Article: Parallelism at HotPar 2010
By: Richard Cownie (tich.delete@this.pobox.com), August 2, 2010 1:48 pm
Room: Moderated Discussions
mpx (mpx@nomail.pl) on 8/2/10 wrote:
---------------------------
>That's why OpenCL is important. You can create a program in >OpenCL and run it on either GPU or CPU.
While OpenCL will allow you to write an app which will
run on various different hardware configurations with
different numbers of cores/threads/gpu resources, I very
much doubt that it's going to give you good performance
on GPU hardware without a fair amount of software effort
dedicated to tuning for the GPU.
And any such software effort would give a benefit only
for the subset of users who have the appropriate GPU
hardware - which gets you back to the original problem.
What *might* break this cycle is the rise of powerful
integrated GPUs in Core i5, SandyBridge, and Ontario/
Llano. It's much more interesting to optimize your
app for hardware that 80% of users will have, than to
optimize for hardware that only 40% will have. But
that only applies if the integrated GPUs really do go
faster than the CPUs for a wide-enough range of apps.
The next couple of years will be interesting.
---------------------------
>That's why OpenCL is important. You can create a program in >OpenCL and run it on either GPU or CPU.
While OpenCL will allow you to write an app which will
run on various different hardware configurations with
different numbers of cores/threads/gpu resources, I very
much doubt that it's going to give you good performance
on GPU hardware without a fair amount of software effort
dedicated to tuning for the GPU.
And any such software effort would give a benefit only
for the subset of users who have the appropriate GPU
hardware - which gets you back to the original problem.
What *might* break this cycle is the rise of powerful
integrated GPUs in Core i5, SandyBridge, and Ontario/
Llano. It's much more interesting to optimize your
app for hardware that 80% of users will have, than to
optimize for hardware that only 40% will have. But
that only applies if the integrated GPUs really do go
faster than the CPUs for a wide-enough range of apps.
The next couple of years will be interesting.