Sharp Zaurus SL-5600 And SL-6000 Benchmark Overview

SL-5600 and SL-6000 Benchmark Data

     For those lucky people who have purchased the SL-6000 (and for those who are thinking of buying it), I have run a benchmark comparison between the SL-5600 and the SL-6000. The comparisons are between the two machines, but among several kernels: two on the SL-5600 and the standard SL-6000 kernel. The table below compares the standard SL-5600 kernel performance with the SL-5600 overclocked (with preemptif) patched kernel and the standard SL-6000 kernel.

     The benchmark that I used has been optimized (using GCC compile time paremeters) for the different ARM processors. The goal was to see which GCC optimizations worked best with which processors for different target operations. I did not compile the software (and I can't seem to find where I got them), but you can run your own tests with them: nbench-arm.tar.gz or nbench-arm.zip.

     The data was collected from the better of two runs on the SL-6000 and the better of two runs on each kernel for the SL-5600. The original data collected for each kernel can be found here: SL-5600 Standard kernel data, SL-5600 Overclocked kernel data and SL-6000 kernel data.

   nBench optimized for ARM version 3 (nbench-armv3).

TEST Iterations/sec.
SL-5600
(no overclock)
Iterations/sec.
SL-5600
(overclocked)
Iterations/sec.
SL-6000
NUMERIC SORT    75.01 141.57 120.16
STRING SORT 5.9102 8.181 6.8655
BITFIELD 1.9658e+07 2.3315e+07 1.9627e+07
FP EMULATION 5.9453 7.1799 6.0618
ASSIGNMENT 0.8597 0.98309 0.93879
IDEA 212.44 250.87 211.13
HUFFMAN 25.349 41.096 34.471

   nBench optimized for ARM version 4 (nbench-armv4).

TEST Iterations/sec.
SL-5600
(no overclock)
Iterations/sec.
SL-5600
(overclocked)
Iterations/sec.
SL-6000
NUMERIC SORT    73.421 140.88 123.44
STRING SORT 5.9195 8.1609 6.8327
BITFIELD 1.9673e+07 2.3343e+07 1.9627e+07
FP EMULATION 8.8374 10.857 9.1597
ASSIGNMENT 0.92285 1.0567 1.0104
IDEA 232.56 274.51 228.14
HUFFMAN 25.595 41.118 34.412

   nBench optimized for ARM version 5te (nbench-armv5te).

TEST Iterations/sec.
SL-5600
(no overclock)
Iterations/sec.
SL-5600
(overclocked)
Iterations/sec.
SL-6000
NUMERIC SORT    73.741 165.32 140.88
STRING SORT 5.9148 8.1739 6.9031
BITFIELD 2.8382e+07 3.3614e+07 2.8415e+07
FP EMULATION 9.4899 11.702 9.8658
ASSIGNMENT 0.92868 1.0604 1.019
IDEA 266.46 314.71 264.75
HUFFMAN 25.53 41.334 34.71

   nBench optimized for Xscale (nbench-cpu-xscale).

TEST Iterations/sec.
SL-5600
(no overclock)
Iterations/sec.
SL-5600
(overclocked)
Iterations/sec.
SL-6000
NUMERIC SORT    76.278 165.24 141.36
STRING SORT 5.9079 8.1544 6.8547
BITFIELD 2.8382e+07 3.377e+07 2.8461e+07
FP EMULATION 9.6 11.982 10.107
ASSIGNMENT 0.94056 1.078 1.0406
IDEA 269.96 318.73 268.61
HUFFMAN 27.686 43.617 36.443

   nBench optimized for Xscale (nbench-tune-xscale).

TEST Iterations/sec.
SL-5600
(no overclock)
Iterations/sec.
SL-5600
(overclocked)
Iterations/sec.
SL-6000
NUMERIC SORT    75.29 165.24 141.72
STRING SORT 5.9172 8.1546 6.9177
BITFIELD 2.8404e+07 3.3604e+07 2.837e+07
FP EMULATION 9.4205 11.706 9.8815
ASSIGNMENT 0.92765 1.0605 1.0183
IDEA 266.36 314.71 264.65
HUFFMAN 25.707 41.345 34.686


You are visitor number: