Me, myself & IT

(Diversions in) Saturating Integer Arithmetic

Introduction
Assembly Implementation for i386 Processors
Assembly Implementation for Pentium Pro Processors
Assembly Implementation for AMD64 Processors
Double-Word Assembly Implementations for i386 and AMD64 Processors
Implementation in ANSI C
Implementation in GNU C

Introduction

In 2010, John Regehr published the blog posts Fun With Saturating Arithmetic and More Saturating Arithmetic, presenting the following i386 assembly implementation for signed and unsigned 32-bit integers:

In a previous post I guessed that 91 bytes was close to the minimum size for implementing signed and unsigned saturating 32-bit addition and subtraction on x86. A commenter rightly pointed out that it should be possible to do better. Attached is my best shot: 83 bytes. Given the crappy x86 calling convention I’m going to guess it’s hard to do much better unless the saturating SSE instructions offer a way. Or can the branches in the signed functions be avoided? It seems like it, but I couldn’t figure out how.

sat_unsigned_add:
	movl	4(%esp), %eax
	xor	%edx, %edx
	not	%edx
	addl	8(%esp), %eax
	cmovb	%edx, %eax
	ret

sat_unsigned_sub:
	movl	4(%esp), %eax
	xor	%edx, %edx
	subl	8(%esp), %eax
	cmovb	%edx, %eax
	ret

sat_signed_add:
	movl	4(%esp), %eax
	movl	$2147483647, %edx
	movl	$-2147483648, %ecx
	addl	8(%esp), %eax
	jno	out1
	cmovg	%edx, %eax
	cmovl	%ecx, %eax
out1:	ret

sat_signed_sub:
	movl	4(%esp), %eax
	movl	$2147483647, %edx
	movl	$-2147483648, %ecx
	subl	8(%esp), %eax
	jno	out2
	cmovg	%edx, %eax
	cmovl	%ecx, %eax
out2:	ret

27 instructions in 84 bytes – not 83 bytes, as John Regehr stated.

Note: due to the (ab)use of conditional move instructions CMOVcc introduced with the Pentium Pro processor, this code but does not run on the original i386 32-bit processor!

The 4 highlighted instructions common to both signed saturating addition and subtraction can of course be relocated, thus saving 4 instructions and 16 bytes for a total of 23 instructions in 68 bytes; additionally the directions of both conditional branch instructions Jcc suit the branch predictor of super-scalar processors better:

sat_signed_add:
	movl	4(%esp), %eax
	addl	8(%esp), %eax
	jo	common
return:	ret

sat_signed_sub:
	movl	4(%esp), %eax
	subl	8(%esp), %eax
	jno	return
common:	movl	$2147483647, %edx
	movl	$-2147483648, %ecx
	cmovg	%edx, %eax
	cmovl	%ecx, %eax
	ret

Finally the branch-free implementation, also using a total of 23 instructions in 68 bytes:

sat_signed_add:
	movl	4(%esp), %eax
	addl	8(%esp), %eax
	cltd
	lea	2147483648(%edx), %edx
	cmovo	%edx, %eax
	ret

sat_signed_sub:
	movl	4(%esp), %eax
	subl	8(%esp), %eax
	cltd
	lea	2147483648(%edx), %edx
	cmovo	%edx, %eax
	ret

Assembly Implementation for i386 Processors

The following straightforward implementation uses but only 21 instructions in just 57 bytes for signed and unsigned saturating addition and subtraction, without conditional move instructions, thus runs on the original i386 32-bit processor; additionally it provides signed and unsigned saturating division, modulus and multiplication, totalling 63 instructions in 165 bytes,

# Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

# Common "cdecl" calling and naming convention for i386 platform:
# - arguments are pushed on stack in reverse order (from right to left),
#   4-byte aligned;
# - 64-bit integer arguments are passed as pair of 32-bit integer arguments,
#   low part below high part;
# - 64-bit integer result is returned in registers EAX (low part) and
#   EDX (high part);
# - 32-bit integer or pointer result is returned in register EAX;
# - registers EAX, ECX and EDX are volatile and can be clobbered;
# - registers EBX, ESP, EBP, ESI and EDI must be preserved;
# - function names are prefixed with an underscore.

.file	"i386.s"
.arch	generic32
.code32
.intel_syntax noprefix
.text
				# [esp+8] = addend
				# [esp+4] = augend
sat_unsigned_add:
	mov	eax, [esp+4]
	add	eax, [esp+8]
	sbb	ecx, ecx
	or	eax, ecx
	ret

.size	sat_unsigned_add, .-sat_unsigned_add
.type	sat_unsigned_add, @function
.global	sat_unsigned_add
				# [esp+8] = divisor
				# [esp+4] = dividend
sat_unsigned_div:
	xor	edx, edx
	mov	eax, [esp+4]
	div	dword ptr [esp+8]
	ret

.size	sat_unsigned_div, .-sat_unsigned_div
.type	sat_unsigned_div, @function
.global	sat_unsigned_div
				# [esp+8] = divisor
				# [esp+4] = dividend
sat_unsigned_mod:
	xor	edx, edx
	mov	eax, [esp+4]
	div	dword ptr [esp+8]
	mov	edx, eax
	ret

.size	sat_unsigned_mod, .-sat_unsigned_mod
.type	sat_unsigned_mod, @function
.global	sat_unsigned_mod
				# [esp+8] = multiplier
				# [esp+4] = multiplicand
sat_unsigned_mul:
	mov	eax, [esp+4]
	mul	dword ptr [esp+8]
	sbb	ecx, ecx
	or	eax, ecx
	ret

.size	sat_unsigned_mul, .-sat_unsigned_mul
.type	sat_unsigned_mul, @function
.global	sat_unsigned_mul
				# [esp+8] = subtrahend
				# [esp+4] = minuend
sat_unsigned_sub:
	mov	eax, [esp+4]
	sub	eax, [esp+8]
	cmc
	sbb	ecx, ecx
	and	eax, ecx
	ret

.size	sat_unsigned_sub, .-sat_unsigned_sub
.type	sat_unsigned_sub, @function
.global	sat_unsigned_sub
				# [esp+8] = addend
				# [esp+4] = augend
sat_signed_add:
	mov	eax, [esp+4]
	add	eax, [esp+8]
	jo	.Loverflow
.Lreturn1:
	ret
				# [esp+8] = subtrahend
				# [esp+4] = minuend
sat_signed_sub:
	mov	eax, [esp+4]
	sub	eax, [esp+8]
	jno	.Lreturn1
.Loverflow:
.if 0
	cdq
	mov	eax, 2147483648
	xor	eax, edx
.else
	sar	eax, 31
	add	eax, 2147483648
.endif
.Lreturn2:
	ret

.size	sat_signed_add, .-sat_signed_add
.type	sat_signed_add, @function
.global	sat_signed_add
.size	sat_signed_sub, .-sat_signed_sub+1
.type	sat_signed_sub, @function
.global	sat_signed_sub
				# [esp+8] = multiplier
				# [esp+4] = multiplicand
sat_signed_mul:
	mov	eax, [esp+4]
	imul	eax, [esp+8]
	jno	.Lreturn2

	mov	eax, [esp+4]
	xor	eax, [esp+8]
.if 0
	cdq
	mov	eax, 2147483647
	xor	eax, edx
.else
	sar	eax, 31
	xor	eax, 2147483647
.endif
	ret

.size	sat_signed_mul, .-sat_signed_mul+1
.type	sat_signed_mul, @function
.global	sat_signed_mul
				# [esp+8] = divisor
				# [esp+4] = dividend
sat_signed_div:
	mov	ecx, [esp+8]
	mov	eax, [esp+4]
	cmp	ecx, -1
	je	.Lspecial

	cdq
	idiv	ecx
.Lreturn3:
	ret
.Lspecial:
	neg	eax
	jno	.Lreturn3

	not	eax
	ret

.size	sat_signed_div, .-sat_signed_div
.type	sat_signed_div, @function
.global	sat_signed_div
				# [esp+8] = divisor
				# [esp+4] = dividend
sat_signed_mod:
	mov	ecx, [esp+8]
	mov	eax, ecx
	inc	eax
	jz	.Ltrivial

	mov	eax, [esp+4]
	cdq
	idiv	ecx
	mov	edx, eax
.Ltrivial:
	ret

.size	sat_signed_mod, .-sat_signed_mod
.type	sat_signed_mod, @function
.global	sat_signed_mod
.end

Assembly Implementation for Pentium Pro Processors

The following branch-free implementation of signed saturating addition, subtraction and multiplication uses conditional move instructions; the complete code for signed and unsigned saturating addition and subtraction now uses 23 instructions in 66 bytes:

# Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

.file	"i686.s"
.code32
.intel_syntax noprefix
.text
				# [esp+8] = addend
				# [esp+4] = augend
sat_signed_add:
	mov	eax, [esp+4]
	add	eax, [esp+8]
	cdq
	lea	edx, [edx+2147483648]
	cmovo	eax, edx
	ret

.size	sat_signed_add, .-sat_signed_add
.type	sat_signed_add, @function
.global	sat_signed_add
				# [esp+8] = multiplier
				# [esp+4] = multiplicand
sat_signed_mul:
	mov	eax, [esp+4]
	mov	edx, [esp+8]
	mov	ecx, eax
	xor	ecx, edx
	sar	ecx, 31
	xor	ecx, 2147483647
	imul	eax, edx
	cmovo	eax, ecx
	ret

.size	sat_signed_mul, .-sat_signed_mul
.type	sat_signed_mul, @function
.global	sat_signed_mul
				# [esp+8] = subtrahend
				# [esp+4] = minuend
sat_signed_sub:
	mov	eax, [esp+4]
	sub	eax, [esp+8]
	cdq
	lea	edx, [edx+2147483648]
	cmovo	eax, edx
	ret

