[pull] master from ruby:master #729
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Resolves Shopify#772 Adds profiling for the `getblockparamproxy` YARV instruction and handles the `nil` block case by pushing `nil` instead of the block proxy object, improves `ratio_in_zjit` a tiny bit (0.1%) Profiling data for `getblockparamproxy` on Lobsters ``` Top-6 getblockparamproxy handler (100.0% of total 3,353,291): polymorphic: 2,337,372 (69.7%) nil: 552,629 (16.5%) iseq: 259,636 ( 7.7%) no_profiles: 156,734 ( 4.7%) proc: 40,223 ( 1.2%) megamorphic: 6,697 ( 0.2%) ``` Lobsters benchmark stats: <details> <summary>Stats before (master):</summary> <p> ``` ❯ ./run_benchmarks.rb --chruby 'ruby-zjit --zjit-stats' lobsters ***ZJIT: Printing ZJIT statistics on exit*** ... Top-20 side exit reasons (100.0% of total 15,338,024): guard_type_failure: 6,889,050 (44.9%) guard_shape_failure: 6,848,898 (44.7%) block_param_proxy_not_iseq_or_ifunc: 1,008,525 ( 6.6%) unhandled_hir_insn: 236,977 ( 1.5%) compile_error: 191,763 ( 1.3%) fixnum_mult_overflow: 50,739 ( 0.3%) block_param_proxy_modified: 28,119 ( 0.2%) patchpoint_stable_constant_names: 18,229 ( 0.1%) unhandled_newarray_send_pack: 14,481 ( 0.1%) unhandled_block_arg: 13,782 ( 0.1%) fixnum_lshift_overflow: 10,085 ( 0.1%) patchpoint_no_ep_escape: 7,815 ( 0.1%) unhandled_yarv_insn: 7,540 ( 0.0%) expandarray_failure: 4,533 ( 0.0%) guard_super_method_entry: 4,475 ( 0.0%) patchpoint_method_redefined: 1,207 ( 0.0%) patchpoint_no_singleton_class: 1,130 ( 0.0%) obj_to_string_fallback: 412 ( 0.0%) guard_less_failure: 163 ( 0.0%) interrupt: 82 ( 0.0%) ... ratio_in_zjit: 82.1% ``` </p> </details> <details> <summary>Stats after:</summary> <p> ``` ❯ ./run_benchmarks.rb --chruby 'ruby-zjit --zjit-stats' lobsters ***ZJIT: Printing ZJIT statistics on exit*** ... Top-20 side exit reasons (100.0% of total 15,061,422): guard_type_failure: 6,892,934 (45.8%) guard_shape_failure: 6,850,512 (45.5%) block_param_proxy_not_iseq_or_ifunc: 549,823 ( 3.7%) unhandled_hir_insn: 236,979 ( 1.6%) compile_error: 191,782 ( 1.3%) unhandled_yarv_insn: 128,695 ( 0.9%) block_param_proxy_not_nil: 68,623 ( 0.5%) fixnum_mult_overflow: 50,739 ( 0.3%) patchpoint_stable_constant_names: 18,568 ( 0.1%) unhandled_newarray_send_pack: 14,481 ( 0.1%) block_param_proxy_modified: 13,819 ( 0.1%) unhandled_block_arg: 13,798 ( 0.1%) fixnum_lshift_overflow: 10,085 ( 0.1%) patchpoint_no_ep_escape: 7,815 ( 0.1%) expandarray_failure: 4,533 ( 0.0%) guard_super_method_entry: 4,475 ( 0.0%) patchpoint_method_redefined: 1,207 ( 0.0%) obj_to_string_fallback: 1,140 ( 0.0%) patchpoint_no_singleton_class: 1,130 ( 0.0%) guard_less_failure: 163 ( 0.0%) ... ratio_in_zjit: 82.2% ``` </p> </details>
Break out the different cases into different blocks in the bytecode to HIR parser. Use a `RefineType` to plumb the case's type through so the type specialization can see it. Then join the logic back to the rest of the current block after each case's send.
lobsters before
<details>
```
***ZJIT: Printing ZJIT statistics on exit***
Top-20 not inlined C methods (58.7% of total 4,476,259):
Hash#fetch: 849,219 (19.0%)
String#start_with?: 328,017 ( 7.3%)
Regexp#match?: 148,149 ( 3.3%)
Hash#key?: 135,034 ( 3.0%)
Kernel#is_a?: 110,030 ( 2.5%)
Set#include?: 97,934 ( 2.2%)
Integer#===: 96,952 ( 2.2%)
Process.clock_gettime: 92,795 ( 2.1%)
String#sub!: 84,940 ( 1.9%)
String.new: 80,730 ( 1.8%)
SQLite3::Statement#done?: 73,532 ( 1.6%)
SQLite3::Statement#step: 73,532 ( 1.6%)
Time#plus_without_duration: 66,724 ( 1.5%)
String#<<: 63,954 ( 1.4%)
Time#to_i: 60,817 ( 1.4%)
Hash#delete: 60,664 ( 1.4%)
Time#subsec: 60,363 ( 1.3%)
String#hash: 51,261 ( 1.1%)
IO#read: 47,753 ( 1.1%)
String#to_sym: 43,915 ( 1.0%)
Top-20 calls to C functions from JIT code (83.7% of total 35,570,418):
rb_vm_opt_send_without_block: 10,516,746 (29.6%)
rb_vm_env_write: 2,382,117 ( 6.7%)
rb_zjit_writebarrier_check_immediate: 2,241,285 ( 6.3%)
rb_hash_aref: 2,189,588 ( 6.2%)
rb_vm_getinstancevariable: 1,762,596 ( 5.0%)
rb_ivar_get_at_no_ractor_check: 1,702,246 ( 4.8%)
rb_vm_send: 1,460,754 ( 4.1%)
rb_hash_aset: 1,151,302 ( 3.2%)
rb_vm_setinstancevariable: 1,029,286 ( 2.9%)
rb_obj_is_kind_of: 1,000,979 ( 2.8%)
rb_vm_opt_getconstant_path: 623,490 ( 1.8%)
rb_vm_invokesuper: 595,831 ( 1.7%)
Hash#fetch: 562,212 ( 1.6%)
rb_vm_invokeblock: 545,744 ( 1.5%)
rb_class_allocate_instance: 422,454 ( 1.2%)
rb_ec_ary_new_from_values: 388,035 ( 1.1%)
String#start_with?: 328,017 ( 0.9%)
rb_hash_new_with_size: 289,130 ( 0.8%)
fetch: 287,007 ( 0.8%)
rb_vm_sendforward: 284,183 ( 0.8%)
Top-1 not optimized method types for send (100.0% of total 428):
null: 428 (100.0%)
Top-3 not optimized method types for send_without_block (100.0% of total 102,413):
optimized_send: 92,837 (90.6%)
null: 8,595 ( 8.4%)
optimized_block_call: 981 ( 1.0%)
Top-3 not optimized method types for super (100.0% of total 517,931):
cfunc: 489,746 (94.6%)
alias: 26,398 ( 5.1%)
attrset: 1,787 ( 0.3%)
Top-4 instructions with uncategorized fallback reason (100.0% of total 868,223):
invokeblock: 545,744 (62.9%)
sendforward: 284,183 (32.7%)
invokesuperforward: 29,713 ( 3.4%)
opt_send_without_block: 8,583 ( 1.0%)
Top-20 send fallback reasons (100.0% of total 13,432,971):
send_without_block_polymorphic: 4,825,641 (35.9%)
singleton_class_seen: 3,257,447 (24.2%)
send_without_block_no_profiles: 1,906,060 (14.2%)
uncategorized: 868,223 ( 6.5%)
send_no_profiles: 806,168 ( 6.0%)
one_or_more_complex_arg_pass: 537,965 ( 4.0%)
super_not_optimized_method_type: 517,931 ( 3.9%)
send_without_block_megamorphic: 158,893 ( 1.2%)
too_many_args_for_lir: 127,160 ( 0.9%)
