I get the point/criticism, but it's rather pointless to me.

Most if not all AAA games released in the past 4 years have either been Vulkan or supported DX12.

Additionally, RDNA2 performs substantially better in DX11 games like the Witcher 3 and GTA 5 then previous AMD architectures. Which reaffirms what AMD are claiming, that if the game is popular and demand is still there, they will optimise their drivers for older games.

I still hope AMD improve DX11 performance overall for RDNA2, but I hardly feel as an owner of one of these cards that it's a pressing or significant issue. If anything, i'd say it's more of a PR/perception issue for AMD, no doubt they'd love to shake that reputation for having poor drivers and this wouldn't help.

 

There isn't any backwards compatibility issue here, just an observation that AMD tends to do relatively better in DX12 than DX11.

NVIDIA has long had better threaded DX11 drivers, it's just much less of an issue now than it was in prior generations.

Of course, if one is looking to maximize performance in specific titles or classes of titles that tend to favor one manufacturer over the other, one should make sure the benchmarks they are looking at are reflective of those titles.

Additionally, RDNA2 performs substantially better in DX11 games like the Witcher 3 and GTA 5 then previous AMD architectures. Which reaffirms what AMD are claiming, that if the game is popular and demand is still there, they will optimise their drivers for older games.

Plenty of niche/obscure games that AMD isn't likely to be optimizing for that run faster on RDNA2 simply because RDNA2 cards tend to be faster than AMD's prior cards.

The main DX11 game I play is Elite: Dangerous, and my 6800 XT is way faster than my 5700 XT here, simply because it's a much faster card. It's also noticeably faster than my RTX 3080 in this game...because it's completely fill rate limited.

 

I was the guy who bought the Radeon HD5870, one for the first, if not the first DX11 card. I never really used the feature, game and driver maturity lead to use the existing, instead of the cutting edge modes.
Funny how the GFX card companies are trying to perform in existing titles, but are neglecting all the performance from previous cards...

 

Backwards compatibility is why I have a hard time seriously considering AMD cards. I play old games, and I used emulators that are more mature with older APIs (like opengl). I don't want to just hope that AMD does right be me. Nvidia is far from perfect either, but at least they get it right early on so I don't need them to fix it later.

 

Read the article and as someone who mainly plays older games......I don't get it. Who is buying $500 msrp cards to play older titles at the highest frame rates? Performance will be there for older titles regardless of which brand within this segment. Who gives two hoots if I get 154fps vs 165fps?

 

Who gives two hoots if I get 154fps vs 165fps?

The OP does.

 

I use DXVK on my 5700XT in certain games like The Witcher II and F.E.A.R. for compatibility issues, pretty sure it supports DX11 as well.

 

Source: Play Older PC Games? Think Twice About AMD's RDNA 2 Cards



Here we go again another case of backwards compatibility causing problems. I bet we wouldn't have to worry about these archaic titles if operating systems followed Apple by ending support for 32 bit apps. Everyone should switch to Vulkan but the Evil Empire keeps engulfing all the developers that were.

If apple didn't have such a tiny segment of the market they wouldn't be able to make a switch like this and someone who spends as much time worrying about this kind of stuff, you should know that already. It's like every thread you start is an opportunity for you to make sure everyone knows what brands you prefer.

Get over yourself.


I don't see any issue with the benchmarks in the article. For one, they didn't test enough games to draw any meaningful conclusions, they found what like 2 games that run perhaps slightly slower than you would expect that came out in the PS3 era? They also only compared it to a 3070 which is supposed to be a good amount faster than a 6700XT. Even the 360Ti beats the 6700XT in a lot of titles. What I see here more than anything else is the 6700XT actually keeps up with the 3070 more than it should in some of the newer titles they tested which could give the impression that the 6700XT doesn't work as well in older titles, though those benchmarks are the only ones that really line up with the performance differentials one should expect when comparing those two cards. If anything, it looks more like PCmag had an issue with their 3070 performance in newer titles rather than AMD issues with older titles.

This articles didn't do enough research by a long shot to prove or even hardly speculate about anything at all and only someone looking to support their predetermined opinions on the matter would see this as evidence of anything.

 

Even on older cpus? eh eh eh

 

the performance difference is merely academic. anything above 120 pretty much feels the same unless you're playing super competitively

 

What is an older game?
Out of all stuff I play regularly, I’d guess only quake champions would blow, but that game isn’t particularly good on nvidia either. Rest either has vulkan/mantle/dx12 or runs on ogl.

