AMD drivers

[Rothhaus]

Hi, so I just watched the amd event and I’m really interested in their new cards, but I’ve heard from a lot of people that amd usually have bad/unstable drivers and that their gpus gets really hot compared to nvidia, can anyone confirm or deny this? I’m not looking to start a huge discussion on which company is better I’m just looking for peoples opinions and experiences with amd cards since I’ve never owned one.

[FloRolf]

I think their drivers aren't as easy and nice to handle. People have had many many issues with their drivers in the past so I'd definitely wait a bit with your purchase. 

 

As for Temps it really depends on what cooler you get. A blower style will be hotter than a big and beefy open air cooler or water cooler. AMD also mostly designs their GPU's to run hot in order to achieve the highest clock speeds possible out of the box. 

[Semper]

Historically, AMD does have a fairly poor track record with their drivers.

Heat generation has been all over the place, from fairly reasonable, to nice and toasty. The same can be said for Nvidia, however.

[BuckGup]

3 minutes ago, FloRolf said:

I think their drivers aren't as easy and nice to handle. People have had many many issues with their drivers in the past so I'd definitely wait a bit with your purchase. 

Can you site a source that isn't anecdotal evidence? I don't want to come across as an AMD fanboy but I have heard this argument for the past 10 years and it's all been word of mouth. The only thing I can think of is Linux drivers and OpenGL support being dog doo doo. 

[FloRolf]

8 minutes ago, BuckGup said:

Can you site a source that isn't anecdotal evidence? I don't want to come across as an AMD fanboy but I have heard this argument for the past 10 years and it's all been word of mouth. The only thing I can think of is Linux drivers and OpenGL support being dog doo doo. 

No, just what I hear. To be fair I never advised against AMD, I'm just saying wait a couple days or weeks to see what other people have to say but I would do that with literally any product. Never pre-order anything. 

[jaslion]

Just now, FloRolf said:

No, just what I hear. To be fair I never advised against AMD, I'm just saying wait a couple days or weeks to see what other people have to say but I would do that with literally any product. Never pre-order anything. 

Amds drivers have always been a bit wonky at launch but that was usually fixed after a week or so with a hotfix. The only series that really had a problem was the rx5000 series and of those it was mainly the 5700 and 5700xt. No clue why but it took them forever to get the drivers right for those cards but all the while the previous gen cards running the same drivers just stayed rock solid. Probably had to do with their rdna transfer but the weird thing was that the 5500xt and 5600(xt) were doing fine.

[BuckGup]

3 minutes ago, FloRolf said:

No, just what I hear. To be fair I never advised against AMD, I'm just saying wait a couple days or weeks to see what other people have to say but I would do that with literally any product. Never pre-order anything. 

I agree with that statement for everything as well. I just haven't seen any proof that AMD drivers are actually worse at all. Maybe someone could compile all the topics on various forums and see which has had more posts in the past. 

[Waifu4Life]

Never tried an AMD graphic card. Last time I tried a Radeon card, they were still made by ATI. I didn't really like the ATI drivers back then, they often crashed. Has this changed?

[ewitte]

43 minutes ago, BuckGup said:

Can you site a source that isn't anecdotal evidence? I don't want to come across as an AMD fanboy but I have heard this argument for the past 10 years and it's all been word of mouth. The only thing I can think of is Linux drivers and OpenGL support being dog doo doo. 

Personal driver issue - I had a 5700xt briefly when the computer came out of sleep the fan wouldn't spin back up.  Launch anything it would race to 100-105C and crash the computer.  RPM meter would glitch at the same time and show something like 80000RPM but it wasn't moving if you looked at it.

I was going to "downgrade" from a strix 2080 but decided to keep it when it came back from RMA.

[Slayer3032]

Everyone has bad drivers at some point. I bought a GTX 260 and the drivers didn't work for 2 months and I went to a HD5870 instead after my EVGA 8800 GT died as most did along with my EVGA Nvidia 680i motherboard.

 

Nvidia really benefit from a fairly mature architecture from Kepler, Maxwell and Pascal mostly being the same while AMD cards were fairly weak with only HBM based cards competing at the higher end. Maxwell and Pascal were really unbeatable at that time. Turing definitely needed some fine wine as far as performance and a Super refresh to even be worthwhile which it still isn't based on Pascal/Maxwell pricing.

