From: "Zhang, Jerry (Junwei)" <Jerry.Zhang-5C7GfCeVMHo@public.gmane.org>
To: "Christian König"
<ckoenig.leichtzumerken-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>,
amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Subject: Re: Optimize VM handling a bit more
Date: Tue, 11 Sep 2018 10:39:26 +0800 [thread overview]
Message-ID: <5B972ADE.7090901@amd.com> (raw)
In-Reply-To: <20180909180339.1910-1-christian.koenig-5C7GfCeVMHo@public.gmane.org>
Apart from Felix comments,
Looks good for me, patch 2 ~ 8 are
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Patch 9 ~ 11 are
Acked-by: Junwei Zhang <Jerry.Zhang@amd.com>
On 09/10/2018 02:03 AM, Christian König wrote:
> Hi everyone,
>
> Especially on Vega and Raven VM handling is rather inefficient while creating PTEs because we originally only supported 2 level page tables and implemented 4 level page tables on top of that.
>
> This patch set reworks quite a bit of that handling and adds proper iterator and tree walking functions which are then used to update PTEs more efficiently.
>
> A totally constructed test case which tried to map 2GB of VRAM on an unaligned address is reduced from 45ms down to ~20ms on my test system.
>
> As a very positive side effect this also adds support for 1GB giant VRAM pages additional to the existing 2MB huge pages on Vega/Raven and also enables all additional power of two values (2MB-2GB) for the L1.
>
> This could be beneficial for applications which allocate very huge amounts of memory because it reduces the overhead of page table walks by 50% (huge pages where 25%).
>
> Please comment and/or review,
> Christian.
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
prev parent reply other threads:[~2018-09-11 2:39 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-09 18:03 Optimize VM handling a bit more Christian König
[not found] ` <20180909180339.1910-1-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-09-09 18:03 ` [PATCH 01/11] drm/amdgpu: try allocating VRAM as power of two Christian König
[not found] ` <20180909180339.1910-2-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-09-11 0:08 ` Felix Kuehling
[not found] ` <60b9c96a-d31d-cd61-df22-c8414f0166dc-5C7GfCeVMHo@public.gmane.org>
2018-09-11 6:49 ` Christian König
[not found] ` <9b61bb4c-4721-c8af-0810-4ecc18ed1ea4-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2018-09-11 7:37 ` Michel Dänzer
[not found] ` <46b7da24-4b04-7daf-e6c3-07e10ecb6fd6-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-09-11 7:46 ` Christian König
[not found] ` <d0af0309-e4f9-9825-fd30-9cf79ed5f823-5C7GfCeVMHo@public.gmane.org>
2018-09-11 7:55 ` Michel Dänzer
[not found] ` <5fd331d0-24ab-7540-2d0a-b227230743c0-otUistvHUpPR7s880joybQ@public.gmane.org>
2018-09-11 8:20 ` Christian König
[not found] ` <a3c91610-e5e2-c666-cfda-344f1f40485c-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2018-09-11 8:25 ` Michel Dänzer
2018-09-11 14:27 ` Kuehling, Felix
[not found] ` <DM5PR12MB17073F8B673AEFC092FBA5B392040-2J9CzHegvk9TCtO+SvGBKwdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org>
2018-09-11 14:48 ` Christian König
[not found] ` <2361b55f-ed56-bd42-b628-fc001d9ea300-5C7GfCeVMHo@public.gmane.org>
2018-09-11 22:25 ` Felix Kuehling
2018-09-11 9:50 ` Huang Rui
2018-09-11 11:16 ` Christian König
2018-09-09 18:03 ` [PATCH 02/11] drm/amdgpu: add amdgpu_vm_pt_parent helper Christian König
2018-09-09 18:03 ` [PATCH 03/11] drm/amdgpu: add amdgpu_vm_update_func Christian König
2018-09-09 18:03 ` [PATCH 04/11] drm/amdgpu: add some VM PD/PT iterators Christian König
[not found] ` <20180909180339.1910-5-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-09-11 1:51 ` Felix Kuehling
2018-09-11 2:38 ` Zhang, Jerry (Junwei)
2018-09-09 18:03 ` [PATCH 05/11] drm/amdgpu: use leaf iterator for allocating PD/PT Christian König
2018-09-09 18:03 ` [PATCH 06/11] drm/amdgpu: use dfs iterator to free PDs/PTs Christian König
2018-09-09 18:03 ` [PATCH 07/11] drm/amdgpu: use the DFS iterator in amdgpu_vm_invalidate_level Christian König
[not found] ` <20180909180339.1910-8-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-09-11 0:19 ` Felix Kuehling
2018-09-09 18:03 ` [PATCH 08/11] drm/amdgpu: use leaf iterator for filling PTs Christian König
2018-09-09 18:03 ` [PATCH 09/11] drm/amdgpu: meld together VM fragment and huge page handling Christian König
2018-09-09 18:03 ` [PATCH 10/11] drm/amdgpu: use the maximum possible fragment size on Vega/Raven Christian König
[not found] ` <20180909180339.1910-11-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-09-11 0:45 ` Felix Kuehling
2018-09-09 18:03 ` [PATCH 11/11] drm/amdgpu: allow fragment processing for invalid PTEs Christian König
2018-09-11 2:17 ` Optimize VM handling a bit more Felix Kuehling
2018-09-11 2:39 ` Zhang, Jerry (Junwei) [this message]
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=5B972ADE.7090901@amd.com \
--to=jerry.zhang-5c7gfcevmho@public.gmane.org \
--cc=amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=ckoenig.leichtzumerken-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.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.