 

With more providers offering subscription services for catalogs of games (Game Pass, EA Play, Uplay (which is actually Ubisoft +)) more gamers will have access to older titles.

However, by and large, PC gamers always had access to older games as PC is an open platform. It just required purchases. Now with all these services, you can simply subscribe and BOOM, you have access to all these older titles.

This is why the topic is a bit more relevant. Sorry, I needed to reiterate the context because I see it's missed from the conversation (even though the article mentions it).

I think most people when purchasing new graphics card are really interested in the future of gaming (DX12 etc.). Also, I do agree with some of the posts above. Does 100 vs 140 fps really matter all that much in Shadow of the Tomb Raider? No because you can simply enable DX12. Why would you OPT for DX11 in a title that has DX12 when you can support DX12 with your fancy new video card?

Now, if we are looking at games exclusively built on DX9, DX10, and DX11, the results would be more interesting but the article doesn't dive into it. Which makes no sense to me. Why not Halo MCC, Mass Effect, or Resident Evil? These are games that are old and are popular on PC.

You want to know why? Because I bet you the 6700XT can already play them at 100 to 300 FPS at 4K. But nah, that doesn't create artificial drama...

 

Now, if we are looking at games exclusively built on DX9, DX10, and DX11, the results would be more interesting but the article doesn't dive into it. Which makes no sense to me. Why not Halo MCC, Mass Effect, or Resident Evil? These are games that are old and are popular on PC.

Because it's a crappy article that shouldn't be taken seriously.

 

This does feel similar to the "8GB is not enough" or "Nvidia runs badly on old CPUs" type article.

It has some core of truth, but at the same time is pretty insignificant compared to the headline. The problem with this one is that we don't even know if it is widely applicable, since it depends on optimizations for particular games it would take a lot of benchmarking of a large Steam library to get a solid dataset.

It is still interesting, but it isn't good ammo in the 'GPU wars'.

 

Exactly. The only real news here is that the OP found a headline that bashes AMD, and keeping his MO intact, he posted it even though it's a weak click bait article.

 

I see this as just another marketing gimmick basically would need much more information. My question is will it run GTA 4 any better? GTA 4 runs like crap no matter what your are using because it was very unoptimized port. I am into older and indi games as of late and I still have a CD of Unreal that will only run on 3DFX's Glide API and basically have to use a old Glide emulator just to even run it.

I have been following some retro gaming communities and I have been getting the itch to build a rig just just for retro PC gaming. I got a few single core CPU's in the parts bin like the Athlon XP, P4 prescott, and the AMD K6-2 but no boards. I wish I had my old 22" CAD CRT which games like resident evil and Doom 3 looked great on

 

GTA IV will never run better I believe at some point Rockstar will revive GTA III, Vice City, San Andres, and IV on PC. I love IV. It's my favourite GTA but damn... it was a pain back in the day.

All my mods are likely lost and its requirement for GWFL was very annoying. I can't recall if they removed that requirement. Maybe with an absurdly overpower CPU/GPU, I can run the game at over 60 FPS...

Make sure to do a build log if you end up proceeding with the retro rig

 

"Rdna 2" are the same memory chips from 5700 series only double the vRam , so were is the huge leep in the tech world, the major "discovery" of theyr sciency ? let me tell you the obvious, the science that they claim to have and big tech jumps are in the guillible people are easily fooled, and they see this and use it for profit, for some years they practice "updates" for improvement, but they improve theyr market profit, i saw how each win10 update slowly chipping performance for years now 1% is not to be noticeble but after 5-10 updates a 10% downgrade again noticeble but not really, then pooof a new hardware appears that is 100% faster, reality is 35-50% faster then the previous one ( previous hardware having latest updates via win10 or hardware drivers, but if u tested tons of mix between w10 updates and a mix of tons of bios/gpu/chipset drivers and get the best performing mix , drawing the line ull see that the new hardware vs previous one is only 20-35% improvement, from the claimed 100% faster as they mouth says, so this improvement majorly is controlled by software and this is happening for quite some years now ) dont get me wrong tech has improved by years passing and they do have top hardwares but not for consumers, its perfectly normal this way, buuut they saw ppl are guillible and they are the ones that dictate in wich direction this goes so many shread theyr humanity and go rabbid after money / power , you can see this happening everywhere but we talk about the rdna2 if u'll be willing to make ur habbit in keeping evidence and testing hardwares with different softwares u'll notice real fast how they force this, saying it cuz last years was hard to point with the finger on what is happening, now they have no shame in doing it and they dont hide it that hard ( its like a norm now and is like time is pressing now on them vs the passing years ) be an example, be a human that respects other humans and theyr rights/intimacy ( until the new generations come and they modify the laws and strip u of every single rights that u dont have )

 

i saw how each win10 update slowly chipping performance for years now 1% is not to be noticeble but after 5-10 updates a 10% downgrade again noticeble but not really, then pooof a new hardware appears that is 100% faster, reality is 35-50% faster then the previous one ( previous hardware having latest updates via win10 or hardware drivers, but if u tested tons of mix between w10 updates and a mix of tons of bios/gpu/chipset drivers and get the best performing mix , drawing the line ull see that the new hardware vs previous one is only 20-35% improvement, from the claimed 100% faster as they mouth says, so this improvement majorly is controlled by software and this is happening for quite some years now

Do you have a data set supporting this theory? It sounds like a mad conspiracy theory to me but if you really had some data to back it up it might be interesting.

How does Linux tie into this? Are the Linux dev in on this conspiracy too? How about Apple? Everyone who writes drivers or works with GPUs at a lower level is adding artificial slowdowns every year until a new card comes out, then they all switch to a new code path without any artificial slowdowns? Then start adding them back in slowly until the next gen comes out?

The organizational abilities of this conspiracy are AMAZING, to think they can get all the Linux devs to act in concert like that.

Yeah, no.

 

Tl;dr Wall-o-text ^^

Edit.....adding

 

Yes paragraphs would've made that post much easier to read rather than stream of consciousness.

 

...

 

I read the article and think it's pointless to test new hardware with older titles and even said they should drop backwards compatibility.

I may not be overly concerned about the exact performance in older titles, as long as it's sufficient, but I definitely expect anything that can be executed on supported OSes to work with the GPUs in question.

Dropping backwards compatibility with standards and APIs that have implicit support on modern hardware would be senseless anyway. Any card fully compliant with D3D12 is automatically compliant all the way back to at least D3D9/SM2.0 and any card capable of running modern OGL apps is complaint with OGL versions back to at least 3.1. That's almost sixteen years in the case of D3D and nearly a dozen for OGL...all supported just by supporting the mandatory features of the newest versions.

 

...

 

Just add me to your block list, you're the only one complaining in this thread. Ever since I posted that article about the RDRAND issue AMD caused that prevented use of Linux and Destiny you've been up my ass. Just add me to your block list. You have some sort of controlling behavior of what you consider "news" I see you all the time harassing people on these forums about it.

Never said Reddit was a news source. I stated if I wanted it to fit some sort of narrative I would have added it to the OP to compliment the article. When hundreds of people complain on Reddit regarding the same issues it's definitely concerning. It's the reason Nvidia RTX 30 launch was first discovered and the AMD USB issue. AMD is already aware of their terrible DX11 performance on RDNA but as mentioned by them they're only going to support popular titles which makes perfect sense.

I bet if Microsoft dropped 32 bit app support we wouldn't even have this discussion.



I was referring to 32bit app support. There are plenty of DX11 titles that run smooth on AMD hardware like Overwatch and Destiny 2 but they're modern popular titles. I'm sure most of the issues RDNA users are complaining about are 32 bit games.

You're welcome to follow your own advise, but a better solution would be for you to just stop posting weak click bait articles like this one, that grasps at straws over ps3 era game performance so you can post something negative about amd.

 

Does anybody actually have a problem running older titles? This is the penthouse of hardware.

 

I personally think Nvidia's DX12 overhead issue is more of a problem. One key reason for this is someone with a decent RTX GPU and a slower GPU, running DLSS will not see a lot of gains because DLSS runs at lower resolutions.

 

Edit: nevermind.

 

Just gonna add to this. This article is 100% factual. As someone who had a 6800, traded up for a 6800 XT and briefly had a 5700xt, rdna/rdna2 cards will have fluctuating utilization and core clocks on dx11 or lower games, especially if they are "less demanding". If you hop on Reddit, you'll see a ton of posts about this problem. The card is a beast in modern games, but does fall behind in older games.

 

Glad to have a RX6800. Nice performance boost from my Rad VII.

 

Yeah, coming from my 1080 ti, the performance gain in dx12/Vulkan titles is stupidly impressive.