 

Ampere while the numbers look great definitely gives me Fermi vibes where we will probably see 300w+ GPUs go into the oven right outside of warranty as OEM models and low end cooler cards start creeping in on top of cards randomly boosting to the moon and crashing as well as overall instability due to pushing the cards to the absolute edge of stability already.

 

Nvidia's Linux drivers are notoriously miserable and they generally reject open source, I've only heard very positive things about AMD's Linux drivers since Polaris. I'm actually very surprised when I update Nvidia drivers and I'm even able to reboot to a usable system.

 

RDNA has been experiencing the same transient load issues Ampere is currently dealing with tripping OCP and crashing due to people splitting a single 8 pin, ect. There was definitely initial driver issues as well but those were fixed fairly early. My HD5870 and R9 280 both experienced poor drivers right at EOL, AMD however has been very committed to the Polaris drivers and RDNA2/Navi is going to be around for the long haul in consoles, APUs and more. End of life means end of life though lol so it's going to be upgrade time anyways by then.

 

I would expect with a less fractured mess of having two different architectures where the high end gaming cards are basically workstation/compute rejects of a different design, that they should be able to reign in what's left of the mess.

 

I really want to see how the RDNA2 encoder does, Turing's NVENC just absolutely blows away AMD's, it's like not even close. No one really supports the AMD encoder either, it would be really great to see Plex and other software to embrace it this generation. I'd really love to see how the lower power RDNA2 cards turn out, Nvidia has basically abandoned the sub $200 market while AMD has up until recently been excelling in it. I think we're seeing everything shift up market though unfortunately.

 

Wait for reviews and don't pre-order hardware, ect.

[SFFDesigns]

14 minutes ago, Slayer3032 said:

Ampere while the numbers look great definitely gives me Fermi vibes where we will probably see 300w+ GPUs go into the oven right outside of warranty as OEM models and low end cooler cards start creeping in on top of cards randomly boosting to the moon and crashing as well as overall instability due to pushing the cards to the absolute edge of stability already.

Honestly, my 3080 has had great temps. Considering it's an FE card in an NZXT H510 with rough airflow config, it never goes above 80°c.

 

Most of that is the case being bad in terms of airflow

[Uzrathixius]

1 hour ago, BuckGup said:

I agree with that statement for everything as well. I just haven't seen any proof that AMD drivers are actually worse at all. Maybe someone could compile all the topics on various forums and see which has had more posts in the past. 

Go use them. All evidence will be "anecdotal" since you're asking for peoples experience. 

In my experience, driver releases were few and far between. IIRC I had to run beta drivers for well over a year just to get updates. Not to mention the re-introduction of bugs like the cursor distortion bug; that was enough for me. 

I've heard they're better, and they are prettier, but still buggy. Can't say how true it is; but between that the lack of a cuda competitor (yes yes, I know they have stuff but if no one uses it...who cares) it's hard to risk 500-1000 on a piece of hardware that has, at least in the past, had lackluster software support. 

[Caeradwyn]

I recently upgraded from an R9 Fury, and I have to say that's I've had some issues from time to time. Drivers crashing the first time I start a game, game crashes, Windows says drivers have been recovered, and after that, game runs fine. I've also had instabilities with the Radeon control panel, so for a period of time, I couldn't access any of the settings. That being said, it seemed like all those issues disappeared with the drivers that were released late August.
And historically, there's been times when Nvidia had shit drivers as well. In fact, I remember a time when Nvidia were a laughing stock on hardware forums etc. due to having the worst drivers. Thanks to that, they really picked their game up and improved. The tables turned, and now it's been AMD who's struggled with less than stellar drivers, but I wouldn't expect that to last. It's not the first time AMD has been declared finished only to come back swinging.

[BuckGup]

1 hour ago, Uzrathixius said:

Go use them. All evidence will be "anecdotal" since you're asking for peoples experience. 

In my experience, driver releases were few and far between. IIRC I had to run beta drivers for well over a year just to get updates. Not to mention the re-introduction of bugs like the cursor distortion bug; that was enough for me. 

