<div dir="ltr"><div class="gmail_quote"><div dir="ltr"></div><div dir="ltr">On Wed, Jan 9, 2019 at 1:41 PM Christian König <<a href="mailto:ckoenig.leichtzumerken@gmail.com">ckoenig.leichtzumerken@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div bgcolor="#FFFFFF">
<div class="gmail-m_-7810855682144448022moz-cite-prefix">Am 09.01.19 um 17:14 schrieb Marek
Olšák:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div dir="ltr">
<div class="gmail_quote">
<div dir="ltr">On Wed, Jan 9, 2019 at 8:09 AM Christian
König <<a href="mailto:ckoenig.leichtzumerken@gmail.com" target="_blank">ckoenig.leichtzumerken@gmail.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div bgcolor="#FFFFFF">
<div class="gmail-m_-7810855682144448022m_8930908994027207900m_-7323258551939539738gmail-m_4443106390953893387gmail-m_-3252015337124169842moz-cite-prefix">Am
09.01.19 um 13:36 schrieb Marek Olšák:<br>
</div>
<blockquote type="cite">
<div dir="auto">
<div><br>
<br>
<div class="gmail_quote">
<div dir="ltr">On Wed, Jan 9, 2019, 5:28 AM
Christian König <<a href="mailto:ckoenig.leichtzumerken@gmail.com" target="_blank">ckoenig.leichtzumerken@gmail.com</a>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Looks good,
but I'm wondering what's the actual
improvement?<br>
</blockquote>
</div>
</div>
<div dir="auto"><br>
</div>
<div dir="auto">No malloc calls and 1 less for loop
copying the bo list.</div>
</div>
</blockquote>
<br>
Yeah, but didn't we want to get completely rid of the bo
list?<br>
</div>
</blockquote>
<div><br>
</div>
If we have multiple IBs (e.g. gfx + compute) that share a BO
list, I think it's faster to send the BO list to the kernel
only once.</div>
</div>
</div>
</blockquote>
<br>
That's not really faster.<br>
<br>
The only thing we safe us is a single loop over all BOs to lockup
the handle into a pointer and that is only a tiny fraction of the
overhead.<br>
<br>
The majority of the overhead is locking the BOs and reserving space
for the submission.<br>
<br>
What could really help here is to submit gfx+comput together in just
one CS IOCTL. This way we would need the locking and space
reservation only once.<br>
<br>
It's a bit of work in the kernel side, but certainly doable.<br></div></blockquote><div><br></div><div>OK. Any objections to this patch?</div><div><br></div><div>Thanks,<br></div><div>Marek</div></div></div>