send_polymorphic: 112,628 ( 0.8%)
send_without_block_not_optimized_need_permission: 100,041 ( 0.7%)
send_without_block_not_optimized_method_type_optimized: 93,818 ( 0.7%)
super_complex_args_pass: 34,022 ( 0.3%)
super_target_complex_args_pass: 25,536 ( 0.2%)
super_polymorphic: 16,853 ( 0.1%)
obj_to_string_not_string: 13,794 ( 0.1%)
argc_param_mismatch: 9,927 ( 0.1%)
send_without_block_not_optimized_method_type: 8,595 ( 0.1%)
send_without_block_direct_keyword_mismatch: 5,568 ( 0.0%)
send_megamorphic: 4,525 ( 0.0%)
Top-4 setivar fallback reasons (100.0% of total 1,029,286):
not_monomorphic: 992,723 (96.4%)
not_t_object: 21,354 ( 2.1%)
too_complex: 15,188 ( 1.5%)
new_shape_needs_extension: 21 ( 0.0%)
Top-2 getivar fallback reasons (100.0% of total 1,790,794):
not_monomorphic: 1,750,108 (97.7%)
too_complex: 40,686 ( 2.3%)
Top-3 definedivar fallback reasons (100.0% of total 81,713):
not_monomorphic: 80,197 (98.1%)
too_complex: 796 ( 1.0%)
not_t_object: 720 ( 0.9%)
Top-6 invokeblock handler (100.0% of total 545,744):
monomorphic_iseq: 249,809 (45.8%)
polymorphic: 217,915 (39.9%)
monomorphic_ifunc: 46,244 ( 8.5%)
monomorphic_other: 27,938 ( 5.1%)
megamorphic: 2,943 ( 0.5%)
no_profiles: 895 ( 0.2%)
Top-8 popular complex argument-parameter features not optimized (100.0% of total 652,565):
param_forwardable: 246,421 (37.8%)
param_block: 198,808 (30.5%)
param_rest: 101,529 (15.6%)
param_kwrest: 44,809 ( 6.9%)
caller_blockarg: 24,596 ( 3.8%)
caller_splat: 15,969 ( 2.4%)
caller_kw_splat: 14,227 ( 2.2%)
caller_kwarg: 6,206 ( 1.0%)
Top-1 compile error reasons (100.0% of total 38,981):
exception_handler: 38,981 (100.0%)
Top-5 unhandled YARV insns (100.0% of total 4,154):
getconstant: 2,566 (61.8%)
checkmatch: 929 (22.4%)
setblockparam: 443 (10.7%)
once: 171 ( 4.1%)
expandarray: 45 ( 1.1%)
Top-3 unhandled HIR insns (100.0% of total 75,904):
throw: 39,721 (52.3%)
invokebuiltin: 35,772 (47.1%)
array_max: 411 ( 0.5%)
Top-20 side exit reasons (100.0% of total 3,770,125):
guard_shape_failure: 1,927,218 (51.1%)
guard_type_failure: 1,395,315 (37.0%)
block_param_proxy_not_iseq_or_ifunc: 257,894 ( 6.8%)
unhandled_hir_insn: 75,904 ( 2.0%)
compile_error: 38,981 ( 1.0%)
patchpoint_stable_constant_names: 25,375 ( 0.7%)
block_param_proxy_modified: 13,713 ( 0.4%)
fixnum_lshift_overflow: 10,085 ( 0.3%)
fixnum_mult_overflow: 8,550 ( 0.2%)
unhandled_yarv_insn: 4,154 ( 0.1%)
unhandled_block_arg: 2,548 ( 0.1%)
unhandled_newarray_send_pack: 2,322 ( 0.1%)
patchpoint_no_singleton_class: 2,008 ( 0.1%)
patchpoint_no_ep_escape: 1,683 ( 0.0%)
obj_to_string_fallback: 1,358 ( 0.0%)
patchpoint_method_redefined: 1,212 ( 0.0%)
expandarray_failure: 837 ( 0.0%)
