Did Nvidia purposely gimp the performance of 50xx series cards with drivers

[JustBenching]

AMD or Nvidia will optimize drivers for the new architectures first and will not care to be everything fine for older Pascals cards for example.

Edit: they optimized newer 9000 series for lower pwr draws in idle and video playback but not for the 7000 series. Why? They want you to upgrade...

My 6700s amd laptop got it's best drivers that fix a lot of stability and performance issues in 2024. That's a 2021 rdna gpu, so clearly it's not true that they let go of older generations.

 

[AusWolf]

As for AMD, the reason why you aren't seeing any changes in power draw etc. on RDNA 3 GPUs is because... the hardware can't do it. It's simple as that, the hardware can't do it. RDNA 3 is still fully supported and in active development, as is RDNA 2 hardware. It's only the 5700 XT that hasn't seen much in the way of changes recently, again, because the hardware can't really do it.

Exactly this, thank you.

AMD admitted that RDNA 4 was going to be an RT+FSR upgrade with hardware bug fixes early on in the development stage. The lower idle power consumption is one of those hardware bug fixes.

The thing is, you can't fix hardware bugs through software.

 

[bruhmoment01]

My 6700s amd laptop got it's best drivers that fix a lot of stability and performance issues in 2024. That's a 2021 rdna gpu, so clearly it's not true that they let go of older generations.

And my 7840U doesn’t function if I try to install drivers newer than 24.4.1, what’s your point?

Anecdotes

 

[AusWolf]

The point is, In the end of the day you'll want to upgrade. So call it something else if you wish, not planned obsolescence but, the result is the same, you will upgrade.

No, you want to upgrade because the new thing offers more performance, better features, or fixes that you want. You don't need any of them. You just want them.

 

[rusty caterpillar]

My 6700s amd laptop got it's best drivers that fix a lot of stability and performance issues in 2024. That's a 2021 rdna gpu, so clearly it's not true that they let go of older generations.

Not in my experience with Pascals, mine doesn't invalidate yours same as yours doesn't invalidate mine.

Another thing, Adrenaline on R9 380 Nitro, despite fan slider being set to 1800 RPM, fans sits at 768 RPM during gaming , card reaches 91 C, fans sits at 768RPM with no hope to ramp up.
Deleted profile, works for a while correctly than same behavior is observed.

ADR version 22.6.1 since card is obsolete. Win 10. Windows driver doesn't do that will do ramp up the fans as need it.

Coincidence twice, issues with the sensor or ADR V 22.6.1? Take you pick but the result is the same I can brick the card if not paying attention.

All sensors on this card shows same temp like are synced, is no way of knowing which is the temp for VRAM.

No, you want to upgrade because the new thing offers more performance, better features, or fixes that you want. You don't need any of them. You just want them.

If users can't afford has to deal with the first 2 options, adding cost in cooling mods or messing eye sight depending on usage hrs.
On one hand new drivers brings lower pwr consumption, on the other hand you have to force your GPU on higher power consumption to get rid of the shimmering.

 

[JustBenching]

And my 7840U doesn’t function if I try to install drivers newer than 24.4.1, what’s your point?

Anecdotes

Your laptop not getting drivers doesn't mean that amd abandons old gpus, since they still release rdna 2 drivers. If they did abandon older gpus then my 6700s wouldn't have gotten better drivers, that's logic 101.

 

[rusty caterpillar]

Your laptop not getting drivers doesn't mean that amd abandons old gpus, since they still release rdna 2 drivers. If they did abandon older gpus then my 6700s wouldn't have gotten better drivers, that's logic 101.

Not abandon completely, small issues there and there which cumulates and at the end you will upgrade sooner or later depending on your BS endurance. Achievement completed.

And my 7840U doesn’t function if I try to install drivers newer than 24.4.1, what’s your point?

Anecdotes

780M is not old can't explain why Adrenaline 24 doesn't work with GPU launched in July 2023, maybe a bug.

 

[Dr. Dro]

And my 7840U doesn’t function if I try to install drivers newer than 24.4.1, what’s your point?

