Page 1 of 1

Grain [RE] quality improves with Hyperthreading on

Posted: 16 Dec 2017
by Re8et
I noticed this. I had some glitches problems playing a melody on the Grain Spectral String patch, poly, 8 voices.
So I switched to legato and 6 voices.
Nothing else in the song, DSP bar is up to two, except multiple BUSes set up already as a proper mixer interface.
Then I switched to Hypethread on and the sound from Grain widened a bit. Not a lot but it was clearly audible.
Anyone noticed differences in sound quality such as this??

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 16 Dec 2017
by Loque
Widening? Maybe phasing due to latency?

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 16 Dec 2017
by Re8et
It's like adding an Ozone maximizer, now I'm earing the difference on a whole arrangement, it's like 24 bit, HD.
It just affects the whole ranges, it's doing something on quality, also the notes are snappier, so latency may be one thing.
More clarity.

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 16 Dec 2017
by normen
Its more probable that for some reason your sound quality DEGRADES when you turn off hyperthreading.

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 16 Dec 2017
by selig
To clarify, you are comparing exports side by side and they sound different? Would be interested in what the difference is if you care to share.

Otherwise, sound memory is difficult to hold for more than a few seconds, which is why doing A/B comparisons with a quick one-click comparison. The more subtle the difference, the more important to compare the same exact measure of music with a single switch - there are a few ways of doing this from looping the audio to offsetting one of the samples by a bar or so to allow switching and hearing the same exact piece of music. If you don’t do this, you can’t be sure of what difference you may be hearing.
:)


Sent from some crappy device using Tapatalk

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 17 Dec 2017
by Re8et
Link to download loop rendering with Hyperthread on/off:
https://www.dropbox.com/s/xmzq75lintezs ... T.zip?dl=0

I opened the two output in audacity and there's differences.
HYPER HTREAD LOOP.JPG
HYPER HTREAD LOOP.JPG (93.29 KiB) Viewed 793 times
non hyperthread analysis of Grain loop
mon hype grain.JPG
mon hype grain.JPG (42.4 KiB) Viewed 787 times
vs Hyper thread analysis
hype grain.JPG
hype grain.JPG (41.63 KiB) Viewed 787 times

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 17 Dec 2017
by avasopht
That's not indicative of anything yet.

Just post some raw waves.

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 17 Dec 2017
by selig
avasopht wrote:That's not indicative of anything yet.

Just post some raw waves.
Agreed, if only because there is no control in your test. The control in this case would be two renders with the exact same settings, which would indicate if there are any random elements in the patch in question, such as a free running LFO or reverb or similar.

Yes your results show a difference, but no, we cannot safely assume the difference is in hyper-threading - must remove all variables first.
:)


Sent from some crappy device using Tapatalk

Re: Grain [RE] quality improves with Hyperthreading on

Posted: 18 Dec 2017
by Re8et
The samples analyzed with scope are from the GRAIN only, however reverb was left on.
Grain sample is part of the loops, fx were taken out except internal lfo and reverbs were forgotten.
Audacity first test was done with the loops provided. Agree it was non indicative. I later added the scope test and the sample used for the Audacity test was dry signal with Grain plus others RE too.
Re-running the test however with a raw sample just yield the same scope results as above.