r600g: Trying to remove the busywait for fence completion, but hitting inexplicable behaviour

2012-01-31 Thread Simon Farnsworth
On Tuesday 31 January 2012, Alan Swanson wrote: > On Tue, 2012-01-31 at 13:16 +, Simon Farnsworth wrote: > > Hello, > > > > When profiling my workload on an AMD E-350 (PALM GPU) to see why it still > > wasn't performing well with Jerome's WIP macrotiling patches, I noticed that > >

r600g: Trying to remove the busywait for fence completion, but hitting inexplicable behaviour

2012-01-31 Thread Alan Swanson
On Tue, 2012-01-31 at 13:16 +, Simon Farnsworth wrote: > Hello, > > When profiling my workload on an AMD E-350 (PALM GPU) to see why it still > wasn't performing well with Jerome's WIP macrotiling patches, I noticed that > r600_fence_finish was taking 10% of my CPU time. I determined

r600g: Trying to remove the busywait for fence completion, but hitting inexplicable behaviour

2012-01-31 Thread Simon Farnsworth
Hello, When profiling my workload on an AMD E-350 (PALM GPU) to see why it still wasn't performing well with Jerome's WIP macrotiling patches, I noticed that r600_fence_finish was taking 10% of my CPU time. I determined experimentally that changing from sched_yield() to os_time_sleep(10) fixed

r600g: Trying to remove the busywait for fence completion, but hitting inexplicable behaviour

2012-01-31 Thread Simon Farnsworth
Hello, When profiling my workload on an AMD E-350 (PALM GPU) to see why it still wasn't performing well with Jerome's WIP macrotiling patches, I noticed that r600_fence_finish was taking 10% of my CPU time. I determined experimentally that changing from sched_yield() to os_time_sleep(10) fixed

Re: r600g: Trying to remove the busywait for fence completion, but hitting inexplicable behaviour

2012-01-31 Thread Simon Farnsworth
On Tuesday 31 January 2012, Alan Swanson swan...@ukfsn.org wrote: On Tue, 2012-01-31 at 13:16 +, Simon Farnsworth wrote: Hello, When profiling my workload on an AMD E-350 (PALM GPU) to see why it still wasn't performing well with Jerome's WIP macrotiling patches, I noticed that

Re: r600g: Trying to remove the busywait for fence completion, but hitting inexplicable behaviour

2012-01-31 Thread Alan Swanson
On Tue, 2012-01-31 at 13:16 +, Simon Farnsworth wrote: Hello, When profiling my workload on an AMD E-350 (PALM GPU) to see why it still wasn't performing well with Jerome's WIP macrotiling patches, I noticed that r600_fence_finish was taking 10% of my CPU time. I determined