AMD Puts Radeon Vega and Polaris GPUs on a Slower Driver Update Track

[TheinsanegamerN]

Yes, it's about the architecture not the release date of the product or if it's still being sold.

You can find cards that are "new" that are even older than 7 years from unsold stock, should someone expect support if they buy a GT210 or something ? Come on dude.

Pro tip: if nvidia released a GT210 as a GT1650 then cut drivers the same year, they would be ROASTED for it.

Guess what AMD is doing? The ryzen 7730 APUs, released THIS YEAR, have vega iGPUs in them.

https://www.amd.com/en/newsroom/press-releases/2023-1-4-amd-extends-its-leadership-with-the-introduction-o.html

Come on dude. You know that these two situations are not the same. If they are, then you should be hammering AMD for releasing a product and dropping it within a single calendar year.

 

[remekra]

It's obvious, the cards are old, and might be not able to run the latest games. But limiting support, is just another implication of abandoning them. As these never had enough support in the first place.

AMD is not the underdog with 3.5B anymore. You should wake up. AMD is $67.967B, and"As of November 2023 AMD has a market cap of $183.50 Billion."(Intel is $159.87 B). They are premium brand. They should not slow down stagnating divisions. They should support each sold product, if it's still sells.
Just look how they invented the "scammy wheel" of the model naming, just to sell the old Vega APU under new disguise with fancy high numbers.

Ok then what is the long outstanding issue/bug with Vega or Polaris that is reproducible and is affecting customers so that slowing down the driver release is actually going to be hurtful to them?
Do you have some info on what changes for example nvidia makes with every driver release that you can still install on old cards? What improvements can a 970 owner get when he installs newest "game ready" driver?
I work for a software company, and slapping a new number on a release and recompiling it is easy, then you just add "General stability improvements" in the release notes and you are golden.
From that point of view then yeah they should still do it. Include all those old cards in the new release, don't make any changes but keep "support".
Doesn't matter if it's an APU, old GPU, newer GPU that was re released, if it's the same driver branch that does not need a new driver every month then it should be put on longer dev cycles.
From a realistic point of view putting them on slower release calendar simply makes sense. If there will be new bugs or security threats they can still fix those when they arise.

 

[Shihab]

If I was AMD, I'd just keep the "support" as is, by only keeping those older cards listed in installer's manifest without doing anything to the actual code wrt said cards, except when actually necessary, then ship.

Everyone obviously don't care about the fact those old cards don't gain anything, so just give them their placebo. </s>

Not like anyone (but us in global south) cares about downlod sizes these days anyway...

 

[Lew Zealand]

All the typical nuanced arguments seem alive and well in this thread with people claiming "dropped" driver support when the statement is:

"Going forward, AMD is providing critical updates for Polaris- and Vega-based products via a separate driver package, including important security and functionality updates as available."

So still getting updates, support not "dropped". Not at the same rate as RDNA but getting them.

Vega APUs are supported but won't get gaming software optimizations because Vega APU gaming is done by how many people? And would a 20% performance improvement in a new game like Alan Wake II from 16 to 19fps at 720p Lowest actually mean anything? Of course not.

 

[Makaveli]

Oh, I totally understand that. The issue I have is that the cards are still available to buy new today. The AMD card with most users in the Steam hardware survey (yeah, I know), is the RX 580.

ASRock Phantom Gaming D Radeon RX 580 Video Card RX580 8G OC - Newegg.com

Buy ASRock Phantom Gaming D Radeon RX 580 8GB GDDR5 PCI Express 3.0 x16 Video Card RX580 8G OC with fast shipping and top-rated customer service. Once you know, you Newegg!

www.newegg.com

AMD does this over and over. The most recent example was just about a year ago when they stopped all driver updates except for RDNA3 for almost three months. Did the world end? No, but it was enough for me to say no more AMD on Windows until they prove they're committed and staffed appropriately to support the products they sell. I guess they still won because I ended up with an RDNA3 card anyway. Just not for Windows.

