Beagle Board Benchmarks.

I don't have a Beagle Board yet let's just say I have too many other
projects right now. But I was wonderiing has anybody done a benchmark
of a Beagle Board compared to an Atom based system?
Since you can run Linux on both it would seem like a good Apples to
Apples test case.

http://www.pengutronix.de/development/kernel/arm-benchmarks-20100702_en.html

Gary Thomas wrote:

I have done several nbench tests on different platforms for work.
http://www.tux.org/~mayer/linux/bmark.html
The test was between a 1.1GHz atom and the Beagleboard running at 600MHz
ATOM 1.1GHZ
Memory Index 5.62
Integer Index 4.74
Floating Point 5.02

Beagleboard 600MHZ
Memory Index 3.643
Integer Index 3.426
Floating Point 0.619

Not exactly a fair fight with the clock differences, but if you scale the performance to 1 clock cycle the beagleboard outperforms the atom in memory and integer indexes. I’m eagerly awaiting the XM to verify this.

If you need the results for the individual nbench tests I can post them.

Kevin

I ran those benchmarks on an xM and I a lot more than 2x improvements, so it looks like the kernel + rootfs they used it quite slow compared to angstrom.

regards,

Koen

They should have throw an AMD Hexacore in there, to be able to write “I would have thought that the newer CortexA8 would indeed be much faster, but the Cortex is still like a boatload times times slower than the hexacore, although at 5 times the clock rate and just a single core”

Vladimir Pantelic <vladoman@gmail.com> writes:

Gary Thomas wrote:

I don't have a Beagle Board yet let's just say I have too many other
projects right now. But I was wonderiing has anybody done a benchmark
of a Beagle Board compared to an Atom based system?
Since you can run Linux on both it would seem like a good Apples to
Apples test case.

http://www.pengutronix.de/development/kernel/arm-benchmarks-20100702_en.html

so we can conclude, BB sucks vs atom :slight_smile:

Those benchmarks look fishy:

- There should be no difference between OMAP3530 and OMAP3503 (same
   ARM core), yet they vary wildly in the tables.
- An ARM11 is reported as much faster than the OMAP3 in several tests.
   This is unreasonable.
- The OMAP3 is clocked at only 500MHz despite being capable of more.
- The Cortex-A8 is listed as ARMv6 which does not inspire confidence.

Whoever made those benchmarks has either no clue or a (not so) hidden agenda.

I don’t know about some special benchmarks, but it is enough to open any flash site at BB and at x86. You will understand what I mean :slight_smile:

2010/8/25 Lioric Z3 <lioriccaymans@live.com>

Maxim Podbereznyy <lisarden@gmail.com> [2010-08-26 11:46:31]:

I don't know about some special benchmarks, but it is enough to open any
flash site at BB and at x86. You will understand what I mean :slight_smile:

Hm, but this is interesting benchmark. It's like benchmarking NEON optimized
codec on x86, you know what I mean :slight_smile:

-- ynezz