See Opcodes Costs

Related publication : "An Energy Consumption Model for an Embedded Java Virtual Machine"

Lecture Notes in Computer Science

Publisher: Springer Berlin / Heidelberg

ISSN: 0302-9743

Subject: Computer Science

Volume 3894 / 2006

Title: Architecture of Computing Systems - ARCS 2006: 19th International Conference, Frankfurt/Main, Germany, March 13-16, 2006. Proceedings

Editors: Werner Grass, Bernhard Sick, Klaus Waldschmidt

ISBN: 3-540-32765-7

DOI: 10.1007/11682127

Chapter: pp. 311 - 325

DOI: 10.1007/11682127_22

Benchmarks for the KVM

All + All -

+ - The Computer Language Shootout Benchmarks
- http://shootout.alioth.debian.org/
+ - Java Microbenchmarks
- http://www-2.cs.cmu.edu/~jch/java/benchmarks.html
+ - A list of benchmark links
- http://www.epcc.ed.ac.uk/javagrande/links.html
- http://www.elec.uq.edu.au/~rae/java.html
+ - Java Grande Benchmarking
- http://www.dhpc.adelaide.edu.au/projects/javagrande/benchmarks/
- http://www.epcc.ed.ac.uk/javagrande/index_1.html
- + - Section 1: Low Level Operations
  - + - Arith: execution of arithmetic operations
    - Works with the minimum of 32Kb
    - 191.330u 0.010s 3:12.02 99.6% 0+0k 0+0io 175pf+0w
  - + - Assign: variable assignment
    - Works with the minimum of 32Kb
    - 302.500u 0.060s 5:03.77 99.6% 0+0k 0+0io 175pf+0
  - + - Create: creating objects and arrays
    - Works with the minimum of 32Kb
    - 919.120u 0.230s 15:29.05 98.9% 0+0k 0+0io 176pf+0w
  - + - Loop: Loop overheads
    - Works with the minimum of 32Kb
    - 100.490u 0.020s 1:40.97 99.5% 0+0k 0+0io 175pf+0w
  - + - Math: execution of maths library operations
    - Works with the minimum of 32Kb
    - 152.100u 0.000s 2:32.76 99.5% 0+0k 0+0io 175pf+0w
  - + - Method: method invocation
    - Works with the minimum of 32Kb
    - 22.560u 0.020s 0:22.69 99.5% 0+0k 0+0io 175pf+0w
  - + - Exception Exception handling
    - Works with the minimum of 32Kb
    - 105.260u 0.000s 1:46.12 99.1% 0+0k 0+0io 175pf+0
  - Serial: Serialisation
  - Cast: casting
- + - Section 2: Kernels
  - Series: Fourier coefficient analysis
  - LUFact: LU Factorisation
  - SOR: Successive over-relaxation
  - HeapSort: Integer sorting
    - + - Need 4Mb of heap for size A
      - with 4Mb : 58.240u 0.020s 0:58.75 99.1% 0+0k 0+0io 179pf+0w
    - + - Need 20Mb of heap for size B
      - with 20Mb :325.400u 0.150s 5:28.73 99.0% 0+0k 0+0io 179pf+0w
    - Can not run problem of size C
  - Crypt: IDEA encryption
    - + - Need 9Mb of heap for size A
      - with 9Mb : 49.130u 0.050s 0:49.39 99.5% 0+0k 0+0io 179pf+0w
    - Can't run size B and C problem
  - FFT: FFT
  - Sparse: Sparse Matrix multiplication
- + - Section 3: Large Scale Applications
  - + - Search: Alpha-beta pruned search
    - Need at least 6Mb of heap for size A
      - with 6Mb : 377.460u 0.090s 6:20.35 99.2% 0+0k 0+0io 179pf+0w
    - Need also at least 6Mb of heap for size B
      - with 6Mb : 1651.730u 0.340s 27:45.65 99.1% 0+0k 0+0io 179pf+0w
    - SizeB : very very slow !
  - Euler: Computational Fluid Dynamics
  - MD: Molecular Dynamics simulation
  - MC: Monte Carlo simulation
  - Ray Tracer: 3D Ray Tracer
+ - The DHPC Java Grande Benchmarks
- http://www.dhpc.adelaide.edu.au/projects/javagrande/benchmarks/
- + - Section1a
  - + - Object Creation
    - The same as Create
  - + - Array Creation
    - The same as Create
  - + - Exception Throwing
    - The same as Exeption
  - + - Thread Switching
    - Remove the call to setName(String) in class TestThread. setName is NOT implemented in J2ME
    - Change t[0].join(10000); into . join(long) method is NOT implemented in J2ME
    - works with the minimum of 32k of heap
    - VERY VERY SLOW !
    - TO SLOW TO BE SIMULATED ON ARMulator > 2700 minutes...
  - + - Method Calling
    - The same as Methos
  - + - Looping
    - Almaos the same as Loop
  - + - Generic Code Segments
    - works with the minimum of 32k of heap
      Generic Code Segments
      Generic:Add
         Incrementing a local variable in a loop.
      Generic:Add:ClassVariable
         Incrementing a static variable in a loop.
      Generic:Add:Method
         Incrementing a static variable using a method call.
      Generic:Add:Loop:n
         Incrementing a local variable in a loop with n operations per loop iteration.
      fast
      21.070u 0.090s 0:24.99 84.6%    0+0k 0+0io 184pf+0w
  - "JGFBenchMark" class is in fact "JGFInstrumentor" class from the JGB
- + - Section2a
  - + - Fibonacci
    - works with the minimum of 32k of heap
    - Slow !
    - 145.050u 0.200s 2:32.12 95.4% 0+0k 0+0io 181pf+0w
  - + - Sieve
    - works with the minimum of 32k of heap
    - 52.760u 0.090s 0:57.68 91.6% 0+0k 0+0io 181pf+0w
  - + - Hanoi
    - works with the minimum of 32k of heap
    - 85.560u 0.120s 1:37.67 87.7% 0+0k 0+0io 181pf+0w
  - + - NAS EP
    - Use fp
  - + - NAS IS
    - Use fp
  - + - NAS FT
    - Use fp
  - "JGFBenchMark" class is in fact "JGFInstrumentor" class from the JGB
+ - Dhrystone Benchmark
- http://www.c-creators.co.jp/okayan/
  need at least 78Kb of heap
  [slafond@exupery output]$ kvm104 -heapsize 1M
  dhry Dhrystone Benchmark, Version 2.1 (Language: Java) Execution starts, 100000 runs through Dhrystone
  total time: 25421ms
  Result: 3933 dhrystone/sec.
  Execution completed successfully
  80107261 bytecodes executed
  0 thread switches
  132 classes in the system (including system classes)
  200510 dynamic objects allocated (4081632 bytes)
  8 garbage collections (3885832 bytes collected)
  
