From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jiada Wang Subject: Re: [PATCH linux-next v3 1/1] spi: imx: dynamic burst length adjust for PIO mode Date: Thu, 1 Jun 2017 00:55:53 -0700 Message-ID: <592FC889.3020001@mentor.com> References: <1495774962-2981-1-git-send-email-jiada_wang@mentor.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8"; format=flowed Content-Transfer-Encoding: 7bit Cc: Mark Brown , "linux-spi@vger.kernel.org" , linux-kernel , Sascha Hauer To: Fabio Estevam Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-spi.vger.kernel.org Hello Fabio On 05/29/2017 04:07 AM, Fabio Estevam wrote: > Hi Jiada, > > On Fri, May 26, 2017 at 2:02 AM, wrote: >> From: Jiada Wang >> >> previously burst length (BURST_LENGTH) is always set to equal >> to bits_per_word, causes a 10us gap between each word in >> transfer, which significantly affects performance. >> >> This patch uses 32 bits transfer to simulate lower bits transfer, >> and adjusts burst length runtimely to use biggeest burst length >> as possible to reduce the gaps in transfer for PIO mode. > Just curious: what is the performance gain you observe with this patch? I did some performance test with 576/384/192 bytes data, following is the result between the test w & w/o this patch (ecspi controller works in PIO and loopback mode) with this patch: bytes bpw time (ms) 576 8 9.2 576 16 9.2 384 8 6.2 192 8 3.1 without this patch bytes bpw time(ms) 576 8 14.4 576 16 11.6 384 8 9.6 384 16 7.8 192 8 4.8 192 16 3.9 Thanks, Jiada