iommu.lists.linux-foundation.org archive mirror
 help / color / mirror / Atom feed
From: Steven Price <steven.price@arm.com>
To: Tomeu Vizoso <tomeu.vizoso@collabora.com>,
	Rob Herring <robh@kernel.org>,
	 Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: Neil Armstrong <narmstrong@baylibre.com>,
	Robin Murphy <robin.murphy@arm.com>,
	dri-devel <dri-devel@lists.freedesktop.org>,
	Linux IOMMU <iommu@lists.linux-foundation.org>,
	Kevin Hilman <khilman@baylibre.com>,
	"open list:ARM/Amlogic Meson..."
	<linux-amlogic@lists.infradead.org>,
	Will Deacon <will@kernel.org>,
	linux-arm-kernel <linux-arm-kernel@lists.infradead.org>,
	Jerome Brunet <jbrunet@baylibre.com>
Subject: Re: [PATCH 0/3] drm: panfrost: Coherency support
Date: Thu, 17 Sep 2020 12:00:58 +0100	[thread overview]
Message-ID: <51a06e78-7b80-c65d-67c4-e026e729e505@arm.com> (raw)
In-Reply-To: <0518e067-3b9a-8fe3-619f-8fce4c6c9677@collabora.com>

On 17/09/2020 11:51, Tomeu Vizoso wrote:
> On 9/17/20 12:38 PM, Steven Price wrote:
>> On 16/09/2020 18:46, Rob Herring wrote:
>>> On Wed, Sep 16, 2020 at 11:04 AM Alyssa Rosenzweig
>>> <alyssa.rosenzweig@collabora.com> wrote:
>>>>
>>>>> So I get a performance regression with the dma-coherent approach, 
>>>>> even if it's
>>>>> clearly the cleaner.
>>>>
>>>> That's bizarre -- this should really be the faster of the two.
>>>
>>> Coherency may not be free. CortexA9 had something like 4x slower
>>> memcpy if SMP was enabled as an example. I don't know if there's
>>> anything going on like that specifically here. If there's never any
>>> CPU accesses mixed in with kmscube, then there would be no benefit to
>>> coherency.
>>
>> The DDK blob has the ability to mark only certain areas of memory as 
>> coherent for performance reasons. For simple things like kmscube I 
>> would expect that it's basically write-only from the CPU and almost 
>> all memory the GPU touches isn't touched by the CPU. I.e. coherency 
>> isn't helping and the coherency traffic is probably expensive. Whether 
>> the complexity is worth it for "real" content I don't know - it may 
>> just be silly benchmarks that benefit.
> 
> Or maybe it's only a problem for applications that do silly things? I 
> don't think kmscube was ever optimized for performance.

Well doing silly things is almost the definition of a benchmark ;) A lot 
of the mobile graphics benchmarks suffer from not being very intelligent 
in how they render (e.g. many have meshes that are far too detailed so 
the triangles are smaller than the pixels).

Of course there are also applications that get things wrong too.

Steve
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu

  reply	other threads:[~2020-09-17 11:01 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-15 23:51 [PATCH 0/3] drm: panfrost: Coherency support Robin Murphy
2020-09-15 23:51 ` [PATCH 1/3] iommu/io-pgtable-arm: Support coherency for Mali LPAE Robin Murphy
2020-09-21 17:57   ` Will Deacon
2020-09-21 21:53     ` Robin Murphy
2020-09-21 22:24       ` Will Deacon
2020-09-15 23:51 ` [PATCH 2/3] drm/panfrost: Support cache-coherent integrations Robin Murphy
2020-09-17 10:37   ` Steven Price
2020-09-15 23:51 ` [PATCH 3/3] arm64: dts: meson: Describe G12b GPU as coherent Robin Murphy
2020-09-16  8:26   ` Neil Armstrong
2020-10-05  8:15     ` Boris Brezillon
2020-10-05  8:34       ` Steven Price
2020-10-05  8:39         ` Boris Brezillon
2020-10-05 10:05           ` Steven Price
2020-09-16 14:54   ` Neil Armstrong
2020-09-16 14:54 ` [PATCH 0/3] drm: panfrost: Coherency support Neil Armstrong
2020-09-16 17:04   ` Alyssa Rosenzweig
2020-09-16 17:46     ` Rob Herring
2020-09-17 10:38       ` Steven Price
2020-09-17 10:51         ` Tomeu Vizoso
2020-09-17 11:00           ` Steven Price [this message]
2020-09-17 12:03         ` Alyssa Rosenzweig
2020-09-17 12:38       ` Robin Murphy

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=51a06e78-7b80-c65d-67c4-e026e729e505@arm.com \
    --to=steven.price@arm.com \
    --cc=alyssa.rosenzweig@collabora.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=iommu@lists.linux-foundation.org \
    --cc=jbrunet@baylibre.com \
    --cc=khilman@baylibre.com \
    --cc=linux-amlogic@lists.infradead.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=narmstrong@baylibre.com \
    --cc=robh@kernel.org \
    --cc=robin.murphy@arm.com \
    --cc=tomeu.vizoso@collabora.com \
    --cc=will@kernel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).