  5.550u 0.000s 0:05.56 99.8% 0+0k 0+0io 185pf+0w
  [slafond@exupery output]$ kvm104 -heapsize 78K
  dhry Dhrystone Benchmark, Version 2.1 (Language: Java) Execution starts, 100000 runs through Dhrystone
  total time: 27521ms
  Result: 3633 dhrystone/sec
  Execution completed successfully
  80107261 bytecodes executed
  0 thread switches
  132 classes in the system (including system classes)
  200510 dynamic objects allocated (4081632 bytes)
  2332 garbage collections (4020280 bytes collected)
  
  5.650u 0.010s 0:06.30 89.8% 0+0k 0+0io 185pf+0w
+ - The Fhourstones 2.0 Benchmark
- http://homepages.cwi.nl/~tromp/c4/fhour2.html
  need at least 6Mb of heap !!!!
  
  [slafond@exupery output]$ kvm104 -heapsize 6M SearchGame
  Java Fhourstones 2.0
  Using 1050011 transposition table entries with 8 probes.
  Solving 17-ply position after 44433337765675566 . . .
  score = -2 (-) work = 15
  43000 pos / 20117 msec = 2 Kpos/sec
  store rate = 1
  - 0 < 0 = 0 > 0 + 0
  1031982 8737    3020    2673    1773    871     490     247
  110     58      30      10      4       3       1       2
  0       0       0       0       0       0       0       0
  0       0       0       0       0       0       0       0
  
  Execution completed successfully
  95061258 bytecodes executed
  0 thread switches
  129 classes in the system (including system classes)
  43463 dynamic objects allocated (7336368 bytes)
  8 garbage collections (2053044 bytes collected)
  
  5.750u 0.020s 0:07.10 81.2%     0+0k 0+0io 186pf+0w
+ - SpecJVM98
- http://www.spec.org/osg/jvm98/
- These cost money!
+ - EEMBC
- ww.eembc.hotdesk.com/
- NOT FREE (1000$...)

See Opcodes Costs

See Benchmarks

Opcodes Costs

Table 1: Opcodes costs, conversion opcodes
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

i2l 0x85 0.928660 2.200260 198 50

l2i 0x88 0.857440 2.037840 184 47

i2b 0x91 0.928540 2.200260 198.0 50

i2c 0x92 0.928520 2.200260 198 50

i2s 0x93 0.928540 2.200260 198 50

**Table 1:** Opcodes costs, conversion opcodes
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
i2l 0x85	0.928660	2.200260	198	50
l2i 0x88	0.857440	2.037840	184	47
i2b 0x91	0.928540	2.200260	198.0	50
i2c 0x92	0.928520	2.200260	198	50
i2s 0x93	0.928540	2.200260	198	50

