I'd love to see a (non-APU) Ryzen CPU with discrete RDNA/RDNA2 GPU with HBM in a similar form factor.
In fact, Navi 12 is meant to be a HBM-using equivalent of Navi 10, but we never got to see it, although linux driver updates still occur mentioning it, and it was meant to be low power despite the large number of CUs.
Sure, it would be 4 chips in a single small package - HBM, GPU, I/O and CPU, but it'd be nice and compact, if it could be pulled off.
I agree it would be interesting but I don't think we'll see it. As of late AMD has been covering their entire product line with a minimum number of different actual chip designs probably to save mask and layout costs. Their CPU has covered everything from 4 core desktop CPUs up to multi-socket server CPUs with 64 cores and their APU has covered laptops and lower end desktops with some overlap with the low end desktop CPUs. This would end up in an awkward middle ground. Not sharing enough in common with the HEDT and Server space to make it economical and too power hungry to be used in the Laptop APU space to make it economical IMO. Maybe if their future GPU line goes a Chiplet route you could set things up so one of those chiplets could be used in this configuration and give us a pretty capable desktop APU with an 8c/16t cpu chiplet, the IO chiplet a GPU chiplet and the HBM for GPU memory.
Exactly. Hades Canyon looked really good. I probably would have pulled the trigger if I weren't already using an alienware alpha. Since then I've been waiting for an upgrade. It could be expensive with HBM and everything but it'll make a really nice halo product, and prove that future is better than the past.
LOL, and how do YOU know other parties are not bidding on console design wins? Are you seriously implying nVidia wouldn't be interested in selling tens of millions of GPU's to Sony\Microsoft? Clearly they are interested in the industry: look at the Switch.
Depends on the margin MS and Sony want to pay. At the time when AMD won all console supply, Nvidia said that the margins were so low that they didn't want to invest so many engineers for that purpose. Yet, as we have seen, console SoC supply is something that has a value beyond the simple money they make. AMD gained the support and optimization for its own architectures while Nvidia more advances features were left behind.
Sony and Microsoft have long switched to semi-custom APUs (no distinct GPU and CPU). Therefore, since Nvidia lack a x86-64 licence, the only way they could bid to make the APU for them is if it was ARM or RISC-V based. Neither of them appears to have any wish to switch to a new ISA, hence Nvidia cannot bid to make the APU for them. Nintendo is different bird.
They didn't switch for any reason but costs and the fact that AMD sold them for a penny. Technically speaking separate CPU and GPU are better but costs more. There's not chance to match the low prices AMD made with their SoC with a discrete combination even thought the latter would be much faster (and easier to cool down). AMD simply killed the competition by price. Something Intel nor Nvidia would like to fight against just for having their devices in the consoles for a penny each.
But as said, while for Intel is not a problem (not a gain, not a loss) for Nvidia the problem is that they have lost the support and optimization for their own architecture. I wish they would supply just one of the two, just to see their advance features supported and see who is really slowing down the gaming evolution with antiquate and obsolete architectures that catch the competition 3 years later. This would really create a competition between the two.
Isn't that a bit chicken-and-egg, though? Imagine a world without AMD (or where AMD went out of business, or where AMD survived but didn't buy ATI, etc.) - it's not like Sony and MS would have just ¯\_(ヅ)_/¯ and stopped making consoles - they'd have chosen something else. Possibly an arch like the original Xbox (which is basically... a PC), possibly have made it worth Nvidia's while to soup up Tegra... custom ARM chip, possibly with a dGPU, or who knows what else.
The time between generations helps a lot in that regard - there's plenty of ARM chips out there that may not be as fast as a Zen 3 design, but would still give a pretty big performance boost over the 2013 Bulldozer-based chips in previous gen consoles.
As amazing as AMD's resiliency in the industry is, from engineering feats to design wins and platform support, their drivers (with a heavy emphasis on the ATi roots here) have always been a weak spot.
Yet part of me agrees with you. It's very, VERY strange for AMD to prioritize such niche developments such as Mac GPU support and even Linux support over a promising partnership with Intel. And that appears to be exactly the case.
I don't think EITHER AMD or Intel thought that Kaby Lake-G was going to be the start of anything long-term. By the time it came out, Intel had to have internally been planning what eventually became Xe, and AMD made no secret of the fact that they were gunning for Intel in as many x86 markets as possible.