.size	sat_signed_sub, .-sat_signed_sub
.type	sat_signed_sub, @function
.global	sat_signed_sub
.end

Assembly Implementation for AMD64 Processors

The following straightforward implementation uses 21 instructions in 56 bytes for signed and unsigned saturating 64-bit addition and subtraction:

# Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

.file	"amd64.s"
.arch	generic64
.code64
.intel_syntax noprefix
.text

# Unix System V calling convention for AMD64 platform:
# - first 6 integer or pointer arguments (from left to right) are passed
#   in registers RDI/R7, RSI/R6, RDX/R2, RCX/R1, R8 and R9
#   (R10 is used as static chain pointer in case of nested functions);
# - surplus arguments are pushed on stack in reverse order (from right to
#   left), 8-byte aligned;
# - 128-bit integer arguments are passed as pair of 64-bit integer arguments,
#   low part before/below high part;
# - 128-bit integer result is returned in registers RAX/R0 (low part) and
#   RDX/R2 (high part);
# - 64-bit integer or pointer result is returned in register RAX/R0;
# - 32-bit integer result is returned in register EAX;
# - registers RBX/R3, RSP/R4, RBP/R5, R12, R13, R14 and R15 must be
#   preserved;
# - registers RAX/R0, RCX/R1, RDX/R2, RSI/R6, RDI/R7, R8, R9, R10 (in
#   case of normal functions) and R11 are volatile and can be clobbered;
# - stack is 16-byte aligned: callee must decrement RSP by 8+n*16 bytes
#   before calling other functions (CALL instruction pushes 8 bytes);
# - a "red zone" of 128 bytes below the stack pointer can be clobbered.

				# rdi = augend
				# rsi = addend
sat_unsigned_add:
	add	rdi, rsi	# rdi = augend + addend
				#     = sum
	sbb	rax, rax	# rax = (sum < 2**64) ? 0 : ~0
	or	rax, rdi	# rax = (sum < 2**64) ? sum : 2**64 - 1
	ret

.size	sat_unsigned_add, .-sat_unsigned_add
.type	sat_unsigned_add, @function
.global	sat_unsigned_add
				# rdi = dividend
				# rsi = divisor
sat_unsigned_div:
	xor	edx, edx
	mov	rax, rdi	# rdx:rax = dividend
	div	rsi		# rax = dividend / divisor
				#     = quotient,
				# rdx = dividend % divisor
				#     = remainder
	ret

.size	sat_unsigned_div, .-sat_unsigned_div
.type	sat_unsigned_div, @function
.global	sat_unsigned_div
.hidden	sat_unsigned_div
				# rdi = dividend
				# rsi = divisor
sat_unsigned_mod:
	xor	edx, edx
	mov	rax, rdi	# rdx:rax = dividend
	div	rsi		# rax = dividend / divisor
				#     = quotient,
				# rdx = dividend % divisor
				#     = remainder
	mov	rax, rdx	# rax = remainder
	ret

.size	sat_unsigned_mod, .-sat_unsigned_mod
.type	sat_unsigned_mod, @function
.global	sat_unsigned_mod
.hidden	sat_unsigned_mod
				# rdi = multiplicand
				# rsi = multiplier
sat_unsigned_mul:
	mov	rax, rdi	# rax = multiplicand
	mul	rsi		# rdx:rax = multiplicand * multiplier
				#         = product
	sbb	rdx, rdx	# rdx = (product < 2**64) ? 0 : ~0
	or	rax, rdx	# rax = (product < 2**64) ? product : 2**64 - 1
	ret

.size	sat_unsigned_mul, .-sat_unsigned_mul
.type	sat_unsigned_mul, @function
.global	sat_unsigned_mul
				# rdi = minuend
				# rsi = subtrahend
sat_unsigned_sub:
.if 0
	sub	rdi, rsi	# rdi = minuend - subtrahend
				#     = difference
	cmc
	sbb	rax, rax	# rax = (difference < 0) ? 0 : ~0
	and	rax, rdi	# rax = (difference < 0) ? 0 : difference
.else
	xor	eax, eax
	sub	rdi, rsi	# rdi = minuend - subtrahend
				#     = difference
	cmovnb	rax, rdi	# rax = (difference < 0) ? 0 : difference
.endif
	ret

.size	sat_unsigned_sub, .-sat_unsigned_sub
.type	sat_unsigned_sub, @function
.global	sat_unsigned_sub
				# rdi = augend
				# rsi = addend
sat_signed_add:
	mov	rax, rdi	# rax = augend
	cqo			# rdx = (augend < 0) ? ~0 : 0
	not	rdx		# rdx = (augend < 0) ? 0 : ~0
	btc	rdx, 63		# rdx = (augend < 0) ? -2**63 : 2**63 - 1
	add	rax, rsi	# rax = augend + addend
				#     = sum
	cmovo	rax, rdx	# rax = clamped sum
	ret

.size	sat_signed_add, .-sat_signed_add
.type	sat_signed_add, @function
.global	sat_signed_add
				# rdi = dividend
				# rsi = divisor
sat_signed_div:
.if 0
	mov	rax, rdi	# rax = dividend
	cmp	rsi, -1
	je	.Lspecial	# divisor = -1?

	cqo			# rdx:rax = dividend
	idiv	rsi		# rax = dividend / divisor
				#     = quotient,
				# rdx = dividend % divisor
				#     = remainder
	ret
.Lspecial:
	neg	rax		# rax = -dividend
	not	rdi		# rdi = ~dividend
				#     = -dividend - 1
	cmovo	rax, rdi	# rax = (dividend = -2**63) ? ~dividend : -dividend
				#     = (dividend = -2**63) ? 2**63 - 1 : -dividend
				#     = clamped quotient
	ret
.else
	mov	rax, rdi	# rax = dividend
	cqo			# rdx:rax = dividend
	btc	rdi, 63		# rdi = (dividend = -2**63) ? 0 : *
	xor	rdx, rsi	# rdx = (divisor = -1) | (divisor = 0) ? 0 : *
	or	rdx, rdi	# rdx = (divisor = -1) & (dividend = -2**63) ? 0 : *
	neg	rdx		# CF = (divisor <> -1) | (dividend <> -2**63)
	adc	rax, -1		# rax = dividend - 1
				#     + ((divisor <> -1) | (dividend <> -2**63))
				#     = dividend'
	cqo			# rdx:rax = dividend'
	idiv	rsi		# rax = dividend' / divisor
				#     = quotient,
				# rdx = dividend' % divisor
				#     = remainder
	ret
.endif
.size	sat_signed_div, .-sat_signed_div
.type	sat_signed_div, @function
.global	sat_signed_div
				# rdi = dividend
				# rsi = divisor
sat_signed_mod:
.if 0
	cmp	rsi, -1
	je	.Ltrivial	# divisor = -1?

	mov	rax, rdi	# rax = dividend
	cqo			# rdx:rax = dividend
	idiv	rsi		# rax = dividend / divisor
				#     = quotient,
				# rdx = dividend % divisor
				#     = remainder
	mov	rax, rdx	# rax = remainder
	ret
.Ltrivial:
	xor	eax, eax	# rax = remainder = 0
	ret
.else
	mov	rax, rdi	# rax = dividend
	cqo			# rdx:rax = dividend
	btc	rdi, 63		# rdi = (dividend = -2**63) ? 0 : *
	xor	rdx, rsi	# rdx = (divisor = -1) | (divisor = 0) ? 0 : *
	or	rdx, rdi	# rdx = (divisor = -1) & (dividend = -2**63) ? 0 : *
	neg	rdx		# CF = (divisor <> -1) | (dividend <> -2**63)
	adc	rax, -1		# rax = dividend - 1
				#     + ((divisor <> -1) | (dividend <> -2**63))
				#     = dividend'
	cqo			# rdx:rax = dividend'
	idiv	rsi		# rax = dividend' / divisor
				#     = quotient,
				# rdx = dividend' % divisor
				#     = remainder
	mov	rax, rdx	# rax = remainder
	ret
.endif
.size	sat_signed_mod, .-sat_signed_mod
.type	sat_signed_mod, @function
.global	sat_signed_mod
				# rdi = multiplicand
				# rsi = multiplier
sat_signed_mul:
	mov	rax, rdi	# rax = multiplicand
	xor	rax, rsi	# rax = multiplicand ^ multiplier
	sar	rax, 63		# rax = (product < 0) ? ~0 : 0
	not	rax		# rax = (product < 0) ? 0 : ~0
	btc	rax, 63		# rax = (product < 0) ? -2**63 : 2**63 - 1
	imul	rdi, rsi	# rdi = multiplicand * multiplier
				#     = product
	cmovno	rax, rdi	# rax = clamped product
	ret

.size	sat_signed_mul, .-sat_signed_mul
.type	sat_signed_mul, @function
.global	sat_signed_mul
				# rdi = minuend
				# rsi = subtrahend
sat_signed_sub:
	mov	rax, rdi	# rax = minuend
	cqo			# rdx = (minuend < 0) ? ~0 : 0
	not	rdx		# rdx = (minuend < 0) ? 0 : ~0
	btc	rdx, 63		# rdx = (minuend < 0) ? -2**63 : 2**63 - 1
	sub	rax, rsi	# rax = minuend - subtrahend
				#     = difference
	cmovo	rax, rdx	# rax = clamped difference
	ret

.size	sat_signed_sub, .-sat_signed_sub
.type	sat_signed_sub, @function
.global	sat_signed_sub
.end

Note: in case of overflow, augend and saturated sum respectively minuend and saturated difference have the same sign!

Double-Word Assembly Implementations for i386 and AMD64 Processors

The real fun begins with signed and unsigned 64-bit integers on 32-bit (i386 alias x86) processors as well as signed and unsigned 128-bit integers on 64-bit (AMD64 alias x64) processors, but ends unfortunately as soon as one compares the following properly optimised assembly to the code generated by so-called optimising compilers:

# Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

.file	"clamp_64.s"
.arch	generic32
.code32
.intel_syntax noprefix
.extern	__divdi3
.extern	__moddi3
.text
				# [esp+16] = high dword of addend
				# [esp+12] = low dword of addend
				# [esp+8] = high dword of augend
				# [esp+4] = low dword of augend
sat_unsigned_add_64:
	mov	eax, [esp+4]
	mov	edx, [esp+8]	# edx:eax = augend
	add	eax, [esp+12]
	adc	edx, [esp+16]	# edx:eax = augend + addend
				#         = sum
	sbb	ecx, ecx	# ecx = (sum < 2**64) ? 0 : ~0
	or	eax, ecx
	or	edx, ecx	# edx:eax = (sum < 2**64) ? sum : 2**64 - 1
	ret

.size	sat_unsigned_add_64, .-sat_unsigned_add_64
.type	sat_unsigned_add_64, @function
.global	sat_unsigned_add_64
				# [esp+16] = high dword of multiplier
				# [esp+12] = low dword of multiplier
				# [esp+8] = high dword of multiplicand
				# [esp+4] = low dword of multiplicand
sat_unsigned_mul_64:
	push	ebx
	xor	edx, edx	# edx = 0
	mov	eax, [esp+12]	# eax = high dword of multiplicand
	cmp	edx, eax
	sbb	ebx, ebx	# ebx = (high dword of multiplicand = 0) ? 0 : ~0
				#     = (multiplicand < 2**32) ? 0 : ~0
	mov	ecx, [esp+20]	# ecx = high dword of multiplier
	cmp	edx, ecx
	sbb	edx, edx	# edx = (high dword of multiplier = 0) ? 0 : ~0
				#     = (multiplier < 2**32) ? 0 : ~0
	and	ebx, edx	# ebx = (multiplicand < 2**32)
				#     & (multiplier < 2**32) ? 0 : ~0
				#     = (product < 2**64) ? 0 : ~0

	mov	edx, [esp+16]	# edx = low dword of multiplier
	mul	edx		# edx:eax = high dword of multiplicand
				#         * low dword of multiplier
	adc	ebx, ebx	# ebx = (product < 2**64) ? 0 : *

	xchg	ecx, eax	# ecx = high dword of multiplicand
				#     * low dword of multiplier,
				# eax = high dword of multiplier
	mov	edx, [esp+8]	# edx = low dword of multiplicand
	mul	edx		# edx:eax = high dword of multiplier
				#         * low dword of multiplicand
	adc	ebx, ebx	# ebx = (product < 2**64) ? 0 : *

	add	ecx, eax	# ecx = high dword of multiplicand
				#     * low dword of multiplier
				#     + high dword of multiplier
				#     * low dword of multiplicand
#	adc	ebx, ebx	# ebx = (product < 2**64) ? 0 : *

	mov	eax, [esp+8]	# eax = low dword of multiplicand
	mov	edx, [esp+16]	# edx = low dword of multiplier
	mul	edx		# edx:eax = low dword of multiplicand
				#         * low dword of multiplier
	add	edx, ecx	# edx:eax = product % 2**64
	adc	ebx, ebx	# ebx = (product < 2**64) ? 0 : *
	jnz	.Loverflow	# product >= 2**64?

	pop	ebx
	ret
.Loverflow:
	stc
	sbb	eax, eax
	cdq			# edx:eax = 2**64 - 1
	pop	ebx
	ret

.size	sat_unsigned_mul_64, .-sat_unsigned_mul_64
.type	sat_unsigned_mul_64, @function
.global	sat_unsigned_mul_64
				# [esp+16] = high dword of subtrahend
				# [esp+12] = low dword of subtrahend
				# [esp+8] = high dword of minuend
				# [esp+4] = low dword of minuend
sat_unsigned_sub_64:
	mov	eax, [esp+4]
	mov	edx, [esp+8]	# edx:eax = minuend
	sub	eax, [esp+12]
	sbb	edx, [esp+16]	# edx:eax = minuend - subtrahend
				#         = difference
	cmc
	sbb	ecx, ecx	# ecx = (difference < 0) ? 0 : ~0
	and	eax, ecx
	and	edx, ecx	# edx:eax = (difference < 0) ? 0 : difference
	ret

.size	sat_unsigned_sub_64, .-sat_unsigned_sub_64
.type	sat_unsigned_sub_64, @function
.global	sat_unsigned_sub_64
				# [esp+16] = high dword of addend
				# [esp+12] = low dword of addend
				# [esp+8] = high dword of augend
				# [esp+4] = low dword of augend
sat_signed_add_64:
	mov	eax, [esp+4]
	mov	edx, [esp+8]	# edx:eax = augend
	add	eax, [esp+12]
	adc	edx, [esp+16]	# edx:eax = augend + addend
				#         = sum
				#         = result
	jo	.Loverflow1	# sum < -2**63?
				# sum >= 2**63?
.Lreturn1:
	ret
				# [esp+16] = high dword of subtrahend
				# [esp+12] = low dword of subtrahend
				# [esp+8] = high dword of minuend
				# [esp+4] = low dword of minuend
sat_signed_sub_64:
	mov	eax, [esp+4]
	mov	edx, [esp+8]	# edx:eax = minuend
	sub	eax, [esp+12]
	sbb	edx, [esp+16]	# edx:eax = minuend - subtrahend
				#         = difference
				#         = result
	jno	.Lreturn1	# -2**63 <= difference < 2**63?
.Loverflow1:
.if 0
	mov	eax, edx
	cdq			# edx = (result < 0) ? ~0 : 0
.else
	sar	edx, 31		# edx = (result < 0) ? ~0 : 0
.endif
	mov	eax, edx	# eax = low dword of clamped result
	xor	edx, 2147483648	# edx = (result < 0) ? -2**31 : 2**31 - 1
				#     = high dword of clamped result
	ret

.size	sat_signed_add_64, .-sat_signed_add_64
.type	sat_signed_add_64, @function
.global	sat_signed_add_64
.size	sat_signed_sub_64, .-sat_signed_sub_64+1
.type	sat_signed_sub_64, @function
.global	sat_signed_sub_64
				# [esp+16] = high dword of multiplier
				# [esp+12] = low dword of multiplier
				# [esp+8] = high dword of multiplicand
				# [esp+4] = low dword of multiplicand
sat_signed_mul_64:
	push	ebx
	mov	ebx, [esp+16]	# ebx = low dword of multiplier
	mov	eax, [esp+8]	# eax = low dword of multiplicand
	mul	ebx		# edx:eax = low dword of multiplicand
				#         * low dword of multiplier
	push	eax
	mov	ecx, edx	# ecx = low dword of "inner" product
	mov	eax, [esp+16]	# eax = high dword of multiplicand
	mul	ebx		# edx:eax = high dword of multiplicand
				#         * low dword of multiplier
	xor	ebx, ebx	# ebx = high dword of "inner" product
	add	ecx, eax
	adc	ebx, edx	# ebx:ecx = intermediate "inner" product
	mov	eax, [esp+12]	# eax = low dword of multiplicand
	mul	dword ptr [esp+24]
				# edx:eax = low dword of multiplicand
				#         * high dword of multiplier
	add	ecx, eax
	adc	ebx, edx	# ebx:ecx = final "inner" product
	push	ecx
	sbb	ecx, ecx	# ecx = 0 - carry from "inner" product
	mov	eax, [esp+20]	# eax = high dword of multiplicand
	mul	dword ptr [esp+28]
				# edx:eax = high dword of multiplicand
				#         * high dword of multiplier
	neg	ecx		# ecx = carry from "inner" product
	add	eax, ebx
	adc	edx, ecx	# edx:eax = high dword of multiplicand
				#         * high dword of multiplier
				#         + high dword of "inner" product
				#         + carry from "inner" product
				#         = high qword of (unsigned) product"

	xor	ebx, ebx	# ebx = sign of multiplier = 0
	cmp	ebx, [esp+28]
	jle	0f		# (high dword of) multiplier >= 0?

	not	ebx		# ebx = 0 - sign of multiplier = -1
	sub	eax, [esp+16]
	sbb	edx, [esp+20]	# edx:eax = high qword of product"
				#         - multiplicand
				#         = high qword of product'
0:
	xor	ecx, ecx
	cmp	ecx, [esp+20]
	jle	1f		# (high dword of) multiplicand >= 0?

