AMD Radeon Software Adrenalin 21.10.3 Released

[btarunr]

AMD today released Radeon Software Adrenalin 21.10.3 beta drivers. These come with optimization for "Marvel's Guardians of the Galaxy," with over 21% performance improvement measured at 4K Ulra settings over the previous driver, as tested using an RX 6900 XT graphics card. Optimization is also added for "Age of Empires IV," where a massive 45% performance improvement was obtained at 4K max settings, using an RX 6800 XT. Other games optimized for is "Riders Republic" and "DOOM Eternal 6.66."

Among the issues fixed are missing CPU tuning feature in Radeon Software for some users with Ryzen 9 5950X processors, elevated disk space utilization with Multimedia Athena, task-switching blackscreens in certain multi-display environments, a game crash noticed in "Battlefield V" with RX 6600 series, and image corruption notived with Radeon Boost enabled in "Cyberpunk 2077" and RX 6700 XT.

DOWNLOAD: AMD Radeon Software Adrenalin 21.10.3 beta



Support For

• Marvel's Guardians of the Galaxy
  • Up to 21% increase in performance in Marvel's Guardians of the Galaxy @ 4K Ultra Settings, using Radeon Software Adrenalin 21.10.3 on the 16 GB Radeon RX 6900 XT graphics card, versus the previous software driver version 21.10.2
  • Up to 21% increase in performance in Marvel's Guardians of the Galaxy @ 4K Ultra Settings, using Radeon Software Adrenalin 21.10.3 on the 16 GB Radeon RX 6800 XT graphics card, versus the previous software driver version 21.10.2
• Riders Republic
• Age of Empires IV: Up to 45% increase in performance in Age of Empires IV @ 4K Max Settings, using Radeon Software Adrenalin 21.10.3 on the 16 GB Radeon RX 6800 XT graphics card, versus the previous software driver version 21.10.2.RS-425
• Doom Eternal: Update 6.66

Fixed Issues

• In Radeon Software, the CPU Tuning feature may be missing for some users with AMD Processors such as Ryzen 9 5950X Processor.
• Some users may experience elevated disk space consumption by the Multimedia Athena Dumps folder.
• While playing a game, some users may experience a black screen flicker if they have multiple displays connected and they attempt to switch between open windows (via Alt + Tab shortcut).
• Battlefield V may crash during gameplay on some AMD Graphics products such as Radeon RX 6600 Graphics.
• Image corruption may be observed on some characters when Radeon Boost is enabled while playing Cyberpunk 2077 on some AMD Graphics products such as Radeon RX 6700 XT Graphics.

Known Issues

• Users may experience display corruption issues when they set their Samsung Odyssey G9 C49G95T monitor to 5120X1440 @ 240 Hz.
• Visual artifacts may be observed while playing Call of Duty : Black Ops Cold War on some AMD Graphics products such as Radeon RX 6800M Graphics.
• AMD Radeon Software may crash or become unresponsive while playing some DirectX 11 games such as PlayerUnknown's Battlegrounds with multiple displays connected in extended mode.
• Enhanced Sync may cause a black screen to occur when enabled on some games and system configurations. Any users who may be experiencing issues with Enhanced Sync enabled should disable it as a temporary workaround.
• Radeon performance metrics and logging features may intermittently report extremely high and incorrect memory clock values.

AMD Link for Windows Known Issues

• Intermittent grey frame corruption might be observed when streaming with HEVC on certain configurations via an Internet connection. A workaround is to use the AVC encoding setting instead.

View at TechPowerUp Main Site

 

[ixi]

21% and 45% improvement in fps, holy moly. I'm not mad, I'm surprised.

 

[stiffmeister]

Damn, just installed a fresh Win11 today and installed these drivers, not even knowing

 

[Makaveli]

Will be installing these in abit.

 

[RJARRRPCGP]

21% and 45% improvement in fps, holy moly. I'm not mad, I'm surprised.

Did you notice this yourself? Which hardware? I'm wondering if Navi10 will get a performance increase now...

 

[RadeonProVega]

Riders Republic for 60 bucks? is the game that damn good lol. looks like a f2p game.

 

[prtskg]

21% and 45% improvement in fps, holy moly. I'm not mad, I'm surprised.

AOE IV was not getting good performance in Radeon initially. So this fixes it.

 

[Liviu Cojocaru]

• Users may experience display corruption issues when they set their Samsung Odyssey G9 C49G95T monitor to 5120X1440 @ 240 Hz.

This is an issue for me since two versions ago not sure why this can't be fixed. AMD drivers are really bad, it is my only complaint with them.

 

[Octopuss]

21% and 45% improvement in fps, holy moly. I'm not mad, I'm surprised.

This must be generally related to the pretty old article where a guy who worked at Nvidia described how incompetent game programmers were.

 

[Zareek]

I usually wait for WHQL because installing new drivers every week or two is tedious but I plan to play AoE4. I wonder if these drivers make a difference in AoE4 on Vega architecture. I may have to look into that very soon.

 

[stiffmeister]

I usually wait for WHQL because installing new drivers every week or two is tedious but I plan to play AoE4. I wonder if these drivers make a difference in AoE4 on Vega architecture. I may have to look into that very soon.

Monthly and is like 2 minutes

 

[Makaveli]

I usually wait for WHQL because installing new drivers every week or two is tedious but I plan to play AoE4. I wonder if these drivers make a difference in AoE4 on Vega architecture. I may have to look into that very soon.

I usually wait for WHQL also but since i've upgraded my videocard i've been installing the beta's since most of the changes directly affect newer hardware compared to old.

 

[Zareek]

Monthly and is like 2 minutes

No, actually if you keep up with the latest it is more like bi-weekly on average. This is the third release this month, hence the 21.10.3 ! Don't get me wrong, it's nice that AMD is regularly releasing updated drivers. I get to game maybe 8 hours a week, sometime more, sometimes not at all, so two minutes is a waste of my time unless it makes a big difference.

 

[opinali]

This must be generally related to the pretty old article where a guy who worked at Nvidia described how incompetent game programmers were.

I hate that we have this de-facto architecture of GPU drivers needing an ever-growing list of hacks to patch performance of specific games. Imagine how ridiculous it would be if every major game required OS vendors to issue kernel updates that fix that game's multithreading code, or disk I/O, or memory management. This is basically what we have today for GPUs, and it's no surprise that everyone complains about bad video drivers, given that every driver needs tons of code like "if we detect Deathloop, then replace this specific shader with our optimized code, and fix this bad parameter when they call this DirectX function, and ignore this unnecessary synchronization call that they do all the time".

 

[Octopuss]

But how is this even possible when the game programmers are able to create an entire game that works?
Are they all SO clueless about graphics?

 

[opinali]

But how is this even possible when the game programmers are able to create an entire game that works?
Are they all SO clueless about graphics?

Not a game dev myself, but these AAA games are extremely complex and often produced on crazy deadlines. Absolute max rendering performance is probably not the highest priority for game projects, they have to deliver many other things. GPU vendors create sophisticated profiling tools that game devs can use to analyze their rendering pipeline and optimize it, but it's still hard work, and not their main concern in the scramble to fix all important bugs before launch, as long as the game performs good-enough.

And when it comes to commercialization, the vast majority of gamers are not #PCMasterRace enthusiasts that want to run everything at 4K / Ultra settings / 240fps. These driver updates that give a 5-20% boost to some games are often mostly serving the GPU vendor's interests, the extra performance may be irrelevant for the majority of gamers with their 1080p/60Hz displays and previous-generation GPUs, but the GPU vendor wants to put their flagship GPU in the top of benchmarks.

Whatever the reason, this is a bad system. If a game is released and it doesn't perform well compared to similar games on the same platform, the pressure should be 100% on the game studio to improve that. Even worse many games are "sponsored" by GPU vendors, sometimes that means that AMD or NVidia will send their GPU experts to work together with the studio and help tune their code for specific GPUs. This can easily result in over-fitting, that makes the game run superb on a specific GPU but suck on any other GPU; e.g. due to hardwiring assumptions about the internal behavior of favored GPUs. When this happens, the vendor of the competing GPU has to issue a driver patch to make the game not suck on their cards. Overall this is not a healthy way to run a software platform. Imagine if Mozilla or Google would release a browser that's super-optimized for Intel NICs and Samsung SSDs, then Realtek and Western Digital had to issue driver updates to make their HW competitive for that browser.

 

[W1zzard]

if we detect Deathloop, then replace this specific shader with our optimized code, and fix this bad parameter when they call this DirectX function, and ignore this unnecessary synchronization call that they do all the time

That's exactly what they do, almost always the reason is performance, or some weird issue that's due to the GPU behaving in slightly different ways than what you would expect if you followed the API spec exactly

Absolute max rendering performance is probably not the highest priority

Especially not for the PC platform