Table 2: Opcodes costs, stack and local variable operations-part 1/2
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

nop 0x0 0.831440 1.989840 178 45

aconst_null 0x1 0.890020 2.126940 190 49

iconst_m1 0x2 0.899160 2.150940 192 50

iconst_0 0x3 0.890020 2.126940 190 49

iconst_1 0x4 0.890020 2.126940 190 49

iconst_2 0x5 0.890020 2.126940 190 49

iconst_3 0x6 0.890020 2.126940 190 49

iconst_4 0x7 0.889760 2.126940 190 49

iconst_5 0x8 0.890020 2.126940 190 49

lconst_0 0x9 0.922300 2.192040 196 50

lconst_1 0xa 0.930960 2.216040 198 51

bipush 0x10 0.926900 2.214420 198 52

sipush 0x11 0.990360 2.373900 212 58

iload 0x15 1.013700 2.434380 216 55

lload 0x16 1.167820 2.815440 248 63

aload 0x19 1.013700 2.434380 216 55

iload_0 0x1a 0.968120 2.322900 206 51

iload_1 0x1b 0.968120 2.322900 206 51

iload_2 0x1c 0.968120 2.322900 206 51

iload_3 0x1d 0.968120 2.322900 206 51

lload_0 0x1e 1.104800 2.655960 234 57

lload_1 0x1f 1.104800 2.655960 234 57

lload_2 0x20 1.104800 2.655960 234 57

lload_3 0x21 1.104800 2.655960 234 57

aload_0 0x2a 0.968120 2.322900 206 51

aload_1 0x2b 0.968120 2.322900 206 51

aload_2 0x2c 0.968120 2.322900 206 51

aload_3 0x2d 0.968120 2.322900 206 51

istore 0x36 1.004140 2.410380 214 54

lstore 0x37 1.148940 2.767440 244 61

astore 0x3a 1.004140 2.410380 214 54

**Table 2:** Opcodes costs, stack and local variable operations-part 1/2
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
nop 0x0	0.831440	1.989840	178	45
aconst_null 0x1	0.890020	2.126940	190	49
iconst_m1 0x2	0.899160	2.150940	192	50
iconst_0 0x3	0.890020	2.126940	190	49
iconst_1 0x4	0.890020	2.126940	190	49
iconst_2 0x5	0.890020	2.126940	190	49
iconst_3 0x6	0.890020	2.126940	190	49
iconst_4 0x7	0.889760	2.126940	190	49
iconst_5 0x8	0.890020	2.126940	190	49
lconst_0 0x9	0.922300	2.192040	196	50
lconst_1 0xa	0.930960	2.216040	198	51
bipush 0x10	0.926900	2.214420	198	52
sipush 0x11	0.990360	2.373900	212	58
iload 0x15	1.013700	2.434380	216	55
lload 0x16	1.167820	2.815440	248	63
aload 0x19	1.013700	2.434380	216	55
iload_0 0x1a	0.968120	2.322900	206	51
iload_1 0x1b	0.968120	2.322900	206	51
iload_2 0x1c	0.968120	2.322900	206	51
iload_3 0x1d	0.968120	2.322900	206	51
lload_0 0x1e	1.104800	2.655960	234	57
lload_1 0x1f	1.104800	2.655960	234	57
lload_2 0x20	1.104800	2.655960	234	57
lload_3 0x21	1.104800	2.655960	234	57
aload_0 0x2a	0.968120	2.322900	206	51
aload_1 0x2b	0.968120	2.322900	206	51
aload_2 0x2c	0.968120	2.322900	206	51
aload_3 0x2d	0.968120	2.322900	206	51
istore 0x36	1.004140	2.410380	214	54
lstore 0x37	1.148940	2.767440	244	61
astore 0x3a	1.004140	2.410380	214	54

Table 3: Opcodes costs, stack and local variable operations-part 2/2
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