Anecdotes

I mean. My laptop's got the R5 5600H, it's been flawless as a processor, but its iGPU's got stability issues that will never be resolved. Vega loves to TDR with embedded videos on web browsers, the APUs have the same bug with the media engine dGPU RX Vega had back in the day. It's utterly unpredictable, you may never run into it, or you could be a lottery winner (and I won with all 3 types of Vega hardware, Vega 10, 20 and iGPU) which makes it quite unreliable for usage in today's media-rich web.

There's 2 workarounds: 1. is disabling hardware acceleration, 2 offloading browser rendering to the RTX 3050 (RIP battery life, which is horrible to begin with).

Not in my experience with Pascals, mine doesn't invalidate yours same as yours doesn't invalidate mine.

Another thing, Adrenaline on R9 380 Nitro, despite fan slider being set to 1800 RPM, fans sits at 768 RPM during gaming , card reaches 91 C, fans sits at 768RPM with no hope to ramp up.
Deleted profile, works for a while correctly than same behavior is observed.

ADR version 22.6.1 since card is obsolete. Win 10. Windows driver doesn't do that will do ramp up the fans as need it.

Coincidence twice, issues with the sensor or ADR V 22.6.1? Take you pick but the result is the same I can brick the card if not paying attention.

All sensors on this card shows same temp like are synced, is no way of knowing which is the temp for VRAM.

View attachment 392252


If users can't afford has to deal with the first 2 options, adding cost in cooling mods or messing eye sight depending on usage hrs.
On one hand new drivers brings lower pwr consumption, on the other hand you have to force your GPU on higher power consumption to get rid of the shimmering.

GCN 3 was a total second class citizen for AMD, and that included R9 Fury. Heavy priority was given to Polaris, GCN 4 had no new instructions and was binary compatible with GCN 3, but it quite a few low level improvements, and end of the day, AMD neglected GCN 3 because it couldn't do these things, leading to a subpar experience overall.

2020-2023 was a dark age overall for AMD's driver teams, but GCN 3 owners really got the short end of the stick. There are many bugs, many unfixed regressions - in fact, you probably shouldn't be using anything newer than 19.9.2 on it, especially if you want to retain the ability to use 4K/60 output through a DisplayPort to HDMI adapter. Unfortunately no fix was ever provided and the issue remained past EOL.

 

[tugrul_LordOfDrinks]

The biggest gimp is motherboard. Too expensive, little capabbility. I remember cheapest mobos having 4 pcie slots and stuff. Today, its like pcie 5.0 16x + pcie 4.0 4x + pcie 2.0 x1. What is this? It cant even work good like having usb failure problems.

 

[Dr. Dro]

The biggest gimp is motherboard. Too expensive, little capabbility. I remember cheapest mobos having 4 pcie slots and stuff. Today, its like pcie 5.0 16x + pcie 4.0 4x + pcie 2.0 x1. What is this? It cant even work good like having usb failure problems.

Meant to upsell HEDT platforms. And then HEDT conveniently went belly up. It never changed, since that also had the side benefit or stopping customers from getting a plenty powerful 16 core chip like a 9950X and making a server with lots of I/O out of it.

 

[Hesychius]

This sounds like just another story for tech-tubers to cash in on the clickbait

 

[AusWolf]

If users can't afford has to deal with the first 2 options, adding cost in cooling mods or messing eye sight depending on usage hrs.
On one hand new drivers brings lower pwr consumption, on the other hand you have to force your GPU on higher power consumption to get rid of the shimmering.

You always make a compromise between what you want and what your budget allows. I don't see how this is new.

Drivers only lower power consumption if they can. Sometimes, it's the hardware that's at fault.

 

[tpa-pr]

Yeah, this was a bit of a funny ride "Guys I in my INFINITE GENIUS have UNLOCKED the TRUE POWER of Nvidia's GPUs which they are HIDING FROM YOU in a bid to STEAL from you-"

The second the guy posted on TPU I was very doubtful, especially when his "expert report" document was a single page of benchmark scores. Still, it made for some good entertainment! Nothing wrong with a good IT crazy guy to bring some levity.

 

[Shihab]

Not in my experience with Pascals, mine doesn't invalidate yours same as yours doesn't invalidate mine.

My experience, however, does invalidate your conclusion. My 1080 hasn't seen performance nor graphical degradation, nor do I recall any significant "bugs" that weren't fixed with subsequent drivers.

