MIPS P5600

Baikal-T1 (MIPS P5600), 1200 MHz, 2 cores, 28 nm, 5 W. RAM: DDR3.

L1 Data cache = 64 KB. 4-way, 32 B/line, 6 LRU bits, VIPT. Up to 9 misses may be outstanding. 32 parity bits per line.
L1 Instruction cache = 64 KB, 4-way. 32 B/line, 6 LRU bits. VIPT, 7 precode bits per 2 instructions, 32 parity bits per line.
L2 Data cache = 1 MB (shared by 2 cores). 32 B/line, 8-way, PIPT. up to 15 misses are outstanding.

L1 Data Cache Latency = 4 cycles
L1 Data Cache Latency = 6 cycles for access with complex address calculation (size_t n, *p; n = p[n]). MIPS ISA doesn't support complex address modes in LOAD instruction.
L2 Cache Latency = 32 cycles
RAM Latency = 32 cycles + 157 ns

Fetch: 4 instructions.
2048-entry global branch history tables (BHT) that are indexed by different combinations of instruction PC and global history.
BTB L1: full-assoc, BTB L2: 4-way.
24-entry instruction buffer decouples the instruction fetch from the execution. Up to 4 instructions can be written into this buffer, and a maximum of 2 instructions can be read from this buffer.
out-of-order (OoO) execution

16-entry Instruction TLB (ITLB) (4 KB or 16 KB pages)
32 dual-entry Data TLB (DTLB) (4 KB or 16 KB pages). Larger pages are split into one of these configured sizes and installed in the DTLB. The DTLB is managed by hardware and is transparent to software.
64 dual-entry Variable Page Size VTLB (4K, 16K, 64K, 256K, 1M, 4M, 16M, 64M, and 256M pages) per entry. Full-assoc.
512 dual-entry 4-way Fixed Page Size FTLB (4 KB or 16 KB pages)
FTLB/VTLB miss causes a TLB exception.

16 KB pages

DTLB L1 size = 32 items (2 pages per item)
FTLB L2 size = 512 items, (2 pages per item)

DTLB L1 miss penalty = 12 cycles, Parallel miss: 2 cycles per read access.
FTLB (L2) miss penalty = 100-120 cycles, Parallel miss: 120 cycles per read access.

  64 K     4                              TLB + L1
 128 K    19               +15            + 28 (L2)
 256 K    26               + 7            
 512 K    29               + 3 
   1 M    31 +  17 ns      + 2 + 17 ns     
   2 M    38 +  87 ns      + 7 + 70 ns    + 157 ns (RAM) + 12 (TLB L1 miss)                
   4 M    41 + 122 ns      + 3 + 35 ns               
   8 M    43 + 140 ns      + 2 + 18 ns     
  16 M    44 + 149 ns      + 1 +  9 ns     
  32 M    94 + 153 ns      +50 +  4 ns    + 120 (FTLB L2 miss)                
  64 M   127 + 155 ns      +33 +  2 ns     
 128 M   150 + 156 ns      +23 +  1 ns     
 256 M   166 + 157 ns      +16 +  1 ns
 512 M   191 + 157 ns      +25
1024 M   236 + 157 ns      +43

Branch misprediction penalty = 11-18 cycles.

4-bytes range cross penalty = 360 cycles

L1 B/W (Parallel Random Read) = 1 cycle per one access
L1 B/W (Parallel Random Read with complex address calculation) = 1.5 cycle per one access

L2 Read B/W (Read, 32 Bytes step, pointer-chasing) = 32 cycles per one access (no hardware prefetch)
L2 Read B/W (Read, 32 Bytes step) = 4.7 cycles per one access
L2 Read B/W (Parallel Random Read) = 4.8 cycles per one access
L2 Write (32 Bytes step, with data skip) = 9.9 cycles per one access

RAM Read B/W (Parallel Random Read) = 32 ns / per cache line access (including L1 TLB miss)
RAM Read B/W (Read, 8-32 Bytes step) = 2600 MB/s
RAM Read B/W (Read, 32 Bytes step, pointer-chasing) = 1060 MB/s (hardware prefetch)
RAM Write (4 Bytes step) = 3900 MB/s
RAM Write (8-32 Bytes step, with data skip) = 1700 MB/s

7-Zip Benchmark

7z b : MIPS values are normalized with Intel Core 2 cpu.


### Linux 4.4
# gcc-7 -O3 -march=p5600 -mtune=74kc

7-Zip (a) 16.04 : Copyright (c) 1999-2016 Igor Pavlov : 2016-10-04
p7zip Version 16.04 (locale=C,Utf16=off,HugeFiles=on,32 bits,2 CPUs LE)

LE
CPU Freq:  1198  1199  1199  1199  1199  1199  1199  1199  1199

RAM size:    1904 MB,  # CPU hardware threads:   2
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:        938   100    914    913  |      11895   100   1015   1016
23:        861   100    879    878  |      11668   100   1009   1010
24:        802   100    863    863  |      11448   100   1005   1005
25:        762   100    871    871  |      11079   100    987    986
----------------------------------  | ------------------------------
Avr:             100    882    881  |              100   1004   1004
Tot:             100    943    943

7-Zip (a) 16.04 : Copyright (c) 1999-2016 Igor Pavlov : 2016-10-04
p7zip Version 16.04 (locale=C,Utf16=off,HugeFiles=on,32 bits,2 CPUs LE)

