ARM Cortex-A55

ARM Cortex-A55, Snapdragon 855, Kryo 485 Silver, 1780 MHz, 7 nm, 73 mm2, 6.7BT, 4 cores. Xiaomi Mi 9T Pro, 6 GB LPDDR4, 4x 16-bit CH

L1 Data cache = 32 KB, 64 B/line, 4-WAY, VIPT,
L1 Instruction cache = 32 KB, 64 B/line, 128-bits read interface to L2,
L2 Cache = 128 KB, 64 B/line, 4-WAY, PIPT, Strictly exclusive with L1 data cache, Pseudo-inclusive with L1 instruction cache. 40-bit PA.
L3 Cache = 2 MB, 64 B/line, ?-WAY, shared by all cores. 1.6GHz ?
L4 (SLC) Cache = 3 MB

L1 Data Cache Latency = 2 cycles for simple access via pointer
L1 Data Cache Latency = 3 cycles for access with complex address calculation (size_t n, *p; n = p[n]).
L2 Cache Latency = 8 cycles (64 KB)
L3 Cache Latency = 24 cycles (1280 KB)
L4 (SLC) Cache Latency = ? cycles
RAM Latency = 170 ns

Instruction TLB L1 : 15 entries. full-assoc. 4KB, 16KB, 64KB, and 2MB pages.
Data TLB L1 : 16 entries. full-assoc, 4KB pages only.
L2 TLB : 1024 entries, 4-way
Walk cache RAM: 64 entries, 4-way
Intermediate Physical Address (IPA) cache RAM: 64 entries, 4-way

4 KB pages (64-bit Android 9, Linux 4.14)

Data TLB L1 (4 KB pages): 16 items. full-assoc. Miss penalty = 3 cycles. Parallel miss: 4 cycles per L1 data cache read access.
TLB L2: 1024 items. 4-way. Miss penalty = 20 cycles. Parallel miss: 22 cycles.

  Size        Latency        Increase   Description

  32 K     2                            
  64 K     5                        3   + 6 (L2) + 
 128 K     9                        4   + 3 (L1 TLB miss)                 
 256 K    17                        8   + 16 (L3)                 
 512 K    22                        5
   1 M    25                        3    
   2 M    35                       10
   4 M          42 ns                           
   8 M          97 ns                   + 20 (L2 TLB miss) +   ns (RAM)        
  16 M         130 ns       
  32 M         147 ns       
  64 M         160 ns       
 128 M         166 ns       
 256 M         190 ns       
 512 M         209 ns       
1024 M         250 ns

MISC

Branch misprediction penalty = 7 cycles

8-bytes range cross penalty = 2 cycles
Page cross - no additional penalty

L1 B/W (Parallel Random Read) = 1 cycle per one access

L2 B/W (Parallel Random Read) = 8 cycles per one access

L3 B/W (Parallel Random Read) = 13 cycles per one access
L3 Read B/W (64 Bytes step) = 15 GB/s
L3 Read B/W (64 Bytes step, pointer chasing) = 15 GB/s
L3 Write B/W (skip, 64 Bytes step) = 7.00 cycles per write

RAM Read B/W (Parallel Random Read) = 80 ns per access (64 bytes cache line)
RAM Read B/W (Read, 8-64 Bytes step) = 7 GB/s
RAM Read B/W (Read, 8-64 Bytes step - pointer chasing) = 6 GB/s (HW prefetch from RAM)
RAM Write B/W (8 Bytes step) = 23 GB/s
RAM Write B/W (skip, 64 Bytes step) = 2.0 GB/s

7-Zip Benchmark


### gcc-6 -O2 -march=armv8-a
# (Kryo 485 Silver, Cortex-A55)
# 7z b : MIPS values are normalized with Intel Core 2 cpu.
#### TWRP, root shell

# gcc-6 -march=armv8-a -O2

7-Zip (a) [64] 16.04 : Copyright (c) 1999-2016 Igor Pavlov : 2016-10-04
p7zip Version 16.04 (locale=C,Utf16=off,HugeFiles=on,64 bits,4 CPUs LE)

LE
CPU Freq:  1766  1775  1776  1774  1774  1774  1773  1774  1774

RAM size:    5524 MB,  # CPU hardware threads:   4
RAM usage:    435 MB,  # Benchmark threads:      1

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KiB/s     %   MIPS   MIPS  |      KiB/s     %   MIPS   MIPS

22:       1114    99   1094   1084  |      22106    99   1903   1887
23:       1059    99   1090   1080  |      21452    99   1872   1857
24:       1018    99   1105   1095  |      20596    99   1823   1808
25:        981    99   1131   1121  |      19448    99   1746   1731
----------------------------------  | ------------------------------
Avr:              99   1105   1095  |               99   1836   1821
Tot:              99   1471   1458


