lekt player

Anything to do with computer audio, hardware, software etc.
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

lekt wrote:tested c7 MOD14. many things improved, but balance a bit still not good, maybe raised by tempo.

uploaded:
v2.99.9 256 noind
v2.99.9 256 noiman

try them.
v2.99.9 256 noind sounds like new best. all things great!
test more later

---
try c7 MOD16 (MOD15 is less bass more ease, treble better)
pls test MOD16 or MOD15. new stack & heap MOD
lekt
Posts: 1102
Joined: Thu Oct 10, 2013 4:52 am

Re: lekt player

Post by lekt »

v2.99.9 256 noind and noiman have very good code, all testing compiled versions in different OS environment have ~ same sound, same good balance.

changed all variables from pointer * type to numeric data type, render loop use almost all variables, not free/redundant variables of function. seems i'm not wrong, pointer is simulation of number, it's simple basically number. compiler may compile it to not good things vs compiling operators with numbers.

noind and noiman: with option /EHsc, not good compiled as c7. will recompile. i sure that it's code better than c7.
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

uploaded lekt.exe2.99.9 256 noind MOD15


MOD15 is my favorite. it makes piano notes faster sharper more solid more body more correct. treble better. ease much better. unforturnately bass notes are lacking -> sometimes dry

original PE is best possible for bass. no need mod.
either original or the other.... difficult to choose.

noind is new best... music more correct. my new favorite.

noind MOD15 all things better > c7 MOD15. except for musicality needs more testing
lekt
Posts: 1102
Joined: Thu Oct 10, 2013 4:52 am

Re: lekt player

Post by lekt »

i feel /EHsc not good, not natural sound. think need remove this option, back to default.
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

lekt wrote:i feel /EHsc not good, not natural sound. think need remove this option, back to default.
don't think /EHsc is needed
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

lekt wrote:...
changed all variables from pointer * type to numeric data type, render loop use almost all variables, not free/redundant variables of function. seems i'm not wrong, pointer is simulation of number, it's simple basically number. compiler may compile it to not good things vs compiling operators with numbers.
.
simple is good! probably makes assembly code more easy to read, and OPTIMIZE in the future!
lekt
Posts: 1102
Joined: Thu Oct 10, 2013 4:52 am

Re: lekt player

Post by lekt »

jesuscheung wrote:uploaded lekt.exe2.99.9 256 noind MOD15


MOD15 is my favorite. it makes piano notes faster sharper more solid more body more correct. treble better. ease much better. unforturnately bass notes are lacking -> sometimes dry

original PE is best possible for bass. no need mod.
either original or the other.... difficult to choose.

noind is new best... music more correct. my new favorite.

noind MOD15 all things better > c7 MOD15. except for musicality needs more testing
very good, 2 sections are removed, sound stage more correct, cleaner. bass lesser, a bit hard but can be accetpted.

in compiler if i use /DYNAMICBASE:NO then section .realloc removed (sound more clean but less organic). => try use this for MOD, coz OS generation will lookup .realloc section, not found and angry!!
-----------------------
using /DYNAMICBASE:NO have big benefit, almost memory addresses are fixed, good aligned. useful for tricks. why lesser organic? will learn more...
for you if remove .realloc then what is needed next? what things are stripped?
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

uploaded MOD17

my best try to keep all bass! try test the bass. can be better. should be tighter more free bass. coz i mod the stack to work harder
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

lekt wrote: for you if remove .realloc then what is needed next? what things are stripped?
in "CFF explorer"
"data directories" -> "relocation.." -> right click -> "remove data directionary"
"section headers" -> right click .reloc -> delete header + section

make sure
relocation stripped = enabled (i see some software=disabled is ok... not sure how to do this yet)
large address = enabled (NO NEED for S4 coz you said pointer is 4bytes or something)

don't rebuild anything -> BAD BAD BAD
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: lekt player

Post by jesuscheung »

is it some way you code makes heap and stack loves the number 16? not 8, 4, 2, 32... 16 is magic

found reason for hardness in MOD15. maybe fixable. stack and heap are correlated in some ways.
Post Reply