LE
CPU Freq:  1198  1199  1199  1199  1199  1199  1199  1199  1199

RAM size:    1904 MB,  # CPU hardware threads:   2
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:       1457   176    805   1418  |      23183   197   1007   1979
23:       1435   182    803   1463  |      22749   197   1002   1969
24:       1342   187    770   1443  |      22311   197    996   1959
25:       1275   190    768   1457  |      21600   196    979   1923
----------------------------------  | ------------------------------
Avr:             184    786   1445  |              197    996   1957
Tot:             190    891   1701

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


7-Zip (a) 16.04 : Copyright (c) 1999-2016 Igor Pavlov : 2016-10-04
p7zip Version 16.04 (locale=C,Utf16=off,HugeFiles=on,32 bits,2 CPUs LE)

freq= 1200
LE
CPU Freq:  1198  1199  1199  1199  1199  1199  1199  1199  1199

RAM size:    1904 MB,  # CPU hardware threads:   2
RAM usage:    225 MB,  # Benchmark threads:      1


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

CPU                      100   1198   1199   100   100
CPU                      100   1201   1199   100   100
CPU                      100   1198   1199   100   100

LZMA:x1           3919   100   1433   1433   119   119
                 12120   100    986    987    82    82
LZMA:x5:mt1        805   100   1006   1006    84    84
                 11449   100    966    966    81    80
LZMA:x5:mt2       1347   187    902   1683    75   140
                 11456   100    966    966    81    81
Deflate:x1        9352   100   1185   1188    99    99
                 32636   100   1016   1014    85    85
Deflate:x5        3084   100   1186   1188    99    99
                 32677   100   1015   1014    85    85
Deflate:x7        1087   100   1205   1205   100   100
                 32818   100   1019   1018    85    85
Deflate64:x5      2842   100   1227   1228   102   102
                 32285   100   1011   1010    84    84
BZip2:x1          1621   100    980    980    82    82
                  6975   100    756    756    63    63
BZip2:x5          1219   100   1018   1018    85    85
                  4248   100    833    834    69    69
BZip2:x5:mt2      2062   193    890   1721    74   143
                  7690   192    786   1509    66   126
BZip2:x7           402   100   1044   1044    87    87
                  4263   100    835    836    70    70
PPMD:x1            998   100   1033   1033    86    86
                   845   100    996    995    83    83
PPMD:x5            605   100   1027   1027    86    86
                   521   100    977    977    81    81
Delta:4         151403   100    930    930    78    78
                127977   100    787    786    66    66
BCJ             294367   100   1203   1206   100   100
                295055   100   1209   1209   101   101
AES256CBC:1      23861   100    587    586    49    49
                 24886   100    611    612    51    51
AES256CBC:2 

CRC32:1         128762   100    937    937    78    78
CRC32:4         281686   100    628    629    52    52
CRC32:8         297287   100    403    403    34    34
CRC64           205858   100    422    422    35    35
SHA256           43204   100    881    881    73    73
SHA1             77056   100    722    721    60    60
BLAKE2sp         58348   100   1283   1284   107   107

CPU                      100   1198   1199   100   100
------------------------------------------------------
Tot:                     112    970   1078    80    90

7-Zip (a) 16.04 : Copyright (c) 1999-2016 Igor Pavlov : 2016-10-04
p7zip Version 16.04 (locale=C,Utf16=off,HugeFiles=on,32 bits,2 CPUs LE)

freq= 1200
LE
CPU Freq:  1198  1199  1199  1199  1199  1199  1199  1199  1199

RAM size:    1904 MB,  # CPU hardware threads:   2
RAM usage:    450 MB,  # Benchmark threads:      2


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

CPU                      196   1201   2357   100   196
CPU                      199   1200   2393   100   199
CPU                      199   1200   2393   100   199

LZMA:x1           5798   197   1075   2120    90   177
                 24186   199    988   1970    82   164
LZMA:x5:mt1       1493   199    938   1866    78   156
                 22617   199    957   1908    80   159
LZMA:x5:mt2       1355   199    852   1693    71   141
                 22635   199    957   1909    80   159
Deflate:x1       16027   196   1038   2035    86   170
                 64502   197   1017   2004    85   167
Deflate:x5        5562   197   1087   2142    91   178
                 64627   197   1017   2006    85   167
Deflate:x7        2064   199   1151   2288    96   191
                 64814   197   1021   2011    85   168
Deflate64:x5      4771   198   1041   2062    87   172
                 64456   200   1010   2016    84   168
BZip2:x1          3018   199    918   1823    76   152
                 13577   200    738   1472    61   123
BZip2:x5          2154   199    904   1798    75   150
                  7764   200    763   1524    64   127
BZip2:x5:mt2      2075   199    869   1732    72   144
                  7605   197    757   1493    63   124
BZip2:x7           725   200    942   1880    79   157
                  7764   200    763   1523    64   127
PPMD:x1           1944   199   1013   2011    84   168
                  1654   199    979   1949    82   162
PPMD:x5           1117   199    951   1893    79   158
                   977   199    919   1831    77   153
Delta:4         302523   200    931   1859    78   155
                256008   199    789   1573    66   131
BCJ             594273   200   1220   2434   102   203
                593542   199   1220   2431   102   203
AES256CBC:1      47682   200    587   1172    49    98
                 49321   198    612   1212    51   101
AES256CBC:2 

CRC32:1         257818   199    941   1877    78   156
CRC32:4         566315   199    634   1264    53   105
CRC32:8         597936   199    406    811    34    68
CRC64           411517   199    424    843    35    70
SHA256           84513   197    874   1724    73   144
SHA1            154055   199    723   1442    60   120
BLAKE2sp        114809   196   1286   2526   107   210

CPU                      199   1200   2393   100   199
------------------------------------------------------
Tot:                     199    916   1819    76   152

Links

MIPS P5600 at mips.com

Baikal-T1 at Baikal Electronics