What *will* be interesting is what becomes of Nvidia using Epyc CPUs in its Ampere-based DGX server. Nvidia seems to have little interest in making an ARM processor for those systems, and with Intel releasing Xe, that means one way or another, Nvidia will have to rely on a CPU vendor with a competing graphics product if it wants to stick with x86....
In the end, this was a marriage made in hell. Reminds me of the last Intel CPU + AMD GPU laptop I bought, all the way back in 2011: a Sony Vaio SC 13.
Drivers stopped coming almost immediately after I bought the laptop. This was in the weird Win7->Win8 transition days, too, so getting drivers that worked for both OSes? Ha! The few Vaio SC 13 owners met online and discussed out options. Wen were lucky we could inf mod drivers back then in Win7, without incurring too much wrath of the driver signature validation gatekeeper in Win7, so that kept out laptops chugging for a little bit. It didn't take long before that stopped working with newer drivers and we were moribund until someone figured out binary mods (hex editing) to bypass the installation check. That worked for a few more driver releases until the laptop's AMD GPU stopped accepting any newer driver updates, whatsoever. Even inplace file replacements and registration stopped working (after 1 update? iirc). Newer games got glitchy real fast.
Eventually, I sold the laptop on that forum and bought it's newer, higher specced replacement in Sony's lineup: The Vaio S13, powered with Nvidia graphics (and Nvidia Optimus). Despite the experiences I just had, I was still a dyed in the wool ATi fanboi and flat out stupid for it. Let's just say, that laptop received working updates, DIRECT from NVIDIA for the next ~7+ years. Even Intel occasionally released CVE-relegated updates.
Every few years, I still buy an AMD Radeon GPU to test the waters. The last one I bought was the Thinkpad E485. AMD Ryzen CPU, AMD Vega GPU, on the same chip. No more excuses, right, AMD? SMH, AMD made of mess of that, too. AMD whined and blamed OEMs for not doing AMD's job, before finally sobering up and releasing driver updates themselves, as they should have. It only took 6 months before AMD stopped acting like a petulant little child.
Either way, I now am taking another 2-4 year hiatus from any AMD GPU purchases again.
IMHO this is not AMD problem, but the seller overall, as AMD cannot support 3rd party driver sets as they can with their first party design (that is, AMD Radeon version XYZ by all means should be getting proper support and driver sets, whereas something from Lenovo cannot be directly supported via the same chain of drivers as they are not "verbatim" design..different wheels, cogs, power limits etc etc)
Now if AMD made the laptop, was a full AMD branded product and they were to have done similar level of crud, then absolutely get right @#$ about it, I would, though when it comes to an ODM/OEM/AIB whatever you would like to call it, the same rules simply cannot apply, they simply do not.
Look at all mighty Intel, they are very adamant about how they do things, a much much larger company, yet even they do not give the same first party support to even full Intel branded designs
ahh well, get the product, use the product, hope it works properly right off the bat and will for a number of years till it gets replaced..these things are always a very hit and miss thing, even more so when the maker (Lenovo, Sony what have you) want you to buy it, but then they really do not at all support it the way they should be.
I personally hate it, only 2 choices, buy it and deal with the crap that likely will happen at some point, or do not buy at all and never have to deal with it ^.^
IMHO this is not AMD problem, but the seller overall, as AMD cannot support 3rd party driver sets as they can with their first party design
This is a ridiculous justification. Nvidia supports its dGPUs mounted into 3rd party laptop. It's up to them to update the drivers for their own product.
Historically AMD has never been a good supporter of its own technology. They tend to fire a new product and forget quite often. They are quite unreliable as partners and that's probably one of the reasons why no OEM wants to put their GPUs inside laptops. Despite the fact that those GPUs are also technically inferior to those of the competitor which also assures better support for years, not just weeks.
AMD changed that situation with regard to laptop drivers, and laptop manufacturers seem to support the change. I have the latest Radeon drivers on my 2700U laptop (which, granted, has plenty of other weaknesses).
I don't think it's a ridiculous justification - AMD gives the OEM a range of operating perameters, and the OEM chooses to go outside of reference for whatever reason. The stock drivers are for stock designs. I can't imagine AMD making an individual driver variant for every OEM variant of their chip. Case in point, my HP 15" Envy with an AMD Ryzen APU couldn't get the latest AMD drivers because not AMD, but HP blocked the install due to possible issues with incompatability. Luckily AMD changed that just recently.
IIRC, it's also the same with intel drivers for laptops. Last I checked, despite newer drivers for my other laptop's (HP) i7, I couldn't install them because the OEM did not validate the newer drivers as my laptop was cTDP up'd from the OEM, and ergo because it wasn't validated wasn't compatible with the install.
This one I place SQUARELY on Intel as they have an established track record of doing this very same thing in the past:
Witness some of their Atom N series CPUs as implemented in the Acer Aspire One series, integrating Intel Atom cores with an iGPU developed from PowerVR IP. Intel contracted PowerVR to supply an initial, operational driver and for a period to supply bug fixes for that one driver. After that initial period of bug fixes was over, complete radio silence! Intel hasn't done a thing to ever touch those drivers again. While the launch drivers mostly work for Windows 7, anything later is relegated to the windows default svga driver, which lacks almost any real acceleration.
PowerVR, when contacted directly, points you to Intel as they own the product and are responsible for its support. Intel points you to PowerVR, stating that the IP in the chip belongs to PowerVR and that, without the contract, they can't (won't) touch it. The core of the issue is that Intel chose to do the absolute bare minimum to get the chip out the door while it was being produced, and then immediately dropped it like a sack of burning animal excrement (which, to be fair, it is). Because they chose not to continue to pay PowerVR to produce driver updates, like, for example, Windows 8, which was released while the laptops that contained the chip were still in retail, PowerVR wasn't going to be a charity for Intel and produce them for free.
What I find most disappointing about this when I look at AMD is that, the chip itself hasn't changed. Its based on a slightly modified version of Polaris, which is still receiving driver updates on the regular. At the VERY least, they could just include a final driver for it in subsequent releases and have it in the release notes that support for that product is feature and maintenance frozen at a certain date. However, there MAY be wording, somewhere deep in the contracts for the driver package, that specifically forbids AMD from doing so in a "planned obsolescence" sort of strategy.
I bet certain OEMs are a bit upset about this though. I believe that some of those processors went into "long life" corporate SKUs that are required to have support for X number of years. That could throw a wrinkle into this one...
Dell is still actively selling the XPS 15 2-in-1 with this in it. I'm sure they aren't moving a ton of those units but I'm sure they aren't happy about it.
I know this is as much on Intel as it's on AMD, but: AMD's graphics driver situation is a trainwreck. Always has been, and likely every will be. At least they've started supporting Ryzen Mobile with drivers a while ago, after not caring for the first year of its existence. But as far as I am concerned, I've regretted buying AMD graphics every single time I did it, all the way back to when it was still ATI before the acquisition. And it was never a hardware issue, the hardware is awesome, has great performance and price/performance ratio, but the drivers are trash :(
Despite their well earned success in recent years this kind of behavior has repeatedly put me off their products. Customers don't get to find out for a couple of years after they have laid down their money whether AMD will deem their product worthy of support. And this has happened repeatedly over many decades and is disassociated with the current fiscal success or not of AMD the company. Fact is you are almost always taking a chance that after some period of time in which you, the consumer are still happily using the product, that it will go unsupported. So AMD can happily wave all those performance numbers they want. But we already have seen supported/unsupported issues with what their chipsets will support for cpu's. Sometimes driver fixes take the better part of a year to fix. And all that puts a serious damper on performance advantages or price advantages when a consumer comes to buy. If they really want to be at the top of the market they have to do much more than simply offer a performing product. Support problems like this need to stop, entirely. You ruin customer trust every time you stop something. And that will, long term, keep AMD from market positions they could have simply because of their own failure.
That's great, except it's ultimately an Intel product. They sold it, so they (or as people bought it as part of a system, the manufacturer of the product it's in) are responsible for support, just like Microsoft is responsible for support of the Xbox which has both AMD graphics *and* CPU cores. According to PCWorld Intel are supporting it five years from release. Chips were sold on that basis. The fact that they're discontinuing sales at the end of the month shouldn't change that - and as a practical matter they've released drivers for HD4000 graphics components all the way to this year.
I don't think this is an Intel thing. Tht BootCamp drivers for my MacBook Pro 16" haven't been updated since it was introduced last November, and AMD still has thesw old drivers on their website. AMD loves to score these design wins for the janky platforms and then ignore them. AMD just isn't good at drivers.
