Bug 147387 – Overflow check elimination fails for a simple test case

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[Filip Pizlo]

The test case:

function foo(o) {
    var result = 0;
    for (var i = 0; i < 100; ++i) // ++i still has an overflow check in the emitted code
        result += o.f;
    return result;
}

noInline(foo);

var p = {f:42};
var o = Object.create(p);

for (var i = 0; i < 10000; ++i) {
    p.f = i;
    foo(o);
}

[Filip Pizlo]

The reason for the bug is that the merge rule assumes TOP when presented two claims about different constants.  For example:

@x == 0
@x < 100

will yield TOP, when it should just yield @x < 100.

[Filip Pizlo]

Created attachment 258001 [details]
Work in progress

This is going to be interesting.  I want to fix this test case but I don't want to introduce too much crazy.  This work-in-progress might be a bit on the crazy side still, and it's not even finished.

[Filip Pizlo]

Created attachment 258002 [details]
maybe it'll work

This is starting to look believable.  I'll have to think about it more though.

[Filip Pizlo]

Created attachment 259443 [details]
more things

I think that this is an even more comprehensive solution.  I haven't even tried compiling it yet, though.

[Filip Pizlo]

Created attachment 259487 [details]
more refined

Still testing this.

[Filip Pizlo]

Speed-up on the two microbenchmarks:

Benchmark report for JSRegress on shakezilla (MacBookPro11,3).

VMs tested:
"TipOfTree" at /Volumes/Data/secondary/OpenSource/WebKitBuild/Release/jsc (r188645)
"BetterRangePhase" at /Volumes/Data/tertiary/OpenSource/WebKitBuild/Release/jsc (r188645)

Collected 8 samples per benchmark/VM, with 8 VM invocations per benchmark. Emitted a call to gc() between
sample measurements. Used 1 benchmark iteration per VM invocation for warm-up. Used the jsc-specific
preciseTime() function to get microsecond-level timing. Reporting benchmark execution times with 95%
confidence intervals in milliseconds.

                                   TipOfTree              BetterRangePhase                                 

hard-overflow-check-equal       37.5336+-0.6439     ^     33.8116+-1.6043        ^ definitely 1.1101x faster
hard-overflow-check             37.2674+-0.7066     ^     32.5268+-0.4279        ^ definitely 1.1457x faster

<geometric>                     37.3964+-0.4636     ^     33.1505+-0.7854        ^ definitely 1.1281x faster

[Filip Pizlo]

Overall perf:


Benchmark report for SunSpider, LongSpider, V8Spider, Octane, Kraken, JSRegress, AsmBench, and CompressionBench on shakezilla (MacBookPro11,3).

VMs tested:
"TipOfTree" at /Volumes/Data/secondary/OpenSource/WebKitBuild/Release/jsc (r188645)
"BetterRangePhase" at /Volumes/Data/tertiary/OpenSource/WebKitBuild/Release/jsc (r188645)

Collected 8 samples per benchmark/VM, with 8 VM invocations per benchmark. Emitted a call to gc() between sample measurements.
Used 1 benchmark iteration per VM invocation for warm-up. Used the jsc-specific preciseTime() function to get microsecond-level
timing. Reporting benchmark execution times with 95% confidence intervals in milliseconds.

                                                        TipOfTree              BetterRangePhase                                 
SunSpider:
   3d-cube                                            4.8236+-0.3991            4.6630+-0.1749          might be 1.0344x faster
   3d-morph                                           5.1740+-0.1408            5.1275+-0.0648        
   3d-raytrace                                        4.9851+-0.0374     ?      5.6385+-1.4123        ? might be 1.1311x slower
   access-binary-trees                                2.0193+-0.0504            2.0042+-0.0449        
   access-fannkuch                                    5.2270+-0.1482            5.2179+-0.0456        
   access-nbody                                       2.4409+-0.1936            2.3753+-0.0203          might be 1.0276x faster
   access-nsieve                                      2.9428+-0.0179     ?      2.9743+-0.0903        ? might be 1.0107x slower
   bitops-3bit-bits-in-byte                           1.4083+-0.0140            1.3948+-0.0147        
   bitops-bits-in-byte                                3.1951+-0.0596     ?      3.2238+-0.0612        ?
   bitops-bitwise-and                                 1.9133+-0.0148     ?      1.9196+-0.0132        ?
   bitops-nsieve-bits                                 2.9024+-0.0209            2.8940+-0.0334        
   controlflow-recursive                              2.0125+-0.0710            1.9435+-0.0323          might be 1.0355x faster
   crypto-aes                                         3.6629+-0.0336     ?      3.6900+-0.0560        ?
   crypto-md5                                         2.2751+-0.0271     ?      2.3149+-0.0385        ? might be 1.0175x slower
   crypto-sha1                                        2.2162+-0.0829            2.1868+-0.1081          might be 1.0134x faster
   date-format-tofte                                  6.4810+-0.0901            6.4090+-0.1364          might be 1.0112x faster
   date-format-xparb                                  4.4182+-0.0787     ?      4.5702+-0.3950        ? might be 1.0344x slower
   math-cordic                                        2.6842+-0.0135     ?      2.7149+-0.0923        ? might be 1.0114x slower
   math-partial-sums                                  5.1099+-0.1114     ?      5.1285+-0.0832        ?
   math-spectral-norm                                 1.7438+-0.0275            1.7392+-0.0240        
   regexp-dna                                         6.0820+-0.0879     ?      6.1329+-0.1491        ?
   string-base64                                      4.2095+-0.0386     ?      4.2466+-0.1689        ?
   string-fasta                                       5.8048+-0.2442            5.6915+-0.0546          might be 1.0199x faster
   string-tagcloud                                    7.7117+-0.0545     ?      7.7816+-0.1479        ?
   string-unpack-code                                17.9664+-0.2744     ?     19.5456+-3.5040        ? might be 1.0879x slower
   string-validate-input                              4.4258+-0.0726            4.4239+-0.1078        

   <arithmetic>                                       4.3783+-0.0232     ?      4.4597+-0.1338        ? might be 1.0186x slower

                                                        TipOfTree              BetterRangePhase                                 
LongSpider:
   3d-cube                                          813.9855+-15.3196         805.6424+-5.0391          might be 1.0104x faster
   3d-morph                                        1490.8229+-2.3010     ?   1491.2573+-6.6030        ?
   3d-raytrace                                      595.5582+-5.0321          595.2532+-2.5105        
   access-binary-trees                              799.6125+-10.5418         796.7921+-7.4698        
   access-fannkuch                                  275.2368+-3.8275     ?    275.6840+-8.7053        ?
   access-nbody                                     510.6766+-3.1072          509.8951+-3.0315        
   access-nsieve                                    357.0611+-4.3922          356.0200+-5.9756        
   bitops-3bit-bits-in-byte                          37.7559+-0.2297           37.5536+-0.2719        
   bitops-bits-in-byte                               75.6636+-2.0708           74.9225+-0.6534        
   bitops-nsieve-bits                               404.6605+-2.3672          404.6030+-4.9826        
   controlflow-recursive                            442.5568+-0.9105     ?    444.2149+-2.1903        ?
   crypto-aes                                       571.8178+-5.9766          568.0103+-7.9767        
   crypto-md5                                       494.9454+-5.3342          493.1496+-3.5585        
   crypto-sha1                                      615.0614+-4.2564     ?    621.4955+-10.5881       ? might be 1.0105x slower
   date-format-tofte                                504.0191+-14.9468         496.7238+-8.5321          might be 1.0147x faster
   date-format-xparb                                616.5331+-2.4952          613.4982+-4.0866        
   hash-map                                         149.9597+-1.0944     ?    150.1344+-3.2742        ?
   math-cordic                                      515.2488+-75.2478         482.0070+-0.6855          might be 1.0690x faster
   math-partial-sums                                490.2735+-2.2433     ?    490.7846+-2.3942        ?
   math-spectral-norm                               548.3604+-2.9451          547.9254+-2.3772        
   string-base64                                    344.3253+-3.6228     ?    352.9198+-5.8084        ? might be 1.0250x slower
   string-fasta                                     367.9363+-2.3383          365.7122+-1.9275        
   string-tagcloud                                  176.3741+-3.0525     ?    176.8751+-2.2641        ?

   <geometric>                                      388.8782+-2.1849          387.5050+-0.8077          might be 1.0035x faster

                                                        TipOfTree              BetterRangePhase                                 
V8Spider:
   crypto                                            47.2592+-0.9358           46.8718+-1.0772        
   deltablue                                         85.9420+-2.6256           85.1528+-2.3473        
   earley-boyer                                      39.1992+-0.5919     ?     39.2281+-0.5339        ?
   raytrace                                          32.1054+-1.5635           30.1890+-1.0699          might be 1.0635x faster
   regexp                                            60.1945+-0.2831     ?     60.3994+-0.4396        ?
   richards                                          64.1570+-1.0411           63.2755+-0.6092          might be 1.0139x faster
   splay                                             34.0770+-1.9339     ?     36.4775+-1.3040        ? might be 1.0704x slower

   <geometric>                                       48.9117+-0.2632           48.7826+-0.4129          might be 1.0026x faster

                                                        TipOfTree              BetterRangePhase                                 
Octane:
   encrypt                                           0.17194+-0.00106    ?     0.17237+-0.00127       ?
   decrypt                                           3.25185+-0.01243    ?     3.25300+-0.00937       ?
   deltablue                                x2       0.15465+-0.00130          0.15197+-0.00169         might be 1.0176x faster
   earley                                            0.27631+-0.00277          0.27623+-0.00176       
   boyer                                             4.20066+-0.10273    ?     4.21329+-0.11973       ?
   navier-stokes                            x2       4.82617+-0.01988    ?     4.91723+-0.19362       ? might be 1.0189x slower
   raytrace                                 x2       1.02384+-0.03019    ?     1.05524+-0.02750       ? might be 1.0307x slower
   richards                                 x2       0.10900+-0.00147          0.10772+-0.00118         might be 1.0119x faster
   splay                                    x2       0.33018+-0.00234    ?     0.33105+-0.00568       ?
   regexp                                   x2      24.41735+-0.30610    ?    24.73577+-0.23839       ? might be 1.0130x slower
   pdfjs                                    x2      36.71915+-0.35206         36.58916+-0.26680       
   mandreel                                 x2      43.46671+-0.46114    ?    43.63361+-0.31383       ?
   gbemu                                    x2      34.89000+-2.22305         33.59036+-0.23926         might be 1.0387x faster
   closure                                           0.55415+-0.00252    ?     0.55733+-0.00256       ?
   jquery                                            7.05812+-0.03970    ?     7.11249+-0.02557       ?
   box2d                                    x2       9.89896+-0.10250    ?     9.96979+-0.10886       ?
   zlib                                     x2     378.73634+-9.97887    ?   385.29796+-6.65037       ? might be 1.0173x slower
   typescript                               x2     638.33517+-8.63410    ?   641.46136+-11.30056      ?

   <geometric>                                       5.53977+-0.03455    ?     5.55368+-0.02200       ? might be 1.0025x slower

                                                        TipOfTree              BetterRangePhase                                 
Kraken:
   ai-astar                                          224.815+-6.703      ?     225.660+-6.294         ?
   audio-beat-detection                               50.138+-0.257      ?      50.531+-0.496         ?
   audio-dft                                          95.676+-1.649      ?      96.139+-1.702         ?
   audio-fft                                          34.613+-0.117      ?      34.676+-0.092         ?
   audio-oscillator                                   58.645+-0.361             58.592+-0.352         
   imaging-darkroom                                   61.352+-1.531             60.631+-0.192           might be 1.0119x faster
   imaging-desaturate                                 49.043+-0.470      ?      50.764+-2.430         ? might be 1.0351x slower
   imaging-gaussian-blur                              84.779+-3.480             84.305+-0.502         
   json-parse-financial                               37.986+-0.294             37.474+-0.262           might be 1.0137x faster
   json-stringify-tinderbox                           21.519+-0.050      !      24.150+-0.207         ! definitely 1.1223x slower
   stanford-crypto-aes                                40.341+-0.607             39.873+-0.138           might be 1.0118x faster
   stanford-crypto-ccm                                34.792+-1.243             34.314+-0.962           might be 1.0139x faster
   stanford-crypto-pbkdf2                             93.400+-0.192      ?      94.099+-1.362         ?
   stanford-crypto-sha256-iterative                   35.667+-0.302      ?      35.836+-0.337         ?

   <arithmetic>                                       65.912+-0.538      ?      66.217+-0.520         ? might be 1.0046x slower

