mirror of https://lore.kernel.org/linux-amlogic/
 help / color / mirror / Atom feed
From: Marc Gonzalez <marc.w.gonzalez@free.fr>
To: AML <linux-amlogic@lists.infradead.org>
Cc: Martin Blumenstingl <martin.blumenstingl@gmail.com>,
	Christian Hewitt <christian@hewittfamily.org.uk>,
	Linux Wireless <linux-wireless@vger.kernel.org>,
	MMC <linux-mmc@vger.kernel.org>,
	Neil Armstrong <neil.armstrong@linaro.org>,
	Ulf Hansson <ulf.hansson@linaro.org>,
	Pierre-Hugues Husson <phh@phh.me>
Subject: Re: [Performance regression] BCM4359/9 on S905X2
Date: Mon, 17 Apr 2023 16:22:36 +0200	[thread overview]
Message-ID: <ac5d9c2e-13c3-7f99-a20c-d857282c6451@free.fr> (raw)
In-Reply-To: <74feca6f-4177-5645-6614-bc5f3dbf0be1@free.fr>

On 06/04/2023 18:35, Marc Gonzalez wrote:

> # curl -o /dev/null http://192.168.1.254:8095/fixed/1G
>   % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                  Dload  Upload   Total   Spent    Left  Speed
> 100 1024M  100 1024M    0     0  9747k      0  0:01:47  0:01:47 --:--:-- 9544k

I was able to import the bcmdhd driver from:
https://android.googlesource.com/kernel/hikey-linaro/+log/refs/heads/android-amlogic-bmeson-6.1/drivers/net/wireless/bcmdhd
as pointed out by Neil.

I switched from brcmfmac to bcmdhd with:

-CONFIG_BRCMFMAC=y
-CONFIG_BRCMDBG=y
+CONFIG_BCMDHD=y
+CONFIG_BCMDHD_FW_PATH=""
+CONFIG_BCMDHD_NVRAM_PATH=""
+CONFIG_BCMDHD_SDIO_IRQ=y

and hard-coded the FW paths in dhd_set_path_params().


Here are the benchmark results on the bcm4359 (AP6398SR3 module)

# curl -o /dev/null http://192.168.1.254:8095/fixed/1G
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 1024M  100 1024M    0     0  17.3M      0  0:00:58  0:00:58 --:--:-- 17.4M

which is noticeably faster (80%) than brcmfmac results.

FTR, system was 18.5% busy for this benchmark:
   81.5653%        195114        0:swapper        
   13.8287%         33080       57:irq/17-ffe03000
    1.3967%          3341      173:curl           
    1.1350%          2715       63:dhd_rxf        
    1.0455%          2501      129:ksdioirqd/mmc2 
    1.0004%          2393       62:dhd_dpc        
    0.0167%            40      171:perf           
    0.0088%            21       19:ksoftirqd/1    


I was not able to get the driver to work with bcm43752 (AP6398SR32 module).

It loops with:

[   15.221622] wl_run_escan: LEGACY_SCAN sync ID: 0, bssidx: 0
[   15.221661] CFG80211-ERROR) wl_run_escan : 
[   15.221664]  Escan set error (-37)
[   15.229067] dhd_ioctl_entry_local invalid parameter
[   15.233902] CFG80211-ERROR) wl_run_escan : 
[   15.233904] error (-37), cnt=1
[   15.241056] CFG80211-ERROR) wl_cfg80211_scan : 
[   15.241059] scan error (-11)

But I may have bungled adding the "new" chip, since the driver architecture
is... [censored].


I suppose the conclusion is:
1) bcmdhd is 80% faster than brcmfmac in one download scenario
2) CPU usage scales linearly with throughput (on this system, 1% ~ 7 Mbps)

Regards



_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

  parent reply	other threads:[~2023-04-17 14:23 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-03-21 10:40 Marc Gonzalez
2023-03-21 10:46 ` Jerome Brunet
2023-03-21 12:37   ` Performance lower than expected with " Marc Gonzalez
2023-03-21 13:54     ` Jerome Brunet
2023-03-21 16:48       ` Marc Gonzalez
     [not found]   ` <EEE9FD80-C106-4A9B-AA8D-5C151E540CFD@hewittfamily.org.uk>
2023-04-04 16:09     ` [Performance regression] " Marc Gonzalez
2023-04-04 21:06       ` Martin Blumenstingl
2023-04-06 16:35         ` Marc Gonzalez
2023-04-06 17:07           ` Christian Hewitt
2023-04-06 17:41             ` Neil Armstrong
2023-04-17 14:22           ` Marc Gonzalez [this message]
2023-03-21 11:45 ` Neil Armstrong

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=ac5d9c2e-13c3-7f99-a20c-d857282c6451@free.fr \
    --to=marc.w.gonzalez@free.fr \
    --cc=christian@hewittfamily.org.uk \
    --cc=linux-amlogic@lists.infradead.org \
    --cc=linux-mmc@vger.kernel.org \
    --cc=linux-wireless@vger.kernel.org \
    --cc=martin.blumenstingl@gmail.com \
    --cc=neil.armstrong@linaro.org \
    --cc=phh@phh.me \
    --cc=ulf.hansson@linaro.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

all inboxes | Powered by JetHome®