istore_0 0x3b 0.958800 2.298900 204 50

istore_1 0x3c 0.958800 2.298900 204 50

istore_2 0x3d 0.958800 2.298900 204 50

istore_3 0x3e 0.958800 2.298900 204 50

lstore_0 0x3f 1.086160 2.607960 230 55

lstore_1 0x40 1.086160 2.607960 230 55

lstore_2 0x41 1.086160 2.607960 230 55

lstore_3 0x42 1.086160 2.607960 230 55

astore_0 0x4b 0.958800 2.298900 204 50

astore_1 0x4c 0.958800 2.298900 204 50

astore_2 0x4d 0.958800 2.298900 204 50

astore_3 0x4e 0.958800 2.298900 204 50

pop 0x57 0.857440 2.037840 184 47

pop2 0x58 0.857440 2.037840 184 47

dup 0x59 0.928740 2.200260 198 50

dup_x1 0x5a 1.040200 2.451780 220 55

dup_x2 0x5b 1.119080 2.638200 236 59

dup2 0x5c 1.026160 2.434680 218 56

dup2_x1 0x5d 1.169000 2.751300 246 62

dup2_x2 0x5e 1.321140 3.100140 276 69

swap 0x5f 0.990280 2.338680 210 52

ldc 0x12 1.022440 2.458380 218 56

ldc_w 0x13 1.085880 2.617860 232 62

ldc2_w 0x14 1.203000 2.870700 254 65

**Table 3:** Opcodes costs, stack and local variable operations-part 2/2
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
istore_0 0x3b	0.958800	2.298900	204	50
istore_1 0x3c	0.958800	2.298900	204	50
istore_2 0x3d	0.958800	2.298900	204	50
istore_3 0x3e	0.958800	2.298900	204	50
lstore_0 0x3f	1.086160	2.607960	230	55
lstore_1 0x40	1.086160	2.607960	230	55
lstore_2 0x41	1.086160	2.607960	230	55
lstore_3 0x42	1.086160	2.607960	230	55
astore_0 0x4b	0.958800	2.298900	204	50
astore_1 0x4c	0.958800	2.298900	204	50
astore_2 0x4d	0.958800	2.298900	204	50
astore_3 0x4e	0.958800	2.298900	204	50
pop 0x57	0.857440	2.037840	184	47
pop2 0x58	0.857440	2.037840	184	47
dup 0x59	0.928740	2.200260	198	50
dup_x1 0x5a	1.040200	2.451780	220	55
dup_x2 0x5b	1.119080	2.638200	236	59
dup2 0x5c	1.026160	2.434680	218	56
dup2_x1 0x5d	1.169000	2.751300	246	62
dup2_x2 0x5e	1.321140	3.100140	276	69
swap 0x5f	0.990280	2.338680	210	52
ldc 0x12	1.022440	2.458380	218	56
ldc_w 0x13	1.085880	2.617860	232	62
ldc2_w 0x14	1.203000	2.870700	254	65

See Benchmarks

Table 4: Opcodes costs, arithmetic opcodes
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

iadd 0x60 0.957860 2.273580 204 51

isub 0x64 0.957360 2.273580 204 51

imul 0x68 0.959500 2.273580 204 51

idiv 0x6c 1.613020 3.851460 348 84

ladd 0x61 1.575480 3.631800 328 76

lsub 0x62 1.575480 3.631800 328 76

lmul 0x69 1.638000 3.865920 348 74

ldiv 0x6d 3.685660 9.344040 820 181

iinc 0x84 1.188360 2.830920 252 63

ineg 0x74 0.920080 2.176260 196 49

lneg 0x75 1.366460 3.136320 286 67

irem 0x70 1.613020 3.851460 348 84

lrem 0x71 3.685660 9.344040 820 181

**Table 4:** Opcodes costs, arithmetic opcodes
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
iadd 0x60	0.957860	2.273580	204	51
isub 0x64	0.957360	2.273580	204	51
imul 0x68	0.959500	2.273580	204	51
idiv 0x6c	1.613020	3.851460	348	84
ladd 0x61	1.575480	3.631800	328	76
lsub 0x62	1.575480	3.631800	328	76
lmul 0x69	1.638000	3.865920	348	74
ldiv 0x6d	3.685660	9.344040	820	181
iinc 0x84	1.188360	2.830920	252	63
ineg 0x74	0.920080	2.176260	196	49
lneg 0x75	1.366460	3.136320	286	67
irem 0x70	1.613020	3.851460	348	84
lrem 0x71	3.685660	9.344040	820	181

Table 5: Opcodes costs, object and arrays-Part1/2
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

new 0xbb (java.lang.Object) 5.456560 12.437760 1146 240

new (java.lang.String) 5.508280 12.561060 1158 240

putfield 0xb5 4.201320 9.604260 872 185

putfield (long)0xb5 4.432900 10.139520 918 196

getfield 0xb4 4.156160 9.515160 864 183

getfield 0xb4(long) 4.324340 9.912000 898 192

putstatic 0xb3 4.100420 9.381000 856 185

putstatic (long) 0xb3 4.334720 9.949800 904 195

getstatic 0xb2 4.083360 9.357000 852 184

getstatic (long)0xb2 4.303640 9.901800 898 193

checkcast 0xc0(is String 'castable' to Object,yes) 3.726300 8.497920 774 156

checkcast 0xc0 3.726300 8.497920 774 156

(is Class 'castable' to Object,yes)

checkcast 0xc0 4.243800 9.677880 888 191

(is SubClass 'castable' to Class,yes)

checkcast(is Class 'castable' to 133.521178 302.378599 28250 7058

subClass,no throws ClassCastException)

checkcast(is Object 'castable' to String,0 133.292438 301.846638 28200 7044

no throws ClassCastException)

instanceof (is String instanceof Object,yes) 0xc1 3.815760 8.733660 792 160

instanceof (is Object instanceof String,no) 0xc1 4.213440 9.600299 880 186

instanceof (is This instance of Object,yes) 0xc1 3.815760 8.733660 792 160

instanceof (is Object instance of This,no) 0xc1 4.213440 9.600299 880 186

instanceof (is This instance of String,no) 0xc1 4.442180 10.132260 930 200

instanceof 4.442180 10.132260 930 200

(is Class instance of subClass,no)