I'm not saying you haven't ran into issues. Issues happen, and I think I spotted someone mentioning regression bugs, these also happen. But something that happened to you or affects a handfull of applications does not automatically translate to a planned obsolescence conspiracy.

 

[Vayra86]

It's not just mental health. Education is in the dumps too. But thats OT. I think this thread is done.

People educating themselves is the bigger issue. The attention span has been conditioned on the length of a search or ai prompt. No critical thought required, just absorb info. In that thought process taking time for understanding is too much effort.

The tyranny of stupidity. Signed, big tech

 

[rusty caterpillar]

My experience, however, does invalidate your conclusion. My 1080 hasn't seen performance nor graphical degradation, nor do I recall any significant "bugs" that weren't fixed with subsequent drivers.

I'm not saying you haven't ran into issues. Issues happen, and I think I spotted someone mentioning regression bugs, these also happen. But something that happened to you or affects a handfull of applications does not automatically translate to a planned obsolescence conspiracy.

You wish to invalidate mine but, doesn't work like that, if that makes you happy, invalidate anyway.
Also you come with 1 example vs 3 of mine. I worked with a lot of hardware, my observations and notes state otherwise. So it happen as I said on 2 different systems 1080 Ti and 1070 and the third I didn't mention 1050 Ti M (combination of high temps in idle and black screen flickering, reverted to old driver fixed both issues) Not planning to dig now my older notes.

Also I didn't talked about planned obsolescence in software but, hardware. The R9 380 is an exception and I will further investigate but, for now twice it failed to put the fan at the desired RPM while card having 91C during gaming. Is enough to reinstall Adrenaline 22.6.1 after DDU, no end user should encounter such issues specially many are not aware of their temps n'or having AMD overlay with metrics.

"you haven't ran into issues" with the shimmering are constant issues for years on different drivers, what are you talking about??? I didn't talked about the extent why assuming is something insignificant?
For example WT Dagor engine(favoring Nvidia cards) had some major updates over the years but didn't fixed anything for Pascal cards shimmering. On the old forum you'll find people complaining about shimmering with NVDA and AMD cards.
Is it Dagor engine at fault? Maybe BUT why shimmering is not present on 3060 cards or newer???

I mentioned few because those are the worst offenders not because are just few.

I don't see how this is new.

You don't see because I didn't say that or maybe you like to put words in my mouth?

You always make a compromise between what you want and what your budget allows.

You completely lost the point.

When you have no budget for upgrade you have option 1 and 2, one overrides the other and users, are at a loss, running high power and load + high temps to correct the shimmering.

Whatever they gained trough the newest driver in terms of less power is now nullified.
Is clear enough now?

 

[Dr. Dro]

People educating themselves is the bigger issue. The attention span has been conditioned on the length of a search or ai prompt. No critical thought required, just absorb info. In that thought process taking time for understanding is too much effort.

The tyranny of stupidity. Signed, big tech

I call it TikTok brain. It's actually a severe illness as far as I'm concerned.

 

[Shihab]

Also you come with 1 example vs 3 of mine. I worked with a lot of hardware, my observations and notes state otherwise. So it happen as I said on 2 different systems 1080 Ti and 1070 and the third I didn't mention 1050 Ti M (combination of high temps in idle and black screen flickering, reverted to old driver fixed both issues) Not planning to dig now my older notes.

Except your claim wasn't about something of a stochastic nature - like failure rates - where one argues sample sizes, it was deterministic; newer driver = artifacts and degraded operational parameters.

It only takes one black swan to disapprove their whiteness claim, and in this case, I have two. Plus an entire web of twitchy gamers and whatnot who would jump at even the slightest hint of what you're proposing being a wide-scale issue. So obviously, it's not a very common observation. Sure, someone would complain about it once in a while, but like I said: issues happen.

 

[_roman_]

why not put in now whataboutsim about amd laptop processosrs and not working gpus mobile trash apus = weak processor - low cache - bad sodimm dram - bad - weak mobile gpu (radeon 6500xt anyone?)

And my 7840U doesn’t function if I try to install drivers newer than 24.4.1, what’s your point?

