ARM Cortex-A17

ARM Cortex-A17, Rockchip RK3288 1800 MHz (Beelink R89), 28 nm, 2 GB DDR3.

L1 Data cache = 32 KB, 64 B/line, 4-WAY, Physically indexed and physically tagged data cache, Pseudo random cache replacement polic
L1 Instruction cache = 32 KB, 64 B/line, 4-WAY
L2 Cache = 1 MB, 64 B/line, 16-WAY, 16-way, Pseudo-random cache replacement polic

L1 Data Cache Latency = 4 cycles for simple access via pointer
L1 Data Cache Latency = 4 cycles for access with complex address calculation (size_t n, *p; n = p[n]).
L2 Cache Latency = 26 cycles
RAM Latency = 26 cycles + 111 ns

1 MB / 2 MB page mode

Micro Data TLB: 32 items (1 MB pages). full-assoc. Miss penalty = 6 cycles for L1 cache access. Parallel miss: 4 cycles per access
intermediate table walk cache: 128 entry, 2-way

  Size       Latency        Increase   Description

  32 K     4                          
  64 K    16              12           + 22 (L2)         
 128 K    21               5           
 256 K    24               3           
 512 K    25 +    ns       1      ns        
   1 M    26 +  3 ns       1 +  3 ns   
   2 M    26 + 60 ns         + 57 ns   + 111 ns (RAM)
   4 M    26 + 86 ns         + 26 ns           
   8 M    26 + 99 ns         + 13 ns   
  16 M    26 +105 ns         +  6 ns   
  32 M    26 +108 ns         +  3 ns
  64 M    30 +109 ns       4 +  1 ns   + 8 (L1 TLB miss)        
 128 M    32 +110 ns       2 +  1 ns   
 256 M    33 +111 ns       1 +  1 ns   
 512 M    42 +111 ns       8 +    ns   + 16 
1024 M    46 +111 ns       4 +    ns

4 KB pages mode

Micro Data TLB: 32 items. full-assoc. Miss penalty = 6 cycles for L1 cache access. Parallel miss: 4 cycles per access
Main TLB: 1024 items. 4-way. Miss penalty = 20. Parallel miss: 20 cycles per access (including L2 cache access latency)
intermediate table walk cache: 128 entry, 2-way

  Size       Latency        Increase   Description

  32 K     4                          
  64 K    16              12           + 22 (L2)         
 128 K    21               5           
 256 K    26               5           + 6 (L1 TLB miss)        
 512 K    29 +    ns       3      ns        
   1 M    31 +  3 ns       2 +  3 ns   
   2 M    42 + 53 ns      10 + 53 ns   + 20 (L2 TLB miss) + 111 ns (RAM)
   4 M    47 + 78 ns       5 + 25 ns           
   8 M    50 + 94 ns       3 + 16 ns   
  16 M    51 +105 ns       1 + 11 ns   
  32 M    52 +113 ns       1 +  8 ns

MISC

Branch misprediction penalty = 12 cycles

4 bytes range cross penalty = 1 cycles
8 bytes range cross penalty = 3 cycles

L1 B/W (Parallel Random Read) = 0.58 cycle per one access

L2->L1 B/W (Parallel Random Read) = 12 cycles per cache line
L2->L1 B/W (Read, 64 bytes step) = 12 cycles per cache line
L2->L1 B/W (Read, 64 Bytes step - pointer chasing) = 12 cycles per cache line (HW prefetch)
L2 Write (Write, 64 bytes step) = 12 cycles per write

RAM Read B/W (Parallel Random Read) = 24 ns / access line
RAM Read B/W (Read, 4-64 Bytes step) = 4.4 GB/s
RAM Read B/W (Read, 12-64 Bytes step - pointer chasing) = 4.4 GB/s (HW prefetch)
RAM Write B/W (Write, 4 Bytes step, full) = 4 GB/s

7-Zip Benchmark


### Linux-5.15, Armbian
## THP on, framebuffer off
# gcc-10 -march=armv-7a -mtune=cortex-a53 -O2

7-Zip (z) 21.06 (arm) : Copyright (c) 1999-2021 Igor Pavlov : 2021-11-24
 32-bit arm_v:7 thumb:2 locale=en_US.UTF-8 Threads:4

 mt=*
Compiler: 10.3.0 GCC 10.3.0
Linux : 5.15.13-rockchip : #trunk SMP PREEMPT Sun Jan 9 19:31:16 MSK 2022 : armv7l
PageSize:4KB THP:always hwcap:3FB8D6:NEON hwcap2:0
LE 

1T CPU Freq (MHz):  1700  1794  1794  1795  1795  1795  1795
2T CPU Freq (MHz): 198% 1795   200% 1795

RAM size:    2006 MB,  # CPU hardware threads:   4
RAM usage:    437 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:       1722   100   1682   1676  |      24013   100   2054   2050
23:       1711   100   1749   1743  |      23588   100   2047   2042
24:       1675   100   1806   1802  |      23207   100   2042   2037
25:       1654   100   1894   1889  |      22767   100   2029   2026
----------------------------------  | ------------------------------
Avr:      1691   100   1783   1778  |      23394   100   2043   2039
Tot:             100   1913   1908


RAM usage:    444 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:       2680   169   1547   2608  |      47028   199   2016   4015
23:       2601   169   1566   2651  |      46380   200   2012   4015
24:       2571   170   1624   2765  |      45585   200   2004   4002
25:       2545   181   1604   2907  |      44722   199   1996   3981
----------------------------------  | ------------------------------
Avr:      2599   172   1585   2733  |      45929   199   2007   4003
Tot:             186   1796   3368


RAM usage:    889 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:       4599   326   1374   4474  |      91905   397   1973   7841
23:       4481   331   1378   4566  |      90520   399   1965   7833
24:       4400   329   1437   4732  |      88943   399   1957   7805
25:       4310   346   1421   4921  |      87381   399   1947   7777
----------------------------------  | ------------------------------
Avr:      4448   333   1402   4673  |      89687   399   1960   7814
Tot:             366   1681   6244


7-Zip (z) 21.06 (arm) : Copyright (c) 1999-2021 Igor Pavlov : 2021-11-24
 32-bit arm_v:7 thumb:2 locale=en_US.UTF-8 Threads:4

 mt=* m=* freq=1800
freq= 1800
Compiler: 10.3.0 GCC 10.3.0
Linux : 5.15.13-rockchip : #trunk SMP PREEMPT Sun Jan 9 19:31:16 MSK 2022 : armv7l
PageSize:4KB THP:always hwcap:3FB8D6:NEON hwcap2:0
LE 

1T CPU Freq (MHz):  1710  1794  1794  1795  1794  1795  1795
2T CPU Freq (MHz): 199% 1791   200% 1795  

RAM size:    2006 MB,  # CPU hardware threads:   4
RAM usage:    227 MB,  # Benchmark threads:      1


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

CPU                      100   1795   1795   100   100
CPU                      100   1797   1795   100   100
CPU                      100   1797   1795   100   100

LZMA:x1           4473   100   1655   1649    92    92
                 25004   100   2004   2000   111   111
LZMA:x3           3140   100   1932   1930   107   107
                 23526   100   1978   1974   110   110
LZMA:x5:mt1       1682   100   2108   2102   117   117
                 23170   100   1958   1954   109   109
LZMA:x5:mt2       2584   171   1892   3229   105   179
                 23219   100   1963   1958   109   109
Deflate:x1       14759   100   1878   1874   104   104
                 64211   100   1997   1995   111   111
Deflate:x5        4767   100   1839   1836   102   102
                 64283   100   2001   1996   111   111
Deflate:x7        1712   100   1901   1898   106   105
                 64927   100   2019   2015   112   112
Deflate64:x5      4311   100   1868   1863   104   104
                 63578   100   1992   1989   111   110
BZip2:x1          2569   100   1556   1552    86    86
                 15997   100   1737   1734    97    96
BZip2:x5          1826   100   1528   1524    85    85
                  9000   100   1769   1766    98    98
BZip2:x5:mt2      3104   195   1330   2591    74   144
                 14167   161   1729   2781    96   154
BZip2:x7           598   100   1553   1550    86    86
                  9047   100   1777   1774    99    99
PPMD:x1           1613   100   1673   1669    93    93
                  1377   100   1624   1622    90    90
PPMD:x5           1086   100   1846   1842   103   102
                   953   100   1791   1787   100    99
Delta:4         632338   100   1943   1943   108   108
                582585   100   2389   2386   133   133
BCJ             350556   100   1436   1436    80    80
                370344   100   1518   1517    84    84
AES256CBC:1      46510   100   1144   1143    64    64
                 44088   100   1085   1084    60    60
AES256CBC:2

AES256CBC:3

CRC32:1         194491   100   1416   1416    79    79
CRC32:4         612778   100   1369   1368    76    76
CRC32:8         724690   100    983    983    55    55
CRC32:32
CRC32:64
CRC64           397024   100    814    813    45    45
SHA256:1         84389   100   1723   1722    96    96
SHA256:2
SHA1:1          154086   100   1444   1442    80    80
SHA1:2
BLAKE2sp         69923   100   1539   1538    86    85

CPU                      100   1797   1795   100   100
------------------------------------------------------
Tot:                     110   1824   2001   101   111


RAM usage:    911 MB,  # Benchmark threads:      4


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

CPU                      400   1798   7187   100   399
CPU                      400   1798   7188   100   399
CPU                      400   1798   7187   100   399

LZMA:x1          11062   398   1025   4078    57   227
                 97494   399   1954   7796   109   433
LZMA:x3           8605   398   1329   5287    74   294
                 90543   399   1902   7596   106   422
LZMA:x5:mt1       5312   399   1664   6636    92   369
                 89114   399   1881   7513   104   417
LZMA:x5:mt2       5008   394   1589   6257    88   348
                 89082   399   1881   7510   105   417
Deflate:x1       29316   398    935   3722    52   207
                253624   398   1978   7880   110   438
Deflate:x5       13579   398   1315   5228    73   290
                254346   398   1982   7896   110   439
Deflate:x7        6152   399   1710   6817    95   379
                255679   398   1991   7934   111   441
Deflate64:x5     11415   399   1236   4933    69   274
                248659   398   1953   7775   108   432
BZip2:x1          7456   394   1144   4505    64   250
                 43318   400   1175   4695    65   261
BZip2:x5          5207   396   1098   4346    61   241
                 27448   399   1350   5386    75   299
BZip2:x5:mt2      5128   395   1083   4280    60   238
                 27922   394   1391   5480    77   304
BZip2:x7          1971   396   1291   5106    72   284
                 27435   400   1346   5380    75   299
PPMD:x1           5699   398   1483   5895    82   328
                  4976   398   1472   5860    82   326
PPMD:x5           2978   399   1265   5047    70   280
                  2732   399   1284   5121    71   285
Delta:4        2499670   399   1922   7679   107   427
               2331854   400   2389   9551   133   531
BCJ            1396990   398   1438   5722    80   318
               1483456   399   1522   6076    85   338
AES256CBC:1     185431   399   1141   4557    63   253
                176431   399   1085   4336    60   241
AES256CBC:2

AES256CBC:3

CRC32:1         777038   399   1416   5657    79   314
CRC32:4        2414517   397   1356   5389    75   299
CRC32:8        2854383   395    981   3871    54   215
CRC32:32
CRC32:64
CRC64          1541332   396    798   3157    44   175
SHA256:1        337629   400   1724   6888    96   383
SHA256:2
SHA1:1          616701   398   1451   5772    81   321
SHA1:2
BLAKE2sp        278557   399   1536   6128    85   340

CPU                      400   1798   7187   100   399
------------------------------------------------------
Tot:                     398   1570   6249    87   347

Links

ARM Cortex-A17 at Wikipedia