltage I understand, 1.35v instead of the normal 1,5v. There are huge price variations between brands and specs... I also suspect some clever merchants are refurbishing chips under the MP specs. that actually don't match them. If it's possible to run the chips at a higher voltage than the specified one, I wonder if it's sensible at all

I need my MP back and, I'm aiming to a 3x16 Gb / 1333 SDRAM / ... set up sticking to the 1.5v

I think I'm buying new. Brands that seem reliable and affordable enough: Samsung and Kingston

Any experiences and / or advice ?
 
As mentioned in the thread linked above:

I am currently using eight (8) 16GB PC3-12800R (PC3L-12800R) DDR3-1600 (1600MHz) ECC 1.35V RDIMM, Dual Rank x4 Module (2Rx4), CL11, 240pin modules that operate at 1333MHz in dual 3.46 MacPro5,1. Samsung M393B2G70DB0-YK0, which were Cisco server pulls. Cisco Part Number 15-14595-01.

With previous single processor on this machine was using three (3) Kingston Technology ValueRAM 16 GB 1333MHz DDR3 PC3-10666 ECC Reg CL9 DIMM (RDIMM) DR x4 (2Rx4) 240pin Intel Certified Server Memory (KVR13R9D4/16I). From specs: standard 2G X 72 ECC 1333MHz 240-Pin Registered DIMM (DDR3, 1.5V, CL9, FBGA, Gold).
 
  • Like
Reactions: handheldgames and HackMacPro
bsbeamer said:
As mentioned in the thread linked above:
Click to expand...

Sorry I went thru that post and others but obviously missed this info, which is good help for reference...

I understand -according to my research that the chips will run at L 1.35v if they are -and all chips are that spec. ,
so the normal or low voltage won't matter if the plan is -optimally , plug all the same brand-make chips

Thanks !
 
HackMacPro said:
I'm seeing many vendors selling 'Mac Pro RAM' and checking the chip's specs, many are labelled as -L *low voltage I understand, 1.35v instead of the normal 1,5v. There are huge price variations between brands and specs... I also suspect some clever merchants are refurbishing chips under the MP specs. that actually don't match them. If it's possible to run the chips at a higher voltage than the specified one, I wonder if it's sensible at all

I need my MP back and, I'm aiming to a 3x16 Gb / 1333 SDRAM / ... set up sticking to the 1.5v

I think I'm buying new. Brands that seem reliable and affordable enough: Samsung and Kingston

Any experiences and / or advice ?
Click to expand...

The 1.35V DIMM has backward compatibility for 1.5V. No need to worry about this parameter on cMP RAM selection. And they will only run at 1.5V on the cMP as well.

Anyway, the thread for cMP 4,1 / 5,1 memory FAQ is here.

https://forums.macrumors.com/thread...memory-upgrade-compatibility-and-faq.2099103/
 
  • Like
Reactions: bsbeamer
bsbeamer said:
....

I am currently using eight (8) 16GB PC3-12800R (PC3L-12800R) DDR3-1600 (1600MHz) ECC 1.35V RDIMM, Dual Rank x4 Module (2Rx4), CL11, 240pin modules that operate at 1333MHz in dual 3.46 MacPro5,1. Samsung M393B2G70DB0-YK0, which were Cisco server pulls....
Click to expand...

I bought 2 of those exact brand and make chips... No chime sound, no boot up. Same with all of the other older chips. I think it's a faulty CPU. I'm going to re-install them 1 by 1
 
HackMacPro said:
I bought 2 of those exact brand and make chips... No chime sound, no boot up. Same with all of the other older chips. I think it's a faulty CPU. I'm going to re-install them 1 by 1
Click to expand...

OWC ram has a lifetime warranty for a good reason. It fails.

2Rx4 RDIMMS are not compatible with your original Apple RAM and OWC RAM. Mac Pro's can only address one type of RAM at a time.

Used 2Rx4 RDIMMS can be found on Ebay for $40/16gb, when buying 2 at a time. RDIMMS also run at 1333mhz whether you are running 3 or 4 banks of RAM per cpu.

You should also consider Macs Fan Control to better manage cooling the 5690's and your RAM. High temps over time is known to effect the lifespan of RAM.
 
Last edited:
Trying to answer everything:

-I think yes, the GTX980 TI it's a EFI GPU version so you get booting up progress screen

-I didn't expect the memory types to be compatible. Rather trying to fix the machine bit by bit, the idea was to have
brand new reliable RAM and discard that area as source of issues. 2 x 16Gb good enough to start

-I swapped the X5690s by the original quad core 2.26Ghz CPUs. Booted up with CPU-A only, didn't do it with both CPUs at the first attempt but it did at the second one. (Too much thermal paste, bad screewing ?). Did boot up with 1 x 16 Gb mentioned RAM. Also with 1 x 8Gb OWC chip

-So now I think the issue is the X5690 themselves. CPU-B had some stains in the pins. I cleaned it but I think I've totally
destroyed too. Cause I used thermal paste remover. Dunno, I think both CPUs had issues and I've made them worse.
You guys know if that liquid can destroy the CPU contacts ?

-Still I have to test the machine fully and hope that's it. And maybe in the near future upgrade CPUs -not second hand this time

Thanks to all for your help
 
-Update... The machine, running on the original CPUs and the mentioned RAM

( 16GB PC3-12800R (PC3L-12800R) DDR3-1600 (1600MHz) ECC 1.35V RDIMM, Dual Rank x4 Module (2Rx4), CL11, 240pin modules that operate at 1333MHz in dual 3.46 MacPro5,1. Samsung M393B2G70DB0-YK0... )

Doesn´t work - fatal crashes every 5 min. - (when managed to boot it up at all )

I presume here that these original (mac Pro 4,1 quad core 2,26 Ghz ) CPUs are ok with these higher speced. RAM chips ??

It´s a desperate situation, after 9 months of an unreliable cMP. If it´s not about the CPUs nor the RAM (too many chips failing concurrently), I can only think:

-It´s the GPU - Nvidia GTX 980 TI . (It´s new )
-It´s the power set-up on it, coming onto both the 6 and the 8 pin connecters from both the 2x6 pin internal AGP and from a spare SATA
-It´s a software issue with OSX 10.11.6, its latest security patches, and the drivers from various hardware parts
-A combination of too many mods and a bad smurf messing up my once solid rock machine


In the need of a working machine, I´m now considering

1-buying a new -second hand MP box
2-building a brand new machine and run linux / hackintosh
3-sell it for scraps and start a new career

What would you do ?
 
HackMacPro said:
-Update... The machine, running on the original CPUs and the mentioned RAM

( 16GB PC3-12800R (PC3L-12800R) DDR3-1600 (1600MHz) ECC 1.35V RDIMM, Dual Rank x4 Module (2Rx4), CL11, 240pin modules that operate at 1333MHz in dual 3.46 MacPro5,1. Samsung M393B2G70DB0-YK0... )

Doesn´t work - fatal crashes every 5 min. - (when managed to boot it up at all )

I presume here that these original (mac Pro 4,1 quad core 2,26 Ghz ) CPUs are ok with these higher speced. RAM chips ??

It´s a desperate situation, after 9 months of an unreliable cMP. If it´s not about the CPUs nor the RAM (too many chips failing concurrently), I can only think:

-It´s the GPU - Nvidia GTX 980 TI . (It´s new )
-It´s the power set-up on it, coming onto both the 6 and the 8 pin connecters from both the 2x6 pin internal AGP and from a spare SATA
-It´s a software issue with OSX 10.11.6, its latest security patches, and the drivers from various hardware parts
-A combination of too many mods and a bad smurf messing up my once solid rock machine


In the need of a working machine, I´m now considering

1-buying a new -second hand MP box
2-building a brand new machine and run linux / hackintosh
3-sell it for scraps and start a new career

What would you do ?
Click to expand...

To test the memory - pull the OWC memory and just run on the Apple memory? Could also try buying a small amount of 1033 memory and run just on that to test stability. I have had OWC memory fail but the 24GB I am running now is all from OWC and has been OK since 2015.

If you suspect a CPU is bad, I think you can pull it and the heat sink entirely and just run the machine as a single-CPU computer. The goal is to approach this scientifically, making one change at a time and evaluating the stability afterward in order to rule out various components as the cause.
 
Ive had OWC memory fail, overheated and failed memory check, when I removed the heat spreaders it was cheap old micron stuff. Ditched the owc rubbish and got me some Hynix cl9 server ram, hasn’t missed a beat and gets no where near as hot as that owc stuff did. What I found was soon as the owc memory cooled down a bit it would work again fine for desktop, start useing it hard and ping fail and big crash. Best thing I did was getting rid of that owc junk.
 
  • You've eliminated RAM as the possibility by using all new RAM.
  • You've eliminated new CPUs as the possibility by swapping to your originals.
  • You can rule out the GPU by swapping to your older one.
  • You can rule out OS and software by installing MacOS clean onto freshly erased spare drive.
4,1 dual CPU upgrades are the most finicky, and people have damaged their boards while doing so. Even if they haven't damaged their board, there can be extremely inconsistent reliability based on how tight or not a single heat sink screw is. To me this is the leading possibility.
 
ActionableMango said:
4,1 dual CPU upgrades are the most finicky, and people have damaged their boards while doing so. Even if they haven't damaged their board, there can be extremely inconsistent reliability based on how tight or not a single heat sink screw is. To me this is the leading possibility.
Click to expand...
Agreed. This is why I asked what method they’d used to upgrade the CPUs. Delidding or washers?
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back