AMD Radeon RX 6000 "Big Navi" RDNA2 Graphics Card Launch Liveblog

[MxPhenom 216]

i could care less about ray tracing. doesn't impress me at all in the games I have seen it in. i would rather turn it off and gain an extra 30-50 frames for the smoothness of high refresh gaming.

I wouldn't care about it at all in a competitive multiplayer game, but in a game that's single player with a good storyline. Id like to have Ray tracing.

 

[Super XP]

All Nvidia has to do is to transfer Ampere to TSMC 7nm euv process because going with Samsung 8nm (aka 10nm) was a shitshow. They'll probably do that in the next 18 months imo.

I've heard RDNA3 is even more impressive over RDNA2 and will see performance gains equal or more as we've seen with ZEN2 to ZEN3. AMD is in a great position at the moment, they just need to hang on and keep pushing forward. They are the only CPU/GPU company that pushes innovation and tech in a competitive way.

 

[petepete]

From 1900 xt to 2900 xt and beyond- I really hope AMD doesn't let me down again with the hype

 

[Khonjel]

BTW can someone tag the news guys? I only remember the names of @btarunr and @Raevenlord. Boy do I have a tasty news piece idea for you.

All (almost) tests are done on a Ryzen 9 5900X system. I say almost cause I saw one slide being tested on Ryzen 5 3600X. AMD gave you numbers for Nvidia and AMD cards. Now you just gotta extrapolate the CPU number.

I know it's trying to hit a pot-shot since the numbers are from 1440p and 4k and most likely GPU-bound but hey why not check it out.

Steve from GamersNexus also confirmed from AMD that normal vs SAM+RAGE Mode benches are also done at same system, likely Ryzen 5900X one.

Only problem currently imo is finding out which slides are which. Slide names are RX-500 something something in the footnotes.

 

[Super XP]

From 1900 xt to 2900 xt and beyond- I really hope AMD doesn't let me down again with the hype

The stakes are far too high for a letdown. Once I heard members of the ZEN3 team assisted with RDNA2 I was confident it would perform well.

 

[MxPhenom 216]

the thing that sucks the most is that AMD is still using 7nm factories ot make ryzen 3000 series and navi 5 series... like seriously, screw it, go all in on this new stuff so it doesn't sell out on day 1...

LOL, TSMC 7nm is being mass produced. Don't worry about it. The volume of 7nm TSMC is able to pump out should be quite adequate. Can't say the same about Samsung's 8nm (10nm+)

 

[randompeep]

That is why, out of the 3 companies, AMD is well known to price its products fairly.

Fairly with the knowledge of people labeling their whole lineup as a better value overall...and people keep that in mind even if the prices are inflated!
Coming at a computer store near you - the guy with IT degree who recommends an Intel laptop because AMD runs too hot...or these who say an AMD FX discounted system is a better value over a pentium curent gen at the same price because *AMD*

 

[BoboOOZ]

Correct me if I'm wrong, but no way AMD would actually make comparable profit from the desktop GPUs as they do with the mobile Ryzens they've released the past years.
As some people say here and there, yeah the console SoC's is where the fat cash at

From what I know, some desktop GPU are pretty profitable, the higher tier ones ( 5700 and up), the margins are higher than with console APUs. Console APUs are a volume business, which they are contractually bound to ensure for many years to come.

However, where the most money is for AMD is EPYC chiplets, that is probably why the 5500XT are so uncompetitively priced, they make way more money selling 7nm wafers as EPYCS then as 5500 or even 5600 GPUs.

Laptop APU shortage is a symptom of AMD growth pains: their market share is increasing too fast, faster than they can reserve capacity at TSMC, so something's gotta give. They cannot reduce the console chips, they don't want to stop the EPYC business, so they make low-end GPU less attractive and they do not fully meet the demand for laptop APUs. The situation will be better next year, probably.

 

[MxPhenom 216]

So its OK for Nvidia to rip people off with high prices but when AMD prices its GPUs at a fair value its too expensive for you?


I have to respectfully disagree with your comment labeling AMD as equally greedy as Nvidia and Intel. Quite the opposite actually, AMD obviously has to answer to its shareholders and is careful to produce reasonable margins. That is why, out of the 3 companies, AMD is well known to price its products fairly.

