AMD Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

[Stu999]

Wow. OK.

Just a bit of fun. Seriously though....

 

[LePhuronn]

Hey, if you're being obnoxious at least you're thinking about the situation, rather than blindly bending over for the big green beast

 

[Xelene]

I have to wonder if the 50% performance per watt is just a median and not a "best case" when it comes to RDNA 2.

My read was it was around 25-30% IPC improvement with a corresponding reduction in power, that would get you a midrange card with 56-60CU (a bit above the xbox, but no integrated CPU) - that clocks like a PS5 on RDNA2; and that feels like a 2080ti level product.

High end looks like it'll be 80CU, but maybe clocked down a little with more memory. That should be 30% over a 2080ti

 

[BigBANGtheory]

High end looks like it'll be 80CU, but maybe clocked down a little with more memory. That should be 30% over a 2080ti

That feels about right maybe a bit more if the clocks are there but its a big die so maybe they have to keep it down.

30% or more over a 2080Ti is putting it ahead of the rumoured 3080 perf and tbh if that is where AMD choose to do battle and disruption that's fine by me but get the price wrong and I'm not looking to line the pockets of AMD shareholders here. Fair price, fair performance in 4k and I'm there with bells and whistles

 

[Participant]

Remember that AMD formally stated that they will disrupt the 4K market with RDNA2.

I don't expect that will turn out to be an empty statement. It can't just mean that 4k will now be just about tolerable with the 5600XT replacement; that's not disruptive and people will protest.

In fact I expect it to mean that they will bring genuinely viable 4K gaming to the mid market, meaning that the 3060 competitor will be 4K capable in the latest games at 'reasonable' settings, even if there is some trickery required to achieve it.

Reasonable I take to be medium settings with no AA.

 

[gpuerrilla]

If the RDNA2 cards are as good / better than the Nvidia cards do you have to be obnoxious to buy one? Do they test you before they let you buy one?

OK Stu.

 

[Calin Banc]

Spoiler: spoiler

WHAT????

Cryengine... Bullet.

Blender... Bullet

@humbug

As per your own examples, no games.

 

[Calin Banc]

It matters a lot whilst there are games that can't run or don't support DLSS, and right now that happens to be pretty every single game bar a handful of exceptions. I'm not against DLSS just that its practical use seems to be a mile away from the hyperbole

True, but hopefully it will catch on or transform somehow, just like Mantle did.

 

[gpuerrilla]

True, but hopefully it will catch on or transform somehow, just like Mantle did.

Mantle was laughed at by most especially the anti-AMD marines. It done a nice job for my FX machine, at least it snowballed into better adoption into Vulkan and DX12. Thing is nvidia dont like to share their toys..

 

[Rroff]

Mantle was laughed at by most especially the anti-AMD marines. It done a nice job for my FX machine, at least it snowballed into better adoption into Vulkan and DX12. Thing is nvidia dont like to share their toys..

It was laughed at because it was inevitable AMD wouldn't put the support behind it to push it into the mainstream and see wider adoption. The folding into Vulkan and the support pushing Vulkan is largely because of... nVidia doing stuff behind the scenes... (I'm not really sure why*). https://developer.nvidia.com/Vulkan


* I'm guessing a lot is motivated by shaping things in their favour.

 

[EastCoastHandle]

It was laughed at because it was inevitable AMD wouldn't put the support behind it to push it into the mainstream and see wider adoption. The folding into Vulkan and the support pushing Vulkan is largely because of... nVidia doing stuff behind the scenes... (I'm not really sure why*). https://developer.nvidia.com/Vulkan


* I'm guessing a lot is motivated by shaping things in their favour.

Roff, that is not what happened. When mantle took traction MS announced DX12. There was some controversy in which some believed MS was sandbagging DX12 until AMD put forth mantle.

In fact, as some of you may recall, an AMD executive publicly stated a year ago that there was no “DirectX 12” on the Microsoft roadmap. Microsoft responded to those comments by affirming that it remained committed to evolving the DirectX standard — and then said nothing more on the topic. Then AMD launched Mantle, with significant support from multiple developers and a bevy of games launching this year — and apparently someone at Microsoft decided to pay attention.

