References: <b9bd16cb-f1fa-34b7-d599-8637cbe5032b@free.fr>
 <1jh6ue74x9.fsf@starbuckisacylon.baylibre.com>
 <ce0bd9a5-e44d-b30b-3434-9d5fd36e251a@free.fr>
User-agent: mu4e 1.8.13; emacs 28.2
From:   Jerome Brunet <jbrunet@baylibre.com>
To:     Marc Gonzalez <marc.w.gonzalez@free.fr>,
        linux-wireless@vger.kernel.org, MMC <linux-mmc@vger.kernel.org>,
        AML <linux-amlogic@lists.infradead.org>
Cc:     Neil Armstrong <neil.armstrong@linaro.org>,
        Arend van Spriel <aspriel@gmail.com>,
        Franky Lin <franky.lin@broadcom.com>,
        Hante Meuleman <hante.meuleman@broadcom.com>,
        Kalle Valo <kvalo@kernel.org>,
        Eric Dumazet <edumazet@google.com>,
        Ulf Hansson <ulf.hansson@linaro.org>,
        Pieter-Paul Giesberts <pieter-paul.giesberts@broadcom.com>,
        Alexander Prutskov <alep@cypress.com>,
        Joseph chuang <jiac@cypress.com>,
        Kevin Hilman <khilman@baylibre.com>,
        Martin Blumenstingl <martin.blumenstingl@gmail.com>,
        Angus Ainslie <angus@akkea.ca>,
        Heiner Kallweit <hkallweit1@gmail.com>,
        Pierre-Hugues Husson <phh@phh.me>
Subject: Re: Performance lower than expected with BCM4359/9 on S905X2
Date:   Tue, 21 Mar 2023 14:54:39 +0100
In-reply-to: <ce0bd9a5-e44d-b30b-3434-9d5fd36e251a@free.fr>
Message-ID: <1jcz526w53.fsf@starbuckisacylon.baylibre.com>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk


>
> I was not asking anyone to reproduce my exact setup.
>
> I find it confusing that some assume there is something different
> about the board I'm using. The more likely hypothesis is that all
> S905X2 boards have the same behavior.

High-speed MMC modes are usually defined/limited per boards.
Check the different  DTs in arch/arm64/boot/dts/amlogic, you'll see
that not all have the same configs.

Yes the S905X2 has the bounce buffer work-around which plays a role on
the overall performance. MMC high-speed modes is critical just has well.

The Amlogic 4.9 downstream does all kinds of fancy (and unsafe) things
to enable higher modes.

So there is no assumption here, merely a tentative to get a clearer
picture.

>
> Regards