locations := map[string][]LocalSlot{}

	// Each time we assign a stack slot to a value v, we remember
	// the slot we used via an index into locations[v.Type].
	slots := f.Cache.allocIntSlice(f.NumValues())
	defer f.Cache.freeIntSlice(slots)
	for i := range slots {
		slots[i] = -1
	}

	// Pick a stack slot for each value needing one.
	used := f.Cache.allocBoolSlice(f.NumValues())
	defer f.Cache.freeBoolSlice(used)

	// The location of each known slot, indexed by SlotID.
	slots []VarLoc
	// The slots present in each register, indexed by register number.
	registers [][]SlotID
}

// reset fills state with the live variables from live.
func (state *stateAtPC) reset(live abt.T) {
	slots, registers := state.slots, state.registers
	for i := range slots {
		slots[i] = VarLoc{}
	}
	for i := range registers {

In Go 1.23, overhead should be in the single digit percentages.

<!-- https://go.dev/issue/62737 , https://golang.org/cl/576681,  https://golang.org/cl/577615 -->
The compiler in Go 1.23 can now overlap the stack frame slots of local variables
accessed in disjoint regions of a function, which reduces stack usage
for Go applications.

<!-- https://go.dev/cl/577935 -->
For 386 and amd64, the compiler will use information from PGO to align certain

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

// multi-consumer queue. The single producer can both push and pop
// from the head, and consumers can pop from the tail.
//
// It has the added feature that it nils out unused slots to avoid
// unnecessary retention of objects. This is important for sync.Pool,
// but not typically a property considered in the literature.
type poolDequeue struct {
	// headTail packs together a 32-bit head index and a 32-bit

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

   * number of CPUS. Table slots remain empty (null) until they are
   * needed.
   *
   * A single spinlock ("busy") is used for initializing and
   * resizing the table, as well as populating slots with new Cells.
   * There is no need for a blocking lock; when the lock is not
   * available, threads try other slots (or the base).  During these

	return &c.partial[sweepgen/2%2]
}

// fullUnswept returns the spanSet which holds unswept spans without any
// free slots for this sweepgen.
func (c *mcentral) fullUnswept(sweepgen uint32) *spanSet {
	return &c.full[1-sweepgen/2%2]
}

// fullSwept returns the spanSet which holds swept spans without any
// free slots for this sweepgen.
func (c *mcentral) fullSwept(sweepgen uint32) *spanSet {
	return &c.full[sweepgen/2%2]
}

	if s != &emptymspan {
		// Mark this span as no longer cached.
		if s.sweepgen != mheap_.sweepgen+3 {
			throw("bad sweepgen in refill")
		}
		mheap_.central[spc].mcentral.uncacheSpan(s)

		// Count up how many slots were used and record it.
		stats := memstats.heapStats.acquire()
		slotsUsed := int64(s.allocCount) - int64(s.allocCountBeforeCache)
		atomic.Xadd64(&stats.smallAllocCount[spc.sizeclass()], slotsUsed)