To be fair, we don't really know his position on Nvidia pricing...

..So your comment looks dumb.

 

[lynx29]

LOL, TSMC 7nm is being mass produced. Don't worry about it. The volume of 7nm TSMC is able to pump out should be quite adequate. Can't say the same about Samsung's 8nm (10nm+)

if I can't buy 5600x and 6800xt on launch day im going to PM you and be like told you so bruh...

 

[windwhirl]

if I can't buy 5600x and 6800xt on launch day im going to PM you and be like told you so bruh...

That's a demand problem lol

 

[Makaveli]

All I'm concerned with is my AMD Stock share value.

Why do you buy NV gpu's if you are an AMD share holder?

 

[randompeep]

From what I know, some desktop GPU are pretty profitable, the higher tier ones ( 5700 and up), the margins are higher than with console APUs. Console APUs are a volume business, which they are contractually bound to ensure for many years to come.

However, where the most money is for AMD is EPYC chiplets, that is probably why the 5500XT are so uncompetitively priced, they make way more money selling 7nm wafers as EPYCS then as 5500 or even 5600 GPUs.

Laptop APU shortage is a symptom of AMD growth pains: their market share is increasing too fast, faster than they can reserve capacity at TSMC, so something's gotta give. They cannot reduce the console chips, they don't want to stop the EPYC business, so they make low-end GPU less attractive and they do not fully meet the demand for laptop APUs. The situation will be better next year, probably.

Epyc needs continous development amd research over the AI machine's (and/or supercomputer) needs. A console SoC is sketched, built and tested. Then they sell millons.
I'd say their biggest bet is on consoles since the first AMD based gaming device wildly spread.

 

[R0H1T]

if I can't buy 5600x and 6800xt on launch day im going to PM you and be like told you so bruh

I hope you keep your trigger ready & aim for the headshot, as soon as they release the krakens!

P.S. Don't be late

 

[lynx29]

I wouldn't care about it at all in a competitive multiplayer game, but in a game that's single player with a good storyline. Id like to have Ray tracing.

next gen consoles are focusing on ray tracing, nvidia is actually going to be left in the dust since they use proprietary ray tracing, most game developers will be focusing on AMD dx12 RT because consoles are the money makers, and we will get those ports. so nvidia has RT crown now, but in future they wont.

 

[windwhirl]

next gen consoles are focusing on ray tracing, nvidia is actually going to be left in the dust since they use proprietary ray tracing, most game developers will be focusing on AMD dx12 RT because consoles are the money makers, and we will get those ports. so nvidia has RT crown now, but in future they wont.

As I understand it, DX games implement raytracing through DirectX, not through Nvidia's own API (if there's such a thing). Vulkan I don't know though, but likely will have its own extension for it.

 

[BoboOOZ]

Epyc needs continous development amd research over the AI machine's (and/or supercomputer) needs. A console SoC is sketched, built and tested. Then they sell millons.

Oh, Epyc uses the same Zen chiplets, it just uses more of them (and potentially higher bins). But the chiplets themselves have the same architecture and come from the same wafers as zen/threadripper chiplets. That's the genius of it.

 

[Khonjel]

the thing that sucks the most is that AMD is still using 7nm factories ot make ryzen 3000 series and navi 5 series... like seriously, screw it, go all in on this new stuff so it doesn't sell out on day 1...

I know it's old but according to this, since Apple moved to 5nm and Qualcomm mostly using Samsung 8nm now (QC's current hot-selling mid-ranger SD720G and upcoming SD875) AMD should be the biggest customer of TSMC 7nm atm.

 

[TheLostSwede]

All Nvidia has to do is to transfer Ampere to TSMC 7nm euv process because going with Samsung 8nm (aka 10nm) was a shitshow. They'll probably do that in the next 18 months imo.

That's it? I think you have no idea what you're asking and how difficult it would be.
It would be quicker and easier for Nvidia to start a new design specifically for whatever TSMC node they want to go with, than to try and move an already taped out design from Samsung to TSMC.

 

[mtcn77]

I know it's old but according to this, since Apple moved to 5nm and Qualcomm mostly using Samsung 8nm now (QC's current hot-selling mid-ranger SD720G and upcoming SD875) AMD should be the biggest customer of TSMC 7nm atm.

