Hi Kumar, I think this patch will improve the performance only when the memory is enough. For if you in a memory tough environment it can not always have such big coherent dma buffer to use.
On Thu, 2008-05-15 at 19:12 +0530, Kumar, Purushotam wrote: > This patch truly enables CONFIG_MMC_BLOCK_BOUNCE feature by setting the > hw/phys segs to 1 without which the feature is not exploited as large/bounce > buffers of size 64KB are not used for any of read/write transaction. Without > this patch, smaller buffers are used whose sizes vary from 512 Bytes to 32KB. > In my testing with a SD Extreme III high speed 2GB SD card using FAT file > system performance improvement of about 10% for write and 40% for reads has > been observed. Write and Read speed without this patch was 5.2Mbytes/sec and > 6.5Mbytes/sec respectively whereas write and read performance with this patch > has jumped to 5.9Mbytes/sec and 9MBytes/sec, respectively. > > Signed-off-by: Purushotam Kumar <[EMAIL PROTECTED]> > --- > Index: linux-davinci-2.6/drivers/mmc/host/davinci_mmc.c > =================================================================== > --- linux-davinci-2.6.orig/drivers/mmc/host/davinci_mmc.c > +++ linux-davinci-2.6/drivers/mmc/host/davinci_mmc.c > @@ -1348,8 +1348,13 @@ static int davinci_mmcsd_probe(struct pl > #endif > mmc->ocr_avail = MMC_VDD_32_33; > > +#ifdef CONFIG_MMC_BLOCK_BOUNCE > + mmc->max_phys_segs = 1; > + mmc->max_hw_segs = 1; > +#else > mmc->max_phys_segs = 2; > mmc->max_hw_segs = 2; > +#endif > mmc->max_blk_size = 4095; /* BLEN is 11 bits */ > mmc->max_blk_count = 65535; /* NBLK is 16 bits */ > mmc->max_req_size = mmc->max_blk_size * mmc->max_blk_count; > _______________________________________________ > Davinci-linux-open-source mailing list > [email protected] > http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source _______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