	not	ebx		# ebx = (0 - sign of multiplier)
				#     ^ (0 - sign of multiplicand)
	sub	eax, [esp+24]
	sbb	edx, [esp+28]	# edx:eax = high qword of product'
				#         - multiplier
				#         = high qword of (signed) product
1:
	xor	eax, ebx
	xor	edx, ebx	# edx:eax = high qword of product
				#         ^ (0 - sign of multiplier)
				#         ^ (0 - sign of multiplicand)
	or	eax, edx	# eax = high qword of product
				#     <> (0 - sign of multiplier)
				#      ^ (0 - sign of multiplicand)
	pop	edx		# edx = high dword of (low qword of) product
	shld	ecx, edx, 1	# ecx = sign of product
	add	ebx, ecx	# ebx = sign of product
				#     <> sign of multiplier
				#      ^ sign of multiplicand
	or	eax, ebx	# eax = (-2**63 <= product < 2**63) ? 0 : *
	pop	eax		# edx:eax = (low qword of) product
	pop	ebx
	jnz	.Loverflow2	# product < -2**63?
				# product >= 2**63?
	ret
.Loverflow2:
	mov	eax, [esp+8]	# eax = high dword of multiplicand
	xor	eax, [esp+16]	# eax = high dword of multiplicand
				#     ^ high dword of multiplier
	sar	eax, 31		# eax = (multiplicand ^ multiplier < 0) ? ~0 : 0
				#     = (product < 0) ? ~0 : 0
				#     = low dword of clamped product
	mov	edx, 2147483648	# edx = ±2**31
	xor	edx, eax	# edx = (product < 0) ? -2**31 : 2**31 - 1
				#     = high dword of clamped product
	ret

.size	sat_signed_mul_64, .-sat_signed_mul_64
.type	sat_signed_mul_64, @function
.global	sat_signed_mul_64
				# [esp+16] = high dword of divisor
				# [esp+12] = low dword of divisor
				# [esp+8] = high dword of dividend
				# [esp+4] = low dword of dividend
sat_signed_mod_64:
	mov	eax, [esp+12]
	and	eax, [esp+16]
	inc	eax
	jz	.Ltrivial	# divisor = -1?

	jmp	__moddi3
.Ltrivial:
	cdq			# edx:eax = remainder = 0
.Lreturn2:
	ret

.size	sat_signed_mod_64, .-sat_signed_mod_64
.type	sat_signed_mod_64, @function
.global	sat_signed_mod_64
				# [esp+16] = high dword of divisor
				# [esp+12] = low dword of divisor
				# [esp+8] = high dword of dividend
				# [esp+4] = low dword of dividend
sat_signed_div_64:
	mov	eax, [esp+12]
	and	eax, [esp+16]
	inc	eax
	jz	.Lspecial	# divisor = -1?

	jmp	__divdi3
.Lspecial:
	cdq			# edx:eax = 0
	sub	eax, [esp+4]
	sbb	edx, [esp+8]	# edx:eax = -dividend
				#         = quotient
	jno	.Lreturn2	# dividend <> -2**63 = 2**63?
.if 0
	add	eax, -1
	adc	edx, -1		# edx:eax = -dividend - 1
				#         = 2**63 - 1
				#         = clamped quotient
.else
	not	eax
	not	edx		# edx:eax = ~dividend
				#         = 2**63 - 1
				#         = clamped quotient
.endif
	ret

.size	sat_signed_div_64, .-sat_signed_div_64+1
.type	sat_signed_div_64, @function
.global	sat_signed_div_64
.end

32 instructions in 96 bytes for signed and unsigned saturating 64-bit addition and subtraction.

Note: my related article Fast(est) Double-Word Integer Division presents properly optimised implementations of the signed 64-bit division routines __divdi3() and __moddi3() called here, while my articles Deficiencies in GCC’s code generator and optimiser and True Lies – or What LLVM claims, but fails to deliver Not quite so optimising Microsoft^® Visual C compiler document their poor implementations in the compiler runtime libraries of both GCC and LLVM.

# Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

.file	"clamp_128.s"
.arch	generic64
.code64
.intel_syntax noprefix
.extern	__divti3
.extern	__modti3
.text
				# rsi:rdi = augend
				# rcx:rdx = addend
sat_unsigned_add_128:
	add	rdi, rdx
	adc	rsi, rcx	# rsi:rdi = augend + addend
				#         = sum
	sbb	rax, rax	# rax = (sum < 2**128) ? 0 : ~0
	cqo			# rdx:rax = (sum < 2**128) ? 0 : 2**128 - 1
	or	rax, rdi
	or	rdx, rsi	# rdx:rax = (sum < 2**128) ? sum : 2**128 - 1
	ret

.size	sat_unsigned_add_128, .-sat_unsigned_add_128
.type	sat_unsigned_add_128, @function
.global	sat_unsigned_add_128
				# rsi:rdi = minuend
				# rcx:rdx = subtrahend
sat_unsigned_sub_128:
	sub	rdi, rdx
	sbb	rsi, rcx	# rsi:rdi = minuend - subtrahend
				#         = difference
	cmc
	sbb	rax, rax	# rax = (difference < 0) ? 0 : ~0
	cqo			# rdx:rax = (difference < 0) ? 0 : 2**128 - 1
	and	rax, rdi
	and	rdx, rsi	# rdx:rax = (difference < 0) ? 0 : difference
	ret

.size	sat_unsigned_sub_128, .-sat_unsigned_sub_128
.type	sat_unsigned_sub_128, @function
.global	sat_unsigned_sub_128
				# rsi:rdi = augend
				# rcx:rdx = addend
sat_signed_add_128:
	mov	r11, rdx
	mov	rax, rsi
	cqo			# rdx = (augend < 0) ? ~0 : 0
	not	rdx		# rdx = (augend < 0) ? 0 : ~0
	mov	rax, rdx	# rdx:rax = (augend < 0) ? 0 : ~0
	btc	rdx, 63		# rdx:rax = (augend < 0) ? -2**127 : 2**127 - 1
	add	rdi, r11
	adc	rsi, rcx	# rsi:rdi = augend + addend
				#         = sum
	cmovno	rax, rdi
	cmovno	rdx, rsi	# rdx:rax = clamped sum
	ret

.size	sat_signed_add_128, .-sat_signed_add_128
.type	sat_signed_add_128, @function
.global	sat_signed_add_128
				# rsi:rdi = dividend
				# rcx:rdx = divisor
sat_signed_div_128:
	mov	rax, rcx
	and	rax, rdx
	inc	rax
	jnz	__divti3	# divisor <> -1?

	mov	edx, eax	# rdx:rax = 0
	sub	rax, rdi
	sbb	rdx, rsi	# rdx:rax = -dividend
	not	rdi
	not	rsi		# rsi:rdi = ~dividend
				#         = -dividend - 1
	cmovo	rax, rdi
	cmovo	rdx, rsi	# rdx:rax = (dividend = -2**127) ? ~dividend : -dividend
				#         = (dividend = -2**127) ? 2**127 - 1 : -dividend
				#         = clamped quotient
	ret

.size	sat_signed_div_128, .-sat_signed_div_128
.type	sat_signed_div_128, @function
.global	sat_signed_div_128
				# rsi:rdi = dividend
				# rcx:rdx = divisor
sat_signed_mod_128:
	mov	rax, rcx
	and	rax, rdx
	inc	rax
	jnz	__modti3	# divisor <> -1?

	mov	edx, eax	# rdx:rax = remainder = 0
	ret

.size	sat_signed_mod_128, .-sat_signed_mod_128
.type	sat_signed_mod_128, @function
.global	sat_signed_mod_128
				# rsi:rdi = multiplicand
				# rcx:rdx = multiplier
sat_signed_mul_128:
	mov	r10, rdx	# r10 = low qword of multiplier
	mov	r11, rcx	# r11 = high qword of multiplier

	mov	rax, rcx	# rax = high qword of multiplier
	cqo			# rdx = (multiplier < 0) ? -1 : 0
	mov	r9, rdx		# r9 = (multiplier < 0) ? -1 : 0
	xor	r10, rdx
	xor	r11, rdx	# r11:r10 = (multiplier < 0) ? ~multiplier : multiplier
	sub	r10, rdx
	sbb	r11, rdx	# r11:r10 = (multiplier < 0) ? -multiplier : multiplier
				#         = |multiplier|
	mov	rax, rsi	# rax = high qword of multiplicand
	cqo			# rdx = (multiplicand < 0) ? -1 : 0
	xor	r9, rdx		# r9 = (multiplicand < 0) <> (multiplier < 0) ? -1 : 0
				#    = (product < 0) ? -1 : 0
	xor	rdi, rdx
	xor	rsi, rdx	# rsi:rdi = (multiplicand < 0) ? ~multiplicand : multiplicand
	sub	rdi, rdx
	sbb	rsi, rdx	# rsi:rdi = (multiplicand < 0) ? -multiplicand : multiplicand
				#         = |multiplicand|
	xor	ecx, ecx
	cmp	rcx, rsi
	sbb	edx, edx	# edx = (high qword of |multiplicand| = 0) ? 0 : -1
				#     = (|multiplicand| < 2**64) ? 0 : -1
	cmp	rcx, r11
	sbb	ecx, ecx	# ecx = (high qword of |multiplier| = 0) ? 0 : -1
				#     = (|multiplier| < 2**64) ? 0 : -1
	and	ecx, edx	# ecx = (|multiplicand| < 2**64)
				#     & (|multiplier| < 2**64) ? 0 : -1
				#     = (|product| < 2**128) ? 0 : -1
	mov	rax, rsi
	mul	r10		# rdx:rax = high qword of |multiplicand|
				#         * low qword of |multiplier|
	adc	ecx, ecx	# ecx = (|product| < 2**128) ? 0 : *

	mov	rsi, rax
	mov	rax, rdi
	mul	r11		# rdx:rax = low qword of |multiplicand|
				#         * high qword of |multiplier|
	adc	ecx, ecx	# ecx = (|product| < 2**128) ? 0 : *

	add	rsi, rax	# rsi = high qword of |multiplicand|
				#     * low qword of |multiplier|
				#     + low qword of |multiplicand|
				#     * high qword of |multiplier|
#	adc	ecx, ecx	# ecx = (|product| < 2**128) ? 0 : *

	mov	rax, rdi
	mul	r10		# rdx:rax = low qword of |multiplicand|
				#         * low qword of |multiplier|
	add	rdx, rsi	# rdx:rax = |product % 2**128|
				#         = |product| % 2**128
	adc	ecx, ecx	# ecx = (|product| < 2**128) ? 0 : *

	add	rax, r9
	adc	rdx, r9		# rdx:rax = (product < 0)
				#         ? ~product % 2**128 : product % 2**128
	mov	rsi, rdx	# rsi = (product % 2**128 < -2**127)
				#     | (product % 2**128 >= 2**127)
				#     ? negative : positive
	xor	rax, r9
	xor	rdx, r9		# rdx:rax = product % 2**128

	not	r9		# r9 = (product < 0) ? 0 : ~0
	mov	r8, r9		# r8:r9 = (product < 0) ? 0 : ~0
	btc	r8, 63		# r8:r9 = (product < 0) ? -2**127 : 2**127 - 1

	add	rsi, rsi
	adc	ecx, ecx	# ecx = (-2**127 <= product < 2**127) ? 0 : *
	cmovnz	rax, r9
	cmovnz	rdx, r8		# rdx:rax = clamped product
	ret


