AMD Software: Adrenalin Edition 23.9.1 - Driver Download and Discussion

But, reflex is not enabled on driver level, but on per game implementation.
AMD Anti-lag for now works everywhere on driver level for RDNA3. But it doesn't mean it won't be able to work as Reflex on per-game basis.

 

Same with Anti-Lag+, it needs to be supported by the game itself. They just put the toggle to turn it on in the driver, which really reeks.

 

That's funny it reminded me of something similar when I enabled CPU core isolation in windows. I was using libre hardware monitor at that time.Some CPU sensors would freak out when core isolation was enabled but they were totally fine when it was off. I disabled core isolation and never enabled it again after that.
Just look at that 65 kW power draw on core 5 lol

 

Oh, it breaks AIDA64 latency measurement as well)
But no, i don't have virtualization support enabled right now, so no. Even if core isolation is enabled, it cannot work without virtualization.

Also parrying you with all the energy in the world. 163*10^30 W on core 6



P.S. Yes, that cache also currently works on 302*10^30 mHz frequency based on HWINFO readings... Insane, right?)

 

Yes, that would/could be it.

I was asking for a friend actually, my AMD laptop only has the integrated graphics and I never had any need for recording.

 

Update to t his driver and had several blue screens had to revert to previous one.

 

no problem for me, great driver for my 6800xt

 

Can confirm - good driver; the downloaded "exe" archive doesn't bother at all.

 

My 6700 XT is more stable with UV with 23.8.2. Slightly less score in Superposition too, reverting.

 

Been awhile since I last updated. This driver seems to be a pretty good fit for my 6950 xt.

 

This drivers on my Vega 64, are no good!
Cannot set a custom oc on performance options, not even original custom stock voltage and clocks!
Always the blue screen of death.

 

I can report high idle memory clock issue is fixed on Linux using rolling zen kernel 6.5.2 with the 7900XTX, paired with 2 Acer 144hz 27" monitors, both at their full speed. It's now using 13 watts idle (application reporting is CoreCtrl). Funny thing is, Windows doesn't have this fixed yet (as I dual-boot), not even with the latest drivers. Possibly the fix will be a combination of chipset + graphics driver update for Windows, but it's not there yet. I think that with Linux having it fixed for my monitors, Windows will soon follow.

Linux has always favored AMD, but this make that even more clear.

 

Well... That took me quite a bit of time to implement (add to my existing overlay) in decently compact and still visually understandable fashion (about 4 hours, i guess?)
RTSS beta added support of PresentMon for that sweet "GPU busy" metric.
Only downside is that by default polling rate is tied to master parameter (200ms for me), and even when i use separator for different polling rate for PresentMon data specifically, it causes ALL data from PresentMon to be updated with stated speed. Meaning i either refresh lower double graph with 200 ms, or refresh everything with, like 50 and lower ms delay making it fast moving mess of a numbers. (And graph still is not updating by frame with this speed, so it feels bit more choppier than native RTSS ones). So i decided to say f this and kept them at 200 ms. It does keep everything noted on graph on per-frame basis though, just updates 5 times a second.


Also, even as my personal delay to PresentMon data is about 0.3-1.2 seconds (cannot be overridden, limitation of PresentMon itself), "LAG" must be enforced to not get consistently dunked on by lack of data on moment of polling. 3 seconds number and sample provided by Unwinder showed themselves well, so i didn't change delay. This LAG is what makes it work consistently even on high poll rates. But with that said lag, everything under PresentMon section is delayed by 3 seconds.

P.S. GPU BUSY RATE is my personal addition, basically what percentage or presentation time GPU busy took. Surprisingly, it looks like this rate can be more than 100% on instant measurements (and quite frequently it seems)? Idk, maybe i have done something wrong. On averaged it doesn't appear to be much of a truth, but on instant measurements it looks like possible result. Well, anyways, most importantly is looking for CPU bottleneck and stuff.

 

Water glitch in Hogwarts Legacy. Still not fixed on RX 570.

