Fresh reports suggest that Nvidia will only launch 8 nm GPUs this year, with a possible 7 nm "Super" refresh scheduled for 1H 2021. This could allow AMD to steal the GPU performance crown with 7 nm Big Navi GPUs this year and jump to 5 nm next year.
https://www.notebookcheck.net/More-evidence-points-towards-AMD-s-RDNA2-Big-Navi-superiority-over-Nvidia-s-high-end-Ampere-RTX-3000-GPUs.482368.0.html
Would be nice if this is close to accurate, as the high end segment is in serious need of competition, but until someone posts leaked benchmarks..
Keep in mind that AMD's current top end parts only roughly equal a 1080ti that was released 3 years ago.
Title says more evidence yet all I read was a bunch of unsubstantiated rumours. Why isn't this tagged as rumour?
when looking at the numbers, this makes sense, but numbers aren't always accurate. But, they're usually not far off and it could be more pronounced then whats stated, in either direction........ I for one as a huge AMD fan, hope its a massive smack down. As a Canadian im proud of AMD's video card division from Ottawa. they've always had product i've loved, as multimedia has always been the first thing ive used my PC for.
So if the gaming cards are on 8nm, the workstation cards will be on 8nm, unless there's a notable twist in which Nvidia starts launching workstation cards from HPC cards??
Riiiiight. I'll believe it when I see it.
AMD are always pumping up their new products before they come out, but I literally cannot recall when they've actually delivered on any of their performance promises so I won't be holding my breath.
Agree with comment above - this is not evidence, it's a big ol' pile of rumour in a environment where we know information is often completely falsified.
I've used NVIDIA and Intel combos my whole life - only recently changed that with Ryzen 7.
So definitely not an AMD fanboy but as a consumer i'd love to see a real challenge to NVIDIA.
AMD has had nanometer advantage for quite sometime now, and that didn't really do much for them, except some minor efficiency advantage over their previous product line.
Miracles can happen, but they usually don't. AMD struck an unprecedented blow to Intel with the Rizen lineup. Expecting them to do the same with their GPU division isn't just unrealistic, it is unfair.
Let us not forget AMD's own claim of targeting a "2080 Super" level of performance with their upcoming products. Unsubstantiated hype has hurt AMD in the past, and it will again. So, errr... just take it easy, and wait for the actual product launch.
Being an Nvidia users for more than 5 years, currently using a 1080 ti, I say it's about time AMD show them who's the boss moving forward, just like how they beat Intel with their CPU lineups.
This is fine and all but DLSS proves to be a gamechanger, AMD needs to come with a solution of their own to combat it.
Quote from: George Agrianitis on July 22, 2020, 07:20:43
This is fine and all but DLSS proves to be a gamechanger, AMD needs to come with a solution of their own to combat it.
FidelityFX CAS anyone? performs exactly like dlss 2.0 quality mode, but has slightly sharper images and is easier to implement. check death stranding on PC, you can compare both there. DLSS is impressive, no question, but the blind fanboying over it has gotten out of hand. It's also funny when people talk like if it was a special hardware feature, you only need a GPU capable of doing fp16 intructions :)
Quote from: CptCosmic on July 22, 2020, 08:07:01
FidelityFX CAS anyone? performs exactly like dlss 2.0 quality mode, but has slightly sharper images and is easier to implement. check death stranding on PC, you can compare both there. DLSS is impressive, no question, but the blind fanboying over it has gotten out of hand. It's also funny when people talk like if it was a special hardware feature, you only need a GPU capable of doing fp16 intructions :)
FidelityFX is nowhere near the quality of DLSS2.0.
FidelityFX CAS
www.hardwareluxx.de/images/cdn01/252D5CC901F049A1AAB5CA7E4D6A50DB/img/711AE36F4CE94DF6935ADF1CCF14D8D5/Death-Stranding-11-DLSS-2CAS_711AE36F4CE94DF6935ADF1CCF14D8D5.jpg
DLSS2.0
www.hardwareluxx.de/images/cdn01/252D5CC901F049A1AAB5CA7E4D6A50DB/img/D1EB5B7367164015A1D6596D3D0E3882/Death-Stranding-11-DLSS-1On_D1EB5B7367164015A1D6596D3D0E3882.jpg
Quote from: CptCosmic on July 22, 2020, 08:07:01
Quote from: George Agrianitis on July 22, 2020, 07:20:43
This is fine and all but DLSS proves to be a gamechanger, AMD needs to come with a solution of their own to combat it.
FidelityFX CAS anyone? performs exactly like dlss 2.0 quality mode, but has slightly sharper images and is easier to implement.
FidelityFX is nothing more than a bunch of postprocessing filters, aka just another technology shortcut by AMD, like their FreeSync garbage. And no, it doesn't perform "exactly like DLSS 2.0". The latter is a hardware implementation, while the former is merely a cheap trick to reel-in the rubes.
Quote from: CptCosmic on July 22, 2020, 08:07:01
Quote from: George Agrianitis on July 22, 2020, 07:20:43
This is fine and all but DLSS proves to be a gamechanger, AMD needs to come with a solution of their own to combat it.
FidelityFX CAS anyone? performs exactly like dlss 2.0 quality mode, but has slightly sharper images and is easier to implement. check death stranding on PC, you can compare both there. DLSS is impressive, no question, but the blind fanboying over it has gotten out of hand. It's also funny when people talk like if it was a special hardware feature, you only need a GPU capable of doing fp16 intructions :)
Does anyone else find this comment extremely ironic given he talked about blind faoism but he is actually the blind fanboy himself?
First of all, FidelityFX is not equal to DLSS Quality 2.0. This has been demonstrated by independent sources like digital foundry who compared both. It is also very evident if you actually play the game or compare screenshots.
imgsli. com/MTk0MTQ/0/1
I hope you can see the difference here. Note there are dynamic effects in this game so the sky might not appear identical (the stream looking thing) but you cleary see there's a big difference in clarify.
"It's also funny when people talk like if it was a special hardware feature, you only need a GPU capable of doing fp16 instructions :)"
You can argue that attracting isn't a hardware feature because Pascal can technically run raytracing games. You do realize that Turing has special cores which allows it to run DLSS without much if any performance hit while if you simply "do fp16 instructions" you would be allocating part of your gpu to do those functions right?
Your post is pretty hillarious though. I am not really sure if you believe what you say or are simply trolling. Regardless of your intent, your humour made my dad.
This article is total clickbait and ill-informed.
It has already been established Nvidia is way ahead on the 5nm game. In fact, it is so far ahead that AMD may not even have a fighting chance for up to two years because Nvidia has purchased almost all 5nm foundry fabrication in advance, globally (not even just at one or two) leaving AMD, literally quite physically, no way to produce 5nm other than on a few handful of ultra premium cards.
In addition, Nvidia's 12/14nm was exceptionally competitive with AMD's 7nm while so Nvidia going to 7nm, even if AMD did try to move to 5nm would still leave Nvidia with the crown unless some truly groundbreaking technology improvements occur on AMD's end, which is well established to be incredibly unlikely.
In addition, current info points at Nvidia Ampere already being a serious performance problem for RDNA2.
At least do some basic research before you post these silly articles.
Quote from: CptCosmic on July 22, 2020, 08:07:01
Quote from: George Agrianitis on July 22, 2020, 07:20:43
This is fine and all but DLSS proves to be a gamechanger, AMD needs to come with a solution of their own to combat it.
FidelityFX CAS anyone? performs exactly like dlss 2.0 quality mode, but has slightly sharper images and is easier to implement. check death stranding on PC, you can compare both there. DLSS is impressive, no question, but the blind fanboying over it has gotten out of hand. It's also funny when people talk like if it was a special hardware feature, you only need a GPU capable of doing fp16 intructions :)
You're comparing apples with oranges, you clearly don't understand what the fuzz about DLSS is a about, it's far more than aliasing, it scales low resolution up to high resolution using AI and without major quality loss, which means you get enormous frame rate boosts as numerous videos have shown already. You clearly dont know anything about it. This is not fanboy talk, the videos prove it how amazing this new technology is, period!
Quote from: Kino on July 23, 2020, 21:58:22
This article is total clickbait and ill-informed.
It has already been established Nvidia is way ahead on the 5nm game. In fact, it is so far ahead that AMD may not even have a fighting chance for up to two years because Nvidia has purchased almost all 5nm foundry fabrication in advance, globally (not even just at one or two) leaving AMD, literally quite physically, no way to produce 5nm other than on a few handful of ultra premium cards.
In addition, Nvidia's 12/14nm was exceptionally competitive with AMD's 7nm while so Nvidia going to 7nm, even if AMD did try to move to 5nm would still leave Nvidia with the crown unless some truly groundbreaking technology improvements occur on AMD's end, which is well established to be incredibly unlikely.
In addition, current info points at Nvidia Ampere already being a serious performance problem for RDNA2.
At least do some basic research before you post these silly articles.
Lmao. It seems like they only thing that you've done is basic research.
Way ahead with 5nm? Just because the started buying capacity at TSMC does not mean that they are really ahead. AMD has a very strong partnership with TSMC, and I am sure that AMD will have the 5nm capacity that they need when it comes time. Nvidia failed secure any of TSMC's 7nm capacity because they thought they could pay a premium for
it, and they were wrong. It is highly likely that Nvidia's Ampere will be stuck on Samsung's 8nm, until TSMC has available 7nm capacity. Node for node, Samsung's 8nm is insufficient to counter TSMC's 7nm + or 7nm EUV (one of which is likely end up in RDNA 2). Nvidia has billions of dollars and a huge work force that they can tap into, AMD's Radeon division is microscopic by comparison, and it was not until recently that AMD had the funds to actually compete with Nvidia. AMD cut into Intel's thick skin and could very well do the same to Nvidia in the next couple of years.