It seems that Radeon RX Vega 10 GPUs used on Radeon RX Vega 64 differ a little, I'll step that up a notch, it seems they differ a little bit in height. And that might be relevant for the AIB cards.
Earlier on in our forums a user already asked if we had a different sample, he noticed that the GPU looks different compared to some other websites. After inspecting our review kit dummy sample and the one used on our graphics card sample I noticed it definitely differ. After further inspection I also noticed that the GPU that is used on our Vega 64 sample, differs from other reviewers.
See that black coating in-between the chips? There should be a dark substrate in-between the HBM2 memory as you can clearly see on the upper photo, the 'dark stuff' is clearly missing
Update: We've received word back from AMD, I'll just leave that here:
We have multiple manufacturing partners for Vega10 package assembly. To meet the high-volume production that we target with Vega10, we need access to most of the world capacity for 2.5D interposer technology. Each partner has unique processes, including one which has an epoxy fill to level the entire SOC surface. This specificity is inconsequential to the system solution and all Vega10 packages are form, fit and function equivalent. As there is no functional difference, any Vega10 package flavor can be binned into a Vega10 XL, XT or XTX.
Here is the official answer from AMD. - "We have multiple manufacturing partners for Vega, one of them uses an epoxy fill to level the SOC surface, otherwise they're functionally identical."
There is no height gap on the GPUs themselves, one has what looks to be a plastic or epoxy surrounding the chips and covering the interposer that makes it look more bulky.
Odd that the manufacturing partners would differ in encapsulate. I know Fiji wasn't encapsulated, but given the fragility of the interposer it seems like a wise step to take. Probably costs two cents to do though, so is probably a target for cost cutting.
Quote:
We are looking at the bottom of the die not the top. The top is face down on the interposer probably connected with tiny solder balls. The silver gray color is just what raw silicon looks like after the manufacturing process.
Two Dutch tech sites did the same discovery, asked AIBs and yes they receive different dies.
Also, apparently AMD uses 2 other companies to build the chips on the interposer. Hence there is a difference. One polished with glue, other bare.
Edit: saw the quote in the OP bit too late.
Anyways, I know most (if not all) of the OCN hwbot team are steering clear of Vega. Even if there is a little variance in Die height, it can easily destroy itself when you go hard wiTh LN2. It would also mean there is even more variance in a good mount and a really good mount on the LN2 pot. Which is a nightmare.
Anyways, I know most (if not all) of the OCN hwbot team are steering clear of Vega. Even if there is a little variance in Die height, it can easily destroy itself when you go hard wiTh LN2. It would also mean there is even more variance in a good mount and a really good mount on the LN2 pot. Which is a nightmare.
Has anyone seen any evidence of a variance in die height?
It's clear that one assembler is encapsulating the entire package, while the other is only using underfill, but this is done before final polishing, and shouldn't result in any difference in height, thickness, or consistency thereof.
What is?
Has anyone seen any evidence of a variance in die height?
It's clear that one assembler is encapsulating the entire package, while the other is only using underfill, but this is done before final polishing, and shouldn't result in any difference in height, thickness, or consistency thereof.
This being known now. I don't see anyone here on OCN that could not adjust easily to this. So what, you might have to grab a micrometer and measure if it came to that...pfft.
It's like saying, god forbid a car mechanic might have to adjust the gap on his new spark plugs. I know, end of the world...
Don't know if i would like to tinker around with a vega die , from what i have read the hbm/interposer is supposed to be very delicate, i read that even when doing simple tasks like cleaning old thermal paste, one has to be very careful not to damage it.
This being known now. I don't see anyone here on OCN that could not adjust easily to this. So what, you might have to grab a micrometer and measure if it came to that...pfft.
It's like saying, god forbid a car mechanic might have to adjust the gap on his new spark plugs. I know, end of the world...
OK at least I hope OCN "Builder" types wont have a problem with this, if it's a problem at all lol.
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Related Threads
?
?
?
?
?
Ask a question
Ask a question
Overclock.net
27.8M posts
541.2K members
Since 2004
A forum community dedicated to overclocking enthusiasts and testing the limits of computing. Come join the discussion about computing, builds, collections, displays, models, styles, scales, specifications, reviews, accessories, classifieds, and more!