Interesting that newegg still carries them I've not seen any polaris or vega cards for sale at the local store I shop at for years now.

 

[RayneYoruka]

Not old in the strict sense, but old in a sense that drivers weren't that good back then. Heck, my 5700 XT had lots of driver issues (among others). The RX 6000 series, though, is rock solid.

Yeah, the difference between 5000 vs 6000 is like day and night

Glad to see my Beloved isn't EoL, yet.

Still, RdN.ID drivers have been lightyears better for these 'legacy' cards (and no VAC bans ).
More-recent 'official' support gives more for the RdN.ID team to work with, too


Speaking from personal experience: Old Vega 64s don't seem to have 'aged well' in a hardware-sense.
Judging from old ActualHardcoreOverclocking vids*, I'd guess it's thermal/current damage over time, and/or TIM 'failure'. (I've seen both TIM 'pump-out' and 'cooking' occur on Vega 10 cards.)
*even a seemingly well-(air)-cooled Vega 64 will be (nearly)constantly 'throttling' to stay stable and thermally safe.

My (WC'd by prev. owner) V64 is a Ball of Yarn of 'issues'. It's temperamental, to say the least; isn't even remotely stable w/o being on the 'high power' vBIOS, and does not 'clock' itself correctly. It often BSODs, crashes, etc. Downclocking it 'helped' but I got fed up with it.
Note: My R9 290 cooked its RAM, my 290X was on/off 'blackscreening' and my RX 580 prev. my Vega 64 and MI25s, was also having sleep and blackscreening issues. Sans the cooked 290, AMD has had 'issues' w/ drivers; however, it's clearly not a software-side-only problem.

My MI25s have had no problems like that (that I haven't directly caused).

Not gonna say there's something 'wrong' with your build, but I can say I've had some 'mirror-like' experiences with Vega 64, that I have not had with other Vega 10-equipped cards.

Totally, my Vega wasn't without it's issues in that sense, suspend, random crashes, driver timouts etc. I've had R7 and temporary R9 200 series and they had some of their issues too.. A friend of mine, their R9 had the same issue as yours, the Vram got crispy even after cooling properly the VRM with extra heatsinks.

My vega was a first revission, it didn't have the "filling" in between the dies, okay that was totally fine.
My VRM got cooked to 95c unless I cranked up the fan speed to 100%.. Guess what?! the first batches of the Strix vega 64 used a 2.8mm thermalpad instead of a 3mm one so it wouldn't get compressed enough to have a good heat transfer. After I replaced the thermalpad it was fine, it will stay barely 10c hotter than the gpu and the HBM, I even had it with 1075mv at the highest voltage.. regardless of minimal, non minimal driver installs, different versions of windows used, LTSB then Pro etc even with the pro drivers, I will still have the same issues. My way around this to do the encoding was to offload everything to a secondary pc, which it consisted of a gt 1030 and an intel igpu, that was my only way, making me feel that I wans't able to play almost any VR and that my investment was a total waste.

Can't blame always the "build" when you start counting that I almost rebuilt my whole pc twice.

Sad to say, because other than that I really enjoyed the vega 64 as a whole, until the issues became something to deal every day, the crashes while streaming, while doing rendering on premiere pro/ vegas pro and other progreamms. I just couldn't bear it so I pulled the trigger to pick a pair of 3080s and I haven't had any issues aside from the random "I run out of vram" or I have 10 programms using the gpu. Other than that which now I got used to how much I can run, I pretty much don't have any issues.

Of course we can call all of this previous messages drama as "someone" didn't even bother reading my bold text to not do it. I wonder why.

 

[NC37]

AMD really screwed up not updating the APUs from Vega to something else. What happens when you are trying to keep profits high. They also milked the Polaris line for far too long. There is probably some big reasons why they want to outright drop these from the drivers. Likely some conflicts that the software engineers say would be better resolved if they dropped this. It isn't like there has never been a moment where they ran different driver sets. Used to be they split drivers for mobile and desktop. Before that each GPU often had their own.

 

[RayneYoruka]

Also, without stating what the nature of the problem was. It's like saying that an extra £1,000 a month would solve all my problems. While the extra cash would certainly help, that statement is incomplete without looking into my finances first. You can't say "driver problem" without looking into what the problem actually is.

There has been some issues detailed but they have been ignored, not in the mood to waste more time with that person anyway. Being dismissive seems to be the default if "you don't experience something" it must be not real.

 

[R-T-B]

Drivers where not even needed if game devs would talk straight to the hardware rather through drivers. But that's a different discussion.

This done today would make most game devs just outright quit, if not all of them.

 

[Craptacular]

That's a REALLY bad argument to use for AMD. Historically, they have been the worst offender of leaving hardware with buggy old drivers.

That doesnt mean that bugs in new software are going to be any slower then they were 6-7 years ago, but it DOES mean that your brand new ryzen 7000 that is actually a rebranded 5000 is going to potentially wait months if not years to get bugfixes that other 7000 series users will get fixed in weeks or days. This, of course, will feed into the story that AMD has terrible drivers because AMD is producing a class system in their userbase.

What do you mean it is a bad argument to use for AMD? It is nothing more than me pointing out that you are fallaciously conflating length of driver support with quality of drivers. They are not the same thing. One cannot claim that because one company releases news drivers for a long time that the drivers are not buggy.

Vega integrated graphics on those 15-watt mobile CPUs that were released last year. Their primary use case is to handle hardware accelerated decoding support of codecs as well as display output support, that is it. It is not for productivity nor for gaming. So, when we talk about bugs what bugs are there right now for the hardware accelerated decoding that they support? How about display output on the Vega architecture?

Look at Nvidia that is still selling the Geforce GT 710 and 730. If Nvidia was to announce that they are ending support and the sale of those cards next year, would you say that even after already nine years of drivers support that the customers that bought those cards this year have been ripped off in terms of drive support? I wouldn't.

In this case, AMD isn't even ending driver support for these products, it is just going to be at a slower cadence, which is perfectly fine.

 

[EatingDirt]

Oh, I totally understand that. The issue I have is that the cards are still available to buy new today. The AMD card with most users in the Steam hardware survey (yeah, I know), is the RX 580.

ASRock Phantom Gaming D Radeon RX 580 Video Card RX580 8G OC - Newegg.com

Buy ASRock Phantom Gaming D Radeon RX 580 8GB GDDR5 PCI Express 3.0 x16 Video Card RX580 8G OC with fast shipping and top-rated customer service. Once you know, you Newegg!

www.newegg.com

AMD does this over and over. The most recent example was just about a year ago when they stopped all driver updates except for RDNA3 for almost three months. Did the world end? No, but it was enough for me to say no more AMD on Windows until they prove they're committed and staffed appropriately to support the products they sell. I guess they still won because I ended up with an RDNA3 card anyway. Just not for Windows.

They're still being sold new, but they're most definitely not being produced. Polaris cards are way past their EOL. This is all just spillover stock from the mining boom.

This isn't even an announcement of them no longer making drivers for them, they're just not making them frequently, which is not a big deal, because you're not going to have an enjoyable experience on AAA games, which is much of the focus of drivers these days. They also, at this point have very few issues, so there are very few issues to patch.

 

[WhitetailAni]

You have it backwards, this is an argument for less need of driver updates not more because things don't change that often and when they do it's usually nothing major that completely breaks compatibility.

No it isn't.

Previously, things didn't get extended driver support because there was no point - the hardware just wasn't capable of anything modern. Now, that's no longer the case.

 

[Fluffmeister]

The age old question... when does a "future proof" card go EOL.

/strokes chin

 

[mechtech]

meh I only update my drive once a year anyway lol

 

[AusWolf]

AMD really screwed up not updating the APUs from Vega to something else. What happens when you are trying to keep profits high.