What I learnt in past few days as I bought a refurbished notebook wiht ryzen 4650u pro.

List of AMD Ryzen processors - Wikipedia

en.wikipedia.org

Lots of renamed processors. Lots of same performance. amd mobile processors are a mess.

I can not say more yet, as laptop #1 goes back to the refurbisher. Laptop #2 = same modell but cheaper is on the way to my home.

laptop #1 worked with the standard windows gpu driver - without updates. speaker dead I did not bother even connecting to the web.

--

https://www.amd.com/en/resources/support-articles/release-notes/RN-RAD-WIN-24-4-1.html

AMD Processors with Radeon Graphics Product Compatibility​

Important Note for Laptop and All-In-One (AIO) PCs​

AMD recommends OEM-provided drivers which are customized and validated for their system-specific features and optimizations.

any questions?

 

[rusty caterpillar]

Except your claim wasn't about something of a stochastic nature - like failure rates - where one argues sample sizes, it was deterministic; newer driver = artifacts and degraded operational parameters.

It only takes one black swan to disapprove their whiteness claim, and in this case, I have two. Plus an entire web of twitchy gamers and whatnot who would jump at even the slightest hint of what you're proposing being a wide-scale issue. So obviously, it's not a very common observation. Sure, someone would complain about it once in a while, but like I said: issues happen.

Wide issue or not, your experience doesn't invalidate mine, the 2 cards with long issues on shimmering doesn't make them in any way "isolated happening". Nor the title is isolated either.
I'm not gonna build an "archive stock" for you to see as much you not gonna do the reverse. So we can agree to disagree.

To add from what I remember, as I'm far from my notes now, on Vulkan is no issues only DX 11 and DX12 which are main drivers for gaming in windows.

A quick search, ppl talked about Negative LOD Bias in relation to shimmering since 2004.


Now days. The fact that shimmering can be on some cards(1000) and on other not(3000- 4000), is very strange since are subjected to same driver. Yet that same driver trigger things only on some cards.
Also AMD cards are affected to but don't know which.

I played with Clamp Negative Lod bias on both cards but, didn't fixed the issue, only pushing very high quality settings did. And as I said not all cards ca be pushed like that do to the setup limitations.

 

[Vayra86]

I call it TikTok brain. It's actually a severe illness as far as I'm concerned.

It was there before TikTok though. TikTok gen isn't voting as we speak.

 

[Dr. Dro]

It was there before TikTok though. TikTok gen isn't voting as we speak.

You sure? Let me put it this way... people born around the time Windows Vista came out in January of 2007 are already of legal age (18 years old) pretty much worldwide. Those born when Windows 7 came out in 2009 are already around 16, and do all the things 16 year olds do (in many cases, vote)

 

[Vayra86]

You sure? Let me put it this way... people born around the time Windows Vista came out in January of 2007 are already of legal age (18 years old) pretty much worldwide. Those born when Windows 7 came out in 2009 are already around 16, and do all the things 16 year olds do (in many cases, vote)

Damn you may be right. But then you look at the overall percentage of young voters and it ain't really impressive compared to the generations prior... Perhaps its not really just the TikTok gen. It has already become far more than that. The Twitter gen. The Facebook gen. Us. Behold the power of peer pressure.

 

[Veseleil]

2020-2023 was a dark age overall for AMD's driver teams, but GCN 3 owners really got the short end of the stick. There are many bugs, many unfixed regressions - in fact, you probably shouldn't be using anything newer than 19.9.2 on it,

Just like you shouldn't be using any post 23.12.1 drivers on an RDNA2. The last rock solid ones for our GPU gen. Learned from my own experience. There's no feature or fix that benefits RDNA2 past that release whatsoever, quite the opposite.
The only issue that might surface are the newer games requiring that special treatment essential oils (AKA support drivers), but they aren't worth it anyway IMO.
That kinda falls into another topic, and a bit heavy one, as too many people have that undisputable "everything newer must be better" mentality I don't have the will nor the time to fight.

 

[R-T-B]

People educating themselves is the bigger issue.

That ties into poor education. I was taught critical thinking as a kid. Not many really were though, I'd hazzard a guess.

But this whole subthread is severely offtopic.