instanceof 4.333260 9.913620 906 195

(is SubClass instance of Class,yes)

newarray 0xbc (of size 0) 4.417460 10.073760 924 192

newarray 0xbc (of 4 bytes) 4.434700 10.114860 928 192

newarray 0xbc (of 8 bytes) 4.451940 10.155960 932 192

newarray 0xbc (of 16 bytes) 4.486420 10.238160 940 192

newarray 0xbc (of 32 bytes) 4.555380 10.402560 956 192

newarray 0xbc (of 64 bytes) 4.770880 10.928640 1006 196

**Table 5:** Opcodes costs, object and arrays-Part1/2
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
new 0xbb (java.lang.Object)	5.456560	12.437760	1146	240
new (java.lang.String)	5.508280	12.561060	1158	240
putfield 0xb5	4.201320	9.604260	872	185
putfield (long)0xb5	4.432900	10.139520	918	196
getfield 0xb4	4.156160	9.515160	864	183
getfield 0xb4(long)	4.324340	9.912000	898	192
putstatic 0xb3	4.100420	9.381000	856	185
putstatic (long) 0xb3	4.334720	9.949800	904	195
getstatic 0xb2	4.083360	9.357000	852	184
getstatic (long)0xb2	4.303640	9.901800	898	193
checkcast 0xc0(is String 'castable' to Object,yes)	3.726300	8.497920	774	156
checkcast 0xc0	3.726300	8.497920	774	156
(is Class 'castable' to Object,yes)
checkcast 0xc0	4.243800	9.677880	888	191
(is SubClass 'castable' to Class,yes)
checkcast(is Class 'castable' to	133.521178	302.378599	28250	7058
subClass,no throws ClassCastException)
checkcast(is Object 'castable' to String,0	133.292438	301.846638	28200	7044
no throws ClassCastException)
instanceof (is String instanceof Object,yes) 0xc1	3.815760	8.733660	792	160
instanceof (is Object instanceof String,no) 0xc1	4.213440	9.600299	880	186
instanceof (is This instance of Object,yes) 0xc1	3.815760	8.733660	792	160
instanceof (is Object instance of This,no) 0xc1	4.213440	9.600299	880	186
instanceof (is This instance of String,no) 0xc1	4.442180	10.132260	930	200
instanceof	4.442180	10.132260	930	200
(is Class instance of subClass,no)
instanceof	4.333260	9.913620	906	195
(is SubClass instance of Class,yes)
newarray 0xbc (of size 0)	4.417460	10.073760	924	192
newarray 0xbc (of 4 bytes)	4.434700	10.114860	928	192
newarray 0xbc (of 8 bytes)	4.451940	10.155960	932	192
newarray 0xbc (of 16 bytes)	4.486420	10.238160	940	192
newarray 0xbc (of 32 bytes)	4.555380	10.402560	956	192
newarray 0xbc (of 64 bytes)	4.770880	10.928640	1006	196

See Benchmarks

Table 6: Opcodes costs, object and arrays-Part2/2
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

anewarray (size=0 non resolved empty class) 0xbd 34.638980 79.455499 7388 1908

anewarray (size=1 non resolved empty class) 0xbd 34.656220 79.496599 7392 1908

anewarray (size=5 non resolved empty class) 0xbd 34.725180 79.660999 7408 1908

anewarray (size=0 resolved empty class)0xbd 32.7967 75.5584 6996 1803

anewarray (size=1 resolved empty class)0xbd 32.8139 75.5995 7000 1803

anewarray (size=5 resolved empty class)0xbd 32.8829 75.7639 7016 1803

anewarray (size=0 resolved java.lang.Object class) 27.806520 64.268619 5936 1502

anewarray (size=1 resolved java.lang.Object class) 27.823760 64.309719 5940 1502

anewarray (size=5 resolved java.lang.Object class) 27.892720 64.474119 5956 1502

anewarray (size=0 resolved java.lang.String class) 27.754860 64.073619 5928 1502

anewarray (size=1 resolved java.lang.String class) 27.772100 64.114719 5932 1502

anewarray (size=5 resolved java.lang.String class) 27.841060 64.279119 5948 1502

multiarray (int 1 dimension,size=0/dim.)0xc5 36.323520 83.483859 7700 1872

multiarray (int 2 dimensions,size=4/dim.)0xc5 49.866399 113.914719 10542 2450

multiarray (int 4 dimension,size=2) 84.130479 190.890879 17702 3934

multiarray (nonResolved 1 dimension,size=0) 36.323520 83.483859 7700 1872

multiarray (nonResolved 1 dimension,size=5) 36.409720 83.689359 7720 1872

multiarray (nonResolved 2 dimension,size=5) 53.042079 121.065519 11212 2580

multiarray (Object 1 dimensions,size=8) 36.461440 83.812659 7732 1872

multiarray (Object 2 dimensions,size=4) 49.866399 113.914719 10542 2450

multiarray (Object 4 dimensions,size=2) 84.130479 190.890879 17702 3934

arraylength 0xbe 0.983920 2.297580 210 52

baload 0x33 1.066680 2.554380 228 60

caload 0x34 1.084380 2.610900 232 62

saload 0x35 1.075780 2.586900 230 61

iaload 0x2e 1.077460 2.588220 230 60

laload 0x2f 1.183560 2.846640 252 67

aaload 0x32 1.077460 2.588220 230 60

bastore 0x54 1.155620 2.748520 246 64

castore 0x55 1.164480 2.787420 248 65

sastore 0x56 1.164480 2.787420 248 65

iastore 0x4f 1.158240 2.774640 246 64

lastore 0x50 1.310000 3.123480 276 71

aastore 0x53 2.003000 4.674360 418 100

**Table 6:** Opcodes costs, object and arrays-Part2/2
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
anewarray (size=0 non resolved empty class) 0xbd	34.638980	79.455499	7388	1908
anewarray (size=1 non resolved empty class) 0xbd	34.656220	79.496599	7392	1908
anewarray (size=5 non resolved empty class) 0xbd	34.725180	79.660999	7408	1908
anewarray (size=0 resolved empty class)0xbd	32.7967	75.5584	6996	1803
anewarray (size=1 resolved empty class)0xbd	32.8139	75.5995	7000	1803
anewarray (size=5 resolved empty class)0xbd	32.8829	75.7639	7016	1803
anewarray (size=0 resolved java.lang.Object class)	27.806520	64.268619	5936	1502
anewarray (size=1 resolved java.lang.Object class)	27.823760	64.309719	5940	1502
anewarray (size=5 resolved java.lang.Object class)	27.892720	64.474119	5956	1502
anewarray (size=0 resolved java.lang.String class)	27.754860	64.073619	5928	1502
anewarray (size=1 resolved java.lang.String class)	27.772100	64.114719	5932	1502
anewarray (size=5 resolved java.lang.String class)	27.841060	64.279119	5948	1502
multiarray (int 1 dimension,size=0/dim.)0xc5	36.323520	83.483859	7700	1872
multiarray (int 2 dimensions,size=4/dim.)0xc5	49.866399	113.914719	10542	2450
multiarray (int 4 dimension,size=2)	84.130479	190.890879	17702	3934
multiarray (nonResolved 1 dimension,size=0)	36.323520	83.483859	7700	1872
multiarray (nonResolved 1 dimension,size=5)	36.409720	83.689359	7720	1872
multiarray (nonResolved 2 dimension,size=5)	53.042079	121.065519	11212	2580
multiarray (Object 1 dimensions,size=8)	36.461440	83.812659	7732	1872
multiarray (Object 2 dimensions,size=4)	49.866399	113.914719	10542	2450
multiarray (Object 4 dimensions,size=2)	84.130479	190.890879	17702	3934
arraylength 0xbe	0.983920	2.297580	210	52
baload 0x33	1.066680	2.554380	228	60
caload 0x34	1.084380	2.610900	232	62
saload 0x35	1.075780	2.586900	230	61
iaload 0x2e	1.077460	2.588220	230	60
laload 0x2f	1.183560	2.846640	252	67
aaload 0x32	1.077460	2.588220	230	60
bastore 0x54	1.155620	2.748520	246	64
castore 0x55	1.164480	2.787420	248	65
sastore 0x56	1.164480	2.787420	248	65
iastore 0x4f	1.158240	2.774640	246	64
lastore 0x50	1.310000	3.123480	276	71
aastore 0x53	2.003000	4.674360	418	100

Table 7: Opcodes costs, method invocation and return
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

invokevirtual+return(empty method)0xb6 12.032760 27.984120 2504 520

invokevirtual+ireturn(empty method) 12.194540 28.380960 2536 529

invokevirtual+lreturn(empty method) 12.363720 28.753800 2570 537

invokevirtual+areturn(empty method,return this)0xb6 12.159580 28.284960 2528 525

invokestatic+return(empty method)0xb8 10.549340 24.648600 2198 455

invokestatic+ireturn(empty method) 10.711120 25.045440 2230 464

invokestatic+lreturn(empty method) 10.880300 25.418280 2264 472

invokespecial+return(empty method)0xb7 10.512680 24.545700 2188 450

invokespecial+ireturn(empty method) 10.674460 24.942540 2220 459

invokespecial+lreturn(empty method) 10.843640 25.315380 2254 467

invokespecial+areturn(empty method,return this) 10.639500 24.846540 2212 455

**Table 7:** Opcodes costs, method invocation and return
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
invokevirtual+return(empty method)0xb6	12.032760	27.984120	2504	520
invokevirtual+ireturn(empty method)	12.194540	28.380960	2536	529
invokevirtual+lreturn(empty method)	12.363720	28.753800	2570	537
invokevirtual+areturn(empty method,return this)0xb6	12.159580	28.284960	2528	525
invokestatic+return(empty method)0xb8	10.549340	24.648600	2198	455
invokestatic+ireturn(empty method)	10.711120	25.045440	2230	464
invokestatic+lreturn(empty method)	10.880300	25.418280	2264	472
invokespecial+return(empty method)0xb7	10.512680	24.545700	2188	450
invokespecial+ireturn(empty method)	10.674460	24.942540	2220	459
invokespecial+lreturn(empty method)	10.843640	25.315380	2254	467
invokespecial+areturn(empty method,return this)	10.639500	24.846540	2212	455

Table 8: Opcodes costs, control flow
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