I can see Intel not wanting to deal with it - they are not going to develop drivers for AMD's GPU - but AMD not making something available for a product they sold is bad - wouldn't take much for the Night Janitor / Head of AMD Software to make a few changes for the sake of their customers - and then put it on EOL time line and forget about it
Also, as some comments here point out, this picks at a barely healed scab with AMD GPUs: driver support. As much as I sometimes dislike NVIDIA, but their drivers tend to not suck as badly as team Red's, and they do update them also for older GPUs. An important factor when spending big bucks on a dGPU, not all of us have the means to buy a new card every 2-3 years.
I have an XPS 15 2in1. You used to be able to install the Dell/Intel driver, and then update the driver with the latest from AMD through device manager, but this quit working at some point last. year. A bit of a non issue now though as the last firmware update from Dell that disabled under voting the CPU bricked it.
I blame all 3 companies equally. The fact you could manually load the AMD default driver and it worked fine shows it's not some insurmountable technical problem, but just a don't give a shit on Intel/AMD's part. I would imagine Dell has enough sway with both companies to put pressure on them to get off their ass as well but don't care enough either. As the above comment mentioned, Bootcamp drivers from AMD are also an afterthought. If the one guy at bootcampdrivers can tweak and modify the generic AMD drivers to get them working, how much effort would it really take for AMD to step up and do it as well?
I guess I understand though. All the money and effort spent on marketing to try and convince everyone that the small performance updates this gen over last are worth it can't leave much left for actual support for the products they already sold.
Could it be that only stable driver versions will be made to support it? The latest Adrenanalin 2020 Edition 20.5.1 is labeled "optional", as is the case for most AMD graphics driver releases.
"Still, I would have expected that driver support is something that would have been hammered out in a contract early on – such that AMD was committed to deliver and paid for the necessary 5 years of drivers – rather than the current situation of Intel and AMD seemingly dancing around the issue."
In the meanwhile they both frack every single Kaby Lake-G buyer, which is utterly disgraceful and unprofessional. I have no idea how much this part sold. Perhaps a class action suit against both Intel and AMD will make them "clarify" their contract clause about driver support?
As someone who owns TWO Hades Canyon NUC's, may I ask:
HOW THE HELL DOES A PRODUCT INTEL IS _STILL_SELLING_ LOSE DRIVER SUPPORT ALREADY!?!
To think I actually bought NUC's because I thought Intel's support and updates would be better than what I get from other DIY motherboard manufacturers.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
46 Comments
Back to Article
EdgeOfDetroit - Monday, June 8, 2020 - link
Weren't we first told about this by [H]ardOCP?DigitalFreak - Monday, June 8, 2020 - link
I miss that site.ND40oz - Wednesday, June 10, 2020 - link
https://www.thefpsreview.com/ has most of their reviewers.psychobriggsy - Monday, June 8, 2020 - link
I'd love to see a (non-APU) Ryzen CPU with discrete RDNA/RDNA2 GPU with HBM in a similar form factor.In fact, Navi 12 is meant to be a HBM-using equivalent of Navi 10, but we never got to see it, although linux driver updates still occur mentioning it, and it was meant to be low power despite the large number of CUs.
Sure, it would be 4 chips in a single small package - HBM, GPU, I/O and CPU, but it'd be nice and compact, if it could be pulled off.
