No Dummies please

Now all we need is someone to tell us which traces need to be pencil modded!
 
Where does it make sense to off load 2 functional dies that would be silly, where AMD can make money on selling them.
 
End point, he thinks there could be more iterations of Threadripper up to 32 core. Some from AMD have already stated that there will be more TR products to come....
 
Not a surprise.

But why did we have to hear this BS first.

The other important thing to know about Threadripper is that the other two die apparently aren’t actually chips at all. They’re apparently structural inserts required for support, but do not represent “bad” Epyc cores or any kind of core recycling
Click to expand...

It includes failed EPYC chips, plain simple.
 
thesmokingman said:
End point, he thinks there could be more iterations of Threadripper up to 32 core. Some from AMD have already stated that there will be more TR products to come....
Click to expand...

The thing that I read was that the X399 motherboards can't handle it
 
Pieter3dnow said:
The thing that I read was that the X399 motherboards can't handle it
Click to expand...

Can't handle a 32 core Epyc sure, but if they make a 32 core TR, it would not necessarily be an Epyc chip right? TR4/x399 boards are not electrically compatible with Epyc due to the difference in pcie lanes and memory channels. From the outside I don't see why they couldn't make it work, a 32 core on a 64 lane/4 channel board.
 
Mad kudos to der8auer for tossing 1k out the window to prove this. That's a month's mortgage for me, lol.
 
Sweet - if it's cost effective, why not? And if AMD releases 32 core Threadripper to give Intel even more reasons to get off of their collective asses and really push the envelope, then all the better.
 
arestavo said:
Sweet - if it's cost effective, why not? And if AMD releases 32 core Threadripper to give Intel even more reasons to get off of their collective asses and really push the envelope, then all the better.
Click to expand...

Oh I'm sure they'll be pushing an envelope... an envelope full of money (er, discounts) across the table to OEM's, like last time.
 
Wow! Then AMD lied us when said there are only two dies on ThreadRipper.

This proves the true origin of ThreadRripper.
 
Last edited:
mnewxcv said:
Now all we need is someone to tell us which traces need to be pencil modded!
Click to expand...

It is simpler to transform a 1900X into a 1950X than transform a 16C 1950X into a hypothetical 32C ThreadRipper.

And the possibilities for the first transformation are zero.
 
juanrga said:
It is simpler to transform a 1900X into a 1950X than transform a 16C 1950X into a hypothetical 32C ThreadRipper.

And the possibilities for the first transformation are zero.
Click to expand...

broken-pencil-clipart-free-clip-art-images.jpg
 
workshop35 said:
So why is this a some kind of controversy?
Click to expand...

der8hauer originally found four dies on an engineering sample. AMD negated the existence of four dies on final ThreadRipper. AMD said to the media and also to der8hauer that ThreadRipper has only two dies and the other two are blanks. Der8hauer didn't believe the official explanation, purchased a retail chip and delidded it. He found that there are four dies in ThreadRipper.
 
juanrga said:
der8hauer originally found four dies on an engineering sample. AMD negated the existence of four dies on final ThreadRipper. AMD said to the media and also to der8hauer that ThreadRipper has only two dies and the other two are blanks. Der8hauer didn't believe the official explanation, purchased a retail chip and delidded it. He found that there are four dies in ThreadRipper.
Click to expand...
I read the articles and didn't understand why its a big deal. The AMD engineer was saying that they aren't functional dies and from what I was reading they aren't even fully connected. Seems like a semantics issue getting blown way out of proportion, unless I'm missing more to the story.
 
workshop35 said:
I read the articles and didn't understand why its a big deal. The AMD engineer was saying that they aren't functional dies and from what I was reading they aren't even fully connected. Seems like a semantics issue getting blown way out of proportion, unless I'm missing more to the story.
Click to expand...


It is weird at the same time as well. There saying that there yield is awesome so how many of these failed dies are there before they have to resort to replace the failed dies with a placeholder.
 
workshop35 said:
I read the articles and didn't understand why its a big deal. The AMD engineer was saying that they aren't functional dies and from what I was reading they aren't even fully connected. Seems like a semantics issue getting blown way out of proportion, unless I'm missing more to the story.
Click to expand...

This topic is completely irrelevant for final users. For final users it doesn't matter if AMD uses four dies or two dies plus two blanks. It is relevant for tech enthusiasts and economic analysts. The first because want to know the fine details of the tech. The second because want to infer more about the economies behind EPYC and ThreadRipper.

If it was something completely irrelevant, then AMD wouldn't lie us in the first place, and guys as der8hauer wouldn't waste time and money on delidding and dismantling until the dead a ThreadRipper CPU.
 
Still wonder why AMD felt the need to blatantly mislead or lie about the make up of TR. Especially considering their terrible track record.
 
juanrga said:
This topic is completely irrelevant for final users. For final users it doesn't matter if AMD uses four dies or two dies plus two blanks. It is relevant for tech enthusiasts and economic analysts. The first because want to know the fine details of the tech. The second because want to infer more about the economies behind EPYC and ThreadRipper.

If it was something completely irrelevant, then AMD wouldn't lie us in the first place, and guys as der8hauer wouldn't waste time and money on delidding and dismantling until the dead a ThreadRipper CPU.
Click to expand...
Because he got the other 2 dies working ? All he did was something which was pretty futile to begin with. He checked if there were dies then he realized that the dies that were in there he could not check anyway. So where is this lying part come from ? Have you already seen someone with 4 dies enabled or swapped the 2 functioning dies to the other 2 ?

jmilcher said:
Still wonder why AMD felt the need to blatantly mislead or lie about the make up of TR. Especially considering their terrible track record.
Click to expand...
You mean the bad track record Intel keep saying they are having, I would put as much faith into that as this https://newsroom.intel.com/chip-sho...d-one-of-the-worlds-most-ethical-companies-2/
 
Pieter3dnow said:
Because he got the other 2 dies working ? All he did was something which was pretty futile to begin with. He checked if there were dies then he realized that the dies that were in there he could not check anyway. So where is this lying part come from ? Have you already seen someone with 4 dies enabled or swapped the 2 functioning dies to the other 2 ?
Click to expand...

AMD initially said us that ThreadRipper has only two dies, and that the other two pieces were "blanks", not actual dies:

Update: AMD has clarified that two of the four silicon die are 'dummy' blanks rather than disabled Ryzen chips. They are included for structural stability. The two die that are enabled sit diagonally from each other to aid in cooling.
Click to expand...

Der8hauer didn't believe the official explanation and tested it by himself:

How far would you go to prove a point? Well, the overclocker and CPU delidding expert der8auer has decided to put AMD's "Dummy Dies" claim to the test, to find out once and for all whether or not AMD is using failed/defective Ryzen dies or silicon spacers.
Click to expand...

Thanks to him we now know there are four ryzen dies on ThreadRipper.

Next logical step would be to check if those extra dies are fully functional or if those dies are failed dies. But checking this will be much more difficult.
 
juanrga said:
AMD initially said us that ThreadRipper has only two dies, and that the other two pieces were "blanks", not actual dies:



Der8hauer didn't believe the official explanation and tested it by himself:



Thanks to him we now know there are four ryzen dies on ThreadRipper.

Next logical step would be to check if those extra dies are fully functional or if those dies are failed dies. But checking this will be much more difficult.
Click to expand...
Then WHAT THE FK is the difference. Seriously are you guys that dense? It could be solid gold or a horse turd or real able Ryzen dies. Without traces and connects they all have the same outcome... THEY DONT DO SHIT!
 
juanrga said:
AMD initially said us that ThreadRipper has only two dies, and that the other two pieces were "blanks", not actual dies:



Der8hauer didn't believe the official explanation and tested it by himself:



Thanks to him we now know there are four ryzen dies on ThreadRipper.

Next logical step would be to check if those extra dies are fully functional or if those dies are failed dies. But checking this will be much more difficult.
Click to expand...
Thank you for outlining this.

At no point did I defend Intel. Funny how the amd fans jump on Intel. I was simply commenting on AMD and their misleading marketing /misinformation blunders toward consumers.
 
JustReason said:
Then WHAT THE FK is the difference. Seriously are you guys that dense? It could be solid gold or a horse turd or real able Ryzen dies. Without traces and connects they all have the same outcome... THEY DONT DO SHIT!
Click to expand...

Basically rocks.
 
jmilcher said:
Thank you for outlining this.

At no point did I defend Intel. Funny how the amd fans jump on Intel. I was simply commenting on AMD and their misleading marketing /misinformation blunders toward consumers.
Click to expand...

Hmm, like wtf how is AMD lying to ppl? Where is the outrage? You didn't buy TR, and you didn't pay for dies you didn't get?
 
juanrga said:
AMD initially said us that ThreadRipper has only two dies, and that the other two pieces were "blanks", not actual dies:
Der8hauer didn't believe the official explanation and tested it by himself:
Thanks to him we now know there are four ryzen dies on ThreadRipper.
Next logical step would be to check if those extra dies are fully functional or if those dies are failed dies. But checking this will be much more difficult.
Click to expand...

Because that makes sense to you. He did not test anything because he still does not know if those are working dies or not so basically he is nothing further then what other people already knew only 2 of them function.

So here goes the German guy wasted money on something he is none the wiser, he still can't tell if the 2 other dies are working or not.