Note: Microsoft has not confirmed “DirectX 12” as an actual brand name for the next-generation of DX technology. It’s used here as the most logical version number.

http://www.extremetech.com/gaming/1...rectx-12-will-imitate-and-destroy-amds-mantle

The issue was that MS was tight lipped on (now called DX12). It had other plans for the API which originally wasn't part of the new os at the time. And, having said it wouldn't be compatible with Win7 found it's way to Win7 only several years later. Also, some believe that DX12 is a copy-paste job of Mantle.

We’ve spoken to several sources with additional information on the topic who have told us that Microsoft’s interest in developing a new API is a recent phenomenon, and that the new DirectX (likely DirectX 12) will substantially duplicate the capabilities of AMD’s Mantle. The two APIs won’t be identical — Microsoft is doing its own implementation — but the end result, for consumers, should be the same: lower CPU overhead and better scaling in modern titles."

http://www.extremetech.com/gaming/1...rectx-12-will-imitate-and-destroy-amds-mantle
Basically added features that are duplicated from Mantle into this unnamed API. So, you cannot trust a word MS tells you.

 

[Satchfanuk]

Mantle was laughed at by most especially the anti-AMD marines. It done a nice job for my FX machine, at least it snowballed into better adoption into Vulkan and DX12. Thing is nvidia dont like to share their toys..

It's not that Nvidia doesn't like to share their toys but rather they create these toys to hamper their competition.

Nvidia are the 'do anything to win' company that took over from Intel as the underhanded shysters of the tech industry

 

[Rroff]

Roff, that is not what happened. When mantle took traction MS announced DX12. There was some controversy in which some believed MS was sandbagging DX12 until AMD put forth mantle.

http://www.extremetech.com/gaming/1...rectx-12-will-imitate-and-destroy-amds-mantle

The issue was that MS was tight lipped on (now called DX12). It had other plans for the API which originally wasn't part of the new os at the time. And, having said it wouldn't be compatible with Win7 found it's way to Win7 only several years later. Also, some believe that DX12 is a copy-paste job of Mantle.



http://www.extremetech.com/gaming/1...rectx-12-will-imitate-and-destroy-amds-mantle
Basically added features that are duplicated from Mantle into this unnamed API. So, you cannot trust a word MS tells you.

Most of that doesn't even relate to what I said - there was like 3 developers and 5 games that actually used Mantle and most of them dropped it pretty quickly after an initial title due to lack of support... "bevy of games and significant support from multiple developers" doesn't hold up to the reality.

Then there was some behind the scenes stuff that saw Mantle folded into Vulkan - which strangely nVidia had a big hand in.

 

[EastCoastHandle]

Most of that doesn't even relate to what I said - there was like 3 developers and 5 games that actually used Mantle and most of them dropped it pretty quickly after an initial title due to lack of support... "bevy of games and significant support from multiple developers" doesn't hold up to the reality.

Then there was some behind the scenes stuff that saw Mantle folded into Vulkan - which strangely nVidia had a big hand in.

Roff, it does relate. Mantle was fairly new back then. By your own words adaption was picking up. And AMD's relation to Khronos had nothing to do with Nvidia! What kind of brain wash talk is this??? LOL

I have no idea what your going on about at this point.
Nvidia was the reason for vulkan, ignoring what now?? Please let us ignore what Johan Andersson and AMD did. Are you serious or are you trolling???

 

[Rroff]

Roff, it does relate. Mantle was fairly new back then. By your own words adaption was picking up. ANd AMD's relation to Khronos had nothing to do with Nvidia? What kind of brain wash talk is this??? LOL
I have no idea what your going on about at this point.
Nvidia was the reason for mantle now?? Are you serious???

You are the one peddling brain washed nonsense here. Much of what you said isn't even related to what I've posted and purely attempts to deflect from the thrust of what I have said.

If you actually read what I said instead of going straight into AMD defence mode you might actually have an idea as to what I'm going on about here:

I never claimed nVidia was the reason for Mantle. nVidia had a big hand behind the scenes in getting Vulkan up and running including the folding of Mantle into what would become Vulkan (hence the amount of documentation on the nVidia side and why it is no accident the amount of nVidia specific feature extensions to it now that AMD doesn't have support for) - ultimately AMD kind of played into their hands there but Mantle would have died a death otherwise anyway.

nVidia literally played a if you can't beat them join them game here and took control of the direction Vulkan was going for their own ends.

