Wednesday, January 29th 2020

AMD to Debut 2nd Gen RDNA Architecture in 2020

AMD CEO Dr Lisa Su, in her Q4-2019 and FY-2019 earnings call, confirmed that the company debut its second-generation RDNA graphics architecture in 2020. "In 2019 we launched our new architecture in GPUs, it's the RDNA architecture, and that was the Navi-based products. You should expect those will be refreshed in 2020, and we will have our new next-generation RDNA architecture that will be part our 2020 lineup."

Second-gen RDNA, or RDNA2, is expected to leverage the new 7 nm+ (EUV) silicon fabrication process at TSMC, to dial up transistor-counts, clock-speeds, and performance. Among the two anticipated feature additions are VRS (variable rate shading) and possibly ray-tracing. The fabled "big Navi" silicon, a GPU larger than "Navi 10," is also on the cards, according to an earlier statement by Dr Su. More details about these upcoming graphics cards are expected to be put out in March, at the 2020 AMD Investor Day conference.
Add your own comment

81 Comments on AMD to Debut 2nd Gen RDNA Architecture in 2020

#76
bug
CasecutterYeah, and that can happen with any player. Considering AMD/RGT are the one on the newest process and partly where Intel has lost headwinds.

AMD/RGT has been on TSMC 7nm for awhile and have been first in line using and asking wanting their fab capacity. Knowing Nvidia's larger market they must supply; they're the one's hoping for "capacity", and now are coming in late asking to get "in on the action". Nvidia are the ones signaling to perhaps go to Samsung, and I'm sure TSMC isn't liking their remarks of thinking of going/giving production to Samsung, while then there's the fact Nvidia didn't get on board for their 7nm production first.

Most of what we consider is baked and going to get today, got into the cakes' batter year(s) ago .
Like @Super XP hinted, the world doesn't revolve around x86, there's a lot of ARM chips competing for fabs as well.
But again, AMD guys are not dumb, if I can foresee this, you can bet they have foreseen it years ago. Short of some last minute surprises, they'll deliver on time.
Posted on Reply
#77
Casecutter
Super XPRadeon Technology Group (RTG).

AMD will also be taking all of Apple's 7nm share to eventually make AMD the largest customer acquiring more than half the TSMC capacity. AMD needs the capacity for Next Generation Gaming Consoles, RDNA2 GPUs and ZEN3 all coming in 2020.

As for Samsung, they've been having issues with there 7nm process node last I checked.
RTG... correct my fingers mix that up more often than I'd like.
If AMD has half, I'd almost think Nvidia need to get stuff from someplace..
bugLike @Super XP hinted, the world doesn't revolve around x86, there's a lot of ARM chips competing for fabs as well.
But again, AMD guys are not dumb, if I can foresee this, you can bet they have foreseen it years ago. Short of some last minute surprises, they'll deliver on time.

I think they'll hit the release road maps, just how much the can flood in the channel would be the consideration.
Posted on Reply
#79
InVasMani
I'd like to see AMD resign it's PCB to shorten the GPU traces to the card slot by angling the chip in the same diagonal manner Asrock took with it's quad raid M.2 PCIe slotted card. That would reduce latency, but also more PCB real estate more flexible in turn to fit more memory chips perhaps or dual/triple/quad gpu chips on a single PCB. I'd probably only do dual and triple chip GPU's for PCIe 4.0. Another interesting option diagonally angle the GPU chip along with two raided M.2 slots in diagonally a V around the GPU chip. That would be a clever means to utilize a lot of that largely untapped PCIe 4.0 bandwidth and up the selling appeal of AMD based GPU's. That said I'd be just as happy if Nvidia were to do such a thing. It's a good means to utilize a lot of wasted slot bandwidth and shouldn't add much to the cost. I think a lot of people would be willing to pay another $25's for that option on a given GPU especially in SFF situations like ITX/Micro ATX form factors where card slots are more scarce in the first place.
Posted on Reply
#80
Bronan
Nice idea but hard to accomplish, in fact i think your idea will not work.
We know that the first gen RDNA does not perform in all cases as AMD wanted, so the refresh was already posted.
That big navi will be released is confirmed by Lisa Su, in the video from CES says enough.
So its clear that the big navi gpu will get that same refresh of a slightly other version.
We know that they are on the right track with this because even driver updates have sometimes a big impact in performance.
But can ofcourse not clear the flaws in the first RDNA chips.
We can not tell if its very fast or not, untill it has been on sale and can be widely tested by alot of people.
But i think the big navi will be a competitive one for nvidia.
And no i do not care what those reviewers post, i only buy something if millions of people confirm it works well and without flaws, and yes i do not care to wait a few years longer :D
Posted on Reply
Add your own comment
Dec 18th, 2024 13:52 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts