<div dir="ltr"><span style="font-size:12.8000001907349px">> Why keeping the assertion commented out?</span><br><div><span style="font-size:12.8000001907349px"><br></span></div><div><span style="font-size:12.8000001907349px">Because to test the image support functionality I'm working on, I have to create an image and pass it to a kernel, which triggers the assertion ATM.</span></div><div><br></div><div><span style="font-size:12.8000001907349px">> </span><span style="font-size:12.8000001907349px">Can you just fix evergreen_set_cs_sampler_view so that it supports views==NULL?</span></div><div><span style="font-size:12.8000001907349px"><br></span></div><div><span style="font-size:12.8000001907349px">Fair point, will do that.</span></div><div><span style="font-size:12.8000001907349px"><br></span></div><div><span style="font-size:12.8000001907349px">> </span><span style="font-size:12.8000001907349px">This will enable the feature for R600, R700, Evegreen, NI, SI, and</span></div><span style="font-size:12.8000001907349px">> CIK, because this function is shared by all of them. I don't think you</span><br style="font-size:12.8000001907349px"><span style="font-size:12.8000001907349px">> intended that.</span><div><span style="font-size:12.8000001907349px"><br></span></div><div><span style="font-size:12.8000001907349px">You are correct, I'll fix this.</span></div><div><span style="font-size:12.8000001907349px"><br></span></div><div><span style="font-size:12.8000001907349px">> </span><span style="font-size:12.8000001907349px">Anyway, I don't what Tom's policy on OpenCL is, but generally we don't</span><br style="font-size:12.8000001907349px"><span style="font-size:12.8000001907349px">> push incomplete features.</span><span style="font-size:12.8000001907349px"><br></span><div><span style="font-size:12.8000001907349px"><br></span></div></div><div><span style="font-size:12.8000001907349px">Fair enough, thanks for the feedback anyway.</span></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Jun 8, 2015 at 1:33 PM, Marek Olšák <span dir="ltr"><<a href="mailto:maraeo@gmail.com" target="_blank">maraeo@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Mon, Jun 8, 2015 at 1:20 PM, Zoltan Gilian <<a href="mailto:zoltan.gilian@gmail.com">zoltan.gilian@gmail.com</a>> wrote:<br> > Allow creation and binding of compute image resources to be able to test the<br> > currently working pieces of OpenCL image support.<br> > ---<br> > src/gallium/drivers/r600/evergreen_compute.c | 2 +-<br> > src/gallium/drivers/r600/evergreen_state.c | 4 ++++<br> > src/gallium/drivers/r600/r600_state_common.c | 1 +<br> > src/gallium/drivers/radeon/r600_pipe_common.c | 2 +-<br> > 4 files changed, 7 insertions(+), 2 deletions(-)<br> ><br> > diff --git a/src/gallium/drivers/r600/evergreen_compute.c b/src/gallium/drivers/r600/evergreen_compute.c<br> > index 4c3c34c..ecad25c 100644<br> > --- a/src/gallium/drivers/r600/evergreen_compute.c<br> > +++ b/src/gallium/drivers/r600/evergreen_compute.c<br> > @@ -666,7 +666,7 @@ void evergreen_set_cs_sampler_view(struct pipe_context *ctx_,<br> > if (resource[i]) {<br> > assert(i+1 < 12);<br> > /* XXX: Implement */<br> > - assert(!"Compute samplers not implemented.");<br> > + //assert(!"Compute samplers not implemented.");<br> <br> </span>Why keeping the assertion commented out?<br> <span class=""><br> > ///FETCH0 = VTX0 (param buffer),<br> > //FETCH1 = VTX1 (global buffer pool), FETCH2... = TEX<br> > }<br> > diff --git a/src/gallium/drivers/r600/evergreen_state.c b/src/gallium/drivers/r600/evergreen_state.c<br> > index 4ddbc0b..5dba577 100644<br> > --- a/src/gallium/drivers/r600/evergreen_state.c<br> > +++ b/src/gallium/drivers/r600/evergreen_state.c<br> > @@ -299,6 +299,10 @@ boolean evergreen_is_format_supported(struct pipe_screen *screen,<br> > if (usage & PIPE_BIND_TRANSFER_WRITE)<br> > retval |= PIPE_BIND_TRANSFER_WRITE;<br> ><br> > + // XXX: review if something is not supported as a compute resource<br> > + if (usage & PIPE_BIND_COMPUTE_RESOURCE)<br> > + retval |= PIPE_BIND_COMPUTE_RESOURCE;<br> > +<br> > return retval == usage;<br> > }<br> ><br> > diff --git a/src/gallium/drivers/r600/r600_state_common.c b/src/gallium/drivers/r600/r600_state_common.c<br> > index 13dc9ee..0566d00 100644<br> > --- a/src/gallium/drivers/r600/r600_state_common.c<br> > +++ b/src/gallium/drivers/r600/r600_state_common.c<br> > @@ -594,6 +594,7 @@ static void r600_set_sampler_views(struct pipe_context *pipe, unsigned shader,<br> > assert(start == 0); /* XXX fix below */<br> ><br> > if (shader == PIPE_SHADER_COMPUTE) {<br> > + /*XXX*/ if (!views) return;<br> <br> </span>Can you just fix evergreen_set_cs_sampler_view so that it supports views==NULL?<br> <span class=""><br> > evergreen_set_cs_sampler_view(pipe, start, count, views);<br> > return;<br> > }<br> > diff --git a/src/gallium/drivers/radeon/r600_pipe_common.c b/src/gallium/drivers/radeon/r600_pipe_common.c<br> > index 3def444..d14791e 100644<br> > --- a/src/gallium/drivers/radeon/r600_pipe_common.c<br> > +++ b/src/gallium/drivers/radeon/r600_pipe_common.c<br> > @@ -631,7 +631,7 @@ static int r600_get_compute_param(struct pipe_screen *screen,<br> > case PIPE_COMPUTE_CAP_IMAGES_SUPPORTED:<br> > if (ret) {<br> > uint32_t *images_supported = ret;<br> > - *images_supported = 0;<br> > + *images_supported = 1;<br> <br> </span>This will enable the feature for R600, R700, Evegreen, NI, SI, and<br> CIK, because this function is shared by all of them. I don't think you<br> intended that.<br> <br> Anyway, I don't what Tom's policy on OpenCL is, but generally we don't<br> push incomplete features. A tessellation is an example of that. We've<br> had patches for about a year, but I can't send them to this mailing<br> list, because they are still not good enough.<br> <span class="HOEnZb"><font color="#888888"><br> Marek<br> </font></span></blockquote></div><br></div>