ifeq 0x99 (branch not taken) 0.957140 2.255160 206 54

ifeq 0x99 (branch taken) 1.142280 2.617860 246 62

ifne 0x9a (branch not taken) 0.957140 2.255160 206 54

ifne 0x9a (branch taken) 1.142280 2.617860 246 62

iflt 0x9b (branch not taken) 0.957140 2.255160 206 54

iflt 0x9b (branch taken) 1.142280 2.617860 246 62

ifle 0x9e (branch not taken) 0.957140 2.255160 206 54

ifle 0x9e (branch taken) 1.142280 2.617860 246 62

ifgt 0x9d (branch not taken) 0.957140 2.255160 206 54

ifgt 0x9d (branch taken) 1.142280 2.617860 246 62

ifge 0x9c (branch not taken) 0.957140 2.255160 206 54

ifge 0x9c (branch taken) 1.142280 2.617860 246 62

if_icmpeq 0x9f (branch not taken) 0.995660 2.352480 214 56

if_icmpeq 0x9f (branch taken) 1.180800 2.715180 254 64

if_icmpne 0xa0 (branch not taken) 0.995660 2.352480 214 56

if_icmpne 0xa0 (branch taken) 1.180800 2.715180 254 64

if_icmplt 0xa1 (branch not taken) 0.995660 2.352480 214 56

if_icmplt 0xa1 (branch taken) 1.180800 2.715180 254 64

if_icmple 0xa4 (branch not taken) 0.995660 2.352480 214 56

if_icmple 0xa4 (branch taken) 1.180800 2.715180 254 64

if_icmpgt 0xa3 (branch not taken) 0.995660 2.352480 214 56

if_icmpgt 0xa3 (branch taken) 1.180800 2.715180 254 64

if_icmpge 0xa2 (branch not taken) 0.995660 2.352480 214 56

if_icmpge 0xa2 (branch taken) 1.180800 2.715180 254 64

lcmp 0x94 (value1<value2) 1.483220 3.404280 310 72

lcmp 0x94 (value1==value2) 1.693000 3.845520 356 79

lcmp 0x94 (value1>value2) 1.693000 3.845520 356 79

ifnull 0xc6(branch not taken) 0.957140 2.255160 206 54

ifnull 0xc6 (branch taken) 1.124720 2.593860 242 61

ifnonnull 0xc7 (branch not taken) 0.957140 2.255160 206 54

ifnonnull 0xc7 (branch taken) 1.124720 2.593860 242 61

if_acmpeq 0xa5 (branch not taken) 0.995660 2.352480 214 56

if_acmpeq 0xa5 (branch taken) 1.180800 2.715180 254 64

if_acmpne 0xa6 (branch not taken) 0.995660 2.352480 214 56

if_acmpne 0xa6 (branch taken) 1.180800 2.715180 254 64

goto 0xa7 1.034380 2.400540 222 55

goto_w 0xc8 1.108140 2.551500 238 60

lookupswitch 0xab (1 iteration) 2.077300 4.759140 438 116

lookupswitch 0xab (2 iterations) 2.721600 6.228240 572 151

lookupswitch 0xab (3 iterations) 3.365900 7.697340 706 186

lookupswitch 0xab (4 iterations) 4.010200 9.166440 840 221

tableswitch 0xaa 1.793420 4.205460 384 108

**Table 8:** Opcodes costs, control flow
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
ifeq 0x99 (branch not taken)	0.957140	2.255160	206	54
ifeq 0x99 (branch taken)	1.142280	2.617860	246	62
ifne 0x9a (branch not taken)	0.957140	2.255160	206	54
ifne 0x9a (branch taken)	1.142280	2.617860	246	62
iflt 0x9b (branch not taken)	0.957140	2.255160	206	54
iflt 0x9b (branch taken)	1.142280	2.617860	246	62
ifle 0x9e (branch not taken)	0.957140	2.255160	206	54
ifle 0x9e (branch taken)	1.142280	2.617860	246	62
ifgt 0x9d (branch not taken)	0.957140	2.255160	206	54
ifgt 0x9d (branch taken)	1.142280	2.617860	246	62
ifge 0x9c (branch not taken)	0.957140	2.255160	206	54
ifge 0x9c (branch taken)	1.142280	2.617860	246	62
if_icmpeq 0x9f (branch not taken)	0.995660	2.352480	214	56
if_icmpeq 0x9f (branch taken)	1.180800	2.715180	254	64
if_icmpne 0xa0 (branch not taken)	0.995660	2.352480	214	56
if_icmpne 0xa0 (branch taken)	1.180800	2.715180	254	64
if_icmplt 0xa1 (branch not taken)	0.995660	2.352480	214	56
if_icmplt 0xa1 (branch taken)	1.180800	2.715180	254	64
if_icmple 0xa4 (branch not taken)	0.995660	2.352480	214	56
if_icmple 0xa4 (branch taken)	1.180800	2.715180	254	64
if_icmpgt 0xa3 (branch not taken)	0.995660	2.352480	214	56
if_icmpgt 0xa3 (branch taken)	1.180800	2.715180	254	64
if_icmpge 0xa2 (branch not taken)	0.995660	2.352480	214	56
if_icmpge 0xa2 (branch taken)	1.180800	2.715180	254	64
lcmp 0x94 (value1<value2)	1.483220	3.404280	310	72
lcmp 0x94 (value1==value2)	1.693000	3.845520	356	79
lcmp 0x94 (value1>value2)	1.693000	3.845520	356	79
ifnull 0xc6(branch not taken)	0.957140	2.255160	206	54
ifnull 0xc6 (branch taken)	1.124720	2.593860	242	61
ifnonnull 0xc7 (branch not taken)	0.957140	2.255160	206	54
ifnonnull 0xc7 (branch taken)	1.124720	2.593860	242	61
if_acmpeq 0xa5 (branch not taken)	0.995660	2.352480	214	56
if_acmpeq 0xa5 (branch taken)	1.180800	2.715180	254	64
if_acmpne 0xa6 (branch not taken)	0.995660	2.352480	214	56
if_acmpne 0xa6 (branch taken)	1.180800	2.715180	254	64
goto 0xa7	1.034380	2.400540	222	55
goto_w 0xc8	1.108140	2.551500	238	60
lookupswitch 0xab (1 iteration)	2.077300	4.759140	438	116
lookupswitch 0xab (2 iterations)	2.721600	6.228240	572	151
lookupswitch 0xab (3 iterations)	3.365900	7.697340	706	186
lookupswitch 0xab (4 iterations)	4.010200	9.166440	840	221
tableswitch 0xaa	1.793420	4.205460	384	108

See Benchmarks

Table 9: Opcodes costs, logic opcodes
Opcode
Inst. Cost in $\mathrm{\mu J}$ Mem. Cost in $\mathrm{\mu J}$ Nb Cycles Nb Proc. Inst.

ishl 0x78 0.976480 2.321580 208 53

ishr 0x7a 0.976360 2.321580 208 53

iushr 0x7c 0.976420 2.321580 208 53

lshl 0x79 1.419040 3.316500 300 72

lshr 0x7b 1.419040 3.316500 300 72

lushr 0x7d 1.545040 3.593340 324 76

iand 0x7e 0.957800 2.273580 204 51

ior 0x80 0.958100 2.273580 204 51

ixor 0x82 0.958120 2.273580 204 51

land 0x7f 1.127820 2.701320 240 63

lor 0x81 1.128420 2.701320 240 63

lxor 0x83 1.128460 2.701320 240 63

**Table 9:** Opcodes costs, logic opcodes
Opcode	Inst. Cost in $\mathrm{\mu J}$	Mem. Cost in $\mathrm{\mu J}$	Nb Cycles	Nb Proc. Inst.
ishl 0x78	0.976480	2.321580	208	53
ishr 0x7a	0.976360	2.321580	208	53
iushr 0x7c	0.976420	2.321580	208	53
lshl 0x79	1.419040	3.316500	300	72
lshr 0x7b	1.419040	3.316500	300	72
lushr 0x7d	1.545040	3.593340	324	76
iand 0x7e	0.957800	2.273580	204	51
ior 0x80	0.958100	2.273580	204	51
ixor 0x82	0.958120	2.273580	204	51
land 0x7f	1.127820	2.701320	240	63
lor 0x81	1.128420	2.701320	240	63
lxor 0x83	1.128460	2.701320	240	63

About this document ...

This document was generated using the LaTeX2HTML translator Version 2002 (1.62)

The command line arguments were:
latex2html -no_subdir -split 0 -show_section_numbers /tmp/lyx_tmpdir21086cAmhPK/lyx_tmpbuf0/tables.tex

The translation was initiated by Sebastien Lafond TUCS on 2005-09-27

Sebastien Lafond TUCS 2005-09-27

Free counter