kpb321 - Monday, June 8, 2020 - link
I agree it would be interesting but I don't think we'll see it. As of late AMD has been covering their entire product line with a minimum number of different actual chip designs probably to save mask and layout costs. Their CPU has covered everything from 4 core desktop CPUs up to multi-socket server CPUs with 64 cores and their APU has covered laptops and lower end desktops with some overlap with the low end desktop CPUs. This would end up in an awkward middle ground. Not sharing enough in common with the HEDT and Server space to make it economical and too power hungry to be used in the Laptop APU space to make it economical IMO. Maybe if their future GPU line goes a Chiplet route you could set things up so one of those chiplets could be used in this configuration and give us a pretty capable desktop APU with an 8c/16t cpu chiplet, the IO chiplet a GPU chiplet and the HBM for GPU memory.grrrgrrr - Monday, June 8, 2020 - link
Exactly. Hades Canyon looked really good. I probably would have pulled the trigger if I weren't already using an alienware alpha. Since then I've been waiting for an upgrade. It could be expensive with HBM and everything but it'll make a really nice halo product, and prove that future is better than the past.Wrong_again - Tuesday, June 9, 2020 - link
You'd still need the APU if the chip was going into a laptop. 99% of the time the dGPU in my laptop is unused.grant3 - Monday, June 8, 2020 - link
Disappointing; I would have expected AMD to give top priority to supporting this product as proof that they can be counted on as a reliable partner.DigitalFreak - Monday, June 8, 2020 - link
I don't think AMD is really interested in extremely low volume one-offs.Deicidium369 - Monday, June 8, 2020 - link
Pretty much all they do...Gigaplex - Monday, June 8, 2020 - link
Their console support shows they can be a reliable partner.Deicidium369 - Monday, June 8, 2020 - link
Easy to get a contract no one else is bidding on...Samus - Tuesday, June 9, 2020 - link
LOL, and how do YOU know other parties are not bidding on console design wins? Are you seriously implying nVidia wouldn't be interested in selling tens of millions of GPU's to Sony\Microsoft? Clearly they are interested in the industry: look at the Switch.CiccioB - Tuesday, June 9, 2020 - link
Depends on the margin MS and Sony want to pay.At the time when AMD won all console supply, Nvidia said that the margins were so low that they didn't want to invest so many engineers for that purpose.
Yet, as we have seen, console SoC supply is something that has a value beyond the simple money they make. AMD gained the support and optimization for its own architectures while Nvidia more advances features were left behind.
Santoval - Tuesday, June 9, 2020 - link
Sony and Microsoft have long switched to semi-custom APUs (no distinct GPU and CPU). Therefore, since Nvidia lack a x86-64 licence, the only way they could bid to make the APU for them is if it was ARM or RISC-V based. Neither of them appears to have any wish to switch to a new ISA, hence Nvidia cannot bid to make the APU for them. Nintendo is different bird.CiccioB - Tuesday, June 9, 2020 - link
They didn't switch for any reason but costs and the fact that AMD sold them for a penny.Technically speaking separate CPU and GPU are better but costs more.
There's not chance to match the low prices AMD made with their SoC with a discrete combination even thought the latter would be much faster (and easier to cool down).
AMD simply killed the competition by price. Something Intel nor Nvidia would like to fight against just for having their devices in the consoles for a penny each.
But as said, while for Intel is not a problem (not a gain, not a loss) for Nvidia the problem is that they have lost the support and optimization for their own architecture.
I wish they would supply just one of the two, just to see their advance features supported and see who is really slowing down the gaming evolution with antiquate and obsolete architectures that catch the competition 3 years later.
This would really create a competition between the two.
sing_electric - Tuesday, June 9, 2020 - link
Isn't that a bit chicken-and-egg, though? Imagine a world without AMD (or where AMD went out of business, or where AMD survived but didn't buy ATI, etc.) - it's not like Sony and MS would have just ¯\_(ヅ)_/¯ and stopped making consoles - they'd have chosen something else. Possibly an arch like the original Xbox (which is basically... a PC), possibly have made it worth Nvidia's while to soup up Tegra... custom ARM chip, possibly with a dGPU, or who knows what else.The time between generations helps a lot in that regard - there's plenty of ARM chips out there that may not be as fast as a Zen 3 design, but would still give a pretty big performance boost over the 2013 Bulldozer-based chips in previous gen consoles.
lmcd - Tuesday, June 9, 2020 - link
Ironically Microsoft's ARM on Windows work could enable a future switch.Samus - Tuesday, June 9, 2020 - link
As amazing as AMD's resiliency in the industry is, from engineering feats to design wins and platform support, their drivers (with a heavy emphasis on the ATi roots here) have always been a weak spot.Yet part of me agrees with you. It's very, VERY strange for AMD to prioritize such niche developments such as Mac GPU support and even Linux support over a promising partnership with Intel. And that appears to be exactly the case.
sing_electric - Tuesday, June 9, 2020 - link
I don't think EITHER AMD or Intel thought that Kaby Lake-G was going to be the start of anything long-term. By the time it came out, Intel had to have internally been planning what eventually became Xe, and AMD made no secret of the fact that they were gunning for Intel in as many x86 markets as possible.What *will* be interesting is what becomes of Nvidia using Epyc CPUs in its Ampere-based DGX server. Nvidia seems to have little interest in making an ARM processor for those systems, and with Intel releasing Xe, that means one way or another, Nvidia will have to rely on a CPU vendor with a competing graphics product if it wants to stick with x86....
lmcd - Tuesday, June 9, 2020 - link
Nvidia's been trying to push POWER10 for this scenario but obviously that isn't an easy sell.jeremyshaw - Monday, June 8, 2020 - link
In the end, this was a marriage made in hell. Reminds me of the last Intel CPU + AMD GPU laptop I bought, all the way back in 2011: a Sony Vaio SC 13.Drivers stopped coming almost immediately after I bought the laptop. This was in the weird Win7->Win8 transition days, too, so getting drivers that worked for both OSes? Ha! The few Vaio SC 13 owners met online and discussed out options. Wen were lucky we could inf mod drivers back then in Win7, without incurring too much wrath of the driver signature validation gatekeeper in Win7, so that kept out laptops chugging for a little bit. It didn't take long before that stopped working with newer drivers and we were moribund until someone figured out binary mods (hex editing) to bypass the installation check. That worked for a few more driver releases until the laptop's AMD GPU stopped accepting any newer driver updates, whatsoever. Even inplace file replacements and registration stopped working (after 1 update? iirc). Newer games got glitchy real fast.
Eventually, I sold the laptop on that forum and bought it's newer, higher specced replacement in Sony's lineup: The Vaio S13, powered with Nvidia graphics (and Nvidia Optimus). Despite the experiences I just had, I was still a dyed in the wool ATi fanboi and flat out stupid for it. Let's just say, that laptop received working updates, DIRECT from NVIDIA for the next ~7+ years. Even Intel occasionally released CVE-relegated updates.
Every few years, I still buy an AMD Radeon GPU to test the waters. The last one I bought was the Thinkpad E485. AMD Ryzen CPU, AMD Vega GPU, on the same chip. No more excuses, right, AMD? SMH, AMD made of mess of that, too. AMD whined and blamed OEMs for not doing AMD's job, before finally sobering up and releasing driver updates themselves, as they should have. It only took 6 months before AMD stopped acting like a petulant little child.
Either way, I now am taking another 2-4 year hiatus from any AMD GPU purchases again.
Dragonstongue - Monday, June 8, 2020 - link
IMHO this is not AMD problem, but the seller overall, as AMD cannot support 3rd party driver sets as they can with their first party design (that is, AMD Radeon version XYZ by all means should be getting proper support and driver sets, whereas something from Lenovo cannot be directly supported via the same chain of drivers as they are not "verbatim" design..different wheels, cogs, power limits etc etc)Now if AMD made the laptop, was a full AMD branded product and they were to have done similar level of crud, then absolutely get right @#$ about it, I would, though when it comes to an ODM/OEM/AIB whatever you would like to call it, the same rules simply cannot apply, they simply do not.
Look at all mighty Intel, they are very adamant about how they do things, a much much larger company, yet even they do not give the same first party support to even full Intel branded designs
ahh well, get the product, use the product, hope it works properly right off the bat and will for a number of years till it gets replaced..these things are always a very hit and miss thing, even more so when the maker (Lenovo, Sony what have you) want you to buy it, but then they really do not at all support it the way they should be.
I personally hate it, only 2 choices, buy it and deal with the crap that likely will happen at some point, or do not buy at all and never have to deal with it ^.^
CiccioB - Tuesday, June 9, 2020 - link
This is a ridiculous justification. Nvidia supports its dGPUs mounted into 3rd party laptop. It's up to them to update the drivers for their own product.
Historically AMD has never been a good supporter of its own technology.
They tend to fire a new product and forget quite often. They are quite unreliable as partners and that's probably one of the reasons why no OEM wants to put their GPUs inside laptops. Despite the fact that those GPUs are also technically inferior to those of the competitor which also assures better support for years, not just weeks.
