Advice for 5.1 upgrade

[pl1984]

I'd just want to know FOR SURE that it was one or the other of the new processors. Of course, you don't conclude this 100% if neither one shows a problem when installed individually, but at least you can be pretty sure if one DOES show a problem. Then you can put your mind at ease that it isn't anything wrong w/ your tray.

OP has twice already re-installed the original CPUs (post #34 and #54) and the system functions as expected. Therefore it's highly unlikely the tray is the issue. From what I have read the only time the system fails to function is with the new CPUs. Therefore it's highly likely one of the two CPUs is bad. Which one? Who cares? Why waste time, thermal paste, and risk damaging the tray just to find a bad one when you're going to have to do an exchange anyway? Just exchange both of them and let the seller do the troubleshooting.

In any case, the amount of time posting here about other thoughts and soliciting add'l opinions after the "install individually" advice had been given could have easily encompassed the 10-15 extra minutes to unscrew and re-screw the heat sinks and try them each X5680 individually.

OP is already waiting on thermal paste so why not do something productive with that time and get two replacement processors on the way instead of waiting for the thermal paste to arrive, troubleshooting the bad processor, and then requesting a replacement for one?

 

[fhturner]

OP has twice already re-installed the original CPUs (post #34 and #54) and the system functions as expected. Therefore it's highly unlikely the tray is the issue. From what I have read the only time the system fails to function is with the new CPUs. Therefore it's highly likely one of the two CPUs is bad. Which one? Who cares? Why waste time, thermal paste, and risk damaging the tray just to find a bad one when you're going to have to do an exchange anyway? Just exchange both of them and let the seller do the troubleshooting.


OP is already waiting on thermal paste so why not do something productive with that time and get two replacement processors on the way instead of waiting for the thermal paste to arrive, troubleshooting the bad processor, and then requesting a replacement for one?

I admit you do not make bad points in the first paragraph. I wouldn't do it that way, but you don't make bad points. As to "Who cares?"...well, I would. I would want to know. He's asking the question, so presumably, at some point, he might have wanted to know also. Regarding wasting time and thermal paste (or waiting on the thermal paste), as I already suggested, he could have done a quick swap to single CPUs in maybe 10-15 min total time, and could have not worried about removing/cleaning/reapplying thermal paste for that testing. I suppose the risk to the tray is valid if one weren't careful or extremely unlucky— I just don't think a couple more careful CPU swaps would be much of a risk.

Another thought that occurs to me is: what if the seller tests the CPUs on his end, doesn't find a problem, and gives Morpheo friction about replacing or refunding? To my mind, if I've conclusively proven that "this CPU right here, #2" is to blame for my problem, I have a little more certainty and leg to stand on. But maybe that's just me. I want to know WHY or WHICH something isn't working, not just THAT it isn't working.

In any case, I hope Morpheo can quickly get this resolved so he can enjoy his upgraded machine.

 

[pl1984]

I admit you do not make bad points in the first paragraph. I wouldn't do it that way, but you don't make bad points. As to "Who cares?"...well, I would. I would want to know. He's asking the question, so presumably, at some point, he might have wanted to know also. Regarding wasting time and thermal paste (or waiting on the thermal paste), as I already suggested, he could have done a quick swap to single CPUs in maybe 10-15 min total time, and could have not worried about removing/cleaning/reapplying thermal paste for that testing. I suppose the risk to the tray is valid if one weren't careful or extremely unlucky— I just don't think a couple more careful CPU swaps would be much of a risk.

You'll note the post of mine you responded to I opened with:

"I'm leaning towards Matty's train of thought."

Neither is wrong which is why I stated "I'm leaning".

If you want to know which CPU is bad that's fine but ultimately he's going to have to go through the exchange process whether he's exchanging one or two CPUs. So my recommendation to him was "Why waste your time? Send them both back and let the seller sort it out." This was because the OP stated he wanted to do the "...most cost-effective and time-saving solution..." My recommendation was with that idea in mind. Your response was to lecture me about proper troubleshooting methodology as if I were clueless.

As for performing the single CPU swap OP has already stated he's out of thermal paste. So his current, at least at the time the recommendation was made, options are to wait for the new thermal paste to arrive (wasting time), troubleshoot the problem (his time and parts), and then exchange a processor. Or he could just reach out to the seller now and get the exchange started so he has the new processors when the thermal paste arrives. I leaned towards the latter as it saves him time. If he still had thermal paste and could immediately perform the single processor testing perhaps my recommendation would have been inline with yours.

Another thought that occurs to me is: what if the seller tests the CPUs on his end, doesn't find a problem, and gives Morpheo friction about replacing or refunding? To my mind, if I've conclusively proven that "this CPU right here, #2" is to blame for my problem, I have a little more certainty and leg to stand on. But maybe that's just me. I want to know WHY or WHICH something isn't working, not just THAT it isn't working.

The situation you describe can arise whether he replaces one or two CPUs. If the seller refuses to exchange the CPUs he can utilize eBay's protection policy. It's extremely effective for the buyer.

 

[F1Mac]

 

[tsialex]

View attachment 790137

So all that trouble was the de-lidded Xeons?

 

[F1Mac]

So all that trouble was the de-lidded Xeons?

No they were not de-lidded. This is a new pair. Took me 20 mn to install them. No problem so far!! Now I'm going to get completely paranoid with my temps, as I did when I upgraded my first Mac Pro (1,1 -> 2,1)
[doublepost=1538169431][/doublepost]GB 4 score with the stock CPUs (2.4): 2050 single, 18055 multi
GB 4 score with the 5680s: 2693 single, 24679 multi




Now...

With the E5645s:

And with the 5680s:

Everything look normal here? Just to make sure!... Computer is almost idle right now.

 

[Matty_TypeR]

No they were not de-lidded. This is a new pair. Took me 20 mn to install them. No problem so far!! Now I'm going to get completely paranoid with my temps, as I did when I upgraded my first Mac Pro (1,1 -> 2,1)
[doublepost=1538169431][/doublepost]GB 4 score with the stock CPUs (2.4): 2050 single, 18055 multi
GB 4 score with the 5680s: 2693 single, 24679 multi




Now...

With the E5645s:

View attachment 790154


And with the 5680s:

View attachment 790152

Everything look normal here? Just to make sure!... Computer is almost idle right now.

So from your original post on a non working CPU, you have a new pair of CPU's that are working and only taken 20 mins to install? might have been better to update the original first post to now working so that others with issues can see your original issues and how getting a new pair of CPU's you were able to get working Mac Pro.

Just goes to show that all the testing in the world isn't going to make a duff CPU work.

Temps all look ok, what is the ambient system temp?

 

[F1Mac]

So from your original post on a non working CPU, you have a new pair of CPU's that are working and only taken 20 mins to install? might have been better to update the original first post to now working so that others with issues can see your original issues and how getting a new pair of CPU's you were able to get working Mac Pro.

First post edited.

Temps all look ok, what is the ambient system temp?

Ambient is 28º C right now.

After 24 hours it almost seems like the 5680s are actually running cooler than the 5645s when idle, although it can be hard tell and I'm not going to look at the temps 24/7 anyway:

I have 2 more questions:
1 - Yesterday while I was doing some test with handbrake, I noticed the ProcHot A was going below 10º (it went to 6 iirc), then when the fans kicked in, it went back to 13-15ºC. I'm not sure I like it when it gets so close to the threshold...should I be concerned?

2 - Also, yesterday there was roughly a 15º difference between proc A and pro B now it's more like 10º. Could this be attributed to the thermal paste settling in?

 

[Matty_TypeR]

First post edited.




Ambient is 28º C right now.

After 24 hours it almost seems like the 5680s are actually running cooler than the 5645s when idle, although it can be hard tell and I'm not going to look at the temps 24/7 anyway:

View attachment 790335

I have 2 more questions:
1 - Yesterday while I was doing some test with handbrake, I noticed the ProcHot A was going below 0º (it went to 6 iirc), then when the fans kicked in, it went back to 13-15ºC. I'm not sure I like it when it gets so close to the threashold...should I be concerned?

2 - Also, yesterday there was roughly a 15º difference between proc A and pro B now it's more like 10º. Could this be attributed to the thermal paste settling in?

CPU A always runs a little bit warmer than CPU B from my experience, and yes sometimes thermal paste can take anything upto 48hrs to settle at a stable temp. depends what paste you used. With PC overclocking we used to flatten the ICH the metal cover over the CPU as sometimes they could be concave or have a high corner.

48c you have CPU A and 35c CPU B so a 13c delta between both CPU's. I would say let it settle in a few days and if you feel you would like to see a better delta, you can always remove CPU A heatsink and re seat it to see if delta improves.

Ambiant temp is fine, and CPU B only 7c higher than ambient, so looks good.

 

[F1Mac]

CPU A always runs a little bit warmer than CPU B from my experience, and yes sometimes thermal paste can take anything upto 48hrs to settle at a stable temp. depends what paste you used.

I went with MX-4 this time.

 

[Matty_TypeR]

I went with MX-4 this time.

MX 4 has a good reputation so I would let it settle a few days and see if you see any difference in the temps. if your not happy then you can reseat the heatsink and see if it improves.

 

[F1Mac]

MX 4 has a good reputation so I would let it settle a few days and see if you see any difference in the temps. if your not happy then you can reseat the heatsink and see if it improves.

Will do. I don't think the temperature difference between A and B is problematic, but I'll continue to monitor them in the coming days.

Thanks for your advices btw.