*whew*, I've made it through the entire series... On Tue, Jul 21, 2015 at 9:38 AM, Francisco Jerez <curroje...@riseup.net> wrote: > Another resend of the i965 compiler-related changes for > ARB_shader_image_load_store, reworked to make use of the SIMD lowering > infrastructure introduced in a previous series [1]. For a > self-contained branch in testable form see [2]. > > [1] http://lists.freedesktop.org/archives/mesa-dev/2015-July/089009.html > [2] http://cgit.freedesktop.org/~currojerez/mesa/log/?h=image-load-store-lower > > [PATCH 01/20] i965/fs: Define logical typed and untyped surface opcodes. > [PATCH 02/20] i965/fs: Hook up SIMD lowering to unroll surface instructions > of unsupported width. > [PATCH 03/20] i965/fs: Implement lowering of logical surface instructions. > [PATCH 04/20] i965/fs: Handle zero-size allocations in fs_builder::vgrf(). > [PATCH 05/20] i965/fs: Import surface message builder helper functions. > [PATCH 06/20] i965/fs: Import image access validity checks.
The above are all Reviewed-by: Jason Ekstrand <jason.ekstr...@intel.com> > [PATCH 07/20] i965/fs: Import image memory offset calculation code. As I said in the e-mail, this needs a *lot* more comments. I'll re-try the review once you've actually explained what it's doing. > [PATCH 08/20] i965/fs: Import image format metadata queries. Reviewed-by: Jason Ekstrand <jason.ekstr...@intel.com> > [PATCH 09/20] i965/fs: Import image format conversion primitives. I had some minor concerns here. > [PATCH 10/20] i965/fs: Implement image load, store and atomic. This looks pretty good. However, I had some concerns about things getting out-of-sync with state setup that I'd like to see addressed. > [PATCH 11/20] i965/fs: Revisit NIR atomic counter intrinsic translation. > [PATCH 12/20] i965/fs: Drop unused untyped surface read and atomic emit > methods. Reviewed-by: Jason Ekstrand <jason.ekstr...@intel.com> > [PATCH 13/20] i965: Teach type_size() about the size of an image uniform. > [PATCH 14/20] i965: Implement logic to set up and upload an image uniform. I'm not particularly familiar with how mesa's uniform setup works. Ken? > [PATCH 15/20] i965/fs: Don't overwrite fs_visitor::uniforms and ::param_size > during the SIMD16 run. > [PATCH 16/20] i965/fs: Execute nir_setup_uniforms, _inputs and _outputs > unconditionally. Reviewed-by: Jason Ekstrand <jason.ekstr...@intel.com> > [PATCH 17/20] i965/fs: Handle image uniforms in NIR programs. Reviewed-by: Jason Ekstrand <jason.ekstr...@intel.com> > [PATCH 18/20] i965/fs: Translate image load, store and atomic NIR intrinsics. I had some comments on this one regarding encapsulation that I'd like to see addressed. > [PATCH 19/20] i965/fs: Translate memory barrier NIR intrinsics. Reviewed-by: Jason Ekstrand <jason.ekstr...@intel.com> > [PATCH 20/20] i965: Expose ARB_shader_image_load_store. I'm going to wait until the rest of it is reviewed to add mine to that little patch. :-) > _______________________________________________ > mesa-dev mailing list > mesa-dev@lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/mesa-dev _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
