<div dir="auto"><div dir="auto"><br></div><div dir="auto">Do we know if Linux kernel driver enables this?</div><div dir="auto"><br></div><div dir="auto">Dave.</div></div><div class="gmail_extra"><br><div class="gmail_quote">On 9 Oct. 2017 6:04 pm, "Samuel Pitoiset" <<a href="mailto:samuel.pitoiset@gmail.com">samuel.pitoiset@gmail.com</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
<br>
On 10/07/2017 10:59 AM, Nicolai Hähnle wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
On 06.10.2017 16:03, Samuel Pitoiset wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Ported from RadeonSI, and -pro seems to enable it as well.<br>
<br>
Signed-off-by: Samuel Pitoiset <<a href="mailto:samuel.pitoiset@gmail.com" target="_blank">samuel.pitoiset@gmail.com</a>><br>
---<br>
  src/amd/vulkan/radv_cmd_buffer<wbr>.c | 11 ++++++++++-<br>
  1 file changed, 10 insertions(+), 1 deletion(-)<br>
<br>
diff --git a/src/amd/vulkan/radv_cmd_buff<wbr>er.c b/src/amd/vulkan/radv_cmd_buff<wbr>er.c<br>
index d9243beae5..c641534fc9 100644<br>
--- a/src/amd/vulkan/radv_cmd_buff<wbr>er.c<br>
+++ b/src/amd/vulkan/radv_cmd_buff<wbr>er.c<br>
@@ -3198,7 +3198,16 @@ radv_emit_dispatch_packets(str<wbr>uct radv_cmd_buffer *cmd_buffer,<br>
          }<br>
      } else {<br>
          unsigned blocks[3] = { info->blocks[0], info->blocks[1], info->blocks[2] };<br>
-        unsigned dispatch_initiator = S_00B800_COMPUTE_SHADER_EN(1);<br>
+        unsigned dispatch_initiator = S_00B800_COMPUTE_SHADER_EN(1) |<br>
+                          S_00B800_FORCE_START_AT_000(1)<wbr>;<br>
+<br>
+        if (cmd_buffer->device->physical_<wbr>device->rad_info.chip_class >= CIK) {<br>
+            /* If the KMD allows it (there is a KMD hw register for<br>
+             * it), allow launching waves out-of-order. (same as<br>
+             * Vulkan)<br>
</blockquote>
<br>
The comment in parenthesis seems a bit like the driver is having an identity crisis :)<br>
</blockquote>
<br>
Ah right :) I will remove that.<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Other than that,<br>
<br>
Acked-by: Nicolai Hähnle <<a href="mailto:nicolai.haehnle@amd.com" target="_blank">nicolai.haehnle@amd.com</a>><br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
+             */<br>
+            dispatch_initiator |= S_00B800_ORDER_MODE(1);<br>
+        }<br>
          if (info->unaligned) {<br>
              unsigned *cs_block_size = compute_shader->info.cs.block_<wbr>size;<br>
<br>
</blockquote>
<br>
<br>
</blockquote>
______________________________<wbr>_________________<br>
mesa-dev mailing list<br>
<a href="mailto:mesa-dev@lists.freedesktop.org" target="_blank">mesa-dev@lists.freedesktop.org</a><br>
<a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev" rel="noreferrer" target="_blank">https://lists.freedesktop.org/<wbr>mailman/listinfo/mesa-dev</a><br>
</blockquote></div></div>