lmcd - Tuesday, June 9, 2020 - link
AMD changed that situation with regard to laptop drivers, and laptop manufacturers seem to support the change. I have the latest Radeon drivers on my 2700U laptop (which, granted, has plenty of other weaknesses).Sushisamurai - Tuesday, June 9, 2020 - link
I don't think it's a ridiculous justification - AMD gives the OEM a range of operating perameters, and the OEM chooses to go outside of reference for whatever reason. The stock drivers are for stock designs. I can't imagine AMD making an individual driver variant for every OEM variant of their chip. Case in point, my HP 15" Envy with an AMD Ryzen APU couldn't get the latest AMD drivers because not AMD, but HP blocked the install due to possible issues with incompatability. Luckily AMD changed that just recently.Sushisamurai - Tuesday, June 9, 2020 - link
IIRC, it's also the same with intel drivers for laptops. Last I checked, despite newer drivers for my other laptop's (HP) i7, I couldn't install them because the OEM did not validate the newer drivers as my laptop was cTDP up'd from the OEM, and ergo because it wasn't validated wasn't compatible with the install.mocseg - Monday, June 8, 2020 - link
Lenovo with apu 3500U & Linux here. Buttery smooth performance. Couldn't be happier.4000U series are even better.
boozed - Monday, June 8, 2020 - link
I don't think there are many happy stories out there that begin with "I bought a Sony VAIO laptop"Samus - Tuesday, June 9, 2020 - link
I think a lot of those issues could be summed up to Sony having historically poor support for their electronics, from phones to PC's.lightningz71 - Monday, June 8, 2020 - link
This one I place SQUARELY on Intel as they have an established track record of doing this very same thing in the past:Witness some of their Atom N series CPUs as implemented in the Acer Aspire One series, integrating Intel Atom cores with an iGPU developed from PowerVR IP. Intel contracted PowerVR to supply an initial, operational driver and for a period to supply bug fixes for that one driver. After that initial period of bug fixes was over, complete radio silence! Intel hasn't done a thing to ever touch those drivers again. While the launch drivers mostly work for Windows 7, anything later is relegated to the windows default svga driver, which lacks almost any real acceleration.
PowerVR, when contacted directly, points you to Intel as they own the product and are responsible for its support. Intel points you to PowerVR, stating that the IP in the chip belongs to PowerVR and that, without the contract, they can't (won't) touch it. The core of the issue is that Intel chose to do the absolute bare minimum to get the chip out the door while it was being produced, and then immediately dropped it like a sack of burning animal excrement (which, to be fair, it is). Because they chose not to continue to pay PowerVR to produce driver updates, like, for example, Windows 8, which was released while the laptops that contained the chip were still in retail, PowerVR wasn't going to be a charity for Intel and produce them for free.
What I find most disappointing about this when I look at AMD is that, the chip itself hasn't changed. Its based on a slightly modified version of Polaris, which is still receiving driver updates on the regular. At the VERY least, they could just include a final driver for it in subsequent releases and have it in the release notes that support for that product is feature and maintenance frozen at a certain date. However, there MAY be wording, somewhere deep in the contracts for the driver package, that specifically forbids AMD from doing so in a "planned obsolescence" sort of strategy.
I bet certain OEMs are a bit upset about this though. I believe that some of those processors went into "long life" corporate SKUs that are required to have support for X number of years. That could throw a wrinkle into this one...
diehardmacfan - Monday, June 8, 2020 - link
Dell is still actively selling the XPS 15 2-in-1 with this in it. I'm sure they aren't moving a ton of those units but I'm sure they aren't happy about it.MenhirMike - Monday, June 8, 2020 - link
I know this is as much on Intel as it's on AMD, but: AMD's graphics driver situation is a trainwreck. Always has been, and likely every will be. At least they've started supporting Ryzen Mobile with drivers a while ago, after not caring for the first year of its existence. But as far as I am concerned, I've regretted buying AMD graphics every single time I did it, all the way back to when it was still ATI before the acquisition. And it was never a hardware issue, the hardware is awesome, has great performance and price/performance ratio, but the drivers are trash :(DigitalFreak - Monday, June 8, 2020 - link
AMD/ATI have always had crap drivers. AMD's takeover certainly didn't change that.Deicidium369 - Monday, June 8, 2020 - link
Way back in ATI's time - all driver sucked - especially when you tried to install and use themFXi - Monday, June 8, 2020 - link