Let me explain it to you Juanrga when you buy Threadripper only 2 dies will function when some German dude tested this according to you he can not prove otherwise so what exactly did he test ?
Before the test he knew only 2 dies were working. After the test he knows only 2 dies are working , see the problem with the test ?

There is no next logical step, unless you can rig a motherboard to support those 2 extra dies it is all rather far fetched.

jmilcher said:
Thank you for outlining this.
At no point did I defend Intel. Funny how the amd fans jump on Intel. I was simply commenting on AMD and their misleading marketing /misinformation blunders toward consumers.
Click to expand...
You did however used the same lingo as what Intel says about AMD
jmilcher said:
Especially considering their terrible track record
Click to expand...

You want some Intel slides to go with that ?


It is futile to think Juanrga did anything beside talk non sense. AMD just used broken dies they had otherwise chucked in the bin. People that purchased Threadripper don't mind this ?
 
How is this in any way a [H]ard member issue? AMD said something that was fact-checked and turned out to be, to put it charitably, a turn of phrase that stretches the truth. Jumping on people for pointing it out doesn't change the fact that this development occurred.
 
mnewxcv said:
Now all we need is someone to tell us which traces need to be pencil modded!
Click to expand...

There is nothing to pencil mod.

thesmokingman said:
End point, he thinks there could be more iterations of Threadripper up to 32 core. Some from AMD have already stated that there will be more TR products to come....
Click to expand...

It's probable that this is the case, but I wouldn't expect 32 cores at 4.0GHz any time soon.

Shintai said:
Not a surprise.

But why did we have to hear this BS first.

It includes failed EPYC chips, plain simple.
Click to expand...

BS? Hardly. You have dies that don't do anything. Whether they have any integrated circuitry or were lumps of silicon or plastic with nothing in them, what would it matter? They are both blanks, dummies, or whatever.

gigaxtreme1 said:
They would have to duplex the memory channels though
Click to expand...

Epyc supports twice as many memory channels and 128 PCIe lanes. 64 of which are used to connect to the Infinity fabric. Simply put, X399 and TR4 are lacking in their ability to handle such a CPU.

arestavo said:
Sweet - if it's cost effective, why not? And if AMD releases 32 core Threadripper to give Intel even more reasons to get off of their collective asses and really push the envelope, then all the better.
Click to expand...

The 16 core 32 thread 1950X is already $1,000. If you doubled the dies, you would have a greater cost. Threadripper already runs at the edge of the clock speeds it can handle with half that many cores / functional dies in it. You would double the TDP on that fucker which the TR4 motherboards almost certainly can't handle. Now, it's possible that it might work if they ran four dies at 2.0GHz or something like that but I doubt you'd be able to keep thermals under control at 2.8GHz or beyond if all four dies were actually enabled.

juanrga said:
Wow! Then AMD lied us when said there are only two dies on ThreadRipper.

This proves the true origin of ThreadRripper.
Click to expand...

There are only two FUNCTIONAL dies on Threadripper. Dummy, non-functional, blank, etc. are just splitting hairs when what's in the spot doesn't work.

workshop35 said:
I read the articles and didn't understand why its a big deal. The AMD engineer was saying that they aren't functional dies and from what I was reading they aren't even fully connected. Seems like a semantics issue getting blown way out of proportion, unless I'm missing more to the story.
Click to expand...

Nope, you are right on target.
 
A nice resume of key points is found on seekingalpha article about this topic

https://seekingalpha.com/article/4107952-amd-threadripper-mysteries?page=2

AMD dissembles about Threadripper

When der8auer earlier delidded a Threadripper engineering sample, he found, not surprisingly, four Zeppelin dies. At the time, according to Tom's Hardware:

AMD responded that two of the die on shipping Threadripper models are not the same as those Der8auer found in the engineering sample - the additional two die are "dummy" die that the company uses to provide structural stability for the package. These additional die prevent the IHS (Integrated Heat Spreader) from caving in when you tighten down the heatsink.
Click to expand...

AMD continued to call the non-functional Zeppelins "dummy" dies after the retail Threadripper delidding, according to Tom's:

We followed up with AMD and inquired if the die were functional. AMD responded that two of the die are in fact non-functional "dummy" die that aren't connected electrically to the substrate.
Click to expand...

This seems to be dissembling on AMD's part. Reviewers have typically assumed (as did Anandtech) that the "dummies" were mere silicon blanks, not fabricated Zeppelin dies.

I find it rather curious that AMD has gone to such lengths to obscure the fact that the dies are fabricated parts, working or not. Tom's concludes:

Mounting dies to a substrate is like any other manufacturing process; there can be defects. It's possible that the Threadripper processors are simply EPYC processors with die that weren't successfully mated to the substrate. Conversely, it's also possible that AMD is using defective die for the fillers, but in either case, it's doubtful that the company is wasting functioning silicon.
Click to expand...