Also, the entire console business model rests on AMD's shoulders, so the industry will do its best to make it work in any way possible.

 

[InVasMani]

About time knew AMD was going to get more and more aggressively competitive in the GPU market, but this is actually feels a touch sooner than I anticipated. I figured we wouldn't see this type of rebound until the next GPU generation in all honesty. Lisa Su for president 2020.

I've heard RDNA3 is even more impressive over RDNA2 and will see performance gains equal or more as we've seen with ZEN2 to ZEN3. AMD is in a great position at the moment, they just need to hang on and keep pushing forward. They are the only CPU/GPU company that pushes innovation and tech in a competitive way.

This makes me so excited about the Xilinix acquisition. FPGA's with infinity cache and Ryzen CPU to compress all the data before sending it to the main OS CPU and loaded into system memory yeah that'll be amazing in a nice PCIE 4.0 or maybe PCIE 5.0 by that point x16 slot with quad M.2 devices in raid-5 and raid-10 if you wanted to raid 8 together with two cards on like a ThreadRipper platform. The future is wild to think about right now. I want to see these RNDA2 cards in crossfire I'm very interested to see if the infinity cache along with PCIE 4.0 can minimize micro stutter significantly especially when you take into account AMD's anti lag as well. These new RNDA2 cards with a Ryzen 5600X are going to be amazing together. It's hard to fathom what will come next for ThreadRipper along with RDNA3 and once they transition to 5nm as well exciting times for the tech world after like what felt like a decade of drought stuck in Intel/Nvidia limbo. I'm really excited to see the fine grain clock gating from AMD that's super important to eek out better efficiency for AMD's designs. Luckily any progress AMD means in that region on CPU or GPU side could be applied to the other in essence which is a huge perk.

The infinity cache is a perfect example of innovation on the CPU side being materialized on the GPU side for the company as well this is exactly the sort of thing people envisioned and hoped to see with the ATI merger so many years ago leveraging the IP strengths from each. It's really going to be great to see what will happen with the combination of FPGA's and it's future APU's along with infinity cache and the I/O die. I think in the not so distant future AMD could do away with the motherboard chipset and probably just incorporate a dual socket design in place of it in some kind of bigLITTLE infinity cache design leveraging both combined for the overall chipset functionality and features

 

[mtcn77]

That's it? I think you have no idea what you're asking and how impossible it would be. You clearly have no clue how chip fabrication works.
It would be quicker and easier for Nvidia to start a new design specifically for whatever TSMC node they want to go with, than to try and move an already taped out design from Samsung to TSMC.

It is great that we have your fine discretion. I cannot think of a better analogy for Nvidia's situation. I had a tough time explaining it.

 

[TheLostSwede]

It is great that we have your fine discretion. I cannot think of a better analogy for Nvidia's situation. I had a tough time explaining it.

Sorry? What are you on about?

 

[mtcn77]

Sorry? What are you on about?

The usual stuff.

RTX 3080 Crash to Desktop Problems Likely Connected to AIB-Designed Capacitor Choice

My advice don't pull someone up for using AIBS and then rant to us telling us we have to use the same abbreviation you just pulled someone up for. You are not the English language police, you can tell me how to do nothing, sir. ... . And I'm English not American. You're responding to someone...

www.techpowerup.com

 

[cueman]

big and so on.,,nope

same as nvidia it rtx 3080,but, rtx 3080 goes ALL games over 60fps for 4K games, except FS 2020,let see real test for rx 6800 and rx 6900

also, i like to see both tested same ryzen 5950 cpu...remembe its 11% faster for gaming than any cpu,single core....11%


summarumm,average,rtx 3080 is 10% faster, belive it,and nvidia gpu have real RT support, and beat easily amd 6000 series for that case...you see? amd hasnone of info RT performance.


also,rtx 3080 ti coming and also rtx 3070 ti..


but i see that 7nm TSMC is better than samsung 8nm, an nvidia new gpus coming with that.


both top 2 gpus maded for 4K gaming and that is moust important battlefield.

...these days many gpu can running 60fps for WQHD resolution, thats not problem..doesent matter if its 124fps or 60fps,.. 60 fps is enough...its price and options.

4K is battle is between rx 6800xt/rx5900xtx vs rtx 3080/3090

who is king? nvidia looks,still



we seen it 18 of november.