What probably happened (in my opinion) was that the CPU development team had to work with whatever was available at the time. You can't develop a new APU with an unfinished GPU architecture. That's why Zen 4 CPUs have RDNA 2 iGPUs instead of RDNA 3. The CPU and GPU architectures are developed by separate teams.

meh I only update my drive once a year anyway lol

Pretty much the same here.

 

[mechtech]

What do you mean it is a bad argument to use for AMD? It is nothing more than me pointing out that you are fallaciously conflating length of driver support with quality of drivers. They are not the same thing. One cannot claim that because one company releases news drivers for a long time that the drivers are not buggy.

Vega integrated graphics on those 15-watt mobile CPUs that were released last year. Their primary use case is to handle hardware accelerated decoding support of codecs as well as display output support, that is it. It is not for productivity nor for gaming. So, when we talk about bugs what bugs are there right now for the hardware accelerated decoding that they support? How about display output on the Vega architecture?

Look at Nvidia that is still selling the Geforce GT 710 and 730. If Nvidia was to announce that they are ending support and the sale of those cards next year, would you say that even after already nine years of drivers support that the customers that bought those cards this year have been ripped off in terms of drive support? I wouldn't.

In this case, AMD isn't even ending driver support for these products, it is just going to be at a slower cadence, which is perfectly fine.

Ya talk about wrong and that's it lol

 

[mplayerMuPDF]

Come on over to the Linux side guys. Even the r600 driver (used by my Llano and Richland ProBooks) still got some improvements not too long ago... The RADV driver (and radeon/amdgpu kernel drivers) will treat your Polaris and Vega cards as first class citizens and even give them software raytracing support.

 

[tabascosauz]

AMD really screwed up not updating the APUs from Vega to something else. What happens when you are trying to keep profits high. They also milked the Polaris line for far too long. There is probably some big reasons why they want to outright drop these from the drivers. Likely some conflicts that the software engineers say would be better resolved if they dropped this. It isn't like there has never been a moment where they ran different driver sets. Used to be they split drivers for mobile and desktop. Before that each GPU often had their own.

What probably happened (in my opinion) was that the CPU development team had to work with whatever was available at the time. You can't develop a new APU with an unfinished GPU architecture. That's why Zen 4 CPUs have RDNA 2 iGPUs instead of RDNA 3. The CPU and GPU architectures are developed by separate teams.

Pretty much the same here.

They had plenty to work with, neither the tech nor engineers are to blame. APUs haven't properly been reliant on Vega for a long time now. Hello, Rembrandt?

Raphael having 4CU RDNA2 is besides the point - it's the bottom of the barrel, it's there for 2D output, and RDNA2 does everything it needs to. They even intentionally resurrected Zen 2 and RDNA2 in a new design on Rembrandt's node (N6) for the ultra budget market.

They clearly can, they just won't, because they're greedy: Phoenix is to be seen by consumers as a "premium" lineup for premium devices, and even rebranded Rembrandt is seen as being above what midrange devices deserve. Everyone else gets relegated to shitty Vega rebrands for minimal cost to AMD. You can use the supply argument to explain Phoenix (although has there really been any real evidence that Phoenix woes are clearly supply-related?) - you cannot convince me that last year's Rembrandt line on N6P is somehow too "expensive" for mass market until 2024.

 

[Prima.Vera]

I am never buying any other video cards other than Vega ones because of AMD Fluid motion being removed in the newer AMD cards. AMD Fluid motion Is like slice bread, its amazing. Its impossible for me to watch TV Shows, Movies (Blu-ray/4K/Motion2/True Theater) videos in general without having AMD Fluid motion enable. I have a 5000/6000 card, but i probably get rid of them somehow eventually.

I bought a Vega 56 early this year and my next card I am buying soon as an upgrade is a Radeon Pro VII. As for Drivers, I'm using May 2022 preview since i bought the Vega 56 early this year and i have no problems with any games or anything.

Not sure what you are saying, but Fluid motion is supported on all cards, including the recent ones.

