[Info-vax] Accuweather new contract

[Craig A. Berry]

On 3/31/15 7:25 PM, johnson.eric at gmail.com wrote:
> On Tuesday, March 31, 2015 at 7:53:14 PM UTC-4, Craig A. Berry wrote:
>
>>     Charged CPU time:        0 00:00:09.93      Elapsed time:       0
>> 00:00:19.48
>
> So that was about 20 seconds? Ouch. :-(
>
>> While it was running, my open ssh session would not respond at all, so
>> it looks like it mighty be a handy way to block all other network
>> activity :-(.
>
> Yeah, it has that characteristic too. Interestingly enough, if you point the
> packet abuse at another VMS box, it is very likely it will be even slower.
> But if you point it at a Linux box (or nothing), it is likely that won't happen.
>
> I'll leave that as a puzzle for now. :-)


On a recent quad-core i7 MackBook Pro running Yosemite, I get:

$ time ./udp_test
Bytes sent:3194872

real	0m2.845s
user	0m0.093s
sys	0m2.750s

Somewhat odd that it did 3 million+ bytes instead of the 10 million 
requested.  Must've done some coalescing somewhere.

On an ancient Power Mac G5 running OS X 10.5.8 (Leopard) I see:

% time ./udp_test
Bytes sent:10000000
0.867u 12.905s 0:13.79 99.7%	0+0k 0+0io 0pf+0w

So for a 10- or 12-year-old somewhat high-end consumer desktop running
Apple's BSD variant it took 14 seconds and for a 10- or 12-year-old low
end HP Integrity server running VMS it took 20 seconds. At least for
these two rather old but equivalent vintage machines we're on the same
order of magnitude though the non-VMS box is somewhat under twice as fast.