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

'09 Mac Pro (5,1 firmware) wont boot with 2 x X5680

S

steenamaroo

macrumors member
Original poster
May 11, 2015
34
1
Hello all,
I'm hoping someone can offer some advice.
I've had an 09 Mac Pro with B03 firmware and 2 delidded x5650s for some time, working well.

I recently decided to upgrade to a pair of x5680, but can't get them to boot together.
I did delid the new cpus, carefully, and have no reason to believe I damaged either.

At present I can boot my pair of X5650 without issue, or either of the new X5680s in slot A without issue,
so I'm ruling out totally defective chips and damaged logic board pins,
but I can not boot the new cpus as a pair. No chime or anything thereafter.

Fans spin high unless I reset SMC in which case the system just purrs quietly, doing nothing.

I have picked up a third X5680, and did not delid, for better troubleshooting.
It also boots fine on its own but refuses to boot as one of a pair with either of the other X5680s.

This leaves me wondering about my power supply, but I was hoping someone here might know better. ;)

My memory is 3 x 8gb 1333mhz, in slots 1,2,3.

Many thanks in advance,
Steen.
 
Last edited:
steenamaroo said:
Hello all,
I'm hoping someone can offer some advice.
I've had an 09 Mac Pro with B03 firmware and 2 delidded x5650s for some time, working well.

I recently decided to upgrade to a pair of x5680, but can't get them to boot together.
I did delid the new cpus, carefully, and have no reason to believe I damaged either.

At present I can boot my pair of X5650 without issue, or either of the new X5680s in slot A without issue,
so I'm ruling out totally defective chips and damaged logic board pins,
but I can not boot the new cpus as a pair. No chime or anything thereafter.

Fans spin high unless I reset SMC in which case the system just purrs quietly, doing nothing.

I have picked up a third X5680, and did not delid, for better troubleshooting.
It also boots fine on its own but refuses to boot as one of a pair with either of the other X5680s.

This leaves me wondering about my power supply, but I was hoping someone here might know better. ;)

My memory is 3 x 8gb 1333mhz, in slots 1,2,3.

Many thanks in advance,
Steen.
Click to expand...
MP4,1 firmware don’t support Westmere processors, you have to upgrade your firmware to MP5,1. What is your full firmware version?
 
Hi,
Thanks for the reply but the x5650s I've been running for years are Westmere too.
The firmware is 5,1 - MP51.007F.B03
 
steenamaroo said:
Hi,
Thanks for the reply but the x5650s I've been running for years are Westmere too.
The firmware is 5,1 - MP51.007F.B03
Click to expand...
Ok, so your problem is not firmware. MP4,1 dual tray is very troublesome to upgrade, unfortunately.

Some Chinese X5680 processors only work with uniprocessor boards, they are defective. Some sellers tell this, some not.
 
I know the heatsink tension can be finicky but I feel I've done the base-level troubleshooting.

I was hoping someone might know something that relates specifically to x5680 (A, B or C) being fine solo in socket A, but not booting with any pair combination.
As I say, the pair of x5650s still work fine, ruling out socket damage/dirt etc.
[doublepost=1550163484][/doublepost]
tsialex said:
Some Chinese X5680 processors only work with uniprocessor boards, they are defective. Some sellers tell this, some not.
Click to expand...

Damn, really?
Seems that's what I've got then. I hadn't heard of this anywhere else.
The third bought x5680 has not been delidded or modified at all, so I'll speak to the seller and see if he knows anything about that.
Thank you.
 
steenamaroo said:
Seems that's what I've got then. I hadn't heard of this anywhere else.
The third bought x5680 has not been delidded or modified at all, so I'll speak to the seller and see if he knows anything about that.
Thank you.
Click to expand...
Yes, sometime ago AliExpress sellers had loads of these defective ones. Now seems they are selling fake X5690.
 
Last edited:
I bought two of my CPUs from one European seller, and the third from a UK seller.
Are these defective chips widespread enough for me to have ended up with three of them, do you think?

I read a lot of threads about MP 12-core cpu upgrades and compatibility; I'm surprised I didn't see anyone talking about this.
 
steenamaroo said:
I bought two of my CPUs from one European seller, and the third from a UK seller.
Are these defective chips widespread enough for me to have ended up with three of them, do you think?

I read a lot of threads about MP 12-core cpu upgrades and compatibility; I'm surprised I didn't see anyone talking about this.
Click to expand...
I don't know if it's widespread, probably not, but for sure there are defective X5680 and fake X5690 circulating. Since 2009 dual tray is so finicky to upgrade, most people probably thinks that the delid process damaged the Xeons. I always test them on a 2010 tray first.
 
  • Like
Reactions: JedNZ and skizzo
tsialex said:
I don't know if it's widespread, probably not, but for sure there are defective X5680 and fake X5690 circulating. Since 2009 dual tray is so finicky to upgrade, most people probably thinks that the delid process damaged the Xeons. I always test them on a 2010 tray first.
Click to expand...

I have only done half a dozen delids but every time I checked that the CPU worked in a single processor 2009 tray before delid and install into a dual processor 2009 tray. Otherwise you have zero idea if the darn things worked prior to your modification. I get that not everyone would have access to that if they only have a 2009 dual processor system but it gives you 100% proof and piece of mind having the chance to test them prior to delidding.
 
  • Like
Reactions: tsialex
What's the advantage, there, testing on a single CPU board?

I could understand the logic of testing a pair, as a pair, before delidding, given what I've just learned.
The annoying thing is I actually did that with very light tension on the heatsinks, just for a quick DOA test.

Foolishly I immediately shutdown once I got to the desktop, assuming all was well, but now realise it's entirely possible for one CPU to not be recognised at all and for the machine to boot on a single, particularly when heatsink tension is very light.
I'm now assuming that my DOA test was really just a single CPU boot... :(

Really not sure what the best next-step is.
Thinking about it, if we assume that the first bought (and delidded pair) are somehow defective, I have no way of knowing if the UK sourced 3rd one is fine or not (as one of a pair).

I can't confidently sell that third one, or request a refund, and I'm not sure how much I want to buy a 4th CPU.

Maybe I can find someone with a working dual x5680 setup who'd be prepared to test my chip(s)...
 
Well I am going on the assumption you have ONLY a 2009 dual processor system. Which as you know uses delidded CPUs. So this means you are destructively modifying this component before you check to see if it functions as it should. If you have a 2009 single processor tray, they require the CPU to have the IHS so you could just drop in the CPU with the IHS still on to verify the CPU works before you remove the IHS. This would rule out any funny business such as getting defective, fake, damaged in transit, etc, type of CPUs. It's just another methodical way of troubleshooting
 
skizzo said:
Well I am going on the assumption you have ONLY a 2009 dual processor system. Which as you know uses delidded CPUs. So this means you are destructively modifying this component before you check to see if it functions as it should. If you have a 2009 single processor tray, they require the CPU to have the IHS so you could just drop in the CPU with the IHS still on to verify the CPU works before you remove the IHS. This would rule out any funny business such as getting defective, fake, damaged in transit, etc, type of CPUs. It's just another methodical way of troubleshooting
Click to expand...
But he said he tested them individually in slot A and they work properly with a single CPU installed on the dual tray. My question would be are they matched pair? How that affects dual CPU performance (if not)
 
and as far as I know, the "matched pair" thing is a myth (if that is what you're talking about in regards to buying pairs). it's not like you need serial #1000 and #1001 because those two were only designed to work together to have dual CPUs work. they simply need to be the same model. As long as the text on the IHS indicates they are both X5680s you are good. To fake that, someone would need to remove the IHS and then resolder on a wrong one and I am not sure how to check that (physically looking at the CPU) other than seeing the CPU info under the System Report page
 
Oh, I see. That makes sense.
I didn't realise the single-cpu 09 used CPUs with lids.
I'm comfortable enough testing a with-lid CPU in my dual tray...carefully!
As long as you're aware of the difference and very careful, it's easy enough to fire one up for testing.

That's how I know that the first paid (delidded) and the third (still lidded) all work perfectly as single CPUs in slot A.
[doublepost=1550166823][/doublepost]I don't know anything about 'matched' pairs but they are the same code number and stepping.
Each chip boots as a solo CPU in slot A, looks good temperature wise, and geekbenches at approx 3k single and 14k multi.

In short all 3 x5680 chips appear perfect, but no combination will boot as a pair.
 
personally I think testing them with their IHS still on in a 2009 dual tray is asking for trouble. it's a risk I was unwilling to take. also the OP is not sure if booth CPUs were recognized during that particular boot. one question I can ask is if the CPU B (boost B) fan was running like nuts? If it only recognized a single CPU the other fan would ramp up as a sign of it not being seen
 
skizzo said:
personally I think testing them with their IHS still on in a 2009 dual tray is asking for trouble. it's a risk I was unwilling to take.
Click to expand...

Can't fault that. ^

skizzo said:
also the OP is not sure if booth CPUs were recognized during that particular boot. one question I can ask is if the CPU B (boost B) fan was running like nuts? If it only recognized a single CPU the other fan would ramp up as a sign of it not being seen
Click to expand...

Honestly I've done so much switching/tightening/loosening, I've lost track.
I can't confidently say if the fans ramped up or not during that initial pre-delid test boot.
 
I stand corrected, I mixed up the initial testing he tried. Seems like there have been a few threads about this recently. That being two CPUs do not want to boot together, but will both boot OK if installed by themselves on socket A. This makes me scratch my head.....I am not sure how to proceed with more troubleshooting
 
skizzo said:
and as far as I know, the "matched pair" thing is a myth (if that is what you're talking about in regards to buying pairs). it's not like you need serial #1000 and #1001 because those two were only designed to work together to have dual CPUs work. they simply need to be the same model. As long as the text on the IHS indicates they are both X5680s you are good. To fake that, someone would need to remove the IHS and then resolder on a wrong one and I am not sure how to check that (physically looking at the CPU) other than seeing the CPU info under the System Report page
Click to expand...

Well here @flowrider says:
"As long as the two processors are the same model of Intel processor and are production units, not samples, all should be well"
Also they should be the same stepping according to Intel
So how do we find whether it is a production unit and not a sample which is known to be a problem?
 
Last edited:
steenamaroo said:
They're all SLBV5. Thank you for all the info and help so far!
Click to expand...
One thing of note, you should upgrade your firmware, microcode had several updates since MP51.007F.B03 - there are 8 firmware updates after your current firmware.

MP51.0084.B00
MP51.0085.B00
MP51.0087.B00
MP51.0089.B00
138.0.0.0.0
139.0.0.0.0
140.0.0.0.0
141.0.0.0.0
 
Oh, really?!
I'll have to look into this; Thank you.
I used "Mac Pro 2009-2010 Firmware tool" V1.0. I guess that was a good few years ago.
 
Damn. I don't have an apple efi GPU.
Thank you for the information and links, though. I'm reading through now.

Should I have my hopes up, here, or are there people happily running dual x5680s on MP51.007F.B03 ?
 
steenamaroo said:
Damn. I don't have an apple efi GPU.
Thank you for the information and links, though. I'm reading through now.

Should I have my hopes up, here, or are there people happily running dual x5680s on MP51.007F.B03 ?
Click to expand...
I don't recall anyone. If you can't borrow an original GPU, I can reconstruct your BootROM, but try to get one first.
 
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