CPU underclock had more clear benefits than RAM underclock, these benefits were easier to appreciate when other areas of the PC were improved but with lower speeds the sound gets softer in a way I dont like, I think there is an overlap of benefits and negatives, smoother, more natural but lower clock speed reduces energy or ...speed
of the music.

After some more tweaking I think voltage plays a very big part in how the CPU affects sound.
With simple underclocking there must be some undervolting happening too, with minimum 800MHz CPU underclock and ''auto'' setting for core voltage in BIOS, the core voltage reduces to around 0.9V from 1.1V @ 3.5GHz. Only with a fixed core voltage can you hear the actual effect of clock speed on sound.
With manual control it can be set much lower to 0.7V at a maximum 1600MHz, which sounds more vibrant and at least as smooth as 0.9V at 800MHz. I have yet to compare 800MHz vs 1600MHz at a fixed voltage or the minimum core voltage 0.6V.

The RAM voltage also has a big influence on sound but its very sensitive and not as clear cut, for some reason only slightly overvolting sounded like an improvement, a 50-100mV increase to default 1.2V.

I hope this general PC testing is ok to post here, if someone will even find it useful (other MoBo manufacturers BIOS may do things differently to Asus).

Anybody prefer not to use upsampling? (the question aimed more at DS DAC users)

I have trouble deciding which I prefer, upsampling is revealing, clinical (in a good way) but.... unnatural? dead? basic terms to describe complex sound is a bit pointless....
there is catch to this, lower sampling rates means different,lower buffer and SFS settings are possible so its hard to fairly compare when there are possible variables to the ''best'' settings depending on rate.
With Kernel streaming it only applies to buffer as lowest possible SFS works no matter what rate (so not a huge deal) but with WASAPI it is limited..
This is something that makes WASAPI special, and makes in my case special sound possible.
with no upsampling 1.51 SFS is possible versus 12.26 with 8X, with KS the choice of SFS resulted in different sound rather than better or worse but with WASAPI I notice distinctive improvement to timbre with the lower SFS, there is also typical SFS differences, I think the higher SFS have more space and open sound, lower more dense presentation.
This could all be imagination though 

I hadnt, thanks. its not working, so that answers that 

foobar resampler doesnt support 705.6 so no other way to test it. there exists some 768kHz audio test file online but no 705.6.
you are correct about number error, screenshot is attached.

A future driver or firmware update for interface could fix the problem, thats the best i can hope for.

I have tried many different tracks, wav, flac, mp3, 88.2kHz, some stored on different drives, nothing works. this was just most recent track in playlist.

It might actually be this USB interface im using https://www.diyinhk.com/shop/audio-kits/107-xmos-dsd-dxd-768khz-high-quality-usb-to-i2sdsd-pcb.html, if you scroll down they list all supported ''I2S signal specification'' but 705.6 is not included, yet the other 44.1 multiples are.

obviously 705.6 is supported in some capacity, it works with KS mode, ASIO and even with WASAPI in JRiver... and the DAC's displays indicates 705.6kHz  in all these cases.

Maybe this is some raw form of WASAPI support which XXHE uses for optimum SQ?

files attached.

768 is working after testing with a few different 48 and 96kHz files

btw this is with Topping D50, using DIYinHK USB to IS2 interface.

''INVALID ARG device can not play'' error pops up when trying to play any file at 705.6kHz with WASAPI, i though it was WASAPI limited to 352.8kHz but Jriver MC will upsample and play at 705.6 with WASAPI. also 768kHz does work in XXHE and it sounds great.
what could the problem be with 705.6?

Being unfamiliar with software would it be likely for current version of XXHe to work on Server 2019 or will that require some work?
and working might not mean sounding good but at least so far impressions of 2019 have been very positive.

btw would OS changes require reactivation of XXHE, so that experimenting with different OS isnt really an option?

it sounds like CPU underclocking is worth another shot then, the PC might have been using integrated graphics at the time which could explain the degraded SQ. as you say it might need experimentation, i think I just set to the lowest possible freq.

A slight RAM underclock like Peter mentions might be best, striking a balance

nah , lower frequency is a lot less resolving, it does give a  pleasant sound unlike an underclocked CPU but not worth it at all.

Since using HQPlayer Ram had been underclocked from 2133MHz to 800MHz after reading about it reducing noise, at the time it was a definite improvement, more natural, smooth etc. no noticeable effect on loading time or PC performance

Today I reverted to 2133MHz to reevaulate with XXHE since it goes so in depth with memory settings.
Now it seems both are viable options, side to side comparison is a bit slow to tell which is best (sticking with 2133MHz for now)

Is there any recommendation or preferences for this?

no luck with that but there is no rush.

I finally started experimenting with the xtweak settings and was blown away, I tried to go to previous track with alt+v (not sure if thats the right command) and it closed, now opening gives the error:
Save or Load - Error at intializing: Value of '49' is not valid for 'Value'. 'Value' should be between 'Minimum' and 'Maximum'.
parameter name: Value Setting: 010

after clicking ok on this another error pops up:
Load Settings - An error occured loading settings

doy you want to load previously autosaved preset settings file

clicking yes and selected any of the files gives the same error.

I would reinstall XXHE but im stuck in minimizedOS, not sure what the best thing to do is


balanced load of 80 and nervous 99 was the current xtweak settings but this was also first time I pressed the alt+V shortcut