 

If anyone really wants that overlay i made (for some reason) - here it is:
Btw borders on screenshot above is layer separator, it isn't visible in actual overlay, so it looks more like this

Spoiler: overlay code

[Master]
Implementation=2
FontFace=Unispace
FontHeight=-8
FontWeight=700
ZoomRatio=2
[Settings]
Name=
EnvVars=PM_RefreshPeriod=200;PM_DisplayDelay=3000
RefreshPeriod=200
EmbeddedImage=
PingAddr=
[General]
Sources=48
Tables=2
Layers=28
[Source0]
Name=CPU usage
Units=%
Format=
Formula=
Provider=HAL
ID=CPU usage
[Source1]
Name=CPU usage max
Units=%
Format=
Formula=
Provider=HAL
ID=CPU usage max
[Source2]
Name=RAM usage
Units=GB
Format=%.2f
Formula=x / 1024
Provider=HAL
ID=RAM usage
[Source3]
Name=RAM usage \ process
Units=GB
Format=%.2f
Formula=x / 1024
Provider=HAL
ID=RAM usage \ process
[Source4]
Name=Ping
Units=ms
Format=
Formula=
Provider=HAL
ID=Ping
[Source5]
Name=GPU1 temperature
Units=°C
Format=
Formula=
Provider=HAL
ID=GPU1 temperature
[Source6]
Name=GPU1 temperature 2
Units=°C
Format=
Formula=
Provider=HAL
ID=GPU1 temperature 2
[Source7]
Name=GPU1 usage
Units=%
Format=
Formula=
Provider=HAL
ID=GPU1 usage
[Source8]
Name=GPU1 power
Units=W
Format=
Formula=
Provider=HAL
ID=GPU1 power
[Source9]
Name=GPU1 memory usage
Units=GB
Format=%.2f
Formula=x / 1024
Provider=HAL
ID=GPU1 memory usage
[Source10]
Name=GPU1 memory usage \ process
Units=GB
Format=%.2f
Formula=x / 1024
Provider=HAL
ID=GPU1 memory usage \ process
[Source11]
Name=GPU1 clock
Units=MHz
Format=
Formula=
Provider=HAL
ID=GPU1 clock
[Source12]
Name=GPU1 memory clock
Units=MHz
Format=
Formula=
Provider=HAL
ID=GPU1 memory clock
[Source13]
Name=GPU1 voltage
Units=mV
Format=
Formula=
Provider=HAL
ID=GPU1 voltage
[Source14]
Name=GPU delta Tj-Te
Units=°C
Format=
Formula=""GPU1 temperature 2"-"GPU1 temperature""
Provider=HAL
ID=GPU temperature delta Tj/Te
[Source15]
Name=CPU temperature
Units=°C
Format=
Formula=
Provider=HAL
ID=CPU temperature
[Source16]
Name=CPU power
Units=W
Format=
Formula=
Provider=HAL
ID=CPU power
[Source17]
Name=CPU clock
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU clock
[Source18]
Name=CPU AVG ECLK
Units=MHz
Format=
Formula=0+(("CPU1 ECLK"+"CPU2 ECLK"+"CPU3 ECLK"+"CPU4 ECLK"+"CPU5 ECLK"+"CPU6 ECLK"+"CPU7 ECLK"+"CPU8 ECLK"+"CPU9 ECLK"+"CPU10 ECLK"+"CPU11 ECLK"+"CPU12 ECLK"+"CPU13 ECLK"+"CPU14 ECLK"+"CPU15 ECLK"+"CPU16 ECLK")/16)
Provider=HAL
ID=CPU1 effective clock
[Source19]
Name=CPU1 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU1 effective clock
[Source20]
Name=CPU2 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU2 effective clock
[Source21]
Name=CPU3 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU3 effective clock
[Source22]
Name=CPU4 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU4 effective clock
[Source23]
Name=CPU5 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU5 effective clock
[Source24]
Name=CPU6 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU6 effective clock
[Source25]
Name=CPU7 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU7 effective clock
[Source26]
Name=CPU8 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU8 effective clock
[Source27]
Name=CPU9 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU9 effective clock
[Source28]
Name=CPU10 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU10 effective clock
[Source29]
Name=CPU11 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU11 effective clock
[Source30]
