Editor's Desk GPUs Hardware Latest news

Why AMD's Board Partners Can't Develop Their Own Designs (Currently) | Editorial

The reason for my heckling today is actually trivial, but it also shows that AMD is doing everything it can to maintain strict secrecy of technical details even to board partners. Now, of course, I can't spoil or leak anything before the launch of the Radeon 7, but a few small glimpses of his then allowed me...

I wrote it from the beginning, among others in the article "For the time being no board partner cards of AMD's Radeon Vega VII in sight – AMD sells the reference card exclusively", that AMD will go exactly this way. It also makes no sense to hop after every false rumour like a misguided bunny (e.g. Powercolor), only to have to revise yourself again a few hours later. Nice, it brings clicks and the readership is hyped again, but you don't even do AMD a favor, let alone yourself.

The reason for my heckling today is actually trivial, but it also shows that AMD is doing everything it can to maintain strict secrecy of technical details even to board partners. Now, of course, I'm not allowed to spoil or leak anything before the launch of the Radeon 7 (yes, AMD also has NDAs), but it is certainly not forbidden to reveal details from my current work. It should be known that I don't just take the cards apart and re-tap details of the PR. Laying on hands is really not quite mine.

The so-called Power Topology of the Radeon 7 has it all. Which brings us back to the above-linked article. If I wrote there about the same board as with the MI50, spitzklicker could immediately object that the Radeon 7 now lacked massive connections, the power supply would be placed differently and in general… But you have to know that I wasn't talking about the same board, but about the same topology. My mistake is not to have been precise.

And what does this have to do with the own designs of the board partners? Lots! The magic word is FAE. Field Applications Engineering (FAE) is nothing more than technical support, e.g. AMD (or Nvidia) to its customers, i.e. the board partners. For example, you can as a reviewer, the topology of the power supply is approximated by close observation, but nothing more. In order to understand the connections and to understand why some results look as they look, you need a little more information.

Normally, this can be discussed with the colleagues of the development departments at the Board partners, who have the exact information they receive from the FAE departments of the respective chip manufacturers (regional contact person). With the Radeon 7, however, even these sources are powerless, because the only statement was that any information about the board and the basic layout is absolutely "confidential". Another source even wrote that in general all the details about professional cards are under wraps.

But how can a board partner who does not even have access to basic technical details develop their own layouts? If you add the time it takes for a company to do so, the whole thing becomes much less likely for the next time. At the time, I had posted a timeline for Vega in the 3DC, which shows very nicely that it takes at least three months for your own product to come off the production line:

Process Description Action / Period
BOM release Bill Of Materials Release Beginning to end of June
Evt Engineering Validation Test End of June to start of July
Dvt Design Validation Test Early to mid-July
Ws Working Sample (all with engineering hardware and software) Mid-July
EMI Test Electromagnetic Interference Test Mid-July
Pvt Production Validation Test End of July to beginning of August
PVT sorting ? Beginning of August
PPBIOS Meals the final source BIOS from AMD Beginning of august (final source-bios should come on august 2nd )
Ramp & MP Start of mass production will be set later, because the appointment by AMD is missing

 

If AMD were to rethink and have enough packages, prices would be such that all parties could still make a profit and the market would react positively to the card, then it would not be expected before May 2019. Computex? Possible, but also probable? I will bring a board analysis to the launch and also calculate what the whole thing could cost. Then some of the subjunctives in this paragraph are likely to die quietly and modestly.

So let us concentrate on the tangible and real existing. I believe that we all have enough work to do with this too.

 

Update:

In the end, I managed to break the wall of silence. You can look forward to an accurate board analysis at launch day, including ALL partial and auxiliary voltages, components and interactions. I can't and won't spoil more, but there will be more videos for the usual detailed review. Let's go!

 

Danke für die Spende



Du fandest, der Beitrag war interessant und möchtest uns unterstützen? Klasse!

Hier erfährst Du, wie: Hier spenden.

Hier kann Du per PayPal spenden.

About the author

Igor Wallossek

Editor-in-chief and name-giver of igor'sLAB as the content successor of Tom's Hardware Germany, whose license was returned in June 2019 in order to better meet the qualitative demands of web content and challenges of new media such as YouTube with its own channel.

Computer nerd since 1983, audio freak since 1979 and pretty much open to anything with a plug or battery for over 50 years.

Follow Igor:
YouTube Facebook Instagram Twitter