guard_super_method_entry: 737 ( 0.0%)
guard_less_failure: 163 ( 0.0%)
interrupt: 49 ( 0.0%)
send_count: 46,003,239
dynamic_send_count: 13,432,971 (29.2%)
optimized_send_count: 32,570,268 (70.8%)
dynamic_setivar_count: 1,029,286 ( 2.2%)
dynamic_getivar_count: 1,790,794 ( 3.9%)
dynamic_definedivar_count: 81,713 ( 0.2%)
iseq_optimized_send_count: 15,117,301 (32.9%)
inline_cfunc_optimized_send_count: 11,837,918 (25.7%)
inline_iseq_optimized_send_count: 884,606 ( 1.9%)
non_variadic_cfunc_optimized_send_count: 2,597,998 ( 5.6%)
variadic_cfunc_optimized_send_count: 2,132,445 ( 4.6%)
compiled_iseq_count: 5,259
failed_iseq_count: 0
compile_time: 1,409ms
profile_time: 10ms
gc_time: 11ms
invalidation_time: 77ms
vm_write_pc_count: 40,924,587
vm_write_sp_count: 40,924,587
vm_write_locals_count: 39,740,467
vm_write_stack_count: 39,740,467
vm_write_to_parent_iseq_local_count: 306,481
vm_read_from_parent_iseq_local_count: 4,841,855
guard_type_count: 48,810,089
guard_type_exit_ratio: 2.9%
guard_shape_count: 19,485,073
guard_shape_exit_ratio: 9.9%
code_region_bytes: 27,262,976
zjit_alloc_bytes: 34,517,324
total_mem_bytes: 61,780,300
side_exit_count: 3,770,125
total_insn_count: 273,152,243
vm_insn_count: 43,926,931
zjit_insn_count: 229,225,312
ratio_in_zjit: 83.9%
```
</details>
lobsters after
<details>
```
***ZJIT: Printing ZJIT statistics on exit***
Top-20 not inlined C methods (61.7% of total 5,220,252):
Hash#fetch: 1,274,409 (24.4%)
String#start_with?: 328,017 ( 6.3%)
Regexp#match?: 147,525 ( 2.8%)
Hash#key?: 139,198 ( 2.7%)
Kernel#is_a?: 110,178 ( 2.1%)
Class#allocate: 107,143 ( 2.1%)
Hash#delete: 106,307 ( 2.0%)
Class#superclass: 98,165 ( 1.9%)
Set#include?: 97,934 ( 1.9%)
Integer#===: 95,874 ( 1.8%)
Process.clock_gettime: 92,795 ( 1.8%)
String#sub!: 80,732 ( 1.5%)
String.new: 80,730 ( 1.5%)
SQLite3::Statement#done?: 73,532 ( 1.4%)
SQLite3::Statement#step: 73,532 ( 1.4%)
Time#plus_without_duration: 66,724 ( 1.3%)
String#<<: 63,954 ( 1.2%)
Kernel#dup: 62,590 ( 1.2%)
Time#to_i: 60,814 ( 1.2%)
Time#subsec: 60,363 ( 1.2%)
Top-20 calls to C functions from JIT code (80.8% of total 33,681,248):
rb_vm_opt_send_without_block: 6,869,559 (20.4%)
rb_hash_aref: 2,487,056 ( 7.4%)
rb_vm_env_write: 2,372,693 ( 7.0%)
rb_zjit_writebarrier_check_immediate: 2,238,890 ( 6.6%)
rb_vm_getinstancevariable: 1,861,700 ( 5.5%)
rb_ivar_get_at_no_ractor_check: 1,702,246 ( 5.1%)
rb_vm_send: 1,468,202 ( 4.4%)
rb_hash_aset: 1,267,469 ( 3.8%)
rb_obj_is_kind_of: 1,126,363 ( 3.3%)
rb_vm_setinstancevariable: 1,055,131 ( 3.1%)
Hash#fetch: 987,402 ( 2.9%)
rb_vm_opt_getconstant_path: 641,779 ( 1.9%)
rb_vm_invokesuper: 603,416 ( 1.8%)