.size	sat_signed_mul_128, .-sat_signed_mul_128
.type	sat_signed_mul_128, @function
.global	sat_signed_mul_128
				# rsi:rdi = minuend
				# rcx:rdx = subtrahend
sat_signed_sub_128:
	mov	r11, rdx
	mov	rax, rsi
	cqo			# rdx = (minuend < 0) ? ~0 : 0
	not	rdx		# rdx = (minuend < 0) ? 0 : ~0
	mov	rax, rdx	# rdx:rax = (minuend < 0) ? 0 : ~0
	btc	rdx, 63		# rdx:rax = (minuend < 0) ? -2**127 : 2**127 - 1
	sub	rdi, r11
	sbb	rsi, rcx	# rsi:rdi = minuend - subtrahend
				#         = difference
	cmovno	rax, rdi
	cmovno	rdx, rsi	# rdx:rax = clamped difference
	ret

.size	sat_signed_sub_128, .-sat_signed_sub_128
.type	sat_signed_sub_128, @function
.global	sat_signed_sub_128
.end

37 instructions in 105 bytes for signed and unsigned saturating 128-bit addition and subtraction.

Note: my related article Fast(est) Double-Word Integer Division presents implementations of the signed 128-bit division routines __divti3() and __modti3() called here, while my articles Deficiencies in GCC’s code generator and optimiser and True Lies – or What LLVM claims, but fails to deliver document their poor implementations in the compiler runtime libraries of both GCC, and LLVM.

Implementation in ANSI C

Sample ANSI C implementation:

// Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

unsigned int sat_unsigned_add(unsigned int augend, unsigned int addend)
{
    const unsigned int sum = augend + addend;

    return -(sum < augend) | sum;
}

unsigned int sat_unsigned_div(unsigned int dividend, unsigned int divisor)
{
    // NOTE: overflow impossible!

    return dividend / divisor;
}

unsigned int sat_unsigned_mod(unsigned int dividend, unsigned int divisor)
{
    // NOTE: overflow impossible!

    return dividend % divisor;
}

unsigned int sat_unsigned_mul(unsigned int multiplicand, unsigned int multiplier)
{
    const unsigned long long product = (unsigned long long) multiplicand * multiplier;
    const unsigned int low = product, high = product >> 32;

    return -(high != 0U) | low;
}

unsigned int sat_unsigned_sub(unsigned int minuend, unsigned int subtrahend)
{
    const unsigned int difference = minuend - subtrahend;

    return -(difference >= minuend) & difference;
}

int sat_signed_add(int augend, int addend)
{
    const int sum = augend + addend;
#ifdef UNDEFINED_BEHAVIOUR
    const int clamp = 2147483647 + (augend < 0);
#else
    const int clamp = 2147483647 ^ (augend >> 31);
#endif
    // overflow if both augend and addend have opposite sign of sum,
    //  which is equivalent to augend has sign of addend
    //   and addend (or augend) has opposite sign of sum

    return ((augend ^ sum) & (addend ^ sum)) < 0 ? clamp : sum;
}

int sat_signed_div(int dividend, int divisor)
{
    // signed division overflows (really: raises 'division exception')
    //  only for -2**31 / -1 = ±2**31!
#if 0
    dividend -= (divisor == -1) && (dividend == ~2147483647);

    return dividend / divisor;
#else
    if (divisor == -1)
        return dividend == ~2147483647 ? ~dividend : -dividend;

    return dividend / divisor;
#endif
}

int sat_signed_mod(int dividend, int divisor)
{
    // signed division overflows (really: raises 'division exception')
    //  only for -2**31 / -1 = ±2**31!
#if 0
    dividend -= (divisor == -1) && (dividend == ~2147483647);

    return dividend % divisor;
#else
    return divisor == -1 ? 0 : dividend % divisor;
#endif
}

int sat_signed_mul(int multiplicand, int multiplier)
{
    const long long product = (long long) multiplicand * multiplier;
#ifdef UNDEFINED_BEHAVIOUR
    const int clamp = 2147483647 + ((multiplicand ^ multiplier) < 0);
#else
    const int clamp = 2147483647 ^ ((multiplicand ^ multiplier) >> 31);
#endif
    const int low = product, high = product >> 32;

    return (low >> 31) != high ? clamp : product;
}

int sat_signed_sub(int minuend, int subtrahend)
{
    const int difference = minuend - subtrahend;
#ifdef UNDEFINED_BEHAVIOUR
    const int clamp = 2147483647 + (minuend < 0);
#else
    const int clamp = 2147483647 ^ (minuend >> 31);
#endif
    // overflow if minuend has opposite sign of subtrahend
    //  and minuend has opposite sign of difference
    //   (or subtrahend has sign of difference)

    return ((minuend ^ subtrahend) & (minuend ^ difference)) < 0 ? clamp : difference;
}

GCC 11.3: 100 instructions in 252 bytes.

# Compilation provided by Compiler Explorer at https://godbolt.org/
sat_unsigned_add(unsigned int, unsigned int):
	mov	eax, [esp+8]
	add	eax, [esp+4]
	sbb	edx, edx
	or	eax, edx
	ret
sat_unsigned_div(unsigned int, unsigned int):
	mov	eax, [esp+4]
	xor	edx, edx
	div	dword ptr [esp+8]
	ret
sat_unsigned_mod(unsigned int, unsigned int):
	mov	eax, [esp+4]
	xor	edx, edx
	div	dword ptr [esp+8]
	mov	eax, edx
	ret
sat_unsigned_mul(unsigned int, unsigned int):
	push	ebx
	mov	eax, [esp+12]
	mul	dword ptr [esp+8]
	pop	ebx
	neg	edx
	mov	ecx, eax
	sbb	edx, edx
	mov	eax, ecx
	or	eax, edx
	ret
sat_unsigned_sub(unsigned int, unsigned int):
	mov	eax, [esp+4]
	mov	edx, eax
	sub	edx, [esp+8]
	cmp	eax, edx
	setbe	al
	movzx	eax, al
	neg	eax
	and	eax, edx
	ret
sat_signed_add(int, int):
	push	ebx
	mov	eax, [esp+8]
	mov	ecx, [esp+12]
	mov	ebx, eax
	lea	edx, [eax+ecx]
	sar	eax, 31
	xor	ebx, edx
	xor	ecx, edx
	xor	eax, 2147483647
	test	ebx, ecx
	pop	ebx
	cmovns	edx, eax
	mov	eax, edx
	ret
sat_signed_div(int, int):
	mov	ecx, [esp+8]
	mov	eax, [esp+4]
	cmp	ecx, -1
	je	.L17
	cdq
	idiv	ecx
	ret
.L17:
	mov	edx, eax
	neg	edx
	cmp	eax, -2147483648
	mov	eax, 2147483647
	cmovne	eax, edx
	ret
sat_signed_mod(int, int):
	mov	ecx, [esp+8]
	xor	edx, edx
	cmp	ecx, -1
	je	.L18
	mov	eax, [esp+4]
	cdq
	idiv	ecx
.L18:
	mov	eax, edx
	ret
sat_signed_mul(int, int):
	push	ebx
	mov	eax, [esp+8]
	imul	dword ptr [esp+12]
	mov	ecx, eax
	mov	ebx, eax
	sar	ecx, 31
	cmp	ecx, edx
	je	.L22
	mov	eax, [esp+8]
	xor	eax, [esp+12]
	sar	eax, 31
	xor	eax, 2147483647
	mov	ebx, eax
.L22:
	mov	eax, ebx
	pop	ebx
	ret
sat_signed_sub(int, int):
	push	ebx
	mov	eax, [esp+8]
	mov	ecx, [esp+12]
	mov	edx, eax
	mov	ebx, eax
	sub	edx, [esp+12]
	xor	ebx, edx
	xor	ecx, eax
	sar	eax, 31
	xor	eax, 2147483647
	test	ecx, ebx
	pop	ebx
	cmovs	edx, eax
	mov	eax, edx
	ret

Clang 14.0.0: 90 instructions in 236 bytes.