I've heard they're better, and they are prettier, but still buggy. Can't say how true it is; but between that the lack of a cuda competitor (yes yes, I know they have stuff but if no one uses it...who cares) it's hard to risk 500-1000 on a piece of hardware that has, at least in the past, had lackluster software support. 

Interesting, in my experience I have had many more problems with Nvidia drivers than I have ever had with AMD

[TheBahrbarian]

Maybe an unpopular opinion here, but I really don't think AMD drivers are that bad (especially not now). Not bad enough to completely write them off even if they have superior hardware or value, that's for sure. And lets not jump to conclusions about this generation based on past experiences, at the very least wait for the cards/reviews to release to get an idea of any issues.

 

What I can say is I bought a 5700xt almost a year ago (back in November 2019) and have had pretty much 0 issues since then. Aside from enhanced sync not working (it never has) it's been flawless. And I really like radeon's software and usability, it's way better than Nvidia control panel and GeForce experience. My graphics card before this was a GTX 970, and frankly, I had a relatively buggy experience over the ~3 year lifetime I had that card. Plenty of times, even years after I bought it, Nvidia pushed an update that broke features or did not fix features that were broken. I'm not gonna sit here and tell you Nvidia has bad drivers (because they don't), but from my experience there was no clear "winner", both have been a typical software experience. 

 

On top of that, I feel like a lot of the "AMD drivers bad" hype was really just that... hype driven. Pretty much anyone on these forums, or the internet in general, who bought an AMD card and had any issue whatsoever immediately wrote it off to bad drivers and swore off AMD forever. When, in fact, I helped/saw dozens people on this forum, other forums, and IRL who had issues with an AMD card, blamed it on the drivers, but eventually figured out it was something altogether (even if people in the comments were telling them it was drivers). I am a personal example of this. I had issues when I first got the card and everyone was telling me it could be the infamous bad drivers, but it ended up being a simple chipset/bios update which fixed it and I never had problems after that. 

 

Now it's pretty clear AMD had poor drivers from the start, I'd be lying if I said otherwise despite my personally positive experience. But they really worked hard on them and to my knowledge they're up to standards now (not excusing them of the fact that they were bad at launch). And several sources (such as Moore's Law is Dead) have stated that AMD drivers are months ahead then where they were at RDNA 1's launch. And frankly, I find this all ironic considering the absolute mess that has been the RTX 3000 launch (i.e. neither company is perfect) and imo it seems like sometimes Nvidia doesn't get the same heat AMD did for their issues.

 

TL;DR: In my opinion, AMD drivers were rough, but not that bad, simply overblown by hype. Several sources state they will be better at launch then RDNA 1, and current drivers for RDNA 1 show their team is capable of good software. So, with today's retailers strong return policies, there is no reason not to at least give AMD a shot if their hardware is the better value/option for you.

[Mark Kaine]

4 hours ago, BuckGup said:

Can you site a source that isn't anecdotal evidence? I don't want to come across as an AMD fanboy but I have heard this argument for the past 10 years and it's all been word of mouth. The only thing I can think of is Linux drivers and OpenGL support being dog doo doo. 

The thing is : they're nice to look at, but they're clunky. And often you need to basically spam stuff for something to get it to work (aka it is unresponsive as hell) because AMD apparently thinks form goes over function. A terrible approach. 

 

I think most people who think AMD drivers are 'ok' or 'good' probably don't use it much or at all. 

 

In comparison Nvidia is very basic - they didn't change the UI since the 90s? Gfe is trash of the highest magnitude, but shadowplay is actually very good and so is the OSD, easy to set up, easy to read (big nope for AMD) and also nvidia has a selection of cool features like super sampling (DSR) and other stuff that literally just works and on the first try too. 

 

So yeah if things like recording and easily adjusting settings aren't someone's cup they're gonna have little to no issues with AMD drivers... 

 

Anecdotal, and also not because that's what almost anyone who actually uses these things would tell you, and, what else do you expect than anecdotal "evidence" anyway? 

 

This is highly subjective and also depends on use case. 