                                                        TipOfTree              BetterRangePhase                                 
JSRegress:
   abc-forward-loop-equal                            28.9037+-0.1573     ?     29.1570+-0.9520        ?
   abc-postfix-backward-loop                         28.8847+-0.1982     ?     28.9605+-0.1234        ?
   abc-simple-backward-loop                          28.5707+-0.2482     ?     28.7503+-0.5265        ?
   abc-simple-forward-loop                           28.6636+-0.1888           28.4852+-0.1495        
   abc-skippy-loop                                   20.4679+-0.1664           20.4223+-0.1127        
   abs-boolean                                        2.4226+-0.0578            2.3932+-0.0322          might be 1.0123x faster
   adapt-to-double-divide                            15.9863+-0.0669           15.8651+-0.0732        
   aliased-arguments-getbyval                         1.1070+-0.0442            1.0899+-0.0403          might be 1.0157x faster
   allocate-big-object                                2.5492+-0.0689            2.5225+-0.0875          might be 1.0106x faster
   arguments-named-and-reflective                    10.8271+-0.1582           10.6658+-0.1544          might be 1.0151x faster
   arguments-out-of-bounds                            9.9830+-0.1649     ^      9.0731+-0.1457        ^ definitely 1.1003x faster
   arguments-strict-mode                              9.4778+-0.3494     ?      9.4856+-0.3083        ?
   arguments                                          8.4255+-0.0838            8.3662+-0.1668        
   arity-mismatch-inlining                            0.7724+-0.0485     ?      0.7748+-0.0440        ?
   array-access-polymorphic-structure                 5.7525+-0.1012     ?      5.8171+-0.1386        ? might be 1.0112x slower
   array-nonarray-polymorhpic-access                 24.0519+-0.2978           23.9109+-0.2056        
   array-prototype-every                             73.8506+-0.3924           73.1471+-0.6807        
   array-prototype-forEach                           81.7823+-19.9516          72.8767+-1.1907          might be 1.1222x faster
   array-prototype-map                               81.0087+-0.7294           80.0226+-0.5770          might be 1.0123x faster
   array-prototype-reduce                            71.4399+-1.6476           70.1857+-1.3713          might be 1.0179x faster
   array-prototype-reduceRight                       69.8370+-0.2417     ?     69.9685+-0.6222        ?
   array-prototype-some                              73.7130+-0.3903           73.3310+-1.4976        
   array-splice-contiguous                           20.5666+-0.2807           20.4993+-0.3252        
   array-with-double-add                              3.3358+-0.0288            3.3247+-0.0881        
   array-with-double-increment                        2.9946+-0.0702            2.9798+-0.0143        
   array-with-double-mul-add                          4.0659+-0.0307     ?      4.1255+-0.0716        ? might be 1.0147x slower
   array-with-double-sum                              3.1389+-0.0235     ?      3.1516+-0.0820        ?
   array-with-int32-add-sub                           5.7395+-0.0380     ?      5.7650+-0.0467        ?
   array-with-int32-or-double-sum                     3.1998+-0.0623     ?      3.2306+-0.0725        ?
   ArrayBuffer-DataView-alloc-large-long-lived   
                                                     25.9695+-0.4886           25.8633+-0.6022        
   ArrayBuffer-DataView-alloc-long-lived             11.4117+-0.1158     ?     11.5979+-0.1124        ? might be 1.0163x slower
   ArrayBuffer-Int32Array-byteOffset                  3.4760+-0.0421     ?      3.4836+-0.0212        ?
   ArrayBuffer-Int8Array-alloc-large-long-lived   
                                                     25.9904+-0.2496           25.8897+-0.1255        
   ArrayBuffer-Int8Array-alloc-long-lived-buffer   
                                                     19.0933+-0.1552     ?     19.6158+-0.9284        ? might be 1.0274x slower
   ArrayBuffer-Int8Array-alloc-long-lived            10.7897+-0.0690     ?     10.8992+-0.1071        ? might be 1.0101x slower
   ArrayBuffer-Int8Array-alloc                        9.1588+-0.1900     ?      9.2501+-0.0453        ?
   arrowfunction-call                                10.4887+-0.1178           10.4426+-0.0995        
   asmjs_bool_bug                                     6.9951+-0.0184     ?      7.0085+-0.0294        ?
   assign-custom-setter-polymorphic                   2.2263+-0.0163     ?      2.2799+-0.0500        ? might be 1.0241x slower
   assign-custom-setter                               3.1642+-0.0486            3.1531+-0.0387        
   basic-set                                          7.1067+-0.1476     ?      7.1109+-0.1829        ?
   big-int-mul                                        3.3803+-0.0365            3.3650+-0.0125        
   boolean-test                                       2.8326+-0.0555     ?      2.8615+-0.0222        ? might be 1.0102x slower
   branch-fold                                        3.5101+-0.0254     ?      3.5134+-0.0234        ?
   branch-on-string-as-boolean                       16.0586+-0.2417     ?     16.4581+-0.2837        ? might be 1.0249x slower
   by-val-generic                                     5.7831+-0.0655     ?      5.7832+-0.0835        ?
   call-spread-apply                                 25.3715+-0.1098     ?     25.5016+-0.4921        ?
   call-spread-call                                  20.5047+-0.1398     ^     20.2368+-0.0802        ^ definitely 1.0132x faster
   captured-assignments                               0.3817+-0.0082     ?      0.3896+-0.0063        ? might be 1.0207x slower
   cast-int-to-double                                 4.9774+-0.0410            4.9602+-0.0284        
   cell-argument                                      6.6173+-0.2184     ^      6.2421+-0.0985        ^ definitely 1.0601x faster
   cfg-simplify                                       2.7287+-0.0736            2.6918+-0.0367          might be 1.0137x faster
   chain-getter-access                                8.2574+-0.1532            8.2330+-0.0826        
   cmpeq-obj-to-obj-other                            11.3852+-0.8566           10.9670+-1.0490          might be 1.0381x faster
   constant-test                                      4.7656+-0.0336            4.7206+-0.0209        
   create-lots-of-functions                           9.4659+-0.5533     ?     10.0303+-0.6269        ? might be 1.0596x slower
   cse-new-array-buffer                               2.1083+-0.0580     ?      2.1815+-0.0968        ? might be 1.0347x slower
   cse-new-array                                      2.2529+-0.0478            2.2106+-0.0525          might be 1.0191x faster
   DataView-custom-properties                        31.0264+-1.2792           30.6884+-0.3600          might be 1.0110x faster
   delay-tear-off-arguments-strictmode               11.9542+-0.1993     ?     12.0002+-0.3376        ?
   deltablue-varargs                                152.1449+-1.5523     ?    153.6654+-2.0226        ?
   destructuring-arguments                          159.0528+-0.8241     ?    160.2248+-1.5889        ?
   destructuring-parameters-overridden-by-function   
                                                      0.4420+-0.0115     ?      0.4578+-0.0204        ? might be 1.0357x slower
   destructuring-swap                                 4.5984+-0.0557     ?      4.6274+-0.0453        ?
   direct-arguments-getbyval                          1.1038+-0.0259     ?      1.1100+-0.0414        ?
   div-boolean-double                                 5.2251+-0.0707            5.1880+-0.0892        
   div-boolean                                        8.0516+-0.0208     ?      8.0667+-0.0349        ?
   double-get-by-val-out-of-bounds                    4.2578+-0.0633     ?      4.3096+-0.0605        ? might be 1.0122x slower
   double-pollution-getbyval                          8.5764+-0.0339     ?      8.5931+-0.0356        ?
   double-pollution-putbyoffset                       3.6664+-0.0384     ?      3.6689+-0.0429        ?
   double-real-use                                   23.8280+-1.2030     ?     23.9349+-1.3436        ?
   double-to-int32-typed-array-no-inline              1.9467+-0.0794            1.9381+-0.0493        
   double-to-int32-typed-array                        1.7139+-0.0430     ?      1.7188+-0.0277        ?
   double-to-uint32-typed-array-no-inline             2.0134+-0.0350     ?      2.0282+-0.0492        ?
   double-to-uint32-typed-array                       1.8013+-0.0168            1.7815+-0.0344          might be 1.0111x faster
   elidable-new-object-dag                           33.5004+-0.3806           33.3678+-0.3841        
   elidable-new-object-roflcopter                    32.4650+-0.3444     ?     32.6691+-0.1026        ?
   elidable-new-object-then-call                     31.8487+-0.5700           31.1458+-0.4476          might be 1.0226x faster
   elidable-new-object-tree                          36.9577+-0.3630     ?     37.3893+-0.7683        ? might be 1.0117x slower
   empty-string-plus-int                              4.5219+-0.1130            4.4958+-0.0399        
   emscripten-cube2hash                              25.2717+-1.1535     ?     28.4517+-5.4926        ? might be 1.1258x slower
   exit-length-on-plain-object                       11.7951+-0.0755     ?     12.0964+-0.2798        ? might be 1.0255x slower
   external-arguments-getbyval                        1.1560+-0.0694     ?      1.1622+-0.0795        ?
   external-arguments-putbyval                        2.0914+-0.0625            2.0900+-0.0715        
   fixed-typed-array-storage-var-index                1.0833+-0.0173            1.0784+-0.0122        
   fixed-typed-array-storage                          0.8212+-0.0338            0.8078+-0.0146          might be 1.0165x faster
   Float32Array-matrix-mult                           3.7165+-0.0371            3.7128+-0.0389        
   Float32Array-to-Float64Array-set                  45.2482+-0.1478     ?     46.0603+-1.3485        ? might be 1.0179x slower
   Float64Array-alloc-long-lived                     57.0244+-0.1395     ?     57.3615+-0.5755        ?
   Float64Array-to-Int16Array-set                    56.2880+-0.2786           55.8081+-0.3373        
   fold-double-to-int                                12.0505+-0.6176           11.7951+-0.0603          might be 1.0217x faster
   fold-get-by-id-to-multi-get-by-offset-rare-int   
