XO-1.75 relative performance

[Bert Freudenberg]

On 02.11.2011, at 13:59, Sascha Silbe wrote:

> Excerpts from C. Scott Ananian's message of 2011-11-02 05:58:42 +0100:
> 
>> A pure-CPU benchmark (maybe something in Pippy?) would be a little
>> more reliable.
> 
> === Begin facspeed.py ===
> #!/usr/bin/env python
> import time
> 
> def factorial(n):
>    result = 1
>    while n > 1:
>        result *= n
>        n -= 1
>    return result
> 
> n = 100000
> start_time = time.time()
> factorial(n)
> time_diff = time.time() - start_time
> 
> print '%d! computed in %.3fs' % (n, time_diff)
> === End facspeed.py ===
> 
> 
> System		SoC/CPU			OS + arch	time
> XO-1.75		Armada 610 @ 0.8GHz	Debian armel	199.108s
> XO-1.5		VIA C7-M @ 1GHz		Debian i386	199.285s
> OpenRD		88F6281 @ 1.2GHz	Debian armel	181.277s
> Desktop PC	Athlon BE-2300 @ 1.9GHz	Debian amd64	 34.180s
> 
> 
> I must admit I'm surprised by the result. sup (my MUA of choice) feels
> much slower on XO-1.75 than on XO-1.5 - and that's even though dstat
> reports higher SD card write throughput (with a different card). Maybe I
> should do some side-by-side comparisons.


According to "0 tinyBenchmarks" in Squeak Etoys (which estimates bytecodes/sec and message sends/sec), the 1.75 is about 10% faster than the 1.5. Both were running os883.

- Bert -