Robert myers 2008-01-26 14:37:03
Think so? How long has multi-threading been an issue? I’m still
reading discussions from people who I think really ought to have it
all figured out by now that make me think they haven’t.
Hand checking the correctness of multi-threaded code is not something
I want to spend my time doing, and it’s not something I’d have a high
confidence I’d get right, anyway.
The problem is that there are people who can do it in c, use whatever
lame threading tools are provided, and get it right often enough that
there is not a general meltdown.
That is to say I agree that most programmers barely understand
multithreading, and I am not optimistic that programmers who are
reliably good at it will become common enough that we will see a
smooth transition to multi-threaded applications.
*Maybe* as it becomes evident just how much programmer time is being
soaked up by debugging badly-written code we will see a movement to better tools.
The good is the enemy of the best? In this case, I’d say the best is
the enemy of the good. The fact that somebody can do it and get it
right limits the likelihood that Cell will become useful as a general
purpose compute engine, something I’d very much like to see.
Why all the cross-posts to intel groups?