Minority Clean (PC-"Enhancer"-software), a perfect placebo test...

[KSTR]

There is a guy in Japan who wrote a small programm for Windows-PCs which claims to do magical stuff.
Site is http://www.mics.ne.jp/~cdorya/MinorityClean/
Some excerpts of the page via https://www.deepl.com/de/translator

[...]
Improves the image, sound, and print quality of your PC by adjusting the CPU's registers to the memory chip's internal circuitry.
[...]
After executing "MinorityClean.exe", it will immediately change the register status of the CPU to a good one to achieve [High Quality Image], [High Quality Sound] and [High Quality Print]. No need to change the settings of Windows, it will improve the physical state of the CPU and create a better PC environment, even if it is removed by rebooting Windows.
[...]
After exe execution, it functions as a jitter stabilizer for CPU.
The loop process is repeated about every 3 seconds and affects all cores.
[...]
After running the exe, it acts as a jitter stabilizer for the CPU.
The looping process is repeated about every 3 seconds and affects all cores.
Rewrite data 3.91, strength can be changed. Default value 2 (maximum value 240)
Strength 2 gives sufficient quality for normal use. Strength 200 is for printing applications.
This time only, ASM source code for MC126 and Rewrite DLL is attached, with Japanese comments.

And of course, users in forums do swear they hear tons of differences when installed, but also assigning different sound character tags to the many dozens of releases up to now (126!).

My Bullshit-o-Meter is fully maxing out, obviously...

----------------

The interesting thing that the guy disclosed, and continues to disclose, the source code (find latest attached) which is unusual but offers further insight, quite funny actually.

The program is a background process that sleeps most of the time. Every X seconds or so it wakes up and does the "cleaning action".
The code looks quite a hack-job of i686 assembly language to me (wrapped up in PowerBasic) and when I look at it it appears to do almost nothing, mostly copying register contents onto themselves (basically a NOP instruction) in certain sequences and loops, plus other weird incrementing/shifting etc stuff.

Now my knowledge of current x686 architecture is limited (I quit x86 assembly language progamming after the 486/Pentium1 era) but nevertheIess I cannot image any way how the occasional execution of those voodoo loops would make any difference to other running threads (audio player, etc) or the OS, let alone how it could ever reduce hardware signal jitter (USB ports, etc) or "stabilize" the computer in general. wtf?

So chances are very good this all is placebo and we happen to have the best possible setup for such an experiment, by lucky chance, haha!
And of course there's neven been any kind of serious blind testing.

 

[mansr]

The code looks quite a hack-job of i686 assembly language to me (wrapped up in PowerBasic) and when I look at it it appears to do almost nothing, mostly copying register contents onto themselves (basically a NOP instruction) in certain sequences and loops, plus other weird incrementing/shifting etc stuff.

Indeed, it's mostly doing nothing in every possible way. The funniest part is that modern CPUs are likely to notice this and optimise it all away.

I cannot image any way how the occasional execution of those voodoo loops would make any difference to other running threads (audio player, etc) or the OS, let alone how it could ever reduce hardware signal jitter (USB ports, etc) or "stabilize" the computer in general. wtf?

Your imagination is in no way lacking. The most this could possibly do is cause a tiny latency blip whenever it runs.

 

[KSTR]

Forgot to add:
I would definitely advise against installing the software just out of curiosity. While the code image (the .exe) is small and mostly contains the repeated patterns seen in the source code, you never know how clever malware may hide itself -- plus there are always risks you download from a hacked alias site or something.

So, be cautiuos. I would try it only in a truly sandboxed temporary/throwaway Windows install.

 

[PierreV]

You need to see this from the processor's point of view.

The poor chap is working tirelessly, constantly switching contexts, accelerating and decelerating on the whims of its power management overlords. He may be frequently assigned to long tedious repetitive tasks. His to-do list is constantly invalidated, his speculative execution regularly discarded. The list could go on forever. As humans, we can't even imagine his stress level.

Now try to imagine how a violinist would perform if he had to simultaneously be an amazon warehouse worker, an uber driver, a deliveroo rider and, play Tartini's Devil Sonata in equilibrium on a bike, in the Sahara desert. Awful...

Obviously, if you throw in perks such as regular Geisha massages, the violinist would be more relaxed when he needs to play.

All those pointless operations and shifts are the equivalents of a (very) mild processor Geisha massage. A welcome break in the routine. No wonder the music it delivers may sound more agile, palpable, airy, authoritative and, even, musical after such a break!

 

[PierreV]

Forgot to add:
I would definitely advise against installing the software just out of curiosity. While the code image (the .exe) is small and mostly contains the repeated patterns seen in the source code, you never know how clever malware may hide itself -- plus there are always risks you download from a hacked alias site or something.

So, be cautiuos. I would try it only in a truly sandboxed temporary/throwaway Windows install.

FWIW, it is 100% safe. No fancy tricks. Just had a look at the binary because I was curious to see if some compiler/linker optimizations had modified the code.

 

[KSTR]

In earlier version we find pearls like this one:

Macro CuresPartner_4096ms_End
  !XOR Rax, Rax
  !INC Rax
  !XOR Rcx, Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !SHL Rax, cl
  !XCHG Rax, Rcx
  !CALL   PB_Delay
  NOPx16
  !XCHG ah, al
  !XCHG al, ah
  !XCHG Eax, Eax
  !XCHG Rax, Rax
EndMacro

I think I understand now. Rather than loading CX with 4096 and call PB_Delay in one go, the processor has some time to meditate over the very mantra-like RISC-style sequence of generating that number, relaxing big time ;-)

 

[Soniclife]

In earlier version we find pearls like this one:

Macro CuresPartner_4096ms_End
  !XOR Rax, Rax
  !INC Rax
  !XOR Rcx, Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !INC Rcx
  !SHL Rax, cl
  !XCHG Rax, Rcx
  !CALL   PB_Delay
  NOPx16
  !XCHG ah, al
  !XCHG al, ah
  !XCHG Eax, Eax
  !XCHG Rax, Rax
EndMacro

I think I understand now. Rather than loading CX with 4096 and call PB_Delay in one go, the processor has some time to meditate over the very mantra-like RISC-style sequence of generating that number, relaxing big time ;-)

Ah, it's making the CPU to do a meditation chant, so the bits it produces are enlightened. That's got to sound better.

 

[highpurityusbcable]

I like the name.

 

[wric01]

For golden ears only, not for measuring old ppl that are hearing impaired.