cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
22,060
Buildzoid from Actually Hardcore Overclocking on YouTube has performed an in-depth analysis of the PCB and VRM layout of the AMD Radeon VII. Watch him discuss the efficiency and cost of the various exotic components that are found in the design of the AMD Radeon VII. You can view our unboxing and teardown video here.

AMD's Radeon VII card left its initial embargo, which allowed tear-downs (link below), just recently, and that allowed us to look closer at the VRM for analysis.
 
Better components better card. Sounds like a pizza commercial. Well hopefully AMD will not have the failure rates like NVidia cheaper component cards.

Vega 64 pushing 500w plus is not even an issue . An NVidia card you probably have a mushroom cloud inside your case.

Hopefully the performance will be up there with NVidia this time around.
 
Better components better card. Sounds like a pizza commercial. Well hopefully AMD will not have the failure rates like NVidia cheaper component cards.

Vega 64 pushing 500w plus is not even an issue . An NVidia card you probably have a mushroom cloud inside your case.

Hopefully the performance will be up there with NVidia this time around.

All this talk of Pizza and Mushroom clouds got me thinking....

images?q=tbn:ANd9GcTiHH2lKSdqH5wmEa7CIP2CEqWex0M0fp9vGHvfD7-Mgdlkgh_jAw.jpg
:LOL:
 
That board is so clean, very awesome video may have to support him via patron as well!
 
  • Like
Reactions: ecktt
like this
This guy looks like he’s like 12, but he’s so smart. I go to him for vrm info all the time.

some people just get super excited about different things that others may not but yeah he's definitely the go to person for vrm info.

but i'm also starting to understand why the cards priced so high with the VRM costs and HBM costs. hopefully the rumors about volume are false otherwise there probably isn't a whole lot of profit coming from radeon 7 after R&D, manufacturing, and component costs.
 
Last edited:
some people just get super excited about different things that others may not but yeah he's definitely the go to person for vrm info.

but i'm also starting to understand why the cards priced so high with the VRM costs and HBM costs. hopefully the rumors about volume are false otherwise there probably isn't a whole lot of profit coming from radeon 7 after R&D, manufacturing, and component costs.

I am an EE and I don’t know that he is right about the cost of the power stages. Those are models that aren’t sold to normal distributors, so he’s guessing at the price. I would imagine AMD has quite a bit of purchasing power and isn’t going to pay what I would pay even for high quantities. HBM is super expensive though.

The Nvidia design seems smarter if those prices are correct, though. The advantages are minimal in practice.

These regulators aren’t all that exotic still, maybe for consumer stuff. Still simple buck regulators.
 
Better components better card. Sounds like a pizza commercial. Well hopefully AMD will not have the failure rates like NVidia cheaper component cards.

Vega 64 pushing 500w plus is not even an issue . An NVidia card you probably have a mushroom cloud inside your case.

Hopefully the performance will be up there with NVidia this time around.

The Nvidia failures probably aren’t due to underdesign of the VRM. When anything after a voltage regulator fails short it is often the regulator components that smoke. It could be a lot of things. Bad lot of ceramic or tantalum caps, bad GPU die, etc. I’m sure Nvidia know by now though.

Lower current switching regulators are often short circuit protected but I am not sure how well it works in these power stages because their normal working condition is damn close to a short circuit compared to more pedestrian electronics.
 
Depending on whatever was causing 2000-series GPUs to go tits-up, they may have gone too far.

Perhaps, we’ll probably never know.

For most electrical engineers that have been around a while, if your PCB catches on fire it was a tantalum capacitor 90% of the time. They have a nasty habit of spontaneously combusting if operated too close to limits. Now, with that said, most devices like this now are using polymer tantalum instead of MnO2 chemistry, so the risk of fire is pretty much gone. I believe they do tend to fail short though in comparison to aluminum electrolytic caps.

My company tries to avoid tantalum where possible from habit with the non polymer types and the fact that it is a conflict mineral. There are aluminum polymer caps that outperform them mostly but they have capacitance per volume advantages still and probably age better.

It would be a pretty newbie mistake to underspec a VRM for your own chip. Especially for a company with as much experience and top notch engineers as NV. They have simulation tools that cost more than a car per user license. I could definitely see a 3rd party doing it trying to save 5 cents though. I wonder if they do their own reference designs still or they outsourced it. Either way it would be something I’d expect to see on a really low margin product, not the flagship.
 
Last edited:
Buildzoid is the man! Wish he was be more concise though.
I agree, I was getting frustrated at times. "It's this, well, uh, er, no it's not, but.. yeah, uh it's this" He needs to write a script he can roughly follow. Or maybe it's a caffeine issue.
 
This guy looks like he’s like 12, but he’s so smart. I go to him for vrm info all the time.

he's not that smart, but he's for sure not stupid and he is dedicated which is very important!


Better components better card. Sounds like a pizza commercial. Well hopefully AMD will not have the failure rates like NVidia cheaper component cards.

Vega 64 pushing 500w plus is not even an issue . An NVidia card you probably have a mushroom cloud inside your case.

Hopefully the performance will be up there with NVidia this time around.

Nvidia have lots of bad reference designs to answer for, but RTX cards is not one of them :)
AMD usually have awesome reference designs(pcb) maybe not for coilwhine but they can push a lot of power for sure

I've usually said to people:
Nvidia, try to find AIB and amd reference is good for PCB, cooler probably not.
 
Nice video. Agreed that he would benefit MASSIVELY from some editing. Despite that criticism, I learned a lot and am glad that he did this.
 
Let's be clear crystal clear, more than 95% of his audience have no clue about what he is saying, comprised people posting in this thread.

But from a more professional perspective Buildzoid have a pretty basic knowledge and share more than often misleading information in his videos.
The information is okish for a basic user, but deeply wrong when one analyze the video content from a scientific and engineering point of view.

So you can greatly praise Buildzoid for sharing these information making basic users interested about, helping the audience to take care about the overall build quality of the product from an electronics perspective.

But in any case, one shouldn't take Buildzoid thought as real verified information, he is worth as the other reviewers, sometimes biased or not.

Also notice that the extreme oc field is greatly biased and sponsored, with poor information sharing, dictated by the fear someone stole your idea, more or less as companies do.

So again, you can greatly praise him for sharing the basic of overcloking and some advanced technique, he is among the few doing so, the rest of the extreme oc field is pretty rubbish!
 
Last edited:
  • Like
Reactions: ecktt
like this
HardOCP, GamersNexus & Buildzoid. Anyone missing?
I like Adored TV as well. I am not a Patron of GamersNexus, but I do pitch them some bucks when they live stream.
 
  • Like
Reactions: nimer
like this
he's not that smart, but he's for sure not stupid and he is dedicated which is very important!




Nvidia have lots of bad reference designs to answer for, but RTX cards is not one of them :)
AMD usually have awesome reference designs(pcb) maybe not for coilwhine but they can push a lot of power for sure

I've usually said to people:
Nvidia, try to find AIB and amd reference is good for PCB, cooler probably not.


Lolwut? The RTX thing was a shit show! Members on this forum have had two to three bad cards in a row.

Ive had nothing but reference AMD/ATi cards since my first one (7500LE) and even the King of coil whine(7970 GHz Edition) and nothing has compared to the screeching I got from the 6800 Ultra(BFG) and current 1080Ti(EVGA FTW3). That is very micro though, and a small sample.
 
Lolwut? The RTX thing was a shit show! Members on this forum have had two to three bad cards in a row.

Ive had nothing but reference AMD/ATi cards since my first one (7500LE) and even the King of coil whine(7970 GHz Edition) and nothing has compared to the screeching I got from the 6800 Ultra(BFG) and current 1080Ti(EVGA FTW3). That is very micro though, and a small sample.

And there is people who's measure them, they are fine - no shorts, voltages are fine, ripple is good so it's not power delivery.
I was specifically talking about the PCB\VRM, VRM was good and the chips or memory was bad.

every single aspect of the design on the PCB is good, they've changed nothing but Quality control and issues are now gone.

On coilwhine
My Coil whine at 500W on my V64 isn't too bad actually, bad hardware keeps me warm during cold winter months :p
It's audible but far from 7970 Ghz ed, that one should get a reward!
 
Back
Top