[angelmb]

Gentlemen, I've got an issue here. My Mac Pro locks up, fans revving up to the max enthusiastically like they were to drive off.

I was looking at the array of dig-LEDs inside and noticed CPU B had a red light on FAIL.

If I don't recall wrong, a Dual CPU Mac Pro should work with one CPU (although I do not know which socket should be the one to use), so this CPU B error and computer still not working does not sound too good.

Could it be a dead logic board issue.? Does the Mac Pro logic board have any known weaknesses or common failures.?

I don't thing bad RAM could cause these symptoms. I have had a bad stick of RAM before and it wasn't anything like this.

What about the GPU.? I have not removed the video card (and restarted) to see if the video card was the culprit. Should I try it, is this a safe thing to do.?

A dead GPU would make GPU number 3 to dead on me with this computer. An astonishing achievement for someone that doesn't play games at all.

It is not possible to boot from a Lion DVD -one that actually does boot and install the System just fine–. Computer locks up, fans speeding up afterwards. I understand this means the internal hard disk is OK.

This Mac Pro was upgraded with an ATI Radeon 5770 HD Mac Edition a year ago, therefore I can't even try to boot it up with the retail Snow Leopard 10.6.0 DVD which lacks support for such a GPU. For the same reason I can't use the gray disks that came with the Mac Pro, while this would allow me to run the Apple Hardware Test, it can't be done as they include Tiger. Right.?

Nothing (physical damage) has happened to the computer at all. Honest.

I didn't install nor upgrade anything on it.

No weird, unsupported 3rd party hardware is present. Display is an Apple Cinema 30". Keyboard and mouse are Apple BT ones. They do work and are being seen.

No weird software hacks. I don't like playing with fire at all.

This Mac Pro had had a CPU failure two or three years ago, it was repaired under Apple Care coverage. Unfortunately, such warranty has expired.

Should I try to reset PRAM, SMC… or anything else.?

Guessing this is actually a fried logic board issue, I am not sure/keen on paying 800 € to get it replaced, which leads me to my next question: are there any external adaptors for the SATA internal HDs, making it feasible to use any of the three HD drives this Mac Pro has inside.?

I have some Xeon CPUs from another Mac Pro 1,1 at hand. Would you say to get them replaced might be the safest, straight way to get this issue fixed.?

TIA

[SierraDragon]

An Apple Store will diagnose the issue for free if you can get to one (at least the Emeryville, CA and Flatiron, CO stores have done so for me).

-Allen

[angelmb]

Thank you Allen. Unfortunately, I am on the wrong side of the pond.

[cgc]

Check out my thread about the same problem (quoted the symptoms below). Those LEDs are there to trick us. I swapped out my 8800 with a POS 7300 I had lying around and all's wunderbar. Performance sucks but I was waiting to see what is next GPU-wise.

Originally Posted by cgc 

My beloved 2006 MacPro (w/ 2 x 2.66GHz Xeon 5150s) locked up after a bit or Portal 2 playing. I could see the top 1/10th of the screen, the mouse worked fine, but the bottom 9/10ths were FUBARed and mostly black with some white. Everything was blocky, not looking like noise per se.


I restarted and the screen looked normal but after about 45 seconds the spinning nondeterminate progress thing froze. I waited a bit and after no change I restarted again. Same thing. Did this two more times then zapped PRAM with no change. Took out memory in pairs with no change but then I noticed my CPUA and CPUB (#2 and #3 LED from left on motherboard) were lit red. I looked online and saw a SYS-RST has fixed this with others so did that to no avail. Tried booting from CD and at same point it froze again. Booted to single user interface and ran fsck w/ no errors but when it tried rebooting it froze in the CLI (right after loading Parallels plugins if that means anything).


Finally broke down and took out nVidia 8800GT and replaced it with my wonderful nVidia 7300GT but also removed my #2 and #3 hard drives and sleds. System booted up like a champ.