                                                     11.1171+-1.6176     ?     11.4040+-0.9605        ? might be 1.0258x slower
   fold-get-by-id-to-multi-get-by-offset             10.2190+-0.9682     ?     10.9825+-1.4132        ? might be 1.0747x slower
   fold-multi-get-by-offset-to-get-by-offset   
                                                      8.1786+-1.4603            8.1560+-1.5435        
   fold-multi-get-by-offset-to-poly-get-by-offset   
                                                      8.6183+-0.7419            8.0331+-1.4701          might be 1.0729x faster
   fold-multi-put-by-offset-to-poly-put-by-offset   
                                                      8.8198+-0.7171     ?      9.0646+-1.6362        ? might be 1.0278x slower
   fold-multi-put-by-offset-to-put-by-offset   
                                                      3.8672+-0.3299     ?      3.9266+-0.3970        ? might be 1.0153x slower
   fold-multi-put-by-offset-to-replace-or-transition-put-by-offset   
                                                      8.2119+-0.4259            7.8499+-0.1089          might be 1.0461x faster
   fold-put-by-id-to-multi-put-by-offset             10.0599+-0.9381     ?     10.2707+-1.4878        ? might be 1.0209x slower
   fold-put-structure                                 5.5307+-0.8999            4.2529+-0.6295          might be 1.3005x faster
   for-of-iterate-array-entries                      10.9317+-0.1193     ?     11.0255+-0.0991        ?
   for-of-iterate-array-keys                          3.5943+-0.1215     ?      3.6625+-0.0582        ? might be 1.0190x slower
   for-of-iterate-array-values                        3.5314+-0.1807     ?      3.5607+-0.1805        ?
   fround                                            18.4798+-0.1521     ?     18.7742+-0.1957        ? might be 1.0159x slower
   ftl-library-inlining-dataview                     55.3061+-0.6132           55.2417+-0.2421        
   ftl-library-inlining                              92.4497+-6.6021           86.8344+-2.7495          might be 1.0647x faster
   function-call                                     10.6109+-0.1819           10.5941+-0.2050        
   function-dot-apply                                 1.9331+-0.0286     ?      1.9815+-0.0431        ? might be 1.0251x slower
   function-test                                      2.6512+-0.0706            2.6436+-0.0229        
   function-with-eval                               102.9968+-0.7137     ?    103.4445+-0.5444        ?
   gcse-poly-get-less-obvious                        13.8987+-0.1339     ?     13.9673+-0.2282        ?
   gcse-poly-get                                     15.6772+-0.3106           15.5052+-0.1014          might be 1.0111x faster
   gcse                                               3.7655+-0.0439            3.7595+-0.0267        
   get-by-id-bimorphic-check-structure-elimination-simple   
                                                      2.5999+-0.0742            2.5479+-0.0250          might be 1.0204x faster
   get-by-id-bimorphic-check-structure-elimination   
                                                      5.5608+-0.0187     ?      5.5794+-0.0187        ?
   get-by-id-chain-from-try-block                     5.3267+-0.0539     ?      5.3515+-0.0480        ?
   get-by-id-check-structure-elimination              4.2457+-0.0325     ?      4.2887+-0.0349        ? might be 1.0101x slower
   get-by-id-proto-or-self                           14.6232+-0.2448     ?     14.8022+-0.5972        ? might be 1.0122x slower
   get-by-id-quadmorphic-check-structure-elimination-simple   
                                                      2.8303+-0.0266     ?      2.8501+-0.0513        ?
   get-by-id-self-or-proto                           14.8144+-0.5716           14.6454+-0.3828          might be 1.0115x faster
   get-by-val-out-of-bounds                           3.7969+-0.0751     ?      3.8738+-0.2593        ? might be 1.0202x slower
   get-by-val-with-string-bimorphic-check-structure-elimination-simple   
                                                      2.7456+-0.0269            2.7270+-0.0179        
   get-by-val-with-string-bimorphic-check-structure-elimination   
                                                      5.8666+-0.0655            5.8596+-0.0309        
   get-by-val-with-string-chain-from-try-block   
                                                      5.4335+-0.0682     ?      5.5839+-0.2115        ? might be 1.0277x slower
   get-by-val-with-string-check-structure-elimination   
                                                      4.9597+-0.0163     ?      4.9831+-0.0217        ?
   get-by-val-with-string-proto-or-self              14.4390+-0.1878     ?     14.5265+-0.3920        ?
   get-by-val-with-string-quadmorphic-check-structure-elimination-simple   
                                                      3.0858+-0.0099     ?      3.0944+-0.0286        ?
   get-by-val-with-string-self-or-proto              15.0469+-1.1603           14.7658+-0.3241          might be 1.0190x faster
   get-by-val-with-symbol-bimorphic-check-structure-elimination-simple   
                                                      2.9669+-0.0222     ?      2.9828+-0.0532        ?
   get-by-val-with-symbol-bimorphic-check-structure-elimination   
                                                     12.1415+-0.1109           12.1192+-0.0395        
   get-by-val-with-symbol-chain-from-try-block   
                                                      5.4829+-0.0612            5.4197+-0.0651          might be 1.0117x faster
   get-by-val-with-symbol-check-structure-elimination   
                                                     10.9056+-0.0905           10.8503+-0.0162        
   get-by-val-with-symbol-proto-or-self              14.6203+-0.6193     ?     14.8031+-0.5622        ? might be 1.0125x slower
   get-by-val-with-symbol-quadmorphic-check-structure-elimination-simple   
                                                      3.7961+-0.0521     ?      3.8689+-0.0483        ? might be 1.0192x slower
   get-by-val-with-symbol-self-or-proto              14.6764+-0.2914     ?     14.8229+-0.6061        ?
   get_callee_monomorphic                             2.2950+-0.0499            2.2794+-0.0371        
   get_callee_polymorphic                             3.2156+-0.2233            3.1268+-0.0519          might be 1.0284x faster
   getter-no-activation                               4.7671+-0.1723            4.6676+-0.1207          might be 1.0213x faster
   getter-prototype                                   9.7256+-0.1193     ^      8.9443+-0.4066        ^ definitely 1.0873x faster
   getter-richards                                  119.7874+-5.5903          119.1671+-5.3613        
   getter                                             5.3912+-0.5562     ?      5.6847+-0.5771        ? might be 1.0544x slower
   global-object-access-with-mutating-structure   
                                                      5.5844+-0.1102     ?      5.5851+-0.1327        ?
   global-var-const-infer-fire-from-opt               0.7790+-0.0145     ?      0.8057+-0.0397        ? might be 1.0343x slower
   global-var-const-infer                             0.6142+-0.0306     ?      0.6479+-0.0474        ? might be 1.0548x slower
   HashMap-put-get-iterate-keys                      25.0407+-1.3869           24.5154+-1.4802          might be 1.0214x faster
   HashMap-put-get-iterate                           27.8880+-2.8092           27.2531+-0.6689          might be 1.0233x faster
   HashMap-string-put-get-iterate                    22.5271+-0.3421     ?     23.2525+-0.9466        ? might be 1.0322x slower
   hoist-make-rope                                    8.1205+-0.7304     ?      8.3211+-0.7454        ? might be 1.0247x slower
   hoist-poly-check-structure-effectful-loop   
                                                      4.0060+-0.0405     ?      4.0338+-0.0452        ?
   hoist-poly-check-structure                         3.2035+-0.0284            3.2019+-0.0247        
   imul-double-only                                   7.7584+-0.1400            7.5764+-0.0490          might be 1.0240x faster
   imul-int-only                                      8.4671+-0.6795            8.2396+-0.6239          might be 1.0276x faster
   imul-mixed                                         7.0561+-0.3007            6.7286+-0.0580          might be 1.0487x faster
   in-four-cases                                     16.5803+-0.0867     ?     16.8786+-0.5897        ? might be 1.0180x slower
   in-one-case-false                                  9.5507+-0.5010            9.4691+-0.5319        
   in-one-case-true                                   9.4998+-0.4536            9.3073+-0.3029          might be 1.0207x faster
   in-two-cases                                       9.6568+-0.3757     ?      9.8642+-0.4013        ? might be 1.0215x slower
   indexed-properties-in-objects                      2.6785+-0.0806            2.6655+-0.0211        
   infer-closure-const-then-mov-no-inline             2.9851+-0.0338            2.9821+-0.0214        
   infer-closure-const-then-mov                      17.7178+-0.1801     ?     17.8023+-0.1982        ?
   infer-closure-const-then-put-to-scope-no-inline   
                                                     10.6766+-0.0361     ?     10.8716+-0.3256        ? might be 1.0183x slower
   infer-closure-const-then-put-to-scope             20.6394+-0.4365           20.5717+-0.2388        
   infer-closure-const-then-reenter-no-inline   
                                                     49.0954+-0.2759           48.8848+-0.0736        
   infer-closure-const-then-reenter                  21.1635+-0.3558           20.8208+-0.3434          might be 1.0165x faster
   infer-constant-global-property                     3.3355+-0.0346     ?      3.3432+-0.0373        ?
   infer-constant-property                            2.5594+-0.0397     ?      2.5883+-0.0611        ? might be 1.0113x slower
   infer-one-time-closure-ten-vars                    9.1095+-1.3297     ^      7.6592+-0.0927        ^ definitely 1.1893x faster
   infer-one-time-closure-two-vars                    8.1317+-0.1909     ^      7.1965+-0.1528        ^ definitely 1.1299x faster
   infer-one-time-closure                             8.1842+-0.2822     ^      7.2675+-0.4643        ^ definitely 1.1261x faster
   infer-one-time-deep-closure                       13.4090+-2.2046     ^     10.3692+-0.1051        ^ definitely 1.2932x faster
   inline-arguments-access                            3.6855+-0.2182            3.5744+-0.0401          might be 1.0311x faster
   inline-arguments-aliased-access                    3.5566+-0.1100     ?      3.5972+-0.0738        ? might be 1.0114x slower
   inline-arguments-local-escape                      3.8441+-0.1673            3.8383+-0.2291        
   inline-get-scoped-var                              4.4918+-0.0866     ?      4.5332+-0.1608        ?
   inlined-put-by-id-transition                       9.6150+-0.0714     ?      9.6804+-0.2821        ?
   int-or-other-abs-then-get-by-val                   4.6833+-0.0272     ?      4.6927+-0.0663        ?
   int-or-other-abs-zero-then-get-by-val             15.4380+-0.2171           15.3268+-0.0387        
   int-or-other-add-then-get-by-val                   3.9353+-0.0536     ?      4.4395+-1.2984        ? might be 1.1281x slower
   int-or-other-add                                   4.7142+-0.0275     ?      4.7747+-0.0441        ? might be 1.0128x slower
   int-or-other-div-then-get-by-val                   3.5835+-0.0225            3.5733+-0.0194        
   int-or-other-max-then-get-by-val                   3.8144+-0.0185     ?      3.8172+-0.0356        ?
   int-or-other-min-then-get-by-val                   3.8270+-0.0458     ?      3.8441+-0.0535        ?
   int-or-other-mod-then-get-by-val                   3.4200+-0.1456            3.3977+-0.0357        
   int-or-other-mul-then-get-by-val                   3.4755+-0.0380     ?      3.5269+-0.0596        ? might be 1.0148x slower
   int-or-other-neg-then-get-by-val                   4.2778+-0.0707     ?      4.3534+-0.0528        ? might be 1.0177x slower
   int-or-other-neg-zero-then-get-by-val             15.4020+-0.2105           15.2993+-0.0974        
   int-or-other-sub-then-get-by-val                   3.9501+-0.1430            3.9162+-0.0419        
   int-or-other-sub                                   3.2995+-0.0344     ?      3.3750+-0.1129        ? might be 1.0229x slower
   int-overflow-local                                 4.1689+-0.0518     ?      4.1841+-0.0704        ?
   Int16Array-alloc-long-lived                       41.8836+-0.2448     ?     42.4842+-0.6542        ? might be 1.0143x slower
   Int16Array-bubble-sort-with-byteLength            16.8834+-0.1165     ?     17.2839+-1.0060        ? might be 1.0237x slower
   Int16Array-bubble-sort                            17.1178+-0.3292           17.0147+-0.0576        
   Int16Array-load-int-mul                            1.3789+-0.0125     ?      1.4103+-0.0537        ? might be 1.0228x slower
   Int16Array-to-Int32Array-set                      42.5354+-0.1417           42.1305+-0.2710        
   Int32Array-alloc-large                            11.6209+-0.1844     ?     11.7088+-0.2335        ?
   Int32Array-alloc-long-lived                       46.9802+-0.2126     ?     47.2723+-0.3565        ?
   Int32Array-alloc                                   2.8239+-0.0582            2.8005+-0.1386        
   Int32Array-Int8Array-view-alloc                    5.9695+-0.1185     ?      6.1803+-0.3224        ? might be 1.0353x slower
   int52-spill                                        4.4226+-0.0899            4.4168+-0.0794        
   Int8Array-alloc-long-lived                        37.8355+-0.8780           37.7861+-0.5679        
   Int8Array-load-with-byteLength                     3.3862+-0.1160            3.3257+-0.0279          might be 1.0182x faster
   Int8Array-load                                     3.3496+-0.0486            3.3155+-0.0132          might be 1.0103x faster
   integer-divide                                    10.1565+-0.0438     ?     10.2185+-0.2505        ?
   integer-modulo                                     1.5407+-0.0671            1.4753+-0.0190          might be 1.0444x faster
   is-boolean-fold-tricky                             3.7085+-0.0479     ?      3.7287+-0.0541        ?
   is-boolean-fold                                    2.6088+-0.0409            2.5840+-0.0370        
   is-function-fold-tricky-internal-function   
                                                      9.6281+-0.0475            9.6138+-0.0564        
   is-function-fold-tricky                            4.0869+-0.0883            4.0269+-0.0131          might be 1.0149x faster
   is-function-fold                                   2.6113+-0.0273     ?      2.6610+-0.0740        ? might be 1.0190x slower
   is-number-fold-tricky                              3.9680+-0.0238            3.9494+-0.0218        
   is-number-fold                                     2.6102+-0.0902            2.5914+-0.0377        
   is-object-or-null-fold-functions                   2.6352+-0.0367            2.6154+-0.0148        
   is-object-or-null-fold-less-tricky                 4.0263+-0.0318     ?      4.0399+-0.0559        ?
   is-object-or-null-fold-tricky                      5.0967+-0.0810     ?      5.1401+-0.0779        ?
   is-object-or-null-fold                             2.6048+-0.0246     ?      2.6170+-0.0326        ?
   is-object-or-null-trickier-function                4.0530+-0.0259     ?      4.0808+-0.0423        ?
   is-object-or-null-trickier-internal-function   
                                                     10.4087+-0.0245     ?     10.4341+-0.0301        ?
   is-object-or-null-tricky-function                  4.2729+-0.5903            4.0709+-0.0429          might be 1.0496x faster
   is-object-or-null-tricky-internal-function   
                                                      7.6279+-0.0770     ?      7.6668+-0.0391        ?
   is-string-fold-tricky                              3.9740+-0.0375            3.9467+-0.0364        
   is-string-fold                                     2.7062+-0.2669            2.5946+-0.0108          might be 1.0430x faster
   is-undefined-fold-tricky                           3.5298+-0.6523            3.2723+-0.0350          might be 1.0787x faster
   is-undefined-fold                                  2.5952+-0.0328            2.5769+-0.0171        
   JSONP-negative-0                                   0.2306+-0.0066     ?      0.2330+-0.0044        ? might be 1.0106x slower
   large-int-captured                                 3.8182+-0.0770     ?      3.9655+-0.4023        ? might be 1.0386x slower
   large-int-neg                                     13.5965+-0.0957           13.4574+-0.1114          might be 1.0103x faster
   large-int                                         12.8596+-0.2121     ^     12.2834+-0.1043        ^ definitely 1.0469x faster
   load-varargs-elimination                          20.4611+-0.1151           20.3619+-0.5167        
   logical-not-weird-types                            2.7943+-0.0327     ?      2.8130+-0.0388        ?
   logical-not                                        4.1881+-0.0929            4.1317+-0.0920          might be 1.0137x faster
   lots-of-fields                                     9.2768+-0.0351     ?      9.3485+-0.1314        ?
   make-indexed-storage                               2.7622+-0.1255     ?      2.7867+-0.0820        ?
   make-rope-cse                                      3.6142+-0.1947            3.5394+-0.0647          might be 1.0211x faster
   marsaglia-larger-ints                             31.0072+-0.1509           30.9019+-0.1782        
   marsaglia-osr-entry                               20.4827+-0.1966     ?     20.4976+-0.2141        ?
   math-with-out-of-bounds-array-values              21.2871+-0.8852           20.9996+-0.2425          might be 1.0137x faster
   max-boolean                                        2.6620+-0.0255     ?      2.6812+-0.0408        ?
   method-on-number                                  17.6446+-1.0188           17.0884+-0.2900          might be 1.0326x faster
   min-boolean                                        2.6280+-0.0459            2.6038+-0.0382        
   minus-boolean-double                               3.0418+-0.0280            3.0226+-0.0210        
   minus-boolean                                      2.3189+-0.0242            2.2945+-0.0616          might be 1.0106x faster
   misc-strict-eq                                    30.5630+-0.3181           30.3447+-0.2634        
   mod-boolean-double                                10.9570+-0.0672           10.9558+-0.0887        
   mod-boolean                                        8.1826+-0.0254     ?      8.2016+-0.0319        ?
   mul-boolean-double                                 3.5208+-0.0280     ?      3.5422+-0.0254        ?
   mul-boolean                                        2.7512+-0.0292            2.7446+-0.0255        
   neg-boolean                                        3.0407+-0.0375     ?      3.0489+-0.0306        ?
   negative-zero-divide                               0.3189+-0.0101            0.3137+-0.0069          might be 1.0168x faster
   negative-zero-modulo                               0.3207+-0.0174            0.3138+-0.0055          might be 1.0220x faster
   negative-zero-negate                               0.2921+-0.0061     ?      0.2951+-0.0090        ? might be 1.0102x slower
   nested-function-parsing                           43.0659+-0.5473     ?     44.3075+-0.7192        ? might be 1.0288x slower
   new-array-buffer-dead                             87.2181+-0.4743           86.9983+-0.7810        
   new-array-buffer-push                              5.9728+-0.0911     ?      5.9843+-0.3223        ?
   new-array-dead                                    15.2503+-1.1216           14.8850+-0.7108          might be 1.0245x faster
   new-array-push                                     3.3983+-0.1357            3.2422+-0.0998          might be 1.0481x faster
   no-inline-constructor                             30.2000+-0.1538     ?     30.3925+-0.1261        ?
   number-test                                        2.8658+-0.0652            2.8444+-0.0397        
   object-closure-call                                4.7395+-0.0530            4.7134+-0.0202        
   object-get-own-property-symbols-on-large-array   
                                                      4.1789+-0.1802            4.0983+-0.1465          might be 1.0197x faster
   object-test                                        2.5692+-0.0389     ?      2.6155+-0.0464        ? might be 1.0180x slower
   obvious-sink-pathology-taken                      98.0240+-0.5555     ?     98.1800+-0.5243        ?
   obvious-sink-pathology                            93.0514+-0.8599           92.9325+-0.6204        
   obviously-elidable-new-object                     28.0062+-0.2373           27.9471+-0.3017        
   plus-boolean-arith                                 2.3344+-0.0427     ?      2.3521+-0.0695        ?
   plus-boolean-double                                3.0469+-0.0422     ?      3.0691+-0.0865        ?
   plus-boolean                                       2.5334+-0.0668            2.4851+-0.0422          might be 1.0194x faster
   poly-chain-access-different-prototypes-simple   
                                                      3.1917+-0.0301     ?      3.2194+-0.0480        ?
   poly-chain-access-different-prototypes             2.8075+-0.0406     ?      2.8640+-0.0953        ? might be 1.0201x slower
   poly-chain-access-simpler                          3.1803+-0.0360     ?      3.2227+-0.0642        ? might be 1.0133x slower
   poly-chain-access                                  3.1902+-0.0642     ?      3.2209+-0.0367        ?
   poly-stricteq                                     49.1909+-0.0836     ?     49.2596+-0.1537        ?
   polymorphic-array-call                             1.1691+-0.0661     ?      1.1803+-0.0735        ?
   polymorphic-get-by-id                              2.7487+-0.0346            2.7416+-0.0316        
   polymorphic-put-by-id                             24.1470+-0.6430     ?     25.2451+-0.8415        ? might be 1.0455x slower
   polymorphic-structure                             12.6327+-0.0388     ?     12.6855+-0.0607        ?
   polyvariant-monomorphic-get-by-id                  6.7895+-0.8216     ?      6.9848+-1.0614        ? might be 1.0288x slower
   proto-getter-access                                8.2133+-0.0831            8.1457+-0.0848        
   prototype-access-with-mutating-prototype           5.2690+-0.1647     ?      5.3030+-0.1380        ?
   put-by-id-replace-and-transition                   7.8909+-0.3368            7.7716+-0.2225          might be 1.0153x faster
   put-by-id-slightly-polymorphic                     2.5768+-0.0186     ?      2.5801+-0.0438        ?
   put-by-id                                          9.5393+-0.1485            9.3793+-0.1828          might be 1.0171x faster
   put-by-val-direct                                  0.3170+-0.0069     ?      0.3176+-0.0055        ?
   put-by-val-large-index-blank-indexing-type   
                                                      5.2555+-0.1114            5.2181+-0.0968        
   put-by-val-machine-int                             2.2789+-0.1215     ?      2.3076+-0.0578        ? might be 1.0126x slower
   rare-osr-exit-on-local                            14.1663+-0.1829           14.0146+-0.2688          might be 1.0108x faster
   register-pressure-from-osr                        15.9931+-0.0587     ?     16.0367+-0.2525        ?
   repeat-multi-get-by-offset                        21.0583+-0.3340           20.8206+-0.0674          might be 1.0114x faster
   setter-prototype                                   7.6321+-0.3081            7.5180+-0.1405          might be 1.0152x faster
   setter                                             5.5865+-0.6564     ?      5.6295+-0.7255        ?
   simple-activation-demo                            25.2338+-0.0715     ?     25.2539+-0.1613        ?
   simple-getter-access                              10.4686+-0.2639           10.3486+-0.1008          might be 1.0116x faster
   simple-poly-call-nested                            8.8576+-0.8880            8.5523+-0.7655          might be 1.0357x faster
   simple-poly-call                                   1.1551+-0.0312     ?      1.1599+-0.0151        ?
   sin-boolean                                       19.5845+-1.1465           19.4984+-1.2468        
   singleton-scope                                   59.3394+-0.2162           59.3335+-0.1824        
   sink-function                                      9.8853+-0.3878     ?     10.2691+-0.5871        ? might be 1.0388x slower
   sink-huge-activation                              16.6965+-0.3679     ^     15.9382+-0.3735        ^ definitely 1.0476x faster
   sinkable-new-object-dag                           53.1363+-0.8301     ?     53.2522+-0.3817        ?
   sinkable-new-object-taken                         40.2906+-0.3114     !     42.2899+-1.5355        ! definitely 1.0496x slower
   sinkable-new-object                               29.3956+-0.4510           29.3117+-0.3099        
   slow-array-profile-convergence                     2.3485+-0.0232     ?      2.3960+-0.0769        ? might be 1.0202x slower
   slow-convergence                                   2.2192+-0.0258     ?      2.2236+-0.0362        ?
   slow-ternaries                                    16.8416+-0.1292           16.8058+-0.1501        
   sorting-benchmark                                 16.1346+-0.3067           16.1222+-0.1711        
   sparse-conditional                                 1.0703+-0.0250     ?      1.0750+-0.0478        ?
   splice-to-remove                                  11.8197+-0.1963     ?     11.8369+-0.0285        ?
   string-char-code-at                               13.7419+-0.1170           13.7165+-0.0515        
   string-concat-object                               2.1486+-0.0698            2.1321+-0.0623        
   string-concat-pair-object                          2.1322+-0.0985            2.1102+-0.0412          might be 1.0105x faster
   string-concat-pair-simple                          9.2304+-0.1709            9.1721+-0.2021        
   string-concat-simple                               9.3719+-0.2161            9.3657+-0.2099        
   string-cons-repeat                                 6.1075+-0.0452     ?      6.1258+-0.0476        ?
   string-cons-tower                                  6.3469+-0.0645     ?      6.6184+-0.5822        ? might be 1.0428x slower
   string-equality                                   18.2658+-0.0413     ?     18.3156+-0.1398        ?
   string-get-by-val-big-char                         6.4338+-0.0550     ?      6.4703+-0.0742        ?
   string-get-by-val-out-of-bounds-insane             3.0786+-0.0439            3.0509+-0.0565        
   string-get-by-val-out-of-bounds                    3.9401+-0.0407     ?      3.9517+-0.0979        ?
   string-get-by-val                                  2.7562+-0.0264     ?      2.7853+-0.0579        ? might be 1.0106x slower
   string-hash                                        1.7682+-0.0438     ?      1.7694+-0.0181        ?
   string-long-ident-equality                        12.1567+-0.0817     ?     12.2990+-0.2500        ? might be 1.0117x slower
   string-out-of-bounds                              10.7607+-0.1621     ^     10.0345+-0.2769        ^ definitely 1.0724x faster
   string-repeat-arith                               26.8748+-1.3202           25.9759+-0.2009          might be 1.0346x faster
   string-sub                                        54.4029+-0.2548     ^     50.4341+-0.1297        ^ definitely 1.0787x faster
   string-test                                        2.7346+-0.0385     ^      2.6511+-0.0208        ^ definitely 1.0315x faster
   string-var-equality                               25.0081+-0.0847     ?     25.1999+-0.6332        ?
   structure-hoist-over-transitions                   2.3070+-0.0613     ?      2.3373+-0.0217        ? might be 1.0132x slower
   substring-concat-weird                            37.5119+-1.7423           36.1075+-0.3865          might be 1.0389x faster
   substring-concat                                  39.8169+-0.3592           38.9979+-0.7159          might be 1.0210x faster
   substring                                         44.2028+-0.1596           44.0794+-0.2848        
   switch-char-constant                               2.5908+-0.0355            2.5766+-0.0390        
   switch-char                                        5.5349+-0.8283            5.5056+-0.6077        
   switch-constant                                    8.8965+-0.7128            8.1693+-0.3282          might be 1.0890x faster
   switch-string-basic-big-var                       13.4647+-0.0648     ?     13.6749+-0.4676        ? might be 1.0156x slower
   switch-string-basic-big                           15.3930+-3.2755           14.2186+-0.4210          might be 1.0826x faster
   switch-string-basic-var                           12.6572+-0.1176     ?     12.7267+-0.1926        ?
   switch-string-basic                               12.3922+-0.0682     ?     12.3944+-0.1253        ?
   switch-string-big-length-tower-var                17.1714+-0.0378     ?     17.2591+-0.0962        ?
   switch-string-length-tower-var                    12.7102+-0.1795           12.6801+-0.1005        
   switch-string-length-tower                        11.4430+-0.1387           11.4159+-0.0704        
   switch-string-short                               11.4859+-0.0800     ?     11.5287+-0.2144        ?
   switch                                            11.4172+-0.4047           11.0317+-0.3679          might be 1.0349x faster
   tear-off-arguments-simple                          3.0115+-0.0938     ?      3.0135+-0.0891        ?
   tear-off-arguments                                 3.9042+-0.0837     ?      3.9290+-0.0393        ?
   temporal-structure                                11.7858+-0.1488     ?     11.7885+-0.1248        ?
   to-int32-boolean                                  12.1361+-0.0474     ?     12.1410+-0.0487        ?
   try-catch-get-by-val-cloned-arguments             13.6319+-0.1312     ^     13.3566+-0.1058        ^ definitely 1.0206x faster
   try-catch-get-by-val-direct-arguments              6.5285+-0.8618            6.2223+-0.0574          might be 1.0492x faster
   try-catch-get-by-val-scoped-arguments              7.3409+-0.0580            7.2933+-0.0840        
   typed-array-get-set-by-val-profiling              26.2068+-0.4291     ?     26.2997+-0.3264        ?
   undefined-property-access                        224.1710+-1.4354          222.8484+-0.8984        
   undefined-test                                     2.8452+-0.0510            2.7933+-0.0408          might be 1.0186x faster
   unprofiled-licm                                   13.8447+-0.2679           13.4341+-0.1693          might be 1.0306x faster
   varargs-call                                      12.9657+-0.0781     ?     12.9774+-0.0930        ?
   varargs-construct-inline                          21.5682+-0.2146     ?     21.7564+-0.4625        ?
   varargs-construct                                 19.2841+-0.1096           19.2502+-0.0962        
   varargs-inline                                     8.2344+-0.1278     ?      8.2966+-0.0736        ?
   varargs-strict-mode                                8.9471+-0.1015            8.9078+-0.0874        
   varargs                                            8.8835+-0.0719            8.8541+-0.0509        
   weird-inlining-const-prop                          2.0304+-0.1507     ?      2.0457+-0.1232        ?

   <geometric>                                        7.4659+-0.0084     ^      7.4356+-0.0133        ^ definitely 1.0041x faster

                                                        TipOfTree              BetterRangePhase                                 
AsmBench:
   bigfib.cpp                                       457.5862+-5.9917          451.4442+-5.5006          might be 1.0136x faster
   cray.c                                           396.6399+-1.4649     ?    397.1593+-2.9399        ?
   dry.c                                            422.3215+-6.5909     ?    427.4141+-2.5018        ? might be 1.0121x slower
   FloatMM.c                                        684.9059+-2.9661     ?    685.5841+-2.7287        ?
   gcc-loops.cpp                                   3420.6547+-14.2443    ?   3451.4038+-78.5523       ?
   n-body.c                                         826.1157+-4.9183          824.4326+-3.0460        
   Quicksort.c                                      408.1837+-2.5891     ?    408.6560+-2.6194        ?
   stepanov_container.cpp                          3575.2595+-23.9858        3563.6010+-17.5153       
   Towers.c                                         233.4927+-1.2216     ?    234.2791+-2.0311        ?

   <geometric>                                      717.7194+-2.5103     ?    718.4167+-2.2634        ? might be 1.0010x slower

                                                        TipOfTree              BetterRangePhase                                 
CompressionBench:
   huffman                                           59.3516+-0.7379           59.3001+-0.9497        
   arithmetic-simple                                272.1458+-2.0620     ?    273.5842+-3.5567        ?
   arithmetic-precise                               245.2025+-2.5290     ?    246.1637+-2.5285        ?
   arithmetic-complex-precise                       244.1802+-2.8239     ?    249.1379+-4.9275        ? might be 1.0203x slower
   arithmetic-precise-order-0                       281.3200+-1.3554     ?    281.8043+-1.6782        ?
   arithmetic-precise-order-1                       302.0360+-0.7118     ?    328.6138+-61.7783       ? might be 1.0880x slower
   arithmetic-precise-order-2                       352.2339+-3.6777     ?    354.8166+-3.4909        ?
   arithmetic-simple-order-1                        325.4685+-2.6654     ?    326.3462+-2.8219        ?
   arithmetic-simple-order-2                        386.3940+-7.8857          381.1298+-2.9513          might be 1.0138x faster
   lz-string                                        314.2741+-3.4966          310.5375+-4.3828          might be 1.0120x faster

   <geometric>                                      254.5743+-1.2129     ?    256.7024+-4.5282        ? might be 1.0084x slower

                                                        TipOfTree              BetterRangePhase                                 
Geomean of preferred means:
   <scaled-result>                                   50.3760+-0.1162     ?     50.5272+-0.2232        ? might be 1.0030x slower

[Filip Pizlo]

Created attachment 259496 [details]
the patch

[WebKit Commit Bot]

Attachment 259496 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:801:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
ERROR: Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:819:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
ERROR: Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:1641:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
Total errors found: 3 in 11 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Benjamin Poulain]

Comment on attachment 259496 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=259496&action=review

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:527
> +        // This is more vague than other. Other is more specific than this. This is less specific than
> +        // other. Other is less vague than this.
> +        ASSERT(vagueness() > other.vagueness());

Seems redundant with your precondition ASSERTs + the previous branch.

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:808
> +                    static_cast<int64_t>(1) + upper - static_cast<int64_t>(thisRight),

If
    thisRight == upper + 3;
the offset is -1. The resulting relation is:
    left < thisRight - 1
which is different from
    left < upper + 1

I have not yet read what follow but I assume you forbid this by API contract. It would be good to make such bounds explicit through a ASSERT_WITH_MESSAGE on entry.

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:844
> +                    if (m_offset >= -1 && m_offset <= 1)

This is a useful check, it would be nice to have a small helper function to name it.

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:857
> +                makeUpper(std::max(thisEffectiveRight, otherEffectiveRight));
> +                makeLower(std::min(thisEffectiveRight, otherEffectiveRight));

I was wondering if we could have offsets with constants.

It seems that merging:
    @a < 1
    @a > 0
    @a > 100
    @a < 99
would give us that case.

Do we have tests for that?

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:866
> +                // Other is: @x < otherEffectiveRight
> +                // This is @x == thisEffectiveRight

I would flip the comments to match the switch() order.

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:879
> +                // Other is: @x > otherEffectiveRight
> +                // This is: @x == thisEffectiveRight

Ditto.

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:1484
> +            // - @x > @c and @x != @d where @c == @d + k and k <= 0
> +            //
> +            //       @x > @c and @x > @d - k

I don't get this one.

Let say @c = 5, @d = 10;
k = -5, @c = @d + k;
    -> @x > @d - (-5) 
    -> @x > 15

Did you mean @x > @d + k here?

> Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:1492
> +            //     that. Otherwise, find all NotEqual constant oeprations and refine them to be LessThan or

typo: "oeprations"

[Filip Pizlo]

(In reply to comment #10)
> Comment on attachment 259496 [details]
> the patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=259496&action=review
> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:527
> > +        // This is more vague than other. Other is more specific than this. This is less specific than
> > +        // other. Other is less vague than this.
> > +        ASSERT(vagueness() > other.vagueness());
> 
> Seems redundant with your precondition ASSERTs + the previous branch.

Removed.

> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:808
> > +                    static_cast<int64_t>(1) + upper - static_cast<int64_t>(thisRight),
> 
> If
>     thisRight == upper + 3;
> the offset is -1. The resulting relation is:
>     left < thisRight - 1
> which is different from
>     left < upper + 1

It is different, but it's also sound.  During merging, we can create a looser relation - one that is more likely to be true.  It's correct to say left < x + 1 if we know that left < x.  Here, we want to say left < upper + 1, but we want to say it in terms of thisRight while creating a general relationship (offset is one of -1,0,1).  Since thisRight - 1 is:

    thisRight - 1
  = upper + 3 - 1
  = upper + 2

We are saying that left < upper + 2 instead of saying left < upper +1, which is fine, according to the widening rule.

> 
> I have not yet read what follow but I assume you forbid this by API
> contract. It would be good to make such bounds explicit through a
> ASSERT_WITH_MESSAGE on entry.
> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:844
> > +                    if (m_offset >= -1 && m_offset <= 1)
> 
> This is a useful check, it would be nice to have a small helper function to
> name it.

OK, I'll look at if there is a place to put it.

> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:857
> > +                makeUpper(std::max(thisEffectiveRight, otherEffectiveRight));
> > +                makeLower(std::min(thisEffectiveRight, otherEffectiveRight));
> 
> I was wondering if we could have offsets with constants.
> 
> It seems that merging:
>     @a < 1
>     @a > 0
>     @a > 100
>     @a < 99
> would give us that case.
> 
> Do we have tests for that?

The two tests I added trigger this.

> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:866
> > +                // Other is: @x < otherEffectiveRight
> > +                // This is @x == thisEffectiveRight
> 
> I would flip the comments to match the switch() order.

OK!

> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:879
> > +                // Other is: @x > otherEffectiveRight
> > +                // This is: @x == thisEffectiveRight
> 
> Ditto.

Fixed.

> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:1484
> > +            // - @x > @c and @x != @d where @c == @d + k and k <= 0
> > +            //
> > +            //       @x > @c and @x > @d - k
> 
> I don't get this one.
> 
> Let say @c = 5, @d = 10;
> k = -5, @c = @d + k;
>     -> @x > @d - (-5) 
>     -> @x > 15
> 
> Did you mean @x > @d + k here?

It's a good thing I didn't implement this case!  You're right, I fixed the comment.

> 
> > Source/JavaScriptCore/dfg/DFGIntegerRangeOptimizationPhase.cpp:1492
> > +            //     that. Otherwise, find all NotEqual constant oeprations and refine them to be LessThan or
> 
> typo: "oeprations"

Fixed.

[Filip Pizlo]

Landed in http://trac.webkit.org/changeset/188720