# Compilation provided by Compiler Explorer at https://godbolt.org/
sat_unsigned_add(unsigned int, unsigned int):
	mov	ecx, [esp+4]
	add	ecx, [esp+8]
	mov	eax, -1
	cmovae	eax, ecx
	ret
sat_unsigned_div(unsigned int, unsigned int):
	mov	eax, [esp+4]
	xor	edx, edx
	div	dword ptr [esp+8]
	ret
sat_unsigned_mod(unsigned int, unsigned int):
	mov	eax, [esp+4]
	xor	edx, edx
	div	dword ptr [esp+8]
	mov	eax, edx
	ret
sat_unsigned_mul(unsigned int, unsigned int):
	mov	eax, [esp+4]
	mul	dword ptr [esp+8]
	mov	ecx, -1
	cmovo	eax, ecx
	ret
sat_unsigned_sub(unsigned int, unsigned int):
	mov	ecx, [esp+4]
	mov	edx, ecx
	sub	edx, [esp+8]
	xor	eax, eax
	cmp	edx, ecx
	cmovae	eax, edx
	ret
sat_signed_add(int, int):
	push	esi
	mov	eax, [esp+8]
	mov	ecx, [esp+12]
	lea	edx, [ecx + eax]
	mov	esi, edx
	xor	esi, eax
	shr	eax, 31
	add	eax, 2147483647
	xor	ecx, edx
	test	esi, ecx
	cmovns	eax, edx
	pop	esi
	ret
sat_signed_div(int, int):
	mov	ecx, [esp+8]
	mov	eax, [esp+4]
	cmp	ecx, -1
	je	.LBB6_1
	cdq
	idiv	ecx
	ret
.LBB6_1:
	mov	ecx, eax
	neg	ecx
	cmp	eax, -2147483648
	mov	eax, 2147483647
	cmovne	eax, ecx
	ret
sat_signed_mod(int, int):
	mov	ecx, [esp+8]
	cmp	ecx, -1
	je	.LBB7_1
	mov	eax, [esp+4]
	cdq
	idiv	ecx
	mov	eax, edx
	ret
.LBB7_1:
	xor	eax, eax
	ret
sat_signed_mul(int, int):
	push	esi
	mov	ecx, [esp+12]
	mov	esi, [esp+8]
	mov	eax, ecx
	imul	esi
	xor	ecx, esi
	shr	ecx, 31
	add	ecx, 2147483647
	mov	esi, eax
	sar	esi, 31
	xor	esi, edx
	cmovne	eax, ecx
	pop	esi
	ret
sat_signed_sub(int, int):
	push	esi
	mov	ecx, [esp+12]
	mov	edx, [esp+8]
	mov	esi, edx
	sub	esi, ecx
	mov	eax, edx
	shr	eax, 31
	add	eax, 2147483647
	xor	ecx, edx
	xor	edx, esi
	test	edx, ecx
	cmovns	eax, esi
	pop	esi
	ret

Implementation in GNU C

Alternative implementation using __builtin_*_overflow() provided by GCC and LLVM:

// Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

unsigned int sat_unsigned_add(unsigned int augend, unsigned int addend)
{
    unsigned int sum;

    return __builtin_add_overflow(augend, addend, &sum) ? ~0U : sum;
}

unsigned int sat_unsigned_mul(unsigned int multiplicand, unsigned int multiplier)
{
    unsigned int product;

    return __builtin_mul_overflow(multiplicand, multiplier, &product) ? ~0U : product;
}

unsigned int sat_unsigned_sub(unsigned int minuend, unsigned int subtrahend)
{
    unsigned int difference;

    return __builtin_sub_overflow(minuend, subtrahend, &difference) ? 0U : difference;
}

int sat_signed_add(int augend, int addend)
{
    int sum, clamp = 2147483647 ^ (augend >> 31);

    return __builtin_add_overflow(augend, addend, &sum) ? clamp : sum;
}

int sat_signed_mul(int multiplicand, int multiplier)
{
    int product, clamp = 2147483647 ^ ((multiplicand ^ multiplier) >> 31);

    return __builtin_mul_overflow(multiplicand, multiplier, &product) ? clamp : product;
}

int sat_signed_sub(int minuend, int subtrahend)
{
    int difference, clamp = 2147483647 ^ (minuend >> 31);

    return __builtin_sub_overflow(minuend, subtrahend, &difference) ? clamp : difference;
}

GCC 11.3:

# Compilation provided by Compiler Explorer at https://godbolt.org/
sat_unsigned_add(unsigned int, unsigned int):
	mov	eax, [esp+8]
	add	eax, [esp+4]
	jc	.L7
	ret
.L7:
	or	eax, -1
	ret
sat_unsigned_mul(unsigned int, unsigned int):
	mov	eax, [esp+4]
	mul	dword ptr [esp+8]
	jo	.L13
	ret
.L13:
	or	eax, -1
	ret
sat_unsigned_sub(unsigned int, unsigned int):
	mov	eax, [esp+4]
	mov	edx, 0
	sub	eax, [esp+8]
	cmovb	eax, edx
	ret
sat_signed_add(int, int):
	mov	eax, [esp+4]
	add	eax, [esp+8]
	jo	.L24
	ret
.L24:
	mov	eax, [esp+4]
	sar	eax, 31
	xor	eax, 2147483647
	ret
sat_signed_mul(int, int):
	mov	eax, [esp+4]
	imul	eax, [esp+8]
	jo	.L30
	ret
.L30:
	mov	eax, [esp+4]
	xor	eax, [esp+8]
	sar	eax, 31
	add	eax, 2147483647
	ret
sat_signed_sub(int, int):
	mov	edx, [esp+4]
	mov	eax, edx
	sub	eax, [esp+8]
	jo	.L36
	ret
.L36:
	mov	eax, edx
	sar	eax, 31
	xor	eax, 2147483648
	ret

43 instructions in 124 bytes instead of 68 instructions in 171 bytes, but not properly optimised; the first pair of highlighted lines can be replaced with the following branch-free instruction sequence:

	sbb	edx, edx
	or	eax, edx

The second pair of highlighted lines can be replaced with the following instruction sequence:

	sbb	edx, edx
	and	eax, edx

Clang 14.0.0: 41 instructions in 129 bytes.

# Compilation provided by Compiler Explorer at https://godbolt.org/
sat_unsigned_add(unsigned int, unsigned int):
	mov	ecx, [esp+4]
	add	ecx, [esp+8]
	mov	eax, -1
	cmovae	eax, ecx
	ret
sat_unsigned_mul(unsigned int, unsigned int):
	mov	eax, [esp+4]
	mul	dword ptr [esp+8]
	mov	ecx, -1
	cmovo	eax, ecx
	ret
sat_unsigned_sub(unsigned int, unsigned int):
	mov	eax, [esp+4]
	xor	ecx, ecx
	sub	eax, [esp+8]
	cmovb	eax, ecx
	ret
sat_signed_add(int, int):
	mov	eax, [esp+4]
	mov	ecx, [esp+8]
	lea	edx, [eax+ecx]
	sar	edx, 31
	xor	edx, -2147483648
	add	eax, ecx
	cmovo	eax, edx
	ret
sat_signed_mul(int, int):
	mov	eax, [esp+4]
	mov	ecx, [esp+8]
	mov	edx, ecx
	xor	edx, eax
	shr	edx, 31
	add	edx, 2147483647
	imul	eax, ecx
	cmovo	eax, edx
	ret
sat_signed_sub(int, int):
	mov	eax, [esp+4]
	mov	edx, [esp+8]
	xor	ecx, ecx
	cmp	eax, edx
	setns	cl
	add	ecx, 2147483647
	sub	eax, edx
	cmovo	eax, ecx
	ret

// Copyright © 2004-2024, Stefan Kanthak <‍stefan‍.‍kanthak‍@‍nexgo‍.‍de‍>

unsigned long long sat_unsigned_add_64(unsigned long long augend, unsigned long long addend)
{
    unsigned long long sum;

    return __builtin_add_overflow(augend, addend, &sum) ? ~0ULL : sum;
}

unsigned long long sat_unsigned_div_64(unsigned long long dividend, unsigned long long divisor)
{
    // NOTE: overflow impossible!

    return dividend / divisor;
}

unsigned long long sat_unsigned_mod_64(unsigned long long dividend, unsigned long long divisor)
{
    // NOTE: overflow impossible!

    return dividend % divisor;
}

unsigned long long sat_unsigned_mul_64(unsigned long long multiplicand, unsigned long long multiplier)
{
    unsigned long long product;

    return __builtin_mul_overflow(multiplicand, multiplier, &product) ? ~0ULL : product;
}

unsigned long long sat_unsigned_sub_64(unsigned long long minuend, unsigned long long subtrahend)
{
    unsigned long long difference;

    return __builtin_sub_overflow(minuend, subtrahend, &difference) ? 0ULL : difference;
}

long long sat_signed_add_64(long long augend, long long addend)
{
    long long sum;
#if 0
    return __builtin_add_overflow(augend, addend, &sum) ? 9223372036854775807LL ^ (augend >> 63) : sum;
#else
    return __builtin_add_overflow(augend, addend, &sum) ? ~9223372036854775807LL ^ (sum >> 63) : sum;
#endif
}

long long sat_signed_div_64(long long dividend, long long divisor)
{
    // NOTE: signed division overflows (really: raises 'division exception')
    //        only for -2**63 / -1 = ±2**63!
#if 0
    dividend -= (divisor == -1LL) && (dividend == ~9223372036854775807LL);

    return dividend / divisor;
#else
    return (divisor == -1LL) & (dividend == ~9223372036854775807LL) ? ~9223372036854775807LL : dividend / divisor;
#endif
}

long long sat_signed_mod_64(long long dividend, long long divisor)
{
    // NOTE: signed division overflows (really: raises 'division exception')
    //        only for -2**63 / -1 = ±2**63!
#if 0
    dividend -= (divisor == -1LL) && (dividend == ~9223372036854775807LL);

    return dividend % divisor;
#else
    return (divisor == -1LL) & (dividend == ~9223372036854775807LL) ? 0LL : dividend % divisor;
#endif
}

long long sat_signed_mul_64(long long multiplicand, long long multiplier)
{
    long long product;

    return __builtin_mul_overflow(multiplicand, multiplier, &product) ? 9223372036854775807LL ^ ((multiplicand ^ multiplier) >> 63) : product;
}

long long sat_signed_sub_64(long long minuend, long long subtrahend)
{
    long long difference;
#if 0
    return __builtin_sub_overflow(minuend, subtrahend, &difference) ? 9223372036854775807LL ^ (minuend >> 63) : difference;
#else
    return __builtin_sub_overflow(minuend, subtrahend, &difference) ? ~9223372036854775807LL ^ (difference >> 63) : difference;
#endif
}

