• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Not much excitement for RDNA3, compared to the RTX 4000 series

Imagine the uproar if AMD had made a Mount Doom GPU that melted connectors...

JRx0jlP.jpg
 
g67575 said:
Considering what an improvement RDNA and RDNA2 were, there doesn't seem to be much excitement for RDNA3. AMD's GPU roadmap shows a release date of Q4 2022, here:

Last year, AMD's CEO said this about RDNA3 progress:
"We remain on-track to launch next-generation products in 2022, including our ... RDNA 3 GPUs".

According to AMD, RDNA 3 will feature 5nm process tech, a 50% performance per watt improvement, and 'rearchitected compute unit'(s):

2022-06-09%2013_36_23_575px.jpg


The weakest part of RDNA2 was RT performance (relative to the RTX 3000 series), so hopefully AMD will tackle that too. Meanwhile Nvidia has still confirmed nothing about the RTX 4000 series, and apparently plan more RTX 3000 models with GDDR6X soon :confused:
Click to expand...
Its due to leaks been squashed by amd
all the rumor people basically have no new content on the subject
Cards are awesome
 
If I can grab a 3080 equivelant with 12GB or more Vram around £450 then I may be on board to buy Radeon 7k otherwise I am happy holding out for another 2 or so years at this point.

Just drop shadows to medium. HBAO to SSAO is like a 10FPS improvement on it's own.
 
Last edited:
It's 4nm, custom 4nm to be precise but it's also rumoured to nearly be 5nm, as in while it's better than the standard tsmc 5nm process it's only slightly better
 
g67575 said:
4nm for the RTX 4090, I assume it's the same for the whole series.
Click to expand...

40 series is 4N basically TSMC 5nm with Nvidia custom libraries. Nvidia likes to trick customers into thinking they are on a process better than AMD or others, same was true of Samsung's 8N it really was 10nm (30 series).
​

Nvidia clarifies: The TSMC 4N used by the RTX 40 GPU is a 5nm process​



According to Hong Kong media HKEPC, Nvidia clarified today that the RTX 40 GPU uses TSMC’s 4N 5nm process, not the 4nm process, due to a large number of media writing errors.
According to reports, the Nvidia RTX 40 GPU is based on the 5nm TSMC 4N process, the N in “4N” should stand for Nvidia custom.

At present, the official press release of Nvidia and the press release of each graphics card manufacturer basically used the “TSMC 4N” writing and did not specify the specific nm process.
Click to expand...
 
Last edited:
If nvidia can get into 2023 limiting damage from what they have done this past couple of years I will be amazed they maintain an 80+% mindshare tbh.
 
gpuerrilla said:
If nvidia can get into 2023 limiting damage from what they have done this past couple of years I will be amazed they maintain an 80+% mindshare tbh.
Click to expand...

The best recent one is the fake transistor count on 40 series that makes no sense. Just have to look at Hopper that is a much larger die on the same process but has 80 Billion.

Hopper H100 814 mm² 4N 80 Billion Transistors (5nm TSMC)

ADA AD102 608 mm² 4N 75 Billion Transistors (5nm TSMC).

Ampere GA100 826 mm² 54.2 Billion Transistors @ 7nm TSMC. (NO fake N node names).

Ampere GA102 628 mm² 8N 28.3 Billion Transistors (Samsung 10nm)

Something doesn't add up here to me and I'm calling AD102 with 75 Billion BS and a Nvidia made up number as with their fake process sizes they make up.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom