Resolved Red Diagnostic LED DSI510 and Mac Pro not starting - broken Northbridge heatsink rivet

[Synchro3]

Writing this thread on my Mac Mini because my 8 year old Mac Pro 4,1 (single CPU) is now out of order.

My Mac Pro shows red LED DSI510 and is not starting anymore. No chime, but fans running, GPU and hard disks are powered.

According to this manual (picture on page 32) https://www.manualslib.com/manual/889501/Apple-Mac-Pro.html?page=32 the LED DSI510 is for CPU A (W3690 installed).



Update:

Uninstalled the CPU cooler and surprise, there is one broken Northbridge Heatsink rivet:


Thanks to Mac Pro forum I'm prepared, got already heatsink rivets... Especially useful is this thread:https://forums.macrumors.com/thread...at-sink-on-2009-mac-pro-2x2-26.1637891/page-3



Update:

Ok, installed new heatsink rivets, cleaned everything incl. Northbridge chip and CPU, reapplied thermal paste on both, and Mac Pro is running again without the red LED.

I cannot emphasise enough that Mac Pro 4,1/5,1 owners should be prepared for this event in advance; Northbridge Heatsink rivets will break with a high degree of probability, it is only a question of time! In my case it happened sudden without warning (I have iStat menus installed), after 8 years with a Mac Pro 4,1 that never encountered any shocks etc.

By the way: This time I reapplied Cooler Master MasterGel Maker, it's much better than the Arctic Silver 5 i applied before: CPU now 40°C instead of 47°C (idle), and Northbridge (I/O Hub Tdiode) now 59°C instead of 75°C.

 

[Surrat]

This should be stickied to the MacPro forum because its very important.

Thanks so much for finding those clips, I just ordered 20 of them for my MP towers.

 

[thornslack]

Please please please let me have sold my cMP before I run into this xD

 

[fendersrule]

Why? Seems to be a piss-easy fix. The hardest part I bet is finding those fasteners.

 

[mattspace]

anyone happen to know if you can diagnose a broken rivet without unscrewing the CPU cooler? can you give the rivet a tug from below on the CPU board?

 

[William Payne]

I am going to be upgrading my cpu sometime soon. I haven't had any rivet issues yet but I am also not fully pushing the machine yet. Maybe I should do this while I have the cooler off just to be done with it.

 

[thornslack]

Why? Seems to be a piss-easy fix. The hardest part I bet is finding those fasteners.

I have a 4,1 with custom processors, so I am not particularly eager to redo the whole upgrade process. It is easier with 5,1s

 

[h9826790]

anyone happen to know if you can diagnose a broken rivet without unscrewing the CPU cooler? can you give the rivet a tug from below on the CPU board?

Rivet doesn't really matter, if the northbridge chip can be cooled properly. Therefore, you don't need to diagnose the rivet, but just monitor the northbridge temperature. If the rivet is borken, you can expect something like constant shows 108C. Or at least well above the normal 50-70C range. If the northbridge is hot, you have to open up to have a look anyway (regardless if the rivet is really broken or not).

 

[William Payne]

How are you guys monitoring the temperatures?

 

[h9826790]

There are few softwares out there. One of the free, easiest and reliable option is the MacsFanControl. iStat or Hardware monitor also work

 

[William Payne]

There are few softwares out there. One of the free, easiest and reliable option is the MacsFanControl. iStat or Hardware monitor also work

Cool thanks

 

[h9826790]

I have a 4,1 with custom processors, so I am not particularly eager to redo the whole upgrade process. It is easier with 5,1s

You mean the delidded CPU?

 

[William Payne]

There are few softwares out there. One of the free, easiest and reliable option is the MacsFanControl. iStat or Hardware monitor also work

Dumb question, I just downloaded macs fan control. Neat little piece of kit. My dumb question is though while my machine seems to be running fine. Is there a chart or something to get a baseline of what the normal temps should be?

 

[h9826790]

Dumb question, I just downloaded macs fan control. Neat little piece of kit. My dumb question is though while my machine seems to be running fine. Is there a chart or something to get a baseline of what the normal temps should be?

No, but in general, 2 parameters to monitor. The software will only monitor the actual temperature. This is the most critical one.

e.g. If a CPU's limit is 100C, and if you see 105C, then something is very wrong. Also, because difference CPU can have different temperature limit. Therefore, it's hard for the programmer to put a reference at there. It may create more confusion then help anything.

The 2nd parameter that the users have to consider is the Delta temperature between the components and the ambitent.

e.g. In all case, all fans at idle

Machine A, ambient 30C, northbridge running at 60C. Then it's a good temperaure.

Machine B, ambient 10C, northbridge also running at 60C. Then it's nowhere near overheat, but obviously the cooling system is not working properly. It should be at the 40C range. May the heat sink is dusty etc.

 

[William Payne]

I barely ever hear my machine going. Most temps are in the 30's 40's my nortbridge is 67 with the heatsink at 57

 

[h9826790]

I barely ever hear my machine going. Most temps are in the 30's 40's my nortbridge is 67 with the heatsink at 57

That's well within the normal range.

If in doubt, you are always welcome to post some screenshots here to let us have a look. There are plenty of experienced people here willing to help.

 

[William Payne]

That's well within the normal range.

That's cool. Gives me an idea of what to look for.

 

[mattspace]

My I/O hub Tdiode is always 77c, never seems to vary (except when it's coming up to temperature from a cold boot). The I/O Hub Heatsink reads 65c at the moment, system ambient of 28c. Not sure what to make of that, really.

 

[h9826790]

In general, what I look for is

CPU below 80
GPU below 90 (general not avail under MacOS)
Northbridge below 85
RAM below 70
HDD, PSU below 50

NOTE: these are not the hard limit, but just my personal preference.

e.g. The hard limit of north bridge is about 105C, however, there is no point to go over 85 because instability may kick in at any time.

Or for CPU, software usually report the diode temperature. However, for a Xeon like X5690, the Intel official limit is ~78C T-case max, which is the temperature at the outside of the IHS, and cannot be measured under normal situation. But in real world practice, diode temperature below 90 is usually good, an I give it another 10C better to make it 80 as my personal limit.

Same thing for PSU or HDD, 50C is just my personal limit. The actual limit usually well above this. To keep them under 50 is mainly for avoiding pre-mature failure.

 

[JedNZ]

This snapshot is representative of my normal status. I/O Hub TDiode can creep to 81˚C under heavy workloads.

No temp readings for PCIE slots 1, 2, 4 and 5.
PCIE Slot 1 = Sapphire Nitro radeon R9 380X
PCIE Slot 2 = Samsung 850 EVO 500GB - for boot and apps
PCIE Slot 3 = NVMe PCIE adapter with Samsung 960 EVO 500GB M2 blade
PCIE Slot 4 = NVIDIA GT120
PCIE Slot 5 = Is this the Mini PCIE slot where my BCM94360CD is connected?

Also note differences:
• CPU A and CPU B - 10 to 14˚C difference in places.
• DIMM temps also vary by 9˚C

Anything look strange?

I have a bag full of rivets on standby if I ever get a red skyrocketing TDiode temps, crashes or the dreaded MoBo red LED light show.

 

[h9826790]

My I/O hub Tdiode is always 77c, never seems to vary (except when it's coming up to temperature from a cold boot). The I/O Hub Heatsink reads 65c at the moment, system ambient of 28c. Not sure what to make of that, really.

I suspect there is some dust at there blocking the airflow on the IOH heatsink. For 28C ambitent, I expect the IOH temperature is about 10C lower than your current temperature.

If possible, get the CPU tray out. And blow the dust out from the back of the heatsink. When look through the heatsink (from the back to the front), you should able to see through it without any dust clogging between the fins.

