[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [linrad] RE: Kernel 2.6.10.x, multithreading
I took one look at the dance Leif does inside this code to keep track
of all the different processes on his "chain" of algorithms and the
way he managed to keep all of these chains fed with what they need.
I decided that all I could do was provide a framework as it was truly
daunting. If I went to Sweden for a week, he could probably guide
me all the way through it, but I felt this was the better way since
it leaves Leif in control of "quality assurance" until it is broken
apart into threads.
Thanks to Frank Brickle, AB2KT, for teaching me all of this stuff to
begin with. Until I began to work with Frank, I wrote large monolithic
code and used flags to do everything. Frankly, it makes the code
impenetrable. Now I write things in small functions, allowing the
good compilers to do their optimization tricks, inline critical
and hot functions, etc. Aim some thanks for what will come at AB2KT.
After Leif breaks this out into pieces that are threads or modules on
their own, semaphore/mutex/barriered, as Leif says, it will be much
easier to get involved and to smelt this ore to get the 24K gold that
is in it.
[mailto:owner-linrad@xxxxxxxxxxxxxxxxxxxxxx]On Behalf Of leif@xxxxxxxxxx
Sent: Tuesday, February 08, 2005 2:52 AM
Subject: [linrad] RE: Kernel 2.6.10.x, multithreading
> CHANGE OF TOPIC:
> I am very eager to try the new multi-threaded Linrad when it is ready.
> Thanks VERY much for moving forward with this, Bob and Leif!
There should be no difference at all on a Pentium IV. The multi-threading
will just make the same code much easier to work with for others, but
the results should not change at all.
Later on I hope a better structure will bring others into the project
as well as make it easier for others to pick algorithms for GNU-radio
and warious other SDR projects. Some day we might even have a Windows
version of Linrad:-)