On Mon, Jan 28, 2013 at 1:52 PM, Daniel Stenberg <daniel_at_haxx.se> wrote:

> On Fri, 25 Jan 2013, Mark Seger wrote:
>
> oops, I have to fall on my sword and say I accidentally profiled my
>> non-pycurl upload script and it was the one that spent all that time in
>> compression, but only 17% of it's time in crypto. When I uploaded a file
>> with pycurl it didn't compress at all (as I originally suspected) but this
>> time spend 2/3 of it's time in crypto. Either way, both cases are spending
>> a LOT of time in CPU and that just doesn't feel right to me. -mark
>
>
correct, I was just trying to understand why other tools seemed to be using
less CPU, but maybe they really weren't.
on a more positive note I've finally solved the 1-sec gap problem. the
secret was to use http v1 instead of the default. I'm now a very happy
camper...
-mark

> Why not? Sending encrypted data over a very high-bandwidth connection
> should be a lot of CPU activity.
>
> --
>
> / daniel.haxx.se
> ______________________________**_________________
> http://cool.haxx.se/cgi-bin/**mailman/listinfo/curl-and-**python<http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-python>
>

_______________________________________________
http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-python
Received on 2013-01-28