In fact I find the former explanation, that Threadripper parts are simply defective EPYC processors, far more likely. The reason is that even using defective Zeppelins for spacers is very inefficient from a manufacturing standpoint. The dies have to be mounted in some fashion as the functional Zeppelins are, and then they are soldered to the heat spreader as shown in der8auer's video.

That's a lot of work, especially the soldering to the spreader, for dies that are non-functional and won't generate any heat. Probably a simpler spacer scheme could have been devised, such as machining the spacer into the underside of the heat spreader.

This was in fact, the explanation I proffered on July 28, which apparently was so controversial. But the fan reaction to this probably explains AMD's dissembling about it. The fans don't like the idea that they're being sold EPYC rejects. They really shouldn't be upset. Threadripper is a perfectly reasonable way to salvage value from defective EPYC parts.
Click to expand...
 
I get what your saying, and what's said in the above. I still think it's splitting hairs. Who cares if you are getting a failed Epyc processor or a purpose built CPU with two dies and two "blanks" in it? What the fuck is the difference? AMD has done this shit before, as has Intel. I fail to see how this makes any difference to us or anyone else. It doesn't alter how the CPU performs or overclocks. Chances are, using failed Epyc CPUs reduces the cost of Threadripper over a purpose built version.

Let's not forget that Threadripper and X399 weren't in AMD's original plans for Ryzen. It wasn't until the architecture turned out to be better than they originally anticipated that AMD decided to offer products for the HEDT segment. Making use of failed Epyc dies is most likely the best and most cost effective way to do that.
 
Last edited:
Tempest in a Teapot.

This is of absolutely no consequence what so ever.

AMD needed spacers. Rather than use an expensive machining process to leave metal bumps on the bottom of the heatspreader, or spend money on some other material to use as spacers, some engineering doing their job was like "Hey guys, why don't we just use rejected cores as spacers? They are already the perfect size, and are going in the trash anyway"

Jesus Christ you people will get worked up over nothing.

This is just an example of AMD's engineers doing their jobs properly.

Pieter3dnow said:

It is weird at the same time as well. There saying that there yield is awesome so how many of these failed dies are there before they have to resort to replace the failed dies with a placeholder.
Click to expand...

Well, ThreadRipper is a very niche product. The quantities will be much smaller than their other products. While yields might be great, they are never going to be perfect, and AMD manufactures quite large quantities of Ryzen chips overall. The rejects from that process should be more than sufficient to cover the spacer needs for TR. Regardless of how much yields improve, there will probably be more than enough rejects for this purpose.
 
Dan_D said:
The 16 core 32 thread 1950X is already $1,000. If you doubled the dies, you would have a greater cost. Threadripper already runs at the edge of the clock speeds it can handle with half that many cores / functional dies in it. You would double the TDP on that fucker which the TR4 motherboards almost certainly can't handle. Now, it's possible that it might work if they ran four dies at 2.0GHz or something like that but I doubt you'd be able to keep thermals under control at 2.8GHz or beyond if all four dies were actually enabled.
Click to expand...
Maybe not 4 GHz, but it should be easier for them to achieve higher clocks than Intel's big monolithic dies. And the cores are much more efficient in the lower 3.X range.

juanrga said:
A nice resume of key points is found on seekingalpha article about this topic

https://seekingalpha.com/article/4107952-amd-threadripper-mysteries?page=2

"In fact I find the former explanation, thatThreadripper parts are simply defective EPYC processors, far more likely. The reason is that even using defective Zeppelins for spacers is very inefficient from a manufacturing standpoint. The dies have to be mounted in some fashion as the functional Zeppelins are, and then they are soldered to the heat spreader as shown in der8auer's video."
Click to expand...
That doesn't seem likely if there are always the same two functional cores in TR.
 
Dan_D said:
I get what your saying, and what's said in the above. I still think it's splitting hairs. Who cares if you are getting a failed Epyc processor or a purpose built CPU with two dies and two "blanks" in it? What the fuck is the difference? AMD has done this shit before, as has Intel. I fail to see how this makes any difference to us or anyone else. It doesn't alter how the CPU performs or overclocks. Chances are, using failed Epyc CPUs reduces the cost of Threadripper over a purpose built version.
Click to expand...

Of course this discussion about dies is irrelevant for final users, as I wrote in a former post:

juanrga said:
This topic is completely irrelevant for final users. For final users it doesn't matter if AMD uses four dies or two dies plus two blanks. It is relevant for tech enthusiasts and economic analysts. The first because want to know the fine details of the tech. The second because want to infer more about the economies behind EPYC and ThreadRipper.
Click to expand...
 
You must log in or register to reply here.
Back
Top