Name=CPU12 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU12 effective clock
[Source31]
Name=CPU13 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU13 effective clock
[Source32]
Name=CPU14 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU14 effective clock
[Source33]
Name=CPU15 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU15 effective clock
[Source34]
Name=CPU16 ECLK
Units=MHz
Format=
Formula=
Provider=HAL
ID=CPU16 effective clock
[Source35]
Name=msBetweenPresents
Units=ms
Format=%.1f
Formula=
Provider=PresentMon
ID=msBetweenPresents
[Source36]
Name=msGpuActive
Units=ms
Format=%.1f
Formula=
Provider=PresentMon
ID=msGpuActive
[Source37]
Name=msReportingLag
Units=s
Format=%.1f
Formula=
Provider=PresentMon
ID=msReportingLag
[Source38]
Name=IsPresentMonDataAvailable
Units=
Format=
Formula=validate(x)
Provider=PresentMon
ID=msBetweenPresents
[Source39]
Name=GPU Busy rating
Units=%
Format=%.0f
Formula=("msGpuActive"/"msBetweenPresents")*100
Provider=PresentMon
ID=msGpuActive
[Source40]
Name=Status
Units=
Format=
Formula=
Provider=PresentMon
ID=Status
[Source41]
Name=msInputLatency
Units=ms
Format=
Formula=
Provider=PresentMon
ID=msInputLatency
[Source42]
Name=IsGpuLimited75
Units=
Format=
Formula=(msGpuActive / msBetweenPresents) >= 0.75
Provider=HAL
ID=Stub
[Source43]
Name=IsGpuLimited80
Units=
Format=
Formula=(msGpuActive / msBetweenPresents) >= 0.80
Provider=HAL
ID=Stub
[Source44]
Name=IsGpuLimited85
Units=
Format=
Formula=(msGpuActive / msBetweenPresents) >= 0.85
Provider=HAL
ID=Stub
[Source45]
Name=IsGpuLimited90
Units=
Format=
Formula=(msGpuActive / msBetweenPresents) >= 0.90
Provider=HAL
ID=Stub
[Source46]
Name=IsGpuLimited95
Units=
Format=
Formula=(msGpuActive / msBetweenPresents) >= 0.95
Provider=HAL
ID=Stub
[Source47]
Name=LimitSwitch
Units=
Format=
Formula=""IsGpuLimited80"+"IsGpuLimited95""
Provider=HAL
ID=Stub
[Table0]
Name=SampleTable
ValueSize=50
Lines=2
Line0Name=HDD1
Line0Cells=3
Line0Cell0Source=HDD1 usage
Line0Cell1Source=HDD1 read rate
Line0Cell2Source=HDD1 write rate
Line1Name=NET1
Line1Cells=3
Line1Cell0Text=
Line1Cell1Source=NET1 download rate
Line1Cell2Source=NET1 upload rate
[Table1]
Name=Core Loads
Lines=1
Line0Name=Core 2
Line0Cells=3
Line0Cell0Source=CPU1 usage
Line0Cell1Text=
Line0Cell2Source=CPU2 usage
[Layer0]
Name=CPU clock
Text=AVG CPU/ECLK:%CPU clock%/%CPU AVG ECLK% MHz
PositionX=-10
PositionY=-10
ExtentX=-27
ExtentY=-1
ExtentOrigin=3
FixedAlignment=1
TextColor=7FC1FF
MarginLeft=2
[Layer1]
Name=CPU Name
Text=%CPU%:
PositionX=0
PositionY=-9
ExtentX=-17
ExtentY=-1
ExtentOrigin=3
TextColor=80FF
RefreshPeriod=500
MarginLeft=2
[Layer2]
Name=GPU Name
Text=%GPUFull%:
PositionX=0
PositionY=0
ExtentX=-23
ExtentY=-1
ExtentOrigin=3
FixedAlignment=1
TextColor=D37C7C
MarginLeft=2
[Layer3]
Name=Framerate graph
Text=<G=Framerate,-32,-3,2,0,200,0> <S=50><A=-1><FR> FPS<A=0><S>
PositionX=0
PositionY=-4
ExtentX=-37
ExtentY=-1
ExtentOrigin=3
FixedAlignment=1
TextColor=FF8080
BgndColor=28FF8080
[Layer4]
Name=CPU Usage
Text=UTIL:<A=0>%CPU usage%<A=0>%
PositionX=0
PositionY=-10
ExtentX=-10
ExtentY=-1
ExtentOrigin=3
TextColor=7FC1FF
MarginLeft=2
[Layer5]
Name=Application type
Text=<APP>
PositionX=0
PositionY=-2
ExtentX=-9
ExtentY=-1
ExtentOrigin=4
TextColor=D37C7C
[Layer6]
Name=GPU usage
Text=LOAD:%GPU1 usage%%/%GPU1 power% W
PositionX=0
PositionY=-1
ExtentX=-15
ExtentY=-1
ExtentOrigin=3
TextColor=CD6637
MarginLeft=2
[Layer7]
Name=VRAM load
Text=VRAM:%GPU1 memory usage% GB
PositionX=-23
PositionY=0
ExtentX=-14
ExtentY=-1
ExtentOrigin=4
TextColor=A351
[Layer8]
Name=Frametime graph
Text=<G=Frametime,-32,-2,2,0,100,0> <S=50><A=-1><FT> ms<A=0><S>
PositionX=0
PositionY=-7
ExtentX=-37
ExtentY=-2
ExtentOrigin=3
TextColor=FF8080
BgndColor=28FF8080
RefreshPeriod=100
[Layer9]
Name=
Text=GPU/MEM:%GPU1 clock%/%GPU1 memory clock% MHz
PositionX=-15
PositionY=-1
ExtentX=-22
ExtentY=-1
ExtentOrigin=3
TextColor=CD6637
MarginLeft=2
[Layer10]
Name=CPU temp
Text=Tdie:%CPU temperature%°C
PositionX=-27
PositionY=-9
ExtentX=-10
ExtentY=-1
ExtentOrigin=4
TextColor=7FC1FF
[Layer11]
Name=CPU PPT
Text=PPT:%CPU power% W
PositionX=-17
PositionY=-9
ExtentX=-10
ExtentY=-1
ExtentOrigin=4
TextColor=7FC1FF
[Layer12]
Name=
Text=AVG:<FR> FPS
PositionX=0
PositionY=-3
ExtentX=-12
ExtentY=-1
ExtentOrigin=3
TextColor=FFFFFF
MarginLeft=2
[Layer13]
Name=
Text=Te:%GPU1 temperature%°C
PositionX=-9
PositionY=-2
ExtentX=-9
ExtentY=-1
ExtentOrigin=4
TextColor=CD6637
[Layer14]
Name=
Text=Tj:%GPU1 temperature 2%°C
PositionX=-18
PositionY=-2
ExtentX=-9
ExtentY=-1
ExtentOrigin=4
FixedAlignment=1
TextColor=CD6637
[Layer15]
Name=
Text=V:%GPU1 voltage% mV
PositionX=-27
PositionY=-2
ExtentX=-10
ExtentY=-1
ExtentOrigin=4
TextColor=CD6637
[Layer16]
Name=
Text=1%:<FR10L> FPS
PositionX=-12
PositionY=-3
ExtentX=-12
ExtentY=-1
ExtentOrigin=1
TextColor=FFFFFF
[Layer17]
Name=
Text=0.1%:<FR01L> FPS
PositionX=-24
PositionY=-3
ExtentX=-13
ExtentY=-1
ExtentOrigin=1
FixedAlignment=1
TextColor=FFFFFF
[Layer18]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=<G=msBetweenPresents,0,0,2,0,50,0>
PositionX=0
PositionY=-13
ExtentX=-32
ExtentY=-4
ExtentOrigin=0
TextColor=D37C7C
BgndColor=28D37C7C
[Layer19]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=GPU BUSY RATE:%GPU Busy rating%%
PositionX=0
PositionY=-12
ExtentX=-19
ExtentY=-1
ExtentOrigin=3
TextColor=D37C7C
MarginLeft=2
[Layer20]
Name=
Text=<SWITCH Status><CASE 0><IF !IsPresentMonDataAvailable>Waiting for PresentMon data stream...<IF><CASE 1>PresentMon service is not available...<CASE 2>Failed to start PresentMon data stream...<CASE 3>PresentMon data stream failed...
PositionX=0
PositionY=-11
ExtentX=-37
ExtentY=-6
ExtentOrigin=4
FixedAlignment=1
TextColor=D37C7C
[Layer21]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=LAG: %msReportingLag% ms
PositionX=-19
PositionY=-12
ExtentX=-18
ExtentY=-1
ExtentOrigin=1
TextColor=D37C7C
[Layer22]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=GPU BUSY
PositionX=-32
PositionY=-15
ExtentX=-5
ExtentY=-1
ExtentOrigin=7
Size=50
TextColor=A351
BgndColor=2800A351
[Layer23]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=PRESENT
PositionX=-32
PositionY=-14
ExtentX=-5
ExtentY=-1
ExtentOrigin=1
Size=50
TextColor=BD6F6F
BgndColor=28D37C7C
[Layer24]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=%msBetweenPresents% ms
PositionX=-32
PositionY=-13
ExtentX=-5
ExtentY=-1
ExtentOrigin=7
Size=50
TextColor=D37C7C
BgndColor=28D37C7C
[Layer25]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=%msGpuActive% ms
PositionX=-32
PositionY=-16
ExtentX=-5
ExtentY=-1
ExtentOrigin=1
Size=50
TextColor=A351
BgndColor=2800A351
[Layer26]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text="PRESENTMON DATA: "
PositionX=0
PositionY=-11
ExtentX=-37
ExtentY=-1
ExtentOrigin=4
TextColor=D37C7C
[Layer27]
Name=
VisibilitySource=IsPresentMonDataAvailable
Text=<G=msGpuActive,0,0,2,0,50,0>
PositionX=0
PositionY=-13
ExtentX=-32
ExtentY=-4
ExtentOrigin=0
TextColor=A351

