All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
To: Francois Dugast <francois.dugast@intel.com>,
	igt-dev@lists.freedesktop.org
Subject: Re: [PATCH i-g-t 03/11] lib/intel_compute: Use common constant for indirect object size
Date: Thu, 13 Mar 2025 16:35:36 +0100	[thread overview]
Message-ID: <3808b4b77f7cb941e038588e6e5fb842a4bca1fd.camel@linux.intel.com> (raw)
In-Reply-To: <20250311152321.16497-4-francois.dugast@intel.com>

On Tue, 2025-03-11 at 16:21 +0100, Francois Dugast wrote:
> Reduce magic values in the code, homogenize the size which has no
> reason to be different among pipelines, define this value close to
> the addresses to make it easier to spot potential overlaps in the
> future.
> 
> Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>

> ---
>  lib/intel_compute.c | 13 +++++++------
>  1 file changed, 7 insertions(+), 6 deletions(-)
> 
> diff --git a/lib/intel_compute.c b/lib/intel_compute.c
> index b4b022ae7..e3c8cf244 100644
> --- a/lib/intel_compute.c
> +++ b/lib/intel_compute.c
> @@ -30,6 +30,7 @@
>  #define SIZE_BUFFER_OUTPUT		MAX(sizeof(float) *
> SIZE_DATA, 0x10000)
>  #define SIZE_SURFACE_STATE		0x10000
>  #define SIZE_DYNAMIC_STATE		0x100000
> +#define SIZE_INDIRECT_OBJECT		0x10000
>  
>  #define ADDR_SYNC			0x010000ULL
>  #define ADDR_SYNC2			0x020000ULL
> @@ -698,7 +699,7 @@ static void compute_exec(int fd, const unsigned
> char *kernel,
>  		  .size = SIZE_SURFACE_STATE,
>  		  .name = "surface state base" },
>  		{ .addr = ADDR_INDIRECT_OBJECT_BASE +
> OFFSET_INDIRECT_DATA_START,
> -		  .size =  0x10000,
> +		  .size = SIZE_INDIRECT_OBJECT,
>  		  .name = "indirect data start" },
>  		{ .addr = ADDR_INPUT,
>  		  .size = SIZE_BUFFER_INPUT,
> @@ -983,7 +984,7 @@ static void xehp_compute_exec(int fd, const
> unsigned char *kernel,
>  		  .size = SIZE_SURFACE_STATE,
>  		  .name = "surface state base"},
>  		{ .addr = ADDR_GENERAL_STATE_BASE +
> OFFSET_INDIRECT_DATA_START,
> -		  .size =  0x10000,
> +		  .size = SIZE_INDIRECT_OBJECT,
>  		  .name = "indirect object base"},
>  		{ .addr = ADDR_INPUT, .size = SIZE_BUFFER_INPUT,
>  		  .name = "addr input"},
> @@ -1200,7 +1201,7 @@ static void xehpc_compute_exec(int fd, const
> unsigned char *kernel,
>  		{ .addr = ADDR_INSTRUCTION_STATE_BASE +
> OFFSET_KERNEL,
>  		  .name = "instr state base"},
>  		{ .addr = ADDR_GENERAL_STATE_BASE +
> OFFSET_INDIRECT_DATA_START,
> -		  .size =  0x10000,
> +		  .size = SIZE_INDIRECT_OBJECT,
>  		  .name = "indirect object base"},
>  		{ .addr = ADDR_INPUT, .size = SIZE_BUFFER_INPUT,
>  		  .name = "addr input"},
> @@ -1558,7 +1559,7 @@ static void xelpg_compute_exec(int fd, const
> unsigned char *kernel,
>  		  .size = SIZE_SURFACE_STATE,
>  		  .name = "surface state base"},
>  		{ .addr = ADDR_GENERAL_STATE_BASE +
> OFFSET_INDIRECT_DATA_START,
> -		  .size =  0x1000,
> +		  .size = SIZE_INDIRECT_OBJECT,
>  		  .name = "indirect object base"},
>  		{ .addr = ADDR_INPUT, .size = SIZE_BUFFER_INPUT,
>  		  .name = "addr input"},
> @@ -1648,7 +1649,7 @@ static void xe2lpg_compute_exec(int fd, const
> unsigned char *kernel,
>  		  .size = SIZE_SURFACE_STATE,
>  		  .name = "surface state base"},
>  		{ .addr = ADDR_GENERAL_STATE_BASE +
> OFFSET_INDIRECT_DATA_START,
> -		  .size =  0x1000,
> +		  .size = SIZE_INDIRECT_OBJECT,
>  		  .name = "indirect object base"},
>  		{ .addr = ADDR_INPUT, .size = SIZE_BUFFER_INPUT,
>  		  .name = "addr input"},
> @@ -1899,7 +1900,7 @@ static void xe2lpg_compute_preempt_exec(int fd,
> const unsigned char *long_kernel
>  		  .size = SIZE_SURFACE_STATE,
>  		  .name = "surface state base"},
>  		{ .addr = ADDR_GENERAL_STATE_BASE +
> OFFSET_INDIRECT_DATA_START,
> -		  .size =  0x1000,
> +		  .size = SIZE_INDIRECT_OBJECT,
>  		  .name = "indirect object base"},
>  		{ .addr = ADDR_INPUT, .size = SIZE_BUFFER_INPUT,
>  		  .name = "addr input"},


