Squeak G5 Benchmarks

Aaron J Reichow reic0024 at d.umn.edu
Mon Sep 15 19:53:01 UTC 2003


Hey Craig-

> 	It seems like very important constraints, namely memory
> bandwidth/capacity, are not taken into account by your data. And
> wouldn't be 256MB be somewhat spartan for the 1+ GHz machines you cited?
> If you're going to venture off into provocative waters like
> "sends/s/MHz", how about "sends/s/MHz/MB/Mbps" or somesuch? (Yes, it
> gets exponentially provocative very quickly ;)
>
> 	The column I most enjoy watching is just sends/s. :)

Yes, it would be interesting to get into more details.  If you're willing
to provide me with the code (if there is any needed), I'm more than
willing to run the tests.  How do I test sends/s/MHz/MB/Mbps?  If we're
going to do that, we may as well run some better benchmarks that do more
than math on the results of #tinyBenchmarks.  If anyone wants to see some
particular benchmarks, send me the code and I'll run the test.

In fact, this would be a good place to play with the Benchmarking
app/image I wrote up a while back.  Just start the image, click 'start'
and it runs a bunch of benchmarks and saves them to a file or FTP site.

Regards,
Aaron

--
  "civilization is a limitless multiplication of
                unnecessary necessities."                :: mark twain



More information about the Squeak-dev mailing list