My question is what do you think is bad? I'm asking this because the video was only bad once. My thoughts are it could be two things:


1. Bad GPU

2. Bad PSU


For option #2 I'm thinking the GPU does something to draw power at 45 secs-ish into boot; the much less powerful nVidia 7300 doesn't cause the problem as it draws less power. How would I check the PSU or GPU to be certain without having an extra Mac or PSU laying around? Thanks.

[OreoCookie]

Originally Posted by angelmb 

Thank you Allen. Unfortunately, I am on the wrong side of the pond.

There are Apple Stores in Europe. If you are in Germany, for instance, there is also Gravis, a long-time reputable Mac retailer. However, if it is the CPU, it's not going to be a cheap repair and perhaps it makes more financial sense to sell the broken Mac Pro on ebay and get a new machine.

[Waragainstsleep]

If you have spare CPUs that are compatible with this Mac, its probably worth getting them swapped in. Some Mac Pros are very fussy when it comes to thermal paste (they use some really nasty toxic stuff on some models, comes plastered with warnings and pre-applied to the heat sinks)

[angelmb]

Thanks guys, you rule.

[ Check out my thread about the same problem (…)

Thanks cgc, I remember having replying on my faulty 8800. I just can't believe the Radeon 5770 has gone belly up already. Albeit it should still be under warranty… me thinks.

Guessing is the GPU, does booting to single mode user bypass the ATI.?

There are Apple Stores in Europe. (…)

While Spain is still part of Europe, Apple Stores are a scarcity here. The nearest Apple Service Provider to my home is not that great, they charge you just for installing OS X… on a new computer !!; so I can only believe they would charge me as well just to get the Mac Pro diagnosed, and then, I would have to believe their word on it.

If you have spare CPUs that are compatible with this Mac (…)

They are from another Mac Pro 1,1 and are faster that mine, so should it be the CPUs I would get a nice speed bump, 2,66 GHz Vs. 2,00 GHz. As for the toxic elements, should the one doing the swap wear a mask or something.?

[Waragainstsleep]

Gloves are usually sufficient. The trick might be getting hold of the same stuff Apple uses and applying it well enough.

If you suspect the GPU for some reason, boot with it removed and check to see if the CPU B error LED still comes on. This supposedly indicates an instruction error. Basically you want to remove or disconnect as many components as possible and check to see if that LED still comes on. Pull all the drives, remove all but minimum RAM (if you have extra, switch RAM for a second test) and remove any other PCI-E cards. Disconnect external peripherals, basically nothing but the power cable.


if the LED still comes on, press the SYS_RST button with the computer running (Do that carefully and use something non conductive to press the button). If it stays on, power off and hit the SMC-RST. Apple also advocates removing the clock battery and replacing it.

Last step is to switch CPU A with CPU B. If the CPU A error LED comes on, you have a bad CPU, if not its probably the logic board. You might as well switch your new ones in if you are going to mess with the CPUs at this point. I think maybe the toxic stuff was for a later model of Mac Pro. This one shouldn't be so bad.

[angelmb]

I have tracked it down to a bad RAM stick. I found that CPU FAIL LEDs no longer lit. The LED for the RAM stick #3 at raiser A did. In fact, if I placed that faulty RAM stick as #1, no RAM at raiser A was seen.

So, with 7 * 1 GB RAM modules installed, I am running the computer with 6 GB of RAM instead. I understand this makes sense since RAM should be installed in pairs on this particular computer.

All seems to be working fine now, hope it holds that way.

[cgc]

Good to hear you found the problem. Curious why you had CPU LED lit with bad RAM and I had CPU LED lit with failed GPU. Could it be the failed items were consuming too much power?

[Waragainstsleep]

Bad RAM is a known cause of that symptom in the service manual. GPU is not mentioned but it does say to remove it as part of your troubleshooting.

[SierraDragon]

Excellent! It is nice when a zero-cost solution presents for a potentially serious problem.

-Allen