  reply	other threads:[~2025-03-13 15:35 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-11 15:21 [PATCH i-g-t 00/11] Prepare lib/intel_compute for SVM/system allocator, part 2 Francois Dugast
2025-03-11 15:21 ` [PATCH i-g-t 01/11] lib/intel_compute: Use common constant for surface state size Francois Dugast
2025-03-13 14:28   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 02/11] lib/intel_compute: Use common constant for dynamic " Francois Dugast
2025-03-13 14:30   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 03/11] lib/intel_compute: Use common constant for indirect object size Francois Dugast
2025-03-13 15:35   ` Thomas Hellström [this message]
2025-03-11 15:21 ` [PATCH i-g-t 04/11] lib/intel_compute: Use common constant for binding table size Francois Dugast
2025-03-13 14:31   ` Thomas Hellström
2025-03-13 14:34   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 05/11] lib/intel_compute: Use common constant for general state size Francois Dugast
2025-03-13 14:35   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 06/11] lib/intel_compute: Use constant for binding table address Francois Dugast
2025-03-13 14:41   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 07/11] lib/intel_compute: Compact memory map Francois Dugast
2025-03-13 14:44   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 08/11] lib/intel_compute: Relocate input and output objects Francois Dugast
2025-03-13 14:45   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 09/11] lib/intel_compute: Fix enqueued local size in xehp Francois Dugast
2025-03-13 14:48   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 10/11] lib/intel_compute: Use constants for thread groups and local work size Francois Dugast
2025-03-13 15:09   ` Thomas Hellström
2025-03-11 15:21 ` [PATCH i-g-t 11/11] lib/intel_compute: Make array size a dynamic parameter Francois Dugast
2025-03-13 15:33   ` Thomas Hellström
2025-03-12  2:44 ` ✗ Xe.CI.BAT: failure for Prepare lib/intel_compute for SVM/system allocator, part 2 Patchwork
2025-03-12  3:08 ` ✓ i915.CI.BAT: success " Patchwork
2025-03-12 14:16 ` ✓ i915.CI.Full: " Patchwork
2025-03-12 18:21 ` ✗ Xe.CI.Full: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3808b4b77f7cb941e038588e6e5fb842a4bca1fd.camel@linux.intel.com \
    --to=thomas.hellstrom@linux.intel.com \
    --cc=francois.dugast@intel.com \
    --cc=igt-dev@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.