Looking for intrinsic "least significant bit" on Visual Studio

[OliverBr]

Hello,
For gcc/clang both intrinsic functions are available for most and least significant bit (leading/trailing zeros):

Code: Select all

__builtin_clz 
__builtin_ctz

For Visual Studio 2019 I have found an intrinsic function for most significant bit (leading zeros)

Code: Select all

__lzcnt64

A function named "__tzcnt64" doesn't exist. (trailing zeros)

Is there a function to get LSB in Visual Studio?

[towforce]

It looks like a language issue rather than visual studio. In C derived languages, it's normally something like:

b [0] = 0 != (v & 1);
b [1] = 0 != (v & 2);
b [2] = 0 != (v & 4);
b [3] = 0 != (v & 8);


etc.

Basically 0 != (v & Math.Pow(2, bitnumber))

[OliverBr]

It looks like a language issue rather than visual studio. In C derived languages, it's normally something like:

b [0] = 0 != (v & 1);
b [1] = 0 != (v & 2);
b [2] = 0 != (v & 4);
b [3] = 0 != (v & 8);


etc.

Basically 0 != (v & Math.Pow(2, bitnumber))

I fear you haven't got my question correctly.

My issue is not a language issue, it is really a compiler issue:

1) I have already working methods in language C to get the least significant bit.
2) In order to improve performance I am looking for an intrinsic function which is compiler dependent.
3) In gcc there are two functions,
4) In Visual Studio I have just found one.
5) I am looking for the second intrinsic function in Visual Studio other than "__lzcnt64"

[mar]

this is not exactly the same but you may be looking for _BitscanForward64 and _BitscanReverse64, which translate to bsf/bsr x86 instructions
I believe __lzcnt64 is available on relatively new machines only
bsf/bsr both return the index of the least (forward) or most (reverse) bit, but is undefined for zero, so you should explicitly check for 0 yourself

[chrisw]

this is not exactly the same but you may be looking for _BitscanForward64 and _BitscanReverse64, which translate to bsf/bsr x86 instructions
I believe __lzcnt64 is available on relatively new machines only
bsf/bsr both return the index of the least (forward) or most (reverse) bit, but is undefined for zero, so you should explicitly check for 0 yourself

If I remember correct from 1997, we were using beta dev version of what became Visual Studio and CSTal had the asm instruction in some sort of macro. I’ll look up what I do now when I get home, but am fairly sure there’s a standard fast way that works, may well be the Mar instructions above, and I do think you indeed have to check you’re giving them non zero data.

Btw, are you compiling SF NNUE in VS, I got stuck on compile errors demanding c++17 support?

[OliverBr]

this is not exactly the same but you may be looking for _BitscanForward64 and _BitscanReverse64, which translate to bsf/bsr x86 instructions
I believe __lzcnt64 is available on relatively new machines only
bsf/bsr both return the index of the least (forward) or most (reverse) bit, but is undefined for zero, so you should explicitly check for 0 yourself

Thank you. _BitscanForward64 helps me a bit, even it is not as fast as __lzcnt64. (< 10% gain)

I still don't get, why there is no "__tzcnt64" in Visual Studio, as the processors that support "lzcnt" do automatically support "tzcnt".
See also:
https://www.felixcloutier.com/x86/lzcnt
https://www.felixcloutier.com/x86/tzcnt

Again, gcc/clang do support both commands. I get great results in Linux and MacOSX. (> 20% gain)

[towforce]

Btw, are you compiling SF NNUE in VS, I got stuck on compile errors demanding c++17 support?

Some projects require the C++ redistributable. If so, you can get it here.

[OliverBr]

If I remember correct from 1997, we were using beta dev version of what became Visual Studio and CSTal had the asm instruction in some sort of macro. I’ll look up what I do now when I get home, but am fairly sure there’s a standard fast way that works, may well be the Mar instructions above, and I do think you indeed have to check you’re giving them non zero data.

An asm code snippet for Visual Studio would be very helpful. This is the machine command:
https://www.felixcloutier.com/x86/tzcnt

(the parameter is never 0)

Btw, are you compiling SF NNUE in VS, I got stuck on compile errors demanding c++17 support?

No, I am developing OliThink572 using "tzcnt" and looking for a solution for my Windows executables. _BitscanForward64 works, but is slower.

[Gerd Isenberg]

Hello,
For gcc/clang both intrinsic functions are available for most and least significant bit (leading/trailing zeros):

Code: Select all

__builtin_clz 
__builtin_ctz

For Visual Studio 2019 I have found an intrinsic function for most significant bit (leading zeros)

Code: Select all

__lzcnt64

A function named "__tzcnt64" doesn't exist. (trailing zeros)

Is there a function to get LSB in Visual Studio?

Maybe this helps
https://developercommunity.visualstudio ... clang.html
https://github.com/gcc-mirror/gcc/blob/ ... miintrin.h

[Gerd Isenberg]

I still don't get, why there is no "__tzcnt64" in Visual Studio, as the processors that support "lzcnt" do automatically support "tzcnt".
See also:
https://www.felixcloutier.com/x86/lzcnt
https://www.felixcloutier.com/x86/tzcnt

Again, gcc/clang do support both commands. I get great results in Linux and MacOSX. (> 20% gain)

The reason might be that AMD's Advanced Bit Manipulation had only _lzcnt64, and trailing zero count came later with Intel's BMI1.