Hi, Am I correct in my understanding of the OpenCL spec that an out-of-order queue can be emulated by creating a new queue for each kernel? Obviously, it is up to the underlying implementation as to whether it executes them out of order but it is certainly allowable to execute kernels from two different queues simultaneously and/or out-of-order?
Tangentially related, I couldn't find any mailing lists about OpenCL? Is stackoverflow the best place for discussions/questions about OpenCL (i.e. not pyopencl specific) - any other recommended forums? Regards, Jonny PS pyopencl is great - thanks for creating it! On Mon, Oct 29, 2012 at 7:11 PM, Brendan Wood <[email protected]> wrote: > Hi Andreas, > > I just tried it on my Nvidia GPU (driver 304.60, CUDA 4.2 toolkit, > compute capability 2.1) and it ran fine without errors. I did not > verify that out-of-order execution actually happened, but it did run. > > Cheers, > Brendan Wood > > > On 10/29/2012 10:27 PM, Andreas Kloeckner wrote: >> Hi Jonny, >> >> Jonathan Hunt <[email protected]> writes: >>> I apologise if I'm doing something dumb - I've tried googling to no avail. >>> >>> I am having trouble creating an out-of-order queue with pyopencl >>> (2012.2 / from git as of today). >>> >>> queue = cl.CommandQueue(ctx, >>> properties=cl.command_queue_properties.OUT_OF_ORDER_EXEC_MODE_ENABLE) >>> >>> raises LogicError: CommandQueue failed: invalid value >>> >>> It works fine for an ordered queue (properties = 0). >>> >>> This is on OS X 10.8 with Apple drivers and devices: >>> [<pyopencl.Device 'Intel(R) Core(TM) i7-3720QM CPU @ 2.60GHz' on >>> 'Apple' at 0xffffffff>, >>> <pyopencl.Device 'GeForce GT 650M' on 'Apple' at 0x1022600>] >>> (it occurs on both devices). >>> >>> Am I doing something wrong? >> >> AFAIK, the Intel CPU implementation is the only one to support >> out-of-order mode for now. I might be wrong, though. I'd appreciate any >> dissenting opinions. >> >> Andreas >> >> _______________________________________________ >> PyOpenCL mailing list >> [email protected] >> http://lists.tiker.net/listinfo/pyopencl >> > > _______________________________________________ > PyOpenCL mailing list > [email protected] > http://lists.tiker.net/listinfo/pyopencl _______________________________________________ PyOpenCL mailing list [email protected] http://lists.tiker.net/listinfo/pyopencl
