Actions
Forum Thread Details
  • Replies 88 replies
  • Subscribers 63 subscribers
  • Views 8542 views
  • Users 0 members are here
Related

SBC CPU Throughput

morgaine

I notice that people are doing some initial benchmarking of BBB and other boards on the RPF forum.  Results roughly as expected I guess:

 

Using just a simple

 

time echo "scale=2000;4*a(1)" | bc -l

 

as a lightweight benchmark, I see these numbers reported (smaller Time is better):

 

[table now updated with extra datapoints reported in current thread below]

 

Submitter
Time (s)
Board
SoC
Clock (MHz)
O/S
shuckle26.488Raspberry Pi BBCM2835700Raspbian 3.1.9
morgaine25.719Raspberry Pi BBCM2835700Raspbian 3.1.9+ #272
shuckle25.009Raspberry Pi BBCM2835700Raspbian 3.2.27
trn24.280Raspberry Pi BBCM2835700Raspbian ?
morgaine22.456Raspberry Pi BBCM2835800Raspbian 3.1.9+ #272
morgaine21.256Raspberry Pi BBCM2835800Raspbian 3.6.11+ #545, new firmware only
selsinork21.0MinnowboardAtom E640T1000Angstrom minnow-2013.07.10.img
shuckle17.0Raspberry Pi BBCM28351000Raspbian ?
morgaine16.153BB (white)AM3359720Angstrom v2012.01-core 3.2.5+, user-gov
selsinork15.850A20-OLinuXino-MICROA20912Debian 7.0, 3.4.67+
selsinork15.328CubieboardA20912Ubuntu/Debian 7.1
pluggy14.510BBBAM33591000Debian
morgaine14.153BBBAM33591000Debian 7.0, 3.8.13-bone20, perf-gov
selsinork13.927A10-OLinuXino-LIMEA101000Debian 7.0, 3.4.67+
Heydt13.159CubieboardA101000?
selsinork12.8Sabre-litei.MX61000Debian armhf
selsinork12.752CubieboardA20912Ubuntu/Debian 7.1 + Angstrom bc
selsinork12.090BBBAM33591000Angstrom dmnd-gov
pluggy11.923BBBAM33591000Angstrom
selsinork11.86BBBAM33591000Angstrom perf-gov
selsinork9.7Sabre-litei.MX61000Debian armhf + Angstrom bc
selsinork9.606Sabre-litei.MX61000LFS 3.12, gcc-4.8.2, glibc-2.18

 

 

As usual, take benchmarks with a truckload of salt, and evaluate with a suitable mixture of suspicion, snoring, and mirth. Use the numbers wisely, and don't draw inappropriate conclusions. image

 

Morgaine.

Parents

  • While our tests here have been limited to a single very specific and easy to run test, they only show one aspect.

     

    For the Allwinner devices, there's some results at http://sunxi.org/Benchmarks It's interesting to compare the openssl speed numbers between A10 & A20 and note that the A10 is significantly faster (and I've confirmed similar numbers between the A10 lime and the A20 Micro).

    Then compare the A10 & A20 Linpack where the A20 seems significantly faster. Of course the comparison isn't quite fair as, from the compile options, they appear to be comparing NEON on the A10 against VFPv4 on the A20, the A10 doesn't have VFPv4.

     

    However, having tried identical linpack binaries on both A10 & A20 I'm getting results that suggest the A20 is approx 3x faster for these floating point operations and it doesn't seem to matter if I use VFPv3 or NEON on both, the A20 still outperforms the A10.

     

    The Cortex-A9 based i.MX6 still beats both A10 & A20, but the A20 is surprisingly close, approx 120000 KFLOPS for the A20 compared to approx 150000 KFLOPS for the i.MX6 when using VFPv3 or NEON, the A20 manages approx 145000 KFLOPS with VFPv4. Some of this will be down to raw clock speed difference, 912MHz for the A20, 996MHz for the iMX6.

    So it seems that in order to gain feature parity with the Cortex-A15, the Cortex-A7 has been gifted the newer floating point unit in it's entirety. Bonus if you're doing floating point stuff on the A20..

Reply

  • While our tests here have been limited to a single very specific and easy to run test, they only show one aspect.

     

    For the Allwinner devices, there's some results at http://sunxi.org/Benchmarks It's interesting to compare the openssl speed numbers between A10 & A20 and note that the A10 is significantly faster (and I've confirmed similar numbers between the A10 lime and the A20 Micro).

    Then compare the A10 & A20 Linpack where the A20 seems significantly faster. Of course the comparison isn't quite fair as, from the compile options, they appear to be comparing NEON on the A10 against VFPv4 on the A20, the A10 doesn't have VFPv4.

     

    However, having tried identical linpack binaries on both A10 & A20 I'm getting results that suggest the A20 is approx 3x faster for these floating point operations and it doesn't seem to matter if I use VFPv3 or NEON on both, the A20 still outperforms the A10.

     

    The Cortex-A9 based i.MX6 still beats both A10 & A20, but the A20 is surprisingly close, approx 120000 KFLOPS for the A20 compared to approx 150000 KFLOPS for the i.MX6 when using VFPv3 or NEON, the A20 manages approx 145000 KFLOPS with VFPv4. Some of this will be down to raw clock speed difference, 912MHz for the A20, 996MHz for the iMX6.

    So it seems that in order to gain feature parity with the Cortex-A15, the Cortex-A7 has been gifted the newer floating point unit in it's entirety. Bonus if you're doing floating point stuff on the A20..

Children
No Data