llvm dev - Aug 2014 - [LLVMdev] [AMDGPU][PATCH 0/3] barriers/memory-fences related additions

Hello LLVM developers,

This serie of patches aims at defining the new intrinsics necessary to 
my libclc
serie of patches to implement barriers/memory fences on AMDGPUs.
This patch serie only provides stubs implementations of the necessary 
intrinsics.
For example, on evergreen hardware, we would need to modify the read/writes
surrounding operations (placed before or in the same loop) so that they make
an ACK when completed.
On Evergreen and SI, everything is lowered to a barrier instruction in 
fact (for
now).
I plan to add the necessary transformations in a following serie of patches.

Tested on Evergreen (Cedar) only.

(It's my first patches to an open-source project, so don't hesitate to 
point me my
mistakes/errors/indentation problems etc :) ).

Sincerely,
Damien Hilloulin.

[1/3] Addition of the new intrinsics in AMDGPUIntrinsics.td
     This patch adds the definitions of the intrinsics used for
     barriers/memory fences support.
[2/3] Stubs implementation of the new intrinsics on Evergreen
     This patch adds some stubs to provide a first implementation of the
     intrinsics for barriers and memory fences on EG. The barrier.nofence()
     intrinsic is the only intrinsic correctly implemented (for sure)
     with this patch. Maybe the barrier.local() intrinsic can be
     considered ok like this as the LDS memory is atomic. The other
     intrinsics need to use WAIT_ACK in some way and that we modify
     the surrounding memory operations with ACK.
[3/3] Stubs implementation of the new intrinsics on Southern_Islands
      This patch is a first implementation of the newly added
      intrinsics for barriers/memory fences on SI. For ultra-simplicity, 
every
      intrinsic is lowered to a barrier with no fence.

  lib/Target/R600/AMDGPUIntrinsics.td      | 11 +++++
  lib/Target/R600/EvergreenInstructions.td | 69 
+++++++++++++++++++++++++++++++-
  lib/Target/R600/SIInstructions.td        | 16 +++++++-
  3 files changed, 94 insertions(+), 2 deletions(-)

-- 
1.9.1

> On Aug 21, 2014, at 5:12 PM, Hilloulin Damien <damien.hilloulin at
supelec.fr> wrote:
> 
> Hello LLVM developers,
> 
> This serie of patches aims at defining the new intrinsics necessary to my
libclc
> serie of patches to implement barriers/memory fences on AMDGPUs.
> This patch serie only provides stubs implementations of the necessary
intrinsics.
> For example, on evergreen hardware, we would need to modify the read/writes
> surrounding operations (placed before or in the same loop) so that they
make
> an ACK when completed.
> On Evergreen and SI, everything is lowered to a barrier instruction in fact
(for
> now).
> I plan to add the necessary transformations in a following serie of
patches.
> 
> Tested on Evergreen (Cedar) only.
> 
> (It's my first patches to an open-source project, so don't hesitate
to point me my
> mistakes/errors/indentation problems etc :) ).
To point out the obvious, patches should always be accompanied by testcases that
exercise the new functionality.

thanks,
adrian
> 
> Sincerely,
> Damien Hilloulin.
> 
> [1/3] Addition of the new intrinsics in AMDGPUIntrinsics.td
>    This patch adds the definitions of the intrinsics used for
>    barriers/memory fences support.
> [2/3] Stubs implementation of the new intrinsics on Evergreen
>    This patch adds some stubs to provide a first implementation of the
>    intrinsics for barriers and memory fences on EG. The barrier.nofence()
>    intrinsic is the only intrinsic correctly implemented (for sure)
>    with this patch. Maybe the barrier.local() intrinsic can be
>    considered ok like this as the LDS memory is atomic. The other
>    intrinsics need to use WAIT_ACK in some way and that we modify
>    the surrounding memory operations with ACK.
> [3/3] Stubs implementation of the new intrinsics on Southern_Islands
>     This patch is a first implementation of the newly added
>     intrinsics for barriers/memory fences on SI. For ultra-simplicity,
every
>     intrinsic is lowered to a barrier with no fence.
> 
> lib/Target/R600/AMDGPUIntrinsics.td      | 11 +++++
> lib/Target/R600/EvergreenInstructions.td | 69
+++++++++++++++++++++++++++++++-
> lib/Target/R600/SIInstructions.td        | 16 +++++++-
> 3 files changed, 94 insertions(+), 2 deletions(-)
> 
> -- 
> 1.9.1
> 
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev