GPUs Graphics Latest news

Exclusive info about NVIDIA’s Ampere: 3 models, memory size, PCBs, an extremely expensive cooler for the FE and the search for the leak

Recently I had already posted some information about the new graphics card generation from NVIDIA and also processed it in the video, but I was still a little bit reserved concerning the currently readable images of the RTX 3080 Founders Edition (FE). I first wanted to wait for confirmation if and to what extent these are really real pictures. As I have now learned from my own sources, NVIDIA has announced an internal investigation to discover the leak. Here the focus is on Foxconn and BYD (Build Your Dreams) as contract manufacturer of the Founders Edition, where the source of the leak is suspected. This means that the authenticity of the images can be confirmed.

As a small side note, not even NVIDIA’s product and sales managers are said to have seen these images before the leak. It remains to be seen if NVIDIA will make some changes to punish the leak a little bit. The remaining 3 months are easily sufficient for mechanical changes, but this is more difficult with the very complex PCB of the FE. Speaking of coolers – the industry is currently offering a production price of 150 USD for the entire FE design, i.e. housing and cooler in one piece. This then also suggests the positioning and pricing of the Founders Edition, which is to be significantly upgraded.

But I have more information for you, which goes a little bit more beyond the picture leak. The somewhat more complex cooler design of the FE requires a special circuit board with an additional cutout. The internal designation is PG133 and the board should be able to be used for all three versions on launch day. The use of flexible power supply cables that lead to the separate sockets at the end of the card is also correct here. The PCB and these cables alone will be a real challenge for the manufacturers of GPU waterblocks.


Also reference and probably only 21.3 cm long is the rectangular board design PG132 without cutout, which is offered to the board partners. and exactly at this point it becomes interesting, because I was also able to find out the possible memory allocation and interface. All three expected launch cards will definitely be equipped with GDDR6X, with the GA102 determining the chip (SKU) used, but not the exact nomenclature. Summarized from several sources, the following structure emerges, which still contains question marks. Those who know Nvidia also know that changes are often made at the last moment to fool the world.

Part PCB Chip Model Extension Memory Interface TBP Connectors
SKU10 PG132 GA102 RTX 3090 none 24 GB GDDR6X
384-bt 350 W 3x DP, HDMI
SKU20 PG132 GA102 RTX 3080 (Ti/Super)* 11/12 GB GDDR6X* 352-/384-bit* 320 W 3x DP, HDMI
SKU30 PG132 GA102 RTX 3080 none 10 GB GDDR6X 320-bit 320 W 3x DP, HDMI

As currently reported, the SKU20 and SKU30 will apparently only differ in the memory capacity and interface, as well as the possible clock rate, although the same TBP was specified. this all sounds a bit confusing at the moment, but it would make sense to define a kind of “beginner class” with the SKU30, similar to the non-A chips of the past, without getting back into the reputation of the 2-class chips. This would give them a certain advantage in selling the lower price plausibly.

Sources: Own, pictures

Danke für die Spende

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

Hier erfährst Du, wie: Hier spenden.

Hier kannst 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