rb_vm_invokeblock: 545,743 ( 1.6%)
rb_class_allocate_instance: 415,748 ( 1.2%)
rb_ec_ary_new_from_values: 380,080 ( 1.1%)
String#start_with?: 328,017 ( 1.0%)
rb_hash_new_with_size: 289,172 ( 0.9%)
fetch: 287,007 ( 0.9%)
rb_vm_sendforward: 283,885 ( 0.8%)
Top-1 not optimized method types for send (100.0% of total 428):
null: 428 (100.0%)
Top-3 not optimized method types for send_without_block (100.0% of total 202,329):
optimized_send: 190,504 (94.2%)
null: 10,844 ( 5.4%)
optimized_block_call: 981 ( 0.5%)
Top-3 not optimized method types for super (100.0% of total 517,421):
cfunc: 489,236 (94.6%)
alias: 26,398 ( 5.1%)
attrset: 1,787 ( 0.3%)
Top-4 instructions with uncategorized fallback reason (100.0% of total 867,452):
invokeblock: 545,743 (62.9%)
sendforward: 283,885 (32.7%)
invokesuperforward: 29,713 ( 3.4%)
opt_send_without_block: 8,111 ( 0.9%)
Top-20 send fallback reasons (100.0% of total 9,800,518):
singleton_class_seen: 3,293,078 (33.6%)
send_without_block_no_profiles: 2,142,301 (21.9%)
uncategorized: 867,452 ( 8.9%)
send_no_profiles: 820,538 ( 8.4%)
send_without_block_polymorphic: 780,065 ( 8.0%)
one_or_more_complex_arg_pass: 556,514 ( 5.7%)
super_not_optimized_method_type: 517,421 ( 5.3%)
send_without_block_not_optimized_method_type_optimized: 191,485 ( 2.0%)
send_without_block_megamorphic: 161,550 ( 1.6%)
too_many_args_for_lir: 127,190 ( 1.3%)
send_polymorphic: 111,290 ( 1.1%)
send_without_block_not_optimized_need_permission: 99,526 ( 1.0%)
super_polymorphic: 45,651 ( 0.5%)
super_complex_args_pass: 33,748 ( 0.3%)
obj_to_string_not_string: 13,794 ( 0.1%)
send_without_block_not_optimized_method_type: 10,844 ( 0.1%)
argc_param_mismatch: 9,927 ( 0.1%)
send_without_block_direct_keyword_mismatch: 6,336 ( 0.1%)
super_target_complex_args_pass: 5,108 ( 0.1%)
send_megamorphic: 4,525 ( 0.0%)
Top-4 setivar fallback reasons (100.0% of total 1,123,837):
not_monomorphic: 1,087,274 (96.7%)
not_t_object: 21,354 ( 1.9%)
too_complex: 15,188 ( 1.4%)
new_shape_needs_extension: 21 ( 0.0%)
Top-2 getivar fallback reasons (100.0% of total 2,132,203):
not_monomorphic: 2,092,243 (98.1%)
too_complex: 39,960 ( 1.9%)
Top-3 definedivar fallback reasons (100.0% of total 107,264):
not_monomorphic: 105,748 (98.6%)
too_complex: 796 ( 0.7%)
not_t_object: 720 ( 0.7%)
Top-6 invokeblock handler (100.0% of total 545,743):
monomorphic_iseq: 249,809 (45.8%)
polymorphic: 217,914 (39.9%)
monomorphic_ifunc: 46,244 ( 8.5%)
monomorphic_other: 27,938 ( 5.1%)
megamorphic: 2,943 ( 0.5%)
no_profiles: 895 ( 0.2%)
Top-8 popular complex argument-parameter features not optimized (100.0% of total 651,185):
param_forwardable: 233,989 (35.9%)