AMD Fluid Motion Frames is out now on AMD Radeon™ RX 7000 Series Desktop Graphics

AMD Software: Adrenalin Edition™ Preview Driver brings AMD Fluid Motion Frames for a more fluid gaming experience. At Gamescom, we were able to show you a glimpse of the exciting new technology being created to push our hardware to do more. Today’s tech preview is all about helping you get a...

community.amd.com

 

[A Computer Guy]

Geez I just ordered another 4750G the other day, although it's not for gaming, so I guess I don't have to worry much other than will it run fine in Windows 11 or 12?

 

[AusWolf]

Geez I just ordered another 4750G the other day, although it's not for gaming, so I guess I don't have to worry much other than will it run fine in Windows 11 or 12?

You can always use the drivers that you can now, and it'll still get support for new drivers as well, just not as often as newer architectures do.

 

[DanglingPointer]

Come on over to the Linux side guys. Even the r600 driver (used by my Llano and Richland ProBooks) still got some improvements not too long ago... The RADV driver (and radeon/amdgpu kernel drivers) will treat your Polaris and Vega cards as first class citizens and even give them software raytracing support.

Fun reading all these complaints about AMD drivers! hahaha Well unfortunately ~90% of the sheep are on Windows.

Perhaps some will see the light and move over to Linux. There's a reason 100% of all Super Comps and mainframes run on Linux! Treat your hardware at home like a super-comp and install a Linux Distro of your choice and see your AMD GPUs age like FINE WINE!!!!

I had a R9-290X last me 10 YEARS (liquid cooled)! And it would have continued running if I hadn't decided to upgrade to a 6900XTXH two years back. The 290X ran on Vulkan EVERYTHING you can play today! I'm expecting the same thing from the 6900XTXH (Liquid Devil Ultimate).

I had an old mobile i7 ivybridge laptop running Dolphin Emulator with Vulkan! Vulkan doesn't even exist on Windows for ivybridge!

Ray Tracing is now enabled by default on Mesa 23.2!

The Mesa driver is NOT dependent on AMD developers and $$$$! Have a look at what the Steam Deck can do, and that does it with translation of DXtoVulkan! Imagine native Vulkan!? Have a look at Proton for PCs on Steam.

 

[A Computer Guy]

Fun reading all these complaints about AMD drivers! hahaha Well unfortunately ~90% of the sheep are on Windows.

Perhaps some will see the light and move over to Linux. There's a reason 100% of all Super Comps and mainframes run on Linux! Treat your hardware at home like a super-comp and install a Linux Distro of your choice and see your AMD GPUs age like FINE WINE!!!!

I had a R9-290X last me 10 YEARS (liquid cooled)! And it would have continued running if I hadn't decided to upgrade to a 6900XTXH two years back. The 290X ran on Vulkan EVERYTHING you can play today! I'm expecting the same thing from the 6900XTXH (Liquid Devil Ultimate).

I had an old mobile i7 ivybridge laptop running Dolphin Emulator with Vulkan! Vulkan doesn't even exist on Windows for ivybridge!

Ray Tracing is now enabled by default on Mesa 23.2!

The Mesa driver is NOT dependent on AMD developers and $$$$! Have a look at what the Steam Deck can do, and that does it with translation of DXtoVulkan! Imagine native Vulkan!? Have a look at Proton for PCs on Steam.

Ah so the fine wine is on the Linux side? I was wondering where that came from.

 

[Squared]

It seems like maybe a year ago AMD dropped support for everything older than Polaris, which had me concerned because I had Polaris at the time. So to read now that Polaris is getting fewer updates, instead of simply being unsupported, seems like an improvement. These days Polaris can't really play the kind of games that need launch-day driver updates anyway.

 

[Tomorrow]

What do you mean it is a bad argument to use for AMD? It is nothing more than me pointing out that you are fallaciously conflating length of driver support with quality of drivers. They are not the same thing. One cannot claim that because one company releases news drivers for a long time that the drivers are not buggy.

An argument could be made for the opposite actually. If a company releases frequent driver updates that could indicate that there are many problems that need constant fixes.
Im not saying that's the case for Nvidia or AMD...more like Intel who still has a lot of work to do.