May 20, 2025 · May 20, 2025 · May 20, 2025 · May 20, 2025 · May 20, 2025 · May 20, 2025
diff --git a/Include/internal/pycore_uop_ids.h b/Include/internal/pycore_uop_ids.h
diff --git a/Include/internal/pycore_uop_metadata.h b/Include/internal/pycore_uop_metadata.h
diff --git a/Python/bytecodes.c b/Python/bytecodes.c
            PyStackRef_CLOSE(value);
        }

        tier2 op(_POP_TWO, (nos, tos --)) {
            PyStackRef_CLOSE(tos);
            PyStackRef_CLOSE(nos);
        }

        pure inst(PUSH_NULL, (-- res)) {
            res = PyStackRef_NULL;
        }
diff --git a/Python/executor_cases.c.h b/Python/executor_cases.c.h
diff --git a/Python/optimizer_analysis.c b/Python/optimizer_analysis.c

 }

 const uint16_t op_without_pop[MAX_UOP_ID] = {
    [_POP_TOP] = _NOP,
    [_POP_TOP_LOAD_CONST_INLINE] = _LOAD_CONST_INLINE,
    [_POP_TOP_LOAD_CONST_INLINE_BORROW] = _LOAD_CONST_INLINE_BORROW,
    [_POP_TWO_LOAD_CONST_INLINE_BORROW] = _POP_TOP_LOAD_CONST_INLINE_BORROW,
 };

 const uint16_t remove_push[MAX_UOP_ID] = {
    [_COPY] = _NOP,
    [_LOAD_CONST_INLINE] = _NOP,
    [_LOAD_CONST_INLINE_BORROW] = _NOP,
    [_LOAD_FAST] = _NOP,
    [_LOAD_FAST_BORROW] = _NOP,
    [_LOAD_SMALL_INT] = _NOP,
    [_POP_TOP_LOAD_CONST_INLINE] = _POP_TOP,
    [_POP_TOP_LOAD_CONST_INLINE_BORROW] = _POP_TOP,
    [_POP_TWO_LOAD_CONST_INLINE_BORROW] = _POP_TWO,
 };


 static int
 remove_unneeded_uops(_PyUOpInstruction *buffer, int buffer_size)
                    buffer[pc].opcode = _NOP;
                }
                break;
            case _POP_TOP:
            case _POP_TOP_LOAD_CONST_INLINE:
            case _POP_TOP_LOAD_CONST_INLINE_BORROW:
            case _POP_TWO_LOAD_CONST_INLINE_BORROW:
            optimize_pop_top_again:
            default:
            {
                _PyUOpInstruction *last = &buffer[pc-1];
                while (last->opcode == _NOP) {
                    last--;
                }
                switch (last->opcode) {
                    case _POP_TWO_LOAD_CONST_INLINE_BORROW:
                        last->opcode = _POP_TOP;
                // Cancel out pushes and pops, repeatedly. So:
                //     _LOAD_FAST + _POP_TWO_LOAD_CONST_INLINE_BORROW + _POP_TOP
                // ...becomes:
                //     _NOP + _POP_TOP + _NOP
                while (op_without_pop[opcode]) {
                    _PyUOpInstruction *last = &buffer[pc - 1];
                    while (last->opcode == _NOP) {
                        last--;
                    }
                    if (!remove_push[last->opcode]) {
                        break;
                    case _POP_TOP_LOAD_CONST_INLINE:
                    case _POP_TOP_LOAD_CONST_INLINE_BORROW:
                        last->opcode = _NOP;
                        goto optimize_pop_top_again;
                    case _COPY:
                    case _LOAD_CONST_INLINE:
                    case _LOAD_CONST_INLINE_BORROW:
                    case _LOAD_FAST:
                    case _LOAD_FAST_BORROW:
                    case _LOAD_SMALL_INT:
                        last->opcode = _NOP;
                        if (opcode == _POP_TOP) {
                            opcode = buffer[pc].opcode = _NOP;
                        }
                        else if (opcode == _POP_TOP_LOAD_CONST_INLINE) {
                            opcode = buffer[pc].opcode = _LOAD_CONST_INLINE;
                        }
                        else if (opcode == _POP_TOP_LOAD_CONST_INLINE_BORROW) {
                            opcode = buffer[pc].opcode = _LOAD_CONST_INLINE_BORROW;
                        }
                        else {
                            assert(opcode == _POP_TWO_LOAD_CONST_INLINE_BORROW);
                            opcode = buffer[pc].opcode = _POP_TOP_LOAD_CONST_INLINE_BORROW;
                            goto optimize_pop_top_again;
                        }
                    }
                    last->opcode = remove_push[last->opcode];
                    opcode = buffer[pc].opcode = op_without_pop[opcode];
                }
                _Py_FALLTHROUGH;
            }
            default:
            {
                /* _PUSH_FRAME doesn't escape or error, but it
                 * does need the IP for the return address */
                bool needs_ip = opcode == _PUSH_FRAME;
diff --git a/Python/optimizer_cases.c.h b/Python/optimizer_cases.c.h
Original file line number	Diff line number	Diff line change
Expand Up		@@ -393,6 +393,11 @@ dummy_func(
		PyStackRef_CLOSE(value);
		}

		tier2 op(_POP_TWO, (nos, tos --)) {
		PyStackRef_CLOSE(tos);
		PyStackRef_CLOSE(nos);
		}

		pure inst(PUSH_NULL, (-- res)) {
		res = PyStackRef_NULL;
		}
Expand Down
Original file line number	Diff line number	Diff line change
Expand Up		@@ -523,6 +523,25 @@ optimize_uops(

		}

		const uint16_t op_without_pop[MAX_UOP_ID] = {
		[_POP_TOP] = _NOP,
		[_POP_TOP_LOAD_CONST_INLINE] = _LOAD_CONST_INLINE,
		[_POP_TOP_LOAD_CONST_INLINE_BORROW] = _LOAD_CONST_INLINE_BORROW,
		[_POP_TWO_LOAD_CONST_INLINE_BORROW] = _POP_TOP_LOAD_CONST_INLINE_BORROW,
		};

		const uint16_t remove_push[MAX_UOP_ID] = {
		[_COPY] = _NOP,
		[_LOAD_CONST_INLINE] = _NOP,
		[_LOAD_CONST_INLINE_BORROW] = _NOP,
		[_LOAD_FAST] = _NOP,
		[_LOAD_FAST_BORROW] = _NOP,
		[_LOAD_SMALL_INT] = _NOP,
		[_POP_TOP_LOAD_CONST_INLINE] = _POP_TOP,
		[_POP_TOP_LOAD_CONST_INLINE_BORROW] = _POP_TOP,
		[_POP_TWO_LOAD_CONST_INLINE_BORROW] = _POP_TWO,
		};


		static int
		remove_unneeded_uops(_PyUOpInstruction *buffer, int buffer_size)
Expand DownExpand Up		@@ -551,50 +570,23 @@ remove_unneeded_uops(_PyUOpInstruction *buffer, int buffer_size)
		buffer[pc].opcode = _NOP;
		}
		break;
		case _POP_TOP:
		case _POP_TOP_LOAD_CONST_INLINE:
		case _POP_TOP_LOAD_CONST_INLINE_BORROW:
		case _POP_TWO_LOAD_CONST_INLINE_BORROW:
		optimize_pop_top_again:
		default:
		{
		_PyUOpInstruction *last = &buffer[pc-1];
		while (last->opcode == _NOP) {
		last--;
		}
		switch (last->opcode) {
		case _POP_TWO_LOAD_CONST_INLINE_BORROW:
		last->opcode = _POP_TOP;
		// Cancel out pushes and pops, repeatedly. So:
		// _LOAD_FAST + _POP_TWO_LOAD_CONST_INLINE_BORROW + _POP_TOP
		// ...becomes:
		// _NOP + _POP_TOP + _NOP
		while (op_without_pop[opcode]) {
		_PyUOpInstruction *last = &buffer[pc - 1];
		while (last->opcode == _NOP) {
		last--;
		}
		if (!remove_push[last->opcode]) {
		break;
		case _POP_TOP_LOAD_CONST_INLINE:
		case _POP_TOP_LOAD_CONST_INLINE_BORROW:
		last->opcode = _NOP;
		goto optimize_pop_top_again;
		case _COPY:
		case _LOAD_CONST_INLINE:
		case _LOAD_CONST_INLINE_BORROW:
		case _LOAD_FAST:
		case _LOAD_FAST_BORROW:
		case _LOAD_SMALL_INT:
		last->opcode = _NOP;
		if (opcode == _POP_TOP) {
		opcode = buffer[pc].opcode = _NOP;
		}
		else if (opcode == _POP_TOP_LOAD_CONST_INLINE) {
		opcode = buffer[pc].opcode = _LOAD_CONST_INLINE;
		}
		else if (opcode == _POP_TOP_LOAD_CONST_INLINE_BORROW) {
		opcode = buffer[pc].opcode = _LOAD_CONST_INLINE_BORROW;
		}
		else {
		assert(opcode == _POP_TWO_LOAD_CONST_INLINE_BORROW);
		opcode = buffer[pc].opcode = _POP_TOP_LOAD_CONST_INLINE_BORROW;
		goto optimize_pop_top_again;
		}
Comment on lines -568 to -592 Copy link Member tomasr8May 20, 2025 Choose a reason for hiding this comment The reason will be displayed to describe this comment to others.Learn more. feels good 🤩 brandtbucher reacted with hooray emoji
		}
		last->opcode = remove_push[last->opcode];
		opcode = buffer[pc].opcode = op_without_pop[opcode];
		}
		_Py_FALLTHROUGH;
		}
		default:
		{
		/* _PUSH_FRAME doesn't escape or error, but it
		* does need the IP for the return address */
		bool needs_ip = opcode == _PUSH_FRAME;
Expand Down