Hi, > On Monday 08 September 2008, Kamat, Nishant wrote: > > +config MTD_NAND_OMAP_PREFETCH > > + bool "GPMC prefetch support for NAND Flash device" > > + depends on MTD_NAND && MTD_NAND_OMAP2 > > + default n > > + help > > + The NAND device can be accessed for Read/Write using GPMC PREFETCH engine > > + to improve the performance. > > Is there a reason this isn't always enabled, and the choice is just > whether to use DMA or not? Currently I had kept this for experimentation purpose. I will enable it in next patch. > I'd be interested to see performance measurements for the three cases. > > - current/slow code > - the readsw() update I sent previously > - this "prefetch engine" I will do these measurements. > And I'd expect this "prefetch engine" code would also be improved by > using readsw() and writesw() when DMA isn't in use. > > On one ARM926 I observed a 16% speedup *just* from switching from > slow byte-at-a-time PIO to using readsb ... this current code has > that same pessimization. > - Dave-- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