 

[EastCoastHandle]

You are the one peddling brain washed nonsense here. Much of what you said isn't even related to what I've posted and purely attempts to deflect from the thrust of what I have said.

If you actually read what I said instead of going straight into AMD defence mode you might actually have an idea as to what I'm going on about here:

I never claimed nVidia was the reason for Mantle. nVidia had a big hand behind the scenes in getting Vulkan up and running including the folding of Mantle into what would become Vulkan (hence the amount of documentation on the nVidia side and why it is no accident the amount of nVidia specific feature extensions to it now that AMD doesn't have support for) - ultimately AMD kind of played into their hands there but Mantle would have died a death otherwise anyway.

I completely understand what you are doing Roff. You are attempting to down play AMD involvement in Vulkan when they presented mantle to Khronos Group. Trying to up-play Nvidia's second hand involvment in Vulkan after AMD's collaboration with Khornos based on some tin foil theory of yours will get called out.

Neither Mantle nor Vulkan excluded Nvidia from participating. So there is no magic receipt from Nvidia. It's only nvidia that close source every API they get their hands on. If nvidia had their way they would have locked out AMD from ever using it again. Then lock down the api as their own, a repeat in history for them.

That is what could have happened if Nvidia was that "involved".

 

[Rroff]

I completely understand what you are doing Roff. You are attempting to down play AMD involvement in Vulkan when they presented mantle to Khronos Group. Trying to up-play Nvidia's second hand involvment in Vulkan after AMD's collaboration with Khornos based on some tin foil theory of yours will get called out.

Actually no - I was only clarify the nature of people laughing at Mantle - which wasn't about the merits of Mantle but that people knew AMD would quickly fall back on the efforts to support it resulting in it going nowhere in the longer run.

The rest is just commentary on where things have gone since - which is quite factual - hence the difference in documentation for Vulkan and why it tends to be used for things like Quake 2 RTX.

 

[doody]

Eastcosthandle 1 / Rroff 0

 

[gpuerrilla]

Roff, that is not what happened. When mantle took traction MS announced DX12. There was some controversy in which some believed MS was sandbagging DX12 until AMD put forth mantle.

I remember it well, thought they definitely sandbagged that one - possibly because they were going to play a trump card only to realise they were beaten to it. Anyhow it was a decent thing and proved you can squeeze much more performance if you get away from the bloat.

It's not that Nvidia doesn't like to share their toys but rather they create these toys to hamper their competition.

Nvidia are the 'do anything to win' company that took over from Intel as the underhanded shysters of the tech industry

Correct. This is the only reason why I refrain from just buying brand green every time I am wanting an upgrade. These days its - right what can I afford? Then check out the two brands and go for what I think is reasonable at the time.

 

[Rroff]

Neither Mantle nor Vulkan excluded Nvidia from participating. So there is no magic receipt from Nvidia. It's only nvidia that close source every API they get their hands on. If nvidia had their way they would have locked out AMD from ever using it again. Then lock down the api as their own, a repeat in history for them.

That is what could have happened if Nvidia was that "involved".

This is kind of aside yet again to what I'm talking about but this is likely largely nVidia's intentions with getting involved with Vulkan - despite an initial natural advantage to AMD's GCN architecture the API increasingly favours nVidia in feature set and performance - quite a few areas with newer features currently AMD doesn't even have a render path for.

Given the nature of the Khronos group and Open GL there are limits to how far nVidia can push it or no doubt it would be locked down by now :s

Anyhow it was a decent thing and proved you can squeeze much more performance if you get away from the bloat.

Problem is - both Mantle [Vulkan] and DX12 are (and hence the relatively poor adoption) a misunderstanding of what most developers look for from a rendering API - most want a fairly abstracted API but with more ability to dip into the inner workings when they need to rather than as with those two APIs where you are forced to do a lot of low level stuff from scratch including a lot of memory management that most developers don't want to go to the hassle of and don't take much advantage from.