GCC 11.3:

# Compilation provided by Compiler Explorer at https://godbolt.org/
sat_unsigned_add_64(unsigned long long, unsigned long long):
	push	ebx
	mov	ecx, [esp+16]
	mov	ebx, [esp+20]
	add	ecx, [esp+8]
	adc	ebx, [esp+12]
	jc	.L6
	mov	edx, ebx
	mov	eax, ecx
	pop	ebx
	ret
.L6:
	mov	eax, -1
	mov	edx, -1
	pop	ebx
	ret
sat_unsigned_div_64(unsigned long long, unsigned long long):
	sub	esp, 12
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	call	__udivdi3
	add	esp, 28
	ret
sat_unsigned_mod_64(unsigned long long, unsigned long long):
	sub	esp, 12
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	call	__umoddi3
	add	esp, 28
	ret
sat_unsigned_mul_64(unsigned long long, unsigned long long):
	push	ebp
	push	edi
	push	esi
	push	ebx
	sub	esp, 20
	mov	edi, [esp+40]
	mov	ecx, [esp+44]
	mov	dword ptr [esp], 0
	mov	ebp, [esp+48]
	mov	ebx, [esp+52]
	mov	[esp+4], edi
	test	ecx, ecx
	jne	.L16
	test	ebx, ebx
	jne	.L17
	mov	eax, edi
	mul	ebp
	mov	ebx, eax
.L14:
	cmp	byte ptr [esp], 0
	mov	ecx, -1
	mov	eax, ebx
	cmovne	eax, ecx
	cmovne	edx, ecx
	add	esp, 20
	pop	ebx
	pop	esi
	pop	edi
	pop	ebp
	ret
.L16:
	test	ebx, ebx
	jne	.L20
	mov	esi, ecx
	mov	edi, ebp
.L18:
	mov	eax, [esp+4]
	mul	ebp
	mov	[esp+8], eax
	mov	eax, edi
	mov	[esp+12], edx
	mul	esi
	mov	edi, [esp+12]
	mov	esi, edi
	xor	edi, edi
	add	esi, eax
	adc	edi, edx
	test	edi, edi
	jne	.L20
	mov	edx, esi
	mov	esi, [esp+8]
	xor	edi, edi
	or	edx, edi
	mov	ebx, esi
	jmp	.L14
.L17:
	mov	esi, ebx
	jmp	.L18
.L20:
	mov	esi, [esp+4]
	imul	ecx, ebp
	mov	dword ptr [esp], 1
	imul	ebx, esi
	mov	eax, esi
	mul	ebp
	add	ecx, ebx
	lea	edx, [edx+ecx]
	mov	ebx, eax
	jmp	.L14
sat_unsigned_sub_64(unsigned long long, unsigned long long):
	push	ebx
	mov	eax, [esp+8]
	mov	edx, [esp+12]
	mov	ecx, eax
	sub	ecx, [esp+16]
	mov	ebx, edx
	sbb	ebx, [esp+20]
	cmp	eax, ecx
	mov	eax, edx
	sbb	eax, ebx
	jc	.L29
	mov	edx, ebx
	mov	eax, ecx
	pop	ebx
	ret
.L29:
	xor	eax, eax
	xor	edx, edx
	pop	ebx
	ret
sat_signed_add_64(long long, long long):
	mov	eax, [esp+12]
	mov	edx, [esp+16]
	add	eax, [esp+4]
	adc	edx, [esp+8]
	jno	.L31
	mov	eax, edx
	sar	eax, 31
	lea	edx, [eax-2147483648]
.L31:
	ret
sat_signed_div_64(long long, long long):
	push	edi
	push	esi
	push	ebx
	mov	eax, [esp+24]
	mov	edx, [esp+28]
	mov	ecx, [esp+16]
	mov	esi, eax
	mov	ebx, [esp+20]
	and	esi, edx
	cmp	esi, -1
	jne	.L42
	mov	esi, ecx
	lea	edi, [ebx-2147483648]
	or	esi, edi
	je	.L41
.L42:
	push	edx
	push	eax
	push	ebx
	push	ecx
	call	__divdi3
	add	esp, 16
.L38:
	pop	ebx
	pop	esi
	pop	edi
	ret
.L41:
	xor	eax, eax
	mov	edx, -2147483648
	jmp	.L38
sat_signed_mod_64(long long, long long):
	push	edi
	push	esi
	push	ebx
	sub	esp, 16
	mov	ecx, [esp+40]
	mov	ebx, [esp+44]
	mov	esi, [esp+32]
	mov	edi, [esp+36]
	mov	eax, ecx
	and	eax, ebx
	mov	[esp+8], esi
	mov	[esp+12], edi
	cmp	eax, -1
	jne	.L51
	mov	eax, edi
	mov	edx, esi
	xor	edi, edi
	xor	esi, esi
	add	eax, -2147483648
	or	edx, eax
	jne	.L51
	add	esp, 16
	mov	eax, esi
	mov	edx, edi
	pop	ebx
	pop	esi
	pop	edi
	ret
.L51:
	push	ebx
	push	ecx
	push	[esp+20]
	push	[esp+20]
	call	__moddi3
	add	esp, 16
	add	esp, 16
	mov	esi, eax
	mov	edi, edx
	mov	eax, esi
	mov	edx, edi
	pop	ebx
	pop	esi
	pop	edi
	ret
sat_signed_mul_64(long long, long long):
	push	ebp
	push	edi
	push	esi
	push	ebx
	sub	esp, 36
	mov	ebx, [esp+56]
	mov	esi, [esp+64]
	mov	dword ptr [esp], 0
	mov	edi, [esp+68]
	mov	ecx, [esp+60]
	mov	edx, ebx
	mov	eax, esi
	sar	edx, 31
	mov	[esp+4], edi
	sar	eax, 31
	cmp	edx, ecx
	jne	.L59
	cmp	eax, edi
	jne	.L60
	mov	eax, ebx
	imul	esi
	mov	ebp, edx
.L57:
	mov	edx, [esp]
	test	edx, edx
	jne	.L67
	add	esp, 36
	mov	edx, ebp
	pop	ebx
	pop	esi
	pop	edi
	pop	ebp
	ret
.L67:
	xor	ecx, [esp+4]
	add	esp, 36
	mov	edx, ecx
	pop	ebx
	pop	esi
	sar	edx, 31
	pop	edi
	pop	ebp
	mov	eax, edx
	xor	edx, 2147483647
	not	eax
	ret
.L59:
	cmp	eax, [esp+4]
	jne	.L62
	mov	[esp+8], ebx
	mov	edi, ecx
	mov	ebp, esi
	mov	[esp+12], ecx
.L61:
	mov	eax, ebx
	mul	esi
	mov	[esp+24], eax
	mov	eax, edi
	mov	[esp+28], edx
	mul	ebp
	mov	[esp+16], eax
	mov	[esp+20], edx
	test	edi, edi
	jns	.L63
	xor	eax, eax
	sub	[esp+16], eax
	sbb	[esp+20], ebp
.L63:
	test	ebp, ebp
	jns	.L64
	mov	edi, [esp+8]
	mov	ebp, [esp+12]
	sub	[esp+16], edi
	sbb	[esp+20], ebp
.L64:
	mov	edx, [esp+28]
	mov	eax, edx
	xor	edx, edx
	add	eax, [esp+16]
	adc	edx, [esp+20]
	mov	edi, eax
	mov	[esp+8], eax
	sar	edi, 31
	mov	[esp+12], edx
	cmp	edi, edx
	jne	.L65
	mov	ebp, [esp+8]
	mov	ebx, [esp+24]
	xor	esi, esi
	mov	eax, ebx
	or	ebp, esi
	jmp	.L57
.L60:
	mov	[esp+8], esi
	mov	ebp, ebx
	mov	[esp+12], edi
	jmp	.L61
.L62:
	mov	ebp, ecx
	mov	eax, edi
	imul	eax, ebx
	imul	ebp, esi
	add	ebp, eax
	mov	eax, ebx
	mul	esi
	add	ebp, edx
	lea	edx, [ecx+1]
	cmp	edx, 1
	ja	.L58
	lea	edx, [edi+1]
	cmp	edx, 1
	ja	.L58
	cmp	ecx, edi
	jne	.L66
	xor	esi, esi
	cmp	esi, eax
	sbb	esi, ebp
	jl	.L57
.L58:
	mov	dword ptr [esp], 1
	jmp	.L57
.L66:
	test	ebp, ebp
	js	.L57
	jmp	.L58
.L65:
	mov	eax, [esp+4]
	mov	ebp, ecx
	imul	ebp, esi
	imul	eax, ebx
	add	ebp, eax
	mov	eax, ebx
	mul	esi
	add	ebp, edx
	jmp	.L58
sat_signed_sub_64(long long, long long):
	mov	eax, [esp+4]
	mov	edx, [esp+8]
	sub	eax, [esp+12]
	sbb	edx, [esp+16]
	jno	.L70
	mov	eax, edx
	sar	eax, 31
	lea	edx, [eax-2147483648]
.L70:
	ret

326 instructions in 866 bytes, miserably optimised!