param_block: 205,158 (31.5%)
param_rest: 100,319 (15.4%)
param_kwrest: 44,596 ( 6.8%)
caller_blockarg: 21,863 ( 3.4%)
caller_kw_splat: 20,970 ( 3.2%)
caller_splat: 18,106 ( 2.8%)
caller_kwarg: 6,184 ( 0.9%)
Top-1 compile error reasons (100.0% of total 38,980):
exception_handler: 38,980 (100.0%)
Top-5 unhandled YARV insns (100.0% of total 4,154):
getconstant: 2,566 (61.8%)
checkmatch: 929 (22.4%)
setblockparam: 443 (10.7%)
once: 171 ( 4.1%)
expandarray: 45 ( 1.1%)
Top-3 unhandled HIR insns (100.0% of total 75,633):
throw: 39,447 (52.2%)
invokebuiltin: 35,775 (47.3%)
array_max: 411 ( 0.5%)
Top-20 side exit reasons (100.0% of total 3,734,975):
guard_shape_failure: 1,908,302 (51.1%)
guard_type_failure: 1,391,624 (37.3%)
block_param_proxy_not_iseq_or_ifunc: 246,820 ( 6.6%)
unhandled_hir_insn: 75,633 ( 2.0%)
compile_error: 38,980 ( 1.0%)
patchpoint_stable_constant_names: 25,375 ( 0.7%)
block_param_proxy_modified: 13,713 ( 0.4%)
fixnum_lshift_overflow: 10,085 ( 0.3%)
fixnum_mult_overflow: 8,550 ( 0.2%)
unhandled_yarv_insn: 4,154 ( 0.1%)
unhandled_block_arg: 2,548 ( 0.1%)
unhandled_newarray_send_pack: 2,322 ( 0.1%)
patchpoint_no_singleton_class: 2,008 ( 0.1%)
patchpoint_no_ep_escape: 1,683 ( 0.0%)
obj_to_string_fallback: 1,358 ( 0.0%)
expandarray_failure: 837 ( 0.0%)
patchpoint_method_redefined: 710 ( 0.0%)
guard_less_failure: 163 ( 0.0%)
guard_super_method_entry: 53 ( 0.0%)
interrupt: 38 ( 0.0%)
send_count: 45,128,693
dynamic_send_count: 9,800,518 (21.7%)
optimized_send_count: 35,328,175 (78.3%)
dynamic_setivar_count: 1,123,837 ( 2.5%)
dynamic_getivar_count: 2,132,203 ( 4.7%)
dynamic_definedivar_count: 107,264 ( 0.2%)
iseq_optimized_send_count: 15,891,453 (35.2%)
inline_cfunc_optimized_send_count: 12,866,297 (28.5%)
inline_iseq_optimized_send_count: 1,102,971 ( 2.4%)
non_variadic_cfunc_optimized_send_count: 2,857,775 ( 6.3%)
variadic_cfunc_optimized_send_count: 2,609,679 ( 5.8%)
compiled_iseq_count: 5,268
failed_iseq_count: 0
compile_time: 1,558ms
profile_time: 10ms
gc_time: 13ms
invalidation_time: 84ms
vm_write_pc_count: 39,300,901
vm_write_sp_count: 39,300,901
vm_write_locals_count: 38,133,357
vm_write_stack_count: 38,133,357
vm_write_to_parent_iseq_local_count: 305,249
vm_read_from_parent_iseq_local_count: 4,818,083
guard_type_count: 48,036,224
guard_type_exit_ratio: 2.9%
guard_shape_count: 19,302,903
guard_shape_exit_ratio: 9.9%
code_region_bytes: 29,491,200
zjit_alloc_bytes: 34,932,040
total_mem_bytes: 64,423,240
side_exit_count: 3,734,975
total_insn_count: 272,964,960
vm_insn_count: 46,583,034
zjit_insn_count: 226,381,926
ratio_in_zjit: 82.9%
```
</details>
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )