WITHOUT PATCH


perf stat  ./zstd -b silesia/*
 3# 12 files         : 211938580 ->  66670873 (3.179), 196.4 MB/s , 938.1 MB/s 

 Performance counter stats for './zstd -b silesia/dickens silesia/mozilla silesia/mr silesia/nci silesia/ooffice silesia/osdb silesia/reymont silesia/samba silesia/sao silesia/webster silesia/xml silesia/x-ray':

          8 075,57 msec task-clock                #    1,000 CPUs utilized          
                10      context-switches          #    0,001 K/sec                  
                 0      cpu-migrations            #    0,000 K/sec                  
           155 826      page-faults               #    0,019 M/sec                  
    27 480 485 995      cycles                    #    3,403 GHz                    
    58 080 440 706      instructions              #    2,11  insn per cycle         
     6 521 154 724      branches                  #  807,516 M/sec                  
       151 553 447      branch-misses             #    2,32% of all branches        

       8,075930116 seconds time elapsed

       7,871783000 seconds user
       0,203994000 seconds sys

perf stat  ./zstd -b silesia/*
 3# 12 files         : 211938580 ->  66670873 (3.179), 194.0 MB/s , 929.1 MB/s 

 Performance counter stats for './zstd -b silesia/dickens silesia/mozilla silesia/mr silesia/nci silesia/ooffice silesia/osdb silesia/reymont silesia/samba silesia/sao silesia/webster silesia/xml silesia/x-ray':

          8 047,22 msec task-clock                #    1,000 CPUs utilized          
                18      context-switches          #    0,002 K/sec                  
                 0      cpu-migrations            #    0,000 K/sec                  
           155 825      page-faults               #    0,019 M/sec                  
    27 665 612 223      cycles                    #    3,438 GHz                    
    58 091 565 944      instructions              #    2,10  insn per cycle         
     6 523 276 050      branches                  #  810,625 M/sec                  
       151 562 467      branch-misses             #    2,32% of all branches        

       8,047877012 seconds time elapsed

       7,815618000 seconds user
       0,231988000 seconds sys


AFTER PATCH


perf stat  ./zstd -b silesia/*
 3# 12 files         : 211938580 ->  66670873 (3.179), 196.3 MB/s , 980.2 MB/s 

 Performance counter stats for './zstd -b silesia/dickens silesia/mozilla silesia/mr silesia/nci silesia/ooffice silesia/osdb silesia/reymont silesia/samba silesia/sao silesia/webster silesia/xml silesia/x-ray':

          7 667,27 msec task-clock                #    1,000 CPUs utilized          
                26      context-switches          #    0,003 K/sec                  
                 0      cpu-migrations            #    0,000 K/sec                  
           155 826      page-faults               #    0,020 M/sec                  
    27 099 818 524      cycles                    #    3,534 GHz                    
    56 887 105 199      instructions              #    2,10  insn per cycle         
     6 546 121 505      branches                  #  853,775 M/sec                  
       151 747 904      branch-misses             #    2,32% of all branches        

       7,668226615 seconds time elapsed

       7,427678000 seconds user
       0,239989000 seconds sys

perf stat  ./zstd -b silesia/*
 3# 12 files         : 211938580 ->  66670873 (3.179), 196.7 MB/s , 982.5 MB/s 

 Performance counter stats for './zstd -b silesia/dickens silesia/mozilla silesia/mr silesia/nci silesia/ooffice silesia/osdb silesia/reymont silesia/samba silesia/sao silesia/webster silesia/xml silesia/x-ray':

          7 718,32 msec task-clock                #    1,000 CPUs utilized          
                17      context-switches          #    0,002 K/sec                  
                 0      cpu-migrations            #    0,000 K/sec                  
           155 826      page-faults               #    0,020 M/sec                  
    27 311 037 836      cycles                    #    3,538 GHz                    
    56 888 763 699      instructions              #    2,08  insn per cycle         
     6 546 414 069      branches                  #  848,166 M/sec                  
       151 767 511      branch-misses             #    2,32% of all branches        

       7,718980167 seconds time elapsed

       7,502608000 seconds user
       0,216075000 seconds sys