Clang 14.0.0

# Compilation provided by Compiler Explorer at https://godbolt.org/
sat_unsigned_add_64(unsigned long long, unsigned long long):
	mov	eax, [esp+4]
	mov	edx, [esp+8]
	add	eax, [esp+12]
	adc	edx, [esp+16]
	mov	ecx, -1
	cmovb	edx, ecx
	cmovb	eax, ecx
	ret
sat_unsigned_div_64(unsigned long long, unsigned long long):
	sub	esp, 12
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	call	__udivdi3
	add	esp, 28
	ret
sat_unsigned_mod_64(unsigned long long, unsigned long long):
	sub	esp, 12
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	push	[esp+28]
	call	__umoddi3
	add	esp, 28
	ret
sat_unsigned_mul_64(unsigned long long, unsigned long long):
	push	ebp
	push	ebx
	push	edi
	push	esi
	mov	ebp, [esp+20]
	mov	eax, [esp+24]
	mov	edi, [esp+32]
	test	edi, edi
	setne	dl
	test	eax, eax
	setne	cl
	and	cl, dl
	mul	dword ptr [esp+28]
	mov	esi, eax
	seto	bl
	mov	eax, edi
	mul	ebp
	seto	ch
	or	ch, bl
	or	ch, cl
	add	esi, eax
	mov	eax, ebp
	mul	dword ptr [esp+28]
	add	edx, esi
	setb	cl
	or	cl, ch
	mov	ecx, -1
	cmovne	eax, ecx
	cmovne	edx, ecx
	pop	esi
	pop	edi
	pop	ebx
	pop	ebp
	ret
sat_unsigned_sub_64(unsigned long long, unsigned long long):
	mov	eax, [esp+4]
	xor	ecx, ecx
	sub	eax, [esp+12]
	mov	edx, [esp+8]
	sbb	edx, [esp+16]
	cmovb	edx, ecx
	cmovb	eax, ecx
	ret
sat_signed_add_64(long long, long long):
	push	ebx
	push	edi
	push	esi
	mov	esi, [esp+16]
	mov	ecx, [esp+20]
	mov	edi, -1
	mov	eax, -1
	add	eax, 1
	mov	edx, 2147483647
	adc	edx, 0
	add	esi, [esp+24]
	mov	ebx, 2147483647
	adc	ecx, [esp+28]
	cmovs	edx, ebx
	cmovs	eax, edi
	seto	bl
	test	bl, bl
	cmove	eax, esi
	test	bl, bl
	cmove	edx, ecx
	pop	esi
	pop	edi
	pop	ebx
	ret
sat_signed_div_64(long long, long long):
	push	edi
	push	esi
	push	eax
	mov	esi, [esp+28]
	mov	edx, [esp+24]
	mov	ecx, [esp+20]
	mov	eax, [esp+16]
	mov	edi, ecx
	xor	edi, -2147483648
	or	edi, eax
	jne	.LBB6_3
	mov	edi, edx
	and	edi, esi
	cmp	edi, -1
	jne	.LBB6_3
	mov	edx, -2147483648
	xor	eax, eax
	add	esp, 4
	pop	esi
	pop	edi
	ret
.LBB6_3:
	push	esi
	push	edx
	push	ecx
	push	eax
	call	__divdi3
	add	esp, 16
	add	esp, 4
	pop	esi
	pop	edi
	ret
sat_signed_mod_64(long long, long long):
	push	ebp
	push	ebx
	push	edi
	push	esi
	sub	esp, 12
	mov	ebx, [esp+44]
	mov	edi, [esp+40]
	mov	esi, [esp+36]
	mov	ecx, [esp+32]
	mov	eax, esi
	xor	eax, -2147483648
	or	eax, ecx
	jne	.LBB7_2
	xor	eax, eax
	mov	ebp, edi
	and	ebp, ebx
	mov	edx, 0
	cmp	ebp, -1
	je	.LBB7_3
.LBB7_2:
	push	ebx
	push	edi
	push	esi
	push	ecx
	call	__moddi3
	add	esp, 16
.LBB7_3:
	add	esp, 12
	pop	esi
	pop	edi
	pop	ebx
	pop	ebp
	ret
sat_signed_mul_64(long long, long long):
	push	ebp
	push	ebx
	push	edi
	push	esi
	sub	esp, 12
	mov	eax, [esp+40]
	mov	ebp, [esp+44]
	mov	esi, [esp+36]
	sar	esi, 31
	mov	ecx, eax
	mul	esi
	mov	[esp+4], eax		# 4-byte Spill
	mov	edi, edx
	imul	ecx, esi
	add	edi, ecx
	imul	esi, ebp
	mov	ebx, ebp
	sar	ebx, 31
	mov	eax, ebx
	mul	dword ptr [esp+32]
	add	esi, edi
	mov	ecx, ebx
	mov	edi, [esp+36]
	imul	ecx, edi
	add	edx, ecx
	mov	ecx, [esp+32]
	imul	ebx, ecx
	add	ebx, edx
	add	eax, [esp+4]		# 4-byte Folded Reload
	mov	[esp+8], eax		# 4-byte Spill
	adc	ebx, esi
	mov	eax, ecx
	mov	ecx, [esp+40]
	mul	ecx
	mov	esi, edx
	mov	[esp+4], eax		# 4-byte Spill
	mov	eax, edi
	mul	ecx
	mov	edi, edx
	add	esi, eax
	adc	edi, 0
	mov	eax, [esp+32]
	mul	ebp
	mov	ecx, edx
	add	esi, eax
	adc	ecx, edi
	setb	byte ptr [esp+3]	# 1-byte Folded Spill
	mov	eax, [esp+36]
	mul	ebp
	add	eax, ecx
	movzx	ecx, byte ptr [esp+3]	# 1-byte Folded Reload
	adc	edx, ecx
	add	eax, [esp+8]		# 4-byte Folded Reload
	adc	edx, ebx
	mov	edi, esi
	sar	edi, 31
	xor	edx, edi
	xor	edi, eax
	mov	eax, -1
	add	eax, 1
	mov	ecx, 2147483647
	adc	ecx, 0
	xor	ebp, [esp+36]
	mov	ebx, 2147483647
	cmovs	ecx, ebx
	mov	ebx, -1
	cmovs	eax, ebx
	or	edi, edx
	cmove	eax, [esp+4]		# 4-byte Folded Reload
	cmove	ecx, esi
	mov	edx, ecx
	add	esp, 12
	pop	esi
	pop	edi
	pop	ebx
	pop	ebp
	ret
sat_signed_sub_64(long long, long long):
	push	ebx
	push	edi
	push	esi
	mov	esi, [esp+16]
	mov	ecx, [esp+20]
	mov	edi, -1
	mov	eax, -1
	add	eax, 1
	mov	edx, 2147483647
	adc	edx, 0
	sub	esi, [esp+24]
	mov	ebx, 2147483647
	sbb	ecx, [esp+28]
	cmovs	edx, ebx
	cmovs	eax, edi
	seto	bl
	test	bl, bl
	cmove	eax, esi
	test	bl, bl
	cmove	edx, ecx
	pop	esi
	pop	edi
	pop	ebx
	ret

253 instructions in 676 bytes, also poorly optimised; especially notice the highlighted parts to load the constants -2⁶³ = 2⁶³ = 0x8000000000000000 and 2⁶³ -1 = 0x7FFFFFFFFFFFFFFF in a rather weird way!

Contact and Feedback

If you miss anything here, have additions, comments, corrections, criticism or questions, want to give feedback, hints or tipps, report broken links, bugs, deficiencies, errors, inaccuracies, misrepresentations, omissions, shortcomings, vulnerabilities or weaknesses, …: don’t hesitate to contact me and feel free to ask, comment, criticise, flame, notify or report!

Use the X.509 certificate to send S/MIME encrypted mail.

Note: email in weird format and without a proper sender name is likely to be discarded!

I dislike HTML (and even weirder formats too) in email, I prefer to receive plain text.
I also expect to see your full (real) name as sender, not your nickname.
I abhor top posts and expect inline quotes in replies.

Terms and Conditions

By using this site, you signify your agreement to these terms and conditions. If you do not agree to these terms and conditions, do not use this site!

The software and the documentation on this site are provided as is without any warranty, neither express nor implied.
In no event will the author be held liable for any damage(s) arising from the use of the software or the documentation.
Permission is granted to use the current version of the software and the current version of the documentation solely for personal private and non-commercial purposes.
An individuals use of the software or the documentation in his or her capacity or function as an agent, (independent) contractor, employee, member or officer of a business, corporation or organisation (commercial or non-commercial) does not qualify as personal private and non-commercial purpose.
Without written approval from the author the software or the documentation must not be used for a business, for commercial, corporate, governmental, military or organisational purposes of any kind, or in a commercial, corporate, governmental, military or organisational environment of any kind.
Redistribution of the software and the documentation is allowed only in unmodified form of its current version and free of charge.

Data Protection Declaration

This web page records no (personal) data and stores no cookies in the web browser.

The web service is operated and provided by

Telekom Deutschland GmbH
Business Center
D-64306 Darmstadt
Germany
<‍hosting‍@‍telekom‍.‍de‍>
+49 800 5252033

The web service provider stores a session cookie in the web browser and records every visit of this web site with the following data in an access log on their server(s):

the (pseudonymised) IP address;
the date and time of the request;
the URL of the requested web page or file;
the Referer and User-Agent HTTP headers sent by the web browser;
the result (success or failure) of the request;
the amount of data received and sent.