But if you have no confidence to do that. It's no big deal, the temperature is within the limit.

Also, if you very sure that's reasonably clean. Then may the difference is coming from single / dual processor config. Or you have something like PCIe SSD that can put more stress on the northbridge.

[doublepost=1504785248][/doublepost]This thread is particularly for us to post the cMP's temperature, you may go to have a look. Lots of good reference at there.

https://forums.macrumors.com/threads/post-your-temperature-here.1958354/

Beside, iStat V6 also has new look, which gives you an idea about each parameter how far away from the predetermined max. (again, the exception is the "CPU Relative to ProcHot", which the limit is zero, and the higher the better).

 

[h9826790]

This snapshot is representative of my normal status. I/O Hub TDiode can creep to 81˚C under heavy workloads.

No temp readings for PCIE slots 1, 2, 4 and 5.
PCIE Slot 1 = Sapphire Nitro radeon R9 380X
PCIE Slot 2 = Samsung 850 EVO 500GB - for boot and apps
PCIE Slot 3 = NVMe PCIE adapter with Samsung 960 EVO 500GB M2 blade
PCIE Slot 4 = NVIDIA GT120
PCIE Slot 5 = Is this the Mini PCIE slot where my BCM94360CD is connected?

Also note differences:
• CPU A and CPU B - 10 to 14˚C difference in places.
• DIMM temps also vary by 9˚C

Anything look strange?

I have a bag full of rivets on standby if I ever get a red skyrocketing TDiode temps, crashes or the dreaded MoBo red LED light show.

View attachment 716211

Everything looks pretty normal to me. I suspect your 960 evo cause the northbridge runs a bit warmer, but that's well within normal range.

However, what catching my attention is that you have 2x 850 Evo, and one of them running at 47C. The other one is just 28C (lower then the ambient???). My guess would be your booting 850 Evo is quite busy at this moment, which runs a bit warm. This also help to explain why the IOH is at 75C. And the other 850 Evo is at idle.

 

[Razzerman]

Nice thread. Thanks for the info & pics OP.

I've got a slight issue with one of my 2009 Mac Pros. It's a dual 2.66 quad core. It was literally being thrown out, so I got permission and rescued the poor thing.

Anyhow, the IOH Thermal diode runs too hot. MFC saying 80°C on cold start, and if I don't set the fans to kick in, it rises to 102°C. I've set MFC to kick in based on the IOH temp.

I have recently re-applied silver arctic paste on both processors, but didn't check the rivets. I guess that's my next job.

Cheers,

Razzerman

 

[JedNZ]

Everything looks pretty normal to me. I suspect your 960 evo cause the northbridge runs a bit warmer, but that's well within normal range.

However, what catching my attention is that you have 2x 850 Evo, and one of them running at 47C. The other one is just 28C (lower then the ambient???). My guess would be your booting 850 Evo is quite busy at this moment, which runs a bit warm. This also help to explain why the IOH is at 75C. And the other 850 Evo is at idle.

One EVO 850 is in the Accelsior S PCIe SATA III card, and the other one is in a Direct Connect Bay (SATA II) sled drive. I believe the hotter one is in the PCIE slot because the lower temp one is comparable to my other sled drives.

 

[Synchro3]

Nice thread. Thanks for the info & pics OP.

I've got a slight issue with one of my 2009 Mac Pros. It's a dual 2.66 quad core. It was literally being thrown out, so I got permission and rescued the poor thing.

Anyhow, the IOH Thermal diode runs too hot. MFC saying 80°C on cold start, and if I don't set the fans to kick in, it rises to 102°C. I've set MFC to kick in based on the IOH temp.

I have recently re-applied silver arctic paste on both processors, but didn't check the rivets. I guess that's my next job.

Cheers,

Razzerman

80°C on cold start and 102°C later is absolutely not good. Check the heatsink rivets.