Tír Na HiFi

int add(int x, int y) {
int a, b;
do {
a = x & y; b = x ^ y; x = a << 1; y = b;
}while(b);

printf("here's something is gone?"); // i don't see anything printed.

return b;
}

tried in C, it's wrong function. if it is not wrong then always return 0, coz ...while(b) (it means if(b!=0)) then goto begin of loop, so then go down only in case b=0, and return b. hehe...
i think add operator run in CPU with co-processor, numeric data processor, no who can make function better than CPU, very complex task with bitwise operators. a+=b only can replace with bitwise operator in some simple cases depend on how they are in binary presentation. ie.
57397=1110.0000 0011.0101
1792 =0000.0111 0000.0000
matched 3 bits and can use or/xor, can take one byte for xor 1792.

"two add 1792 to ram buffer pointer instructions..."
you can learn this pointer, maybe its low part or high part always constant.

lekt wrote:int add(int x, int y) {
int a, b;
do {
a = x & y; b = x ^ y; x = a << 1; y = b;
}while(b);

printf("here's something is gone?"); // i don't see anything printed.

return b;
}

tried in C, it's wrong function. if it is not wrong then always return 0, coz ...while(b) (it means if(b!=0)) then goto begin of loop, so then go down only in case b=0, and return b. hehe...
i think add operator run in CPU with co-processor, numeric data processor, no who can make function better than CPU, very complex task with bitwise operators. a+=b only can replace with bitwise operator in some simple cases depend on how they are in binary presentation. ie.
57397=1110.0000 0011.0101
1792 =0000.0111 0000.0000
matched 3 bits and can use or/xor, can take one byte for xor 1792.

"two add 1792 to ram buffer pointer instructions..."
you can learn this pointer, maybe its low part or high part always constant.

yup, should be while a

also can use lea

uploaded:
lekt.exe v2.97.2 160
lekt.exe v2.97.3 16384

use good method for big buffer size 16384, very interesting algorithm for copy data loop. if your pointer 0x...0000 multiple of 65536 then it can work. tried with test poiner version, all big buffer sizes have 0x...0000. v2.94.1 abs can't work this 16384 size coz i used unsigned short (2 bytes) for counter. overflow.
need re-write code.
v2.97.xx don't use any counter. i see they better than v2.97 160. try and enjoy! of course if they can work, hehe...

lekt.exe v2.97.3 16384... !!
steady sound... small details are not very sharp
pops...

2.97.2... !! also steady sound
sound distribution has improved a lot!
compared to 2.97.2, now 2.97 is more like... spotlighted!

style has changed...
need a lot of testing with 2.97.xx...
vocal emotion is definitely better
weight kept
all things ok here...

2.97.2...
piano notes not bright at all!
CD sound with vinyl brightness!

2.97.2 is most easy-going + highest listenable version of all lekt
not hard/stoned
not bright
mellow/smooth while texture is full
all things vibrate
almost cannot detect earaches

this low level of earaches... was only possible with xa+7... now it's happening with my 2012 haha

width is big
bass seems to reach to very bottom
weight is about 99%-100%

today not good compiling environment, mistake. when laptop too cold, sound lost vibration and hard. v2.97.2 160 must be more vibration, in my case it have little imbalance. need review code and recompile.
need only warm CPU, not hot, not cold, fan must be sleep. hard task with my system, hehe... think need setting that nothing can affect SQ of result, only code. do you know this?

pops, weird, maybe try in other laptop/PC, think you have few. i dont' know why my system have stability pointers, lowbase address always no changed for each buffer size, for 2^n like as 256, 512,...,8192, 16384 always 0x...0000. try different sizes of 2^n, JC, if having like as 0x...0000 or 0x..XX00 then very good for lekt method now. 0x...80 (multiple of 128) good for other sizes ie. 160, 352, 224,...

lekt wrote:today not good compiling environment, mistake. when laptop too cold, sound lost vibration and hard. v2.97.2 160 must be more vibration, in my case it have little imbalance. need review code and recompile.
need only warm CPU, not hot, not cold, fan must be sleep. hard task with my system, hehe... think need setting that nothing can affect SQ of result, only code. do you know this?

pops, weird, maybe try in other laptop/PC, think you have few. i dont' know why my system have stability pointers, lowbase address always no changed for each buffer size, for 2^n like as 256, 512,...,8192, 16384 always 0x...0000. try different sizes of 2^n, JC, if having like as 0x...0000 then very good for lekt method now. 0x...80 (multiple of 128) good for other sizes ie. 160, 352, 224,...

my CPU is <40c... with overclocking to 4Ghz
i have no fan in PC case
my case has no LED light
i have 2 USB...

stability pointers..

need to find reason!

i tried turn of memory remap in bios...
my iGPU uses RAM memory... so turned off...
i also tried deleting allowedinmemorysettings from bcdedit
tried pci lock....

doesn't help
what else?

my GPU also shared system RAM, core i3 3110M old technologue.
maybe RAM capacity, or windows version, wasapi version,... i must go to win8.1.

lekt wrote:my GPU also shared system RAM, core i3 3110M old technologue.
maybe RAM capacity, or windows version, wasapi version,... i must go to win8.1.

tried 8.1 core with update... sound is more solid

don't know why i enjoy 2012 more...

8.1 has many changes in GUI... affects SQ
e.g. dpi handling is different
metro is more...
i haven't got time to tweak 8.1 properly

2012 is least to tweak... that's why i use it... i think win8/8.1 better/higher potential

apart from pops...

2.97.3 is rather richer maybe smooth
prefer 2.97.2

they have same style