Tuesday, September 22nd 2020
AMD Radeon "Navy Flounder" Features 40CU, 192-bit GDDR6 Memory
AMD uses offbeat codenames such as the "Great Horned Owl," "Sienna Cichlid" and "Navy Flounder" to identify sources of leaks internally. One such upcoming product, codenamed "Navy Flounder," is shaping up to be a possible successor to the RX 5500 XT, the company's 1080p segment-leading product. According to ROCm compute code fished out by stblr on Reddit, this GPU is configured with 40 compute units, a step up from 14 on the RX 5500 XT, and retains a 192-bit wide GDDR6 memory interface.
Assuming the RDNA2 compute unit on next-gen Radeon RX graphics processors has the same number of stream processors per CU, we're looking at 2,560 stream processors for the "Navy Flounder," compared to 80 on "Sienna Cichlid." The 192-bit wide memory interface allows a high degree of segmentation for AMD's product managers for graphics cards under the $250-mark.
Sources:
VideoCardz, stblr (Reddit)
Assuming the RDNA2 compute unit on next-gen Radeon RX graphics processors has the same number of stream processors per CU, we're looking at 2,560 stream processors for the "Navy Flounder," compared to 80 on "Sienna Cichlid." The 192-bit wide memory interface allows a high degree of segmentation for AMD's product managers for graphics cards under the $250-mark.
135 Comments on AMD Radeon "Navy Flounder" Features 40CU, 192-bit GDDR6 Memory
i havent read anything else, still laughing
The only thing we're reasonably sure of is that AMD will compete in all the segments of the market this time, the details are yet unknown.
On that note, don't expect much stock from them even if RDNA2 is amazing.
The reason the RTX 3080 only has 10GB is quite simple. Micron currently only has the 8Gb chips in mass production. nVidia has to wait for Micron to mass-produce the 16Gb chips to launch a 20GB RTX 3080. You can look this up at Micron.com, it's open data. If you thing what about the RTX 3090 with 24GB? It uses a really expensive option with double the memory chips on both sides of the PCB. Power consumption of the memory is also doubled with the RTX 3090. It's simply not a viable option for a RTX 3080. 16Gb GDDR6X goes in mass production over a couple of months btw.
RAM pricing is also highly dependent on clock rates. $6/GB might be possible for the slowest chips, but anything delivering current-gen speeds will cost more than that.
www.freepatentsonline.com/y2020/0293445.html
As for the paper reality, I was alluding to the specs relative to actual performance of the new consoles. We've seen it every time, its not quite up there with a similar PC setup. Limitations, stuff lacking or sharing resources... and its not looking like this gen is any different. Awesome tricks, but they do come at some cost and while the systems are likely very efficient, its not like they can simply transplant that efficiency to a PC dGPU.
So it would be a capital mistake to consider the console specs as 'the PC spec' or its performance level as 'the PC perf level'. Its not going to be that at all - and its most certainly isn't going to surpass the console's efficiency.
Me, on the other hand, just gonna wait for the release and see what AMD prepared for the consumer market. All those rumors, fake news and opinions make me kinda dizzy. Besides, in the end, all the rumors supposedly true, may turn into crap if AMD decides to change something in the end.
What I'm really hoping for from AMD is, at the launch date the cards are available and if you decide to go with one, it will be there waiting for you to buy it.
Me, I prefer my chips with just the right amount of salt.