Friday, September 15th 2017

AMD Ryzen Threadripper MCM De-lidded and De-packaged

PC enthusiast Der8auer, with access to a Ryzen Threadripper processor, took it completely apart for science. It won't be the first time that a Threadripper HEDT processor was de-lidded (its integrated heatspreader removed), revealing that it has four "Zeppelin" 8-core dies, making it practically identical to AMD's 32-core Epyc processors; however, it's the first time that someone completely removed the dies from the package.

Ryzen Threadripper processors are built by completely disabling two out of four "Zeppelin" dies on an Epyc multi-chip module (MCM). Two diagonally opposite dies are disabled. The disabled dies can't be reenabled, at least not on an X399 chipset motherboard, as the Threadripper HEDT platform lacks DRAM, PCIe, and possibly even power wiring for the disabled dies.
Source: Der8auer
Add your own comment

51 Comments on AMD Ryzen Threadripper MCM De-lidded and De-packaged

#1
Daven
I must admit I was not only completely wrong in some past posts but also completely surprised by this. I guess this method of packaging is very cheap for AMD to waste two entire dies (I assume that at least some of the dies aren't working out of the factory so cannot be EPYC processors).
Posted on Reply
#2
Red_Machine
"Sir, why did you destroy a one thousand dollar CPU?"

"FOR SCIENCE!"
Posted on Reply
#3
cadaveca
My name is Dave
Didn't AMD say that they DIDN'T use non-functional dies, and the non-working dies were just spacers without any transistors? Or was this more BS by sites that AMD should have never supported? :p ROFL :kookoo:

I wonder what would happen if you stuck a TR CPU in an EPYC board? Would there possibly be a way to enable those "EPYC Failures"??

:slap:


:laugh:


:roll:
Posted on Reply
#4
R00kie
cadavecaDidn't AMD say that they DIDN'T use non-functional dies, and the non-working dies were just spacers without any transistors? Or was this more BS by sites that AMD should have never supported? :p ROFL :kookoo:

I wonder what would happen if you stuck a TR CPU in an EPYC board? Would there possibly be a way to enable those "EPYC Failures"??

:slap:


:laugh:


:roll:
Sockets are electrically different, so if you actually tried it wouldn't work.
Posted on Reply
#5
cadaveca
My name is Dave
gdallskSockets are electrically different, so if you actually tried it wouldn't work.
Are we sure about that? I'd be willing to try... :P
Posted on Reply
#6
R00kie
cadavecaAre we sure about that? I'd be willing to try... :p
Yeah, AMD said so themselves, the layout of the pins might be the same, but the roles could be switched... but you're welcome to wreck a 1k dollar CPU and god knows how expensive the motherboard is :roll:
Posted on Reply
#7
cadaveca
My name is Dave
gdallskYeah, AMD said so themselves, the layout of the pins might be the same, but the roles could be switched... but you're welcome to wreck a 1k dollar CPU and god knows how expensive the motherboard is :roll:
From what I have seen, chances are damage aren't that high, actually. And while I understand your trust in AMD's statements, I am not so trusting, simply because it is far more economical for them to use the same package and pinout for both platforms, and would have also allowed for them to have the very short R&D cycle that they have lauded to have had. To control people buying the wrong chips, all they need to do is some BIOS magic (which could then be modded by enthusiasts).

If AMD has sent samples, my costs are nothing, so yeah, I would try for sure. I've done it with Intel CPUs without any loses so far...
Posted on Reply
#8
newtekie1
Semi-Retired Folder
Red_Machine"Sir, why did you destroy a one thousand dollar CPU?"

"FOR SCIENCE!"
I'm pretty sure the CPU was already dead. In his original TH de-lidding video he said that the CPU did not survive the de-lidding process. So I'm guessing he used that dead CPU.

And he wasn't intending to kill the CPU when he did the original de-lidding, it was unfortunate that it died.
Posted on Reply
#9
Slizzo
btarunrPC enthusiast Der8auer, with access to a Ryzen Threadripper processor, took it completely apart for science. It won't be the first time that a Threadripper HEDT processor was de-lidded (its integrated heatspreader removed), revealing that it has four "Zeppelin" 8-core dies, making it practically identical to AMD's 32-core Epyc processors; however, it's the first time that someone completely removed the dies from the package.

Ryzen Threadripper processors are built by completely disabling two out of four "Zeppelin" dies on an Epyc multi-chip module (MCM). Two diagonally opposite dies are disabled. The disabled dies can't be reenabled, at least not on an X399 chipset motherboard, as the Threadripper HEDT platform lacks DRAM, PCIe, and possibly even power wiring for the disabled dies.

It was my understanding that in statements from AMD that two of the "dies" were blank silicon to even out pressure on the heat spreader, and two dies were actual Ryzen dies?
Posted on Reply
#10
VSG
Editor, Reviews & News
AMD's phrasing was that these two dies were non-functional, with transistors that did not work. It remains true here as well, but was vague enough to where some had misinterpreted it, including me.
Posted on Reply
#11
cadaveca
My name is Dave
VSGAMD's phrasing was that these two dies were non-functional, with transistors that did not work. It remains true here as well, but was vague enough to where some had misinterpreted it, including me.
That's really no-one's fault, really. They were purposefully vague.

What's more interesting is the implication that has for Ryzen's yields.
Posted on Reply
#12
OSdevr
cadavecaDidn't AMD say that they DIDN'T use non-functional dies, and the non-working dies were just spacers without any transistors? Or was this more BS by sites that AMD should have never supported?
SlizzoIt was my understanding that in statements from AMD that two of the "dies" were blank silicon to even out pressure on the heat spreader, and two dies were actual Ryzen dies?
AFAIK AMD always referred to them as "disabled" but many people (myself included) thought they were just blank. I thought yields were over 50% so why would you throw away good dies when you could just use spacers?
Posted on Reply
#13
TheoneandonlyMrK
OSdevrAFAIK AMD always referred to them as "disabled" but many people (myself included) thought they were just blank. I thought yields were over 50% so why would you throw away good dies when you could just use spacers?
It was a partner and not Amd that said they were packing, if i recall right pc world, AMD did exactly as i thought they would and im sure i poo pood the two space filler dies comments then.
It's simply the most economical and practical solution, build test bin , drag out of bin invent consumer line.
Posted on Reply
#14
GoldenX
Wouldn't it be great if we could return to the Phenom II days and unlock them?
Posted on Reply
#15
OSdevr
theoneandonlymrkIt was a partner and not Amd that said they were packing, if i recall right pc world, AMD did exactly as i thought they would and im sure i poo pood the two space filler dies comments then.
It's simply the most economical and practical solution, build test bin , drag out of bin invent consumer line.
Unpackaged dies can be tested these days. If they didn't HBM would have a miserable yield before ever being attached to a GPU. Remember that supposedly only the top 5% of Ryzen dies go into TR.
Posted on Reply
#17
TheoneandonlyMrK
OSdevrUnpackaged dies can be tested these days. If they didn't HBM would have a miserable yield before ever being attached to a GPU. Remember that supposedly only the top 5% of Ryzen dies go into TR.
I know ,They could always be tested to a degree , in reality their will be several check stages during production and packaging each presenting a new binning opportunity.
My last line was mild satire, or a try at it.
Posted on Reply
#18
R0H1T
SlizzoIt was my understanding that in statements from AMD that two of the "dies" were blank silicon to even out pressure on the heat spreader, and two dies were actual Ryzen dies?
That's probably because not all TR CPU will have four dies, reminds me of the time when Phenoms could be unlocked, so most likely YMWV.
GoldenXWouldn't it be great if we could return to the Phenom II days and unlock them?
Who knows AMD might do an Intel here & unlock these, for an undisclosed amount.
Posted on Reply
#19
OSdevr
theoneandonlymrkI know ,They could always be tested to a degree , in reality their will be several check stages during production and packaging each presenting a new binning opportunity.
My last line was mild satire, or a try at it.
Ah, I get it now.
Posted on Reply
#20
trparky
Aw... what did the poor processor do to deserve that? :cry:
Posted on Reply
#21
nemesis.ie
I'm sure CaseKing sponsored him rather than it coming out of his own pocket, trade price at least.
Posted on Reply
#22
vega22
nemesis.ieI'm sure CaseKing sponsored him rather than it coming out of his own pocket, trade price at least.
more likely es chips from asus i would bet.
Posted on Reply
#23
nemesis.ie
Well he did specifically say he "went and bought it" and it was not an ES ...
Posted on Reply
#24
MagnyCours
Gold Supporter
newtekie1I'm pretty sure the CPU was already dead. In his original TH de-lidding video he said that the CPU did not survive the de-lidding process. So I'm guessing he used that dead CPU.

And he wasn't intending to kill the CPU when he did the original de-lidding, it was unfortunate that it died.
The one he used in this particular video is a new, retail purchased CPU.
Posted on Reply
#25
Dave65
newtekie1I'm pretty sure the CPU was already dead. In his original TH de-lidding video he said that the CPU did not survive the de-lidding process. So I'm guessing he used that dead CPU.

And he wasn't intending to kill the CPU when he did the original de-lidding, it was unfortunate that it died.
No, if you watch his video he went out and purchased this cpu with his own money, it was NOT the engineering sample from before..
Posted on Reply
Add your own comment
Apr 27th, 2024 06:33 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts