Intel can squeeze XE igpu even in 14nm RKL, ryzen is nope, damn...

For Windows 10, you can install it without license, can use as long as you want if dont customize it and can live with the watermark lol.

blue screen related to ram oc are totally random, there is no fixed error message, each time is different

GPU has more cores and faster than CPU, most expensive instruction are running on GPU, at higher resolution are GPU bound.
hence, games and mining are running on GPU, even your browser already utilize your GPU more than CPU now, because of ever increasingly complex web content.

Directx 12 and Vulkan made a lot of change to ensure low driver overhead, more draw call.
CPU wont be fully utilized by games, it will just need to be fast enough to pass the commands to GPU and run the games codes that are still in x86 instructions.

Holding too much grudge is not good for health 🤔

Welp, design choice, limitation of the game engine, optimization too expensive, deadline constrain etc, thus the cost are passed down to end user, us.
Game developer decided to stick with D3D12 because they already familiar with it, making new engine or integrate new stuff like DXR will save them a lot of time, result in shorter development time,
unless the team are really expert and decided to use alternative like Vulkan, people who played game with vulkan (Bethesha Id Tech 6)enabled will know you can just crank the graphic quality preset to highest level and your gpu still can take it like a champ.

You can even offloading the AI to cloud compute, Azure cloud, like the Titanfall and the sequel, resulted in less strain on end user system.

with the ever expensive hardware and scalper practice, Geforce now and Stadia starting to make more sense now lol

All hardware have bugs, some have negligible impact on performance but for developer are big headache because they have to read the CPU errata (some are really long lol) to get a workaround for the cpu bug, hence do not buy the cpu and mobo when they hit the shelve because rev1 always have bug.
they will fix it on later cpu stepping, driver, mirocode and BIOS update

Just dont be early bird , because your are always their guinea pigs and free QA tester lol

CPU dont die that easily, i had my Phenom II 1050T 2.8Ghz overclocked to 3.7GHz for many years. until first gen Zen came out lol, overclocked with stock voltage.
things will get dangerous if you increase voltage too much.

Singapore here selling at 900 MYR, 329 SGD.
I mean damn, if want cheaper have to look somewhere else, eg newegg or carosell

buy from brick and mortar shop always never cheap because their rental is absurdly high, they still have supporter from local.
Good thing is easy RMA, that's all, sometimes they do give discount if you buy in store, and give you advice if you're clueless.

Carousell pc part buyer usually only for those who know what they wants, but RMA is pain in the butt lol

The first thing you should do is check the power plan you're using, AMD chipset driver install 2 additional powerplan that boost responsiveness at the cost of your power bill
Set it to balanced mode first before fiddle with BIOS.

to avoid ramp up and down, it's always recommend to use pwm fan, because pwm has 0 to 100 fan duty, which is wider than voltage controlled fan,
voltage controlled fan start at 5 to 7v, full speed at 12v, this is where the annoying ramp came from.
I have 4500rpm gentle typhoon lol, most mobi will not be able to tame this wild beast due to high start up voltage which is 7v, 1900rpm...

my old server fan can spin even at 800rpm. I can choose 7v and it will not go nuts when I boot up my pc

What? An android with USB alt mode display out can do it too🤔😹

Music playback use fraction of the cpu power, youtube vp9 already hardware accelerated.

use a chopstick man, or get a tweezers lol, my job most of the time required me to repeatedly install and remove certain stuff but with a lot of connectors attached, i have to use super long tweezers to get the job done, else 30 minutes job become 90 minutes.

no strange at all, architecture specific optimisation is needed for each generation of processors to make full use of the cpu capability.
developer can just use the compiler preset to fine tuning, but inside has a lot of things like enabling and disabling specific parameters. each generation of cpu has different cpu errata that are required developers to read and disable some stuff to avoid the program behave strangely after enabling optimization

yes, depend how much of the optimization you wish to enable, you can just enable i686 and it will run fine from i686 until the latest Rocket Lake and Zen 3, but you wont be able to make a full use of your cpu.
normal application can just use generic optimization aka Pentium 4 or i686, but for some program optimization is needed, trust me dude, like video encoder, operating system.
i've been compiling stuff since x264/ffdshow era until now, i compile gentoo linux lol

compiler optimization still no match for hand tuned optimization (assembly code), without it the program will run 10 times slower.

AMD and Intel doesn't really hide anything, both always publish their cpu optimization, as well as errata guide in PDF.
one of the example https://developer.amd.com/resources/develop...guides-manuals/
Both have developer portal for a reason.

Most program will run fine and is transparent to end user, but internally they detect all the stuff in your PC and enable/disable stuff to make full use of your cpu.

As i said before, you can use generic optimization and your program will still run from P4 until latest gen, but it wont be as fast with architecture specific optimization enabled.
https://www.rarewares.org/ogg-oggenc.php this Vorbis encoder is one example.

every new gen will come with at least few new SIMD, we bought the expensive CPU, ain't we gonna fully utilize it all? other wise what's the point of using new CPU?

bruh, most video are hardware accelerated. unless your gpu and web browser are ancient one.
jav still mostly using h264/AVC standard, should be working with much older gpu.

I have 7 tab of full hd youtube open while benchmarking Borderlands 3, the fps does not drop much because video decoding engine are on separated core and does not interfere with the main 3D rendering.
Forced to h264 as vp9 does not work with old gpu.

you mixed up with compiler error and compiler generated code error.
Unless your code are really complex like ffdshow. most compiler will just work fine whatever you throw at it.
Most dev will just use quick hax to workaround the compiler bug.

for compatibility reason, dev can just write complex cpu detection to avoid certain SIMD crashing on older CPU, like most of the programs we're using.

Node npm use Javascript and python are inteperted language, just update the the compiler would do and they are platform independent.
I'm more on multimedia side and recently just started trying gentoo Linux on vm, and macos bigsur, totally running fine . Except some hiccup such as crash after updating the vm os. At least I don't see any bsod anymore

like car's speedometer, the limit is 300km/h but does not mean you car can really reach the maximum speed
Mine is 3700X on B350M Mortar Arctic.
1000W PPT (9%)
95A TDC (59%)
140A EDC (80%)
Typical usage not really that high, the number on the brackets are actually running prime95 inplace large FFT 16 threads.