Copy spoiler content in new text file, rename it so extension would be *.ovl, then load it as overlay in RTSS.
I couldn't post it as file here, sadly.

You may want to tinker with ECLK provider, as mine is set up for 8c16t CPU

 

There also a shadow glitch with RT
Make sure you bug report it every driver its not fixed and select this driver and previous drivers

 

What is this Presentmon?

 

"PresentMon is a set of tools to capture and analyze the high-level performance characteristics of graphics applications on Windows. PresentMon traces key performance metrics such as the CPU, GPU, and Display frame durations and latencies; and works across different graphics API such as DirectX, OpenGL, and Vulkan, different hardware configurations, and for both desktop and UWP applications."
Basically open-source data polling tool.
Recently Intel added ability to view "GPU Busy" metric through data pulled from PresentMon.
RTSS developer implemented PresentMon tool as data source within it's application. Intel also made own Intel PresentMon overlay. (compatible with every card vendor). Unlike RTSS one which can only grab instant values, Intel overlay is allowing you to show short-timed averages or instananeous values, meaning you will see bit more divided and clear values compared to "raw"
GPU Busy metric shows how much of a time GPU specifically spent processing a frame.
Combined with presentation time (bit different from default frametime presentation in RTSS), but core is similar, as it is delta between two frames), it allows to judge how big of a frametime part GPU took specifically.
I use GPU Busy to Presentation Time ration to show simpler number for this. Closer to 100% better, if above it means that CPU queues frames and potentially increased latency may be expected, if below 100% it means that CPU process frame longer than GPU does, or, probably it can also hint on vsync or framerate lock if such is enabled (because presentation time doesn't mean CPU time, just overall time).

But for specific details present time to gpu busy time graph is also implemented on overlay. Looking at that you can check if issue lies in optimisation, load balancing or something else.
For example vsync pattern


BG3 Act 3 - mostly GPU limitation (basically most of the times if you don't move camera). Still CAN be beter though, but GPU busy time graph is balanced, so game utilises GPU quite evenly.


BG3 Act 3 - turning it into CPU limitation (active camera movement) [do not forget about forced 3 second data collector lag]


Basically it means that either CPU optimisation could be better (if there are complete spikes of abnormality) or that you could benefit from better CPU. In this case it is loading and unloading crowd that loads CPU
PoE graphs - mostly GPU limit (but it can easily turn into CPU limit btw). Spikes are loading of new zone, and CPU is definitely being hit harder, even more so than GPU switching full set of assets.

23.9.2 is being compiled, waiting for release. Can be RX 7700/7800 only

 

It's from the old 23.10.xxx branch, so it could be for the other cards. 7700/7800 cards have 23.9.1 from the new branch with a higher driver version.

 

Well, i am not checking branches... Too tasking for me to do). Thanks for that, then will look forward to.