Despite their well earned success in recent years this kind of behavior has repeatedly put me off their products. Customers don't get to find out for a couple of years after they have laid down their money whether AMD will deem their product worthy of support. And this has happened repeatedly over many decades and is disassociated with the current fiscal success or not of AMD the company. Fact is you are almost always taking a chance that after some period of time in which you, the consumer are still happily using the product, that it will go unsupported. So AMD can happily wave all those performance numbers they want. But we already have seen supported/unsupported issues with what their chipsets will support for cpu's. Sometimes driver fixes take the better part of a year to fix. And all that puts a serious damper on performance advantages or price advantages when a consumer comes to buy. If they really want to be at the top of the market they have to do much more than simply offer a performing product. Support problems like this need to stop, entirely. You ruin customer trust every time you stop something. And that will, long term, keep AMD from market positions they could have simply because of their own failure.GreenReaper - Monday, June 8, 2020 - link
That's great, except it's ultimately an Intel product. They sold it, so they (or as people bought it as part of a system, the manufacturer of the product it's in) are responsible for support, just like Microsoft is responsible for support of the Xbox which has both AMD graphics *and* CPU cores. According to PCWorld Intel are supporting it five years from release. Chips were sold on that basis. The fact that they're discontinuing sales at the end of the month shouldn't change that - and as a practical matter they've released drivers for HD4000 graphics components all the way to this year.TEAMSWITCHER - Monday, June 8, 2020 - link
I don't think this is an Intel thing. Tht BootCamp drivers for my MacBook Pro 16" haven't been updated since it was introduced last November, and AMD still has thesw old drivers on their website. AMD loves to score these design wins for the janky platforms and then ignore them. AMD just isn't good at drivers.eastcoast_pete - Monday, June 8, 2020 - link
I guess that "one off" design has just gotten the support f-off. Really classy, AMD and Intel! Glad I didn't buy one, but this is not okay!Deicidium369 - Monday, June 8, 2020 - link
I can see Intel not wanting to deal with it - they are not going to develop drivers for AMD's GPU - but AMD not making something available for a product they sold is bad - wouldn't take much for the Night Janitor / Head of AMD Software to make a few changes for the sake of their customers - and then put it on EOL time line and forget about iteastcoast_pete - Monday, June 8, 2020 - link
Also, as some comments here point out, this picks at a barely healed scab with AMD GPUs: driver support. As much as I sometimes dislike NVIDIA, but their drivers tend to not suck as badly as team Red's, and they do update them also for older GPUs. An important factor when spending big bucks on a dGPU, not all of us have the means to buy a new card every 2-3 years.cl1020 - Monday, June 8, 2020 - link
I have an XPS 15 2in1. You used to be able to install the Dell/Intel driver, and then update the driver with the latest from AMD through device manager, but this quit working at some point last. year. A bit of a non issue now though as the last firmware update from Dell that disabled under voting the CPU bricked it.I blame all 3 companies equally. The fact you could manually load the AMD default driver and it worked fine shows it's not some insurmountable technical problem, but just a don't give a shit on Intel/AMD's part. I would imagine Dell has enough sway with both companies to put pressure on them to get off their ass as well but don't care enough either. As the above comment mentioned, Bootcamp drivers from AMD are also an afterthought. If the one guy at bootcampdrivers can tweak and modify the generic AMD drivers to get them working, how much effort would it really take for AMD to step up and do it as well?
I guess I understand though. All the money and effort spent on marketing to try and convince everyone that the small performance updates this gen over last are worth it can't leave much left for actual support for the products they already sold.
Hul8 - Monday, June 8, 2020 - link
Could it be that only stable driver versions will be made to support it? The latest Adrenanalin 2020 Edition 20.5.1 is labeled "optional", as is the case for most AMD graphics driver releases.Ryan Smith - Tuesday, June 9, 2020 - link
According to Tom's Hardware, support is missing from the most recent WHQL driver as well.Santoval - Tuesday, June 9, 2020 - link
"Still, I would have expected that driver support is something that would have been hammered out in a contract early on – such that AMD was committed to deliver and paid for the necessary 5 years of drivers – rather than the current situation of Intel and AMD seemingly dancing around the issue."In the meanwhile they both frack every single Kaby Lake-G buyer, which is utterly disgraceful and unprofessional. I have no idea how much this part sold. Perhaps a class action suit against both Intel and AMD will make them "clarify" their contract clause about driver support?
hubick - Tuesday, June 9, 2020 - link
As someone who owns TWO Hades Canyon NUC's, may I ask:HOW THE HELL DOES A PRODUCT INTEL IS _STILL_SELLING_ LOSE DRIVER SUPPORT ALREADY!?!
To think I actually bought NUC's because I thought Intel's support and updates would be better than what I get from other DIY motherboard manufacturers.