On Sun, 3 Apr 2011 18:35:04 -0700, Ben Widawsky wrote:
> Here is the data from ~100 samples while playing playing Armacycles Advanced
> measured off of d-i-f 7f58aabc369014fda3a4a33604ba0a1b63b941ac.
>
> min 02.775us
> max 19.402us
> avg 07.057us
> stddev02.819us
>
> When I do a ca
On Sat, Apr 02, 2011 at 07:46:31AM +0100, Chris Wilson wrote:
> What I guess I was trying to express was that we need to be very clear
> what the interface is for and the limitations about its use.
>
> For the more complicated set of registers, we can and should expose knobs
> in the debugfs to r
On Sat, Apr 02, 2011 at 07:46:31AM +0100, Chris Wilson wrote:
>
> But perhaps we do need to reconsider the performance aspect. intel_gpu_top
> samples the ring HEAD and TAIL at around 10KHz and forcing gt-wake is
> about 50 microseconds... I hope I'm mistaken, because even batched that is
> doomed
On Fri, 01 Apr 2011 11:51:23 -0700, Eric Anholt wrote:
> On Fri, 01 Apr 2011 08:32:09 +0100, Chris Wilson
> wrote:
> > I'm just not happy about haphazard locking. Can we do simple and safe
> > locking and revisit it if a real use-case for brute-forcing the read/write
> > is found?
>
> The conce
On Fri, 01 Apr 2011 08:32:09 +0100, Chris Wilson
wrote:
> On Fri, 1 Apr 2011 00:06:37 -0700, Ben Widawsky wrote:
> > On Fri, Apr 01, 2011 at 07:36:56AM +0100, Chris Wilson wrote:
> > > Nice. The consensus is that this ioctl is required, just a few comments
> > > inline.
> >
> > Nobody likes it,
On Fri, 1 Apr 2011 00:06:37 -0700, Ben Widawsky wrote:
> On Fri, Apr 01, 2011 at 07:36:56AM +0100, Chris Wilson wrote:
> > Nice. The consensus is that this ioctl is required, just a few comments
> > inline.
>
> Nobody likes it, but it could pay off, especially if we end up with
> weird non-mappab
On Fri, Apr 01, 2011 at 07:36:56AM +0100, Chris Wilson wrote:
> Nice. The consensus is that this ioctl is required, just a few comments
> inline.
Nobody likes it, but it could pay off, especially if we end up with
weird non-mappable registers we wish to allow user space to read. I had
actually dec
Nice. The consensus is that this ioctl is required, just a few comments
inline.
On Thu, 31 Mar 2011 18:31:49 -0700, Ben Widawsky wrote:
> With the invention of forcewake (DevGT) , userspace tools to diagnose
> problems are no longer reliable. This will provide userspace a mechanism
> to read regi
With the invention of forcewake (DevGT) , userspace tools to diagnose
problems are no longer reliable. This will provide userspace a mechanism
to read registers through an IOCTL, as well as root permission to write
to registers.
The code tries to be smart about which registers can be read and
writ
