From: Alex Elder <elder@ieee.org>
To: Zhang Zekun <zhangzekun11@huawei.com>,
xuwei5@hisilicon.com, lihuisong@huawei.com,
Jonathan.Cameron@huawei.com
Cc: linux-kernel@vger.kernel.org, liuyongqiang13@huawei.com
Subject: Re: [PATCH 0/2] soc: hisilicon: Add power control support for kunpeng HBM
Date: Sat, 7 Dec 2024 10:50:01 -0600 [thread overview]
Message-ID: <1f597801-62b7-4fb1-9ffe-1b9554c4e23c@ieee.org> (raw)
In-Reply-To: <20241206112812.32618-1-zhangzekun11@huawei.com>
On 12/6/24 5:28 AM, Zhang Zekun wrote:
> Add power control support for High Bandwidth Memory (HBM) for Kunpeng SoC
> platform. HBM devices on Kunpeng SoC can provide higher bandwidth at the
> cost of higher power consumption. Providing power control methods can help
> reducing the power when the workload does not need use HBM.
Could you explain a little more here how HBM is represented in the
system? When it's powered on, it seems like it's "just memory".
And what you're doing here is enabling a power optimization to
allow this type of memory to be powered off when not in use.
How do you know whether it is in use? What entity is meant to
be able to power this memory on and off?
In addition, it looks like there can be more than one instance of
an HBM device, and each is available to be used only for certain
CPUs. Can you provide more information about that sort of
architectural detail?
Finally, the second patch enables "cache" functionality. Maybe
this is something defined by ACPI and is well understood by others
but it's not clear to me what this even means. How is an HBM
used, and how does its cache enabled/disabled state interact
with the device enabled/disabled state?
Is an HBM device something completely different from an HBM cache
device? I guess I just lack a big-picture overview of how this
HBM fits into a system.
-Alex
>
> Zhang Zekun (2):
> soc: hisilicon: kunpeng_hbmdev: Add support for controling the power
> of hbm memory
> soc: hisilicon: kunpeng_hbmcache: Add support for online and offline
> the hbm cache
>
> MAINTAINERS | 7 +
> drivers/soc/hisilicon/Kconfig | 23 +++
> drivers/soc/hisilicon/Makefile | 2 +
> drivers/soc/hisilicon/kunpeng_hbm.h | 31 ++++
> drivers/soc/hisilicon/kunpeng_hbmcache.c | 136 +++++++++++++++
> drivers/soc/hisilicon/kunpeng_hbmdev.c | 210 +++++++++++++++++++++++
> 6 files changed, 409 insertions(+)
> create mode 100644 drivers/soc/hisilicon/kunpeng_hbm.h
> create mode 100644 drivers/soc/hisilicon/kunpeng_hbmcache.c
> create mode 100644 drivers/soc/hisilicon/kunpeng_hbmdev.c
>
prev parent reply other threads:[~2024-12-07 16:50 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-12-06 11:28 Zhang Zekun
2024-12-06 11:28 ` [PATCH 1/2] soc: hisilicon: kunpeng_hbmdev: Add support for controling the power of hbm memory Zhang Zekun
2024-12-07 16:50 ` Alex Elder
2024-12-09 23:56 ` Jeff Johnson
2024-12-06 11:28 ` [PATCH 2/2] soc: hisilicon: kunpeng_hbmcache: Add support for online and offline the hbm cache Zhang Zekun
2024-12-06 21:02 ` kernel test robot
2024-12-07 16:50 ` Alex Elder
2024-12-10 0:01 ` Jeff Johnson
2024-12-07 16:50 ` Alex Elder [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=1f597801-62b7-4fb1-9ffe-1b9554c4e23c@ieee.org \
--to=elder@ieee.org \
--cc=Jonathan.Cameron@huawei.com \
--cc=lihuisong@huawei.com \
--cc=linux-kernel@vger.kernel.org \
--cc=liuyongqiang13@huawei.com \
--cc=xuwei5@hisilicon.com \
--cc=zhangzekun11@huawei.com \
/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
all inboxes | Powered by JetHome®