Compile opt_send for opt_* only when cc has ISeq

because opt_nil/opt_not/opt_eq populates cc even when it doesn't
fallback to opt_send_without_block because of vm_method_cfunc_is.

```
$ benchmark-driver -v --rbenv 'before --jit;after --jit' benchmark/mjit_opt_cc_insns.yml --repeat-count=4
before --jit: ruby 2.8.0dev (2020-06-22T08:11:24Z master d231b8f95b) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-06-22T08:53:27Z master e1125879ed) +JIT [x86_64-linux]
last_commit=Compile opt_send for opt_* only when cc has ISeq
Calculating -------------------------------------
                     before --jit  after --jit
        mjit_nil?(1)      54.106M      73.693M i/s -     40.000M times in 0.739288s 0.542795s
         mjit_not(1)      53.398M      74.477M i/s -     40.000M times in 0.749090s 0.537075s
     mjit_eq(1, nil)       7.427M       6.497M i/s -      8.000M times in 1.077136s 1.231326s

Comparison:
                     mjit_nil?(1)
         after --jit:  73692594.3 i/s
        before --jit:  54106108.4 i/s - 1.36x  slower

                      mjit_not(1)
         after --jit:  74477487.9 i/s
        before --jit:  53398125.0 i/s - 1.39x  slower

                  mjit_eq(1, nil)
        before --jit:   7427105.9 i/s
         after --jit:   6497063.0 i/s - 1.14x  slower
```

Actually opt_eq becomes slower by this. Maybe it's indeed using
opt_send_without_block, but I'll approach that one in another commit.
This commit is contained in:
Takashi Kokubun 2020-06-22 01:44:11 -07:00
Родитель 3238641750
Коммит 78352fb52e
Не найден ключ, соответствующий данной подписи
Идентификатор ключа GPG: 6FFC433B12EE23DD
4 изменённых файлов: 32 добавлений и 8 удалений

Просмотреть файл

@ -0,0 +1,25 @@
# opt_* insns using vm_method_cfunc_is with send-compatible operands:
# * opt_nil_p
# * opt_not
# * opt_eq
type: lib/benchmark_driver/runner/mjit
prelude: |
def mjit_nil?(obj)
obj.nil?
end
def mjit_not(obj)
!obj
end
def mjit_eq(a, b)
a == b
end
benchmark:
- script: mjit_nil?(1)
loop_count: 40000000
- script: mjit_not(1)
loop_count: 40000000
- script: mjit_eq(1, nil)
loop_count: 8000000

Просмотреть файл

@ -97,9 +97,9 @@ captured_cc_entries(const struct compile_status *status)
// Returns true if call cache is still not obsoleted and vm_cc_cme(cc)->def->type is available.
static bool
has_valid_method_type(CALL_CACHE cc)
has_valid_method_type(CALL_CACHE cc, rb_method_type_t type)
{
return vm_cc_cme(cc) != NULL;
return vm_cc_cme(cc) != NULL && vm_cc_cme(cc)->def->type == type;
}
// Returns true if iseq can use fastpath for setup, otherwise NULL. This becomes true in the same condition
@ -439,9 +439,8 @@ precompile_inlinable_iseqs(FILE *f, const rb_iseq_t *iseq, struct compile_status
const struct rb_callcache *cc = captured_cc_entries(status)[call_data_index(cd, body)]; // use copy to avoid race condition
const rb_iseq_t *child_iseq;
if (has_valid_method_type(cc) &&
if (has_valid_method_type(cc, VM_METHOD_TYPE_ISEQ) &&
!(vm_ci_flag(ci) & VM_CALL_TAILCALL) && // inlining only non-tailcall path
vm_cc_cme(cc)->def->type == VM_METHOD_TYPE_ISEQ &&
fastpath_applied_iseq_p(ci, cc, child_iseq = def_iseq_ptr(vm_cc_cme(cc)->def)) &&
// CC_SET_FASTPATH in vm_callee_setup_arg
inlinable_iseq_p(child_iseq->body)) {

Просмотреть файл

@ -20,13 +20,13 @@
const CALL_INFO ci = cd->ci;
int kw_splat = IS_ARGS_KW_SPLAT(ci) > 0;
extern bool rb_splat_or_kwargs_p(const struct rb_callinfo *restrict ci);
if (!status->compile_info->disable_send_cache && has_valid_method_type(captured_cc) && (
if (!status->compile_info->disable_send_cache && (
% # `CC_SET_FASTPATH(cd->cc, vm_call_cfunc_with_frame, ...)` in `vm_call_cfunc`
(vm_cc_cme(captured_cc)->def->type == VM_METHOD_TYPE_CFUNC
(has_valid_method_type(captured_cc, VM_METHOD_TYPE_CFUNC)
&& !rb_splat_or_kwargs_p(ci) && !kw_splat)
% # `CC_SET_FASTPATH(cc, vm_call_iseq_setup_func(...), vm_call_iseq_optimizable_p(...))` in `vm_callee_setup_arg`,
% # and support only non-VM_CALL_TAILCALL path inside it
|| (vm_cc_cme(captured_cc)->def->type == VM_METHOD_TYPE_ISEQ
|| (has_valid_method_type(captured_cc, VM_METHOD_TYPE_ISEQ)
&& fastpath_applied_iseq_p(ci, captured_cc, iseq = def_iseq_ptr(vm_cc_cme(captured_cc)->def))
&& !(vm_ci_flag(ci) & VM_CALL_TAILCALL))
)) {

Просмотреть файл

@ -56,7 +56,7 @@ switch (insn) {
% when *send_compatible_opt_insns
% # To avoid cancel, just emit `opt_send_without_block` instead of `opt_*` insn if call cache is populated.
% cd_index = insn.opes.index { |o| o.fetch(:type) == 'CALL_DATA' }
if (has_valid_method_type(captured_cc_entries(status)[call_data_index((CALL_DATA)operands[<%= cd_index %>], body)])) {
if (has_valid_method_type(captured_cc_entries(status)[call_data_index((CALL_DATA)operands[<%= cd_index %>], body)], VM_METHOD_TYPE_ISEQ)) {
<%= render 'mjit_compile_send', locals: { insn: opt_send_without_block } -%>
<%= render 'mjit_compile_insn', locals: { insn: opt_send_without_block } -%>
break;