[BuckGup]

59 minutes ago, Mark Kaine said:

The thing is : they're nice to look at, but they're clunky. And often you need to basically spam stuff for something to get it to work (aka it is unresponsive as hell) because AMD apparently thinks form goes over function. A terrible approach. 

 

I think most people who think AMD drivers are 'ok' or 'good' probably don't use it much or at all. 

 

In comparison Nvidia is very basic - they didn't change the UI since the 90s? Gfe is trash of the highest magnitude, but shadowplay is actually very good and so is the OSD, easy to set up, easy to read (big nope for AMD) and also nvidia has a selection of cool features like super sampling (DSR) and other stuff that literally just works and on the first try too. 

 

So yeah if things like recording and easily adjusting settings aren't someone's cup they're gonna have little to no issues with AMD drivers... 

 

Anecdotal, and also not because that's what almost anyone who actually uses these things would tell you, and, what else do you expect than anecdotal "evidence" anyway? 

 

This is highly subjective and also depends on use case. 

I'm going to have to disagree. Shadowplay used to be amazing when it first came out. Then they updated it and it took a larger performance hit and the OSD is ugly, janky. and slow. I've been using discrete AMD cards since the 7770 an d have only had minor issues that a reinstall wouldn't fix. On the otherhand I have had Nvidia drivers so broken DDU won't fix it and I had to rollback to earlier drivers. This was mainly on my 760 I had and my 980ti. I don't think we are accomplishing anything as it's pretty up in the air and without any hard evidence one way or another we will just keep going in circles. So I say buy the card you want and also hats off to the driver devs on both sides as that shit is complex 

[Mark Kaine]

4 minutes ago, BuckGup said:

Then they updated it and it took a larger performance hit 

I mean I agree we can disagree on that all, but this is not correct in my experience, shadowplay cost like maybe 50-100 points in superposition at 1080p, that's basically nothing and in game its about 1-4 fps or something, which is also almost nothing. Any other recording method cost me about 10fps minimum. 

 

 

 

 

[BuckGup]

5 minutes ago, Mark Kaine said:

I mean I agree we can disagree on that all, but this is not correct in my experience, shadowplay cost like maybe 50-100 points in superposition at 1080p, that's basically nothing and in game its about 1-4 fps or something, which is also almost nothing. Any other recording method cost me about 10fps minimum. 

 

 

 

 

Right it barely takes any performance hit I was just saying when they updated it compared to what it previously was it performed worse

[TheBahrbarian]

13 minutes ago, Mark Kaine said:

I mean I agree we can disagree on that all, but this is not correct in my experience, shadowplay cost like maybe 50-100 points in superposition at 1080p, that's basically nothing and in game its about 1-4 fps or something, which is also almost nothing. Any other recording method cost me about 10fps minimum. 

 

 

 

 

I also used to use shadowplay all the time to record my games, but I remember they pushed an update that broke it for me (the replay feature or recording wouldn't wok at all). I liked it but never got it to work again for almost a year before I ended up with my 5700xt

[BTGbullseye]

13 hours ago, BuckGup said:

Maybe someone could compile all the topics on various forums and see which has had more posts in the past. 

The last time someone suggested that, there were a greater number of Nvidia issues, even when filtering exclusively to issues with current-gen dedicated desktop GPUs.

9 hours ago, Mark Kaine said:

The thing is : they're nice to look at, but they're clunky. And often you need to basically spam stuff for something to get it to work (aka it is unresponsive as hell) because AMD apparently thinks form goes over function. A terrible approach.

I see you've never actually used the modern drivers then. I've used it on a variety of low and mid-grade systems, and never had as slow of a responsiveness as I have in the Nvidia software. (granted, my experience with Nvidia software is for 700-series and older) My current install of 20.5.1 beta drivers for the system in my signature has no perceptible interaction delay for any control at all.

9 hours ago, Mark Kaine said:

I think most people who think AMD drivers are 'ok' or 'good' probably don't use it much or at all. 

I think people that think AMD drivers are bad are people that haven't actually used the software, or refuse to even consider learning the different UI.

9 hours ago, Mark Kaine said:

also nvidia has a selection of cool features like super sampling (DSR) and other stuff that literally just works and on the first try too. 

AMD also has DSR, and several other features that "just work on the first try".