RAM usage:    441 MB,  # Benchmark threads:      2

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KiB/s     %   MIPS   MIPS  |      KiB/s     %   MIPS   MIPS

22:       2060   166   1204   2004  |      43929   199   1885   3751
23:       2048   173   1204   2087  |      42529   199   1850   3681
24:       2037   179   1223   2190  |      40995   199   1809   3599
25:       2023   184   1254   2310  |      39236   199   1757   3492
----------------------------------  | ------------------------------
Avr:             176   1221   2148  |              199   1825   3631
Tot:             187   1523   2889


RAM usage:    882 MB,  # Benchmark threads:      4

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KiB/s     %   MIPS   MIPS  |      KiB/s     %   MIPS   MIPS

22:       4102   340   1173   3991  |      87341   397   1879   7452
23:       4065   351   1180   4142  |      84400   394   1852   7303
24:       4022   360   1200   4325  |      82261   396   1825   7221
25:       3941   368   1223   4500  |      79541   395   1794   7079
----------------------------------  | ------------------------------
Avr:             355   1194   4239  |              395   1838   7264
Tot:             375   1516   5752


RAM usage:   1765 MB,  # Benchmark threads:      8

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KiB/s     %   MIPS   MIPS  |      KiB/s     %   MIPS   MIPS

22:       4190   372   1096   4077  |      84874   395   1831   7239
23:       4159   381   1111   4238  |      82741   395   1810   7160
24:       4067   394   1109   4374  |      80634   396   1786   7077
25:       3818   394   1106   4360  |      78162   396   1755   6956
----------------------------------  | ------------------------------
Avr:             386   1105   4262  |              396   1796   7108
Tot:             391   1451   5685



# 7z b -mm=* : MIPS and Effectiveness values are normalized with AMD K8 cpu.


7-Zip (a) [64] 16.04 : Copyright (c) 1999-2016 Igor Pavlov : 2016-10-04
p7zip Version 16.04 (locale=C,Utf16=off,HugeFiles=on,64 bits,4 CPUs LE)

LE
CPU Freq:  1773  1772  1773  1775  1772  1774  1774  1773  1774

RAM size:    5524 MB,  # CPU hardware threads:   4
RAM usage:    225 MB,  # Benchmark threads:      1


Method           Speed Usage    R/U Rating   E/U Effec
                 KiB/s     %   MIPS   MIPS     %     %

CPU                       99   1785   1774
CPU                       99   1785   1774
CPU                       99   1785   1774   101   100

LZMA:x1           5791    99   2135   2117   120   119
                 22329    99   1832   1818   103   102
LZMA:x5:mt1       1018    99   1284   1272    72    72
                 20557    99   1749   1734    99    98
LZMA:x5:mt2       2016   178   1418   2519    80   142
                 20620    99   1754   1739    99    98
Deflate:x1       13103    99   1678   1664    95    94
                 53626    99   1679   1666    95    94
Deflate:x5        4716    99   1831   1816   103   102
                 53714    99   1681   1668    95    94
Deflate:x7        1591    99   1777   1763   100    99
                 54016    99   1689   1676    95    94
Deflate64:x5      4496    99   1959   1943   110   110
                 53315    99   1681   1668    95    94
BZip2:x1          2386    99   1453   1442    82    81
                 13138    99   1435   1424    81    80
BZip2:x5          1837    99   1547   1534    87    86
                 10171    99   2015   1996   114   113
BZip2:x5:mt2      3238   193   1403   2703    79   152
                 17975   187   1889   3528   106   199
BZip2:x7           549    99   1436   1424    81    80
                 10242    99   2027   2009   114   113
PPMD:x1           1303    99   1358   1348    77    76
                  1176    99   1396   1385    79    78
PPMD:x5            893    99   1526   1513    86    85
                   810    99   1531   1518    86    86
Delta:4         273894    99   1695   1683    96    95
                208646    99   1291   1282    73    72
BCJ             319116    99   1316   1307    74    74
                319118    99   1316   1307    74    74
AES256CBC:1      48980    99   1212   1204    68    68
                 51228    99   1268   1259    71    71
AES256CBC:2 

CRC32:1         215762    99   1582   1571    89    89
CRC32:4         483536    99   1087   1079    61    61
CRC32:8         672402    99    918    912    52    51
CRC64           479953    99    990    983    56    55
SHA256           82748    99   1701   1688    96    95
SHA1            141583    99   1335   1325    75    75
BLAKE2sp         85346    99   1892   1878   107   106

CPU                       99   1785   1774
------------------------------------------------------
Tot:                     110   1605   1757    90    99

Links

ARM Cortex-A55 at Wikipedia

ARM Cortex-A55 at WikiChip