Hi Olliver, Unfortunately, I have a problem to see and use quoted text so I delete it.
I saw broken-hpi option but currently I use mainly kernel 4.5.3 so not sure if it is available there. About 8-bit staff in my patch it is there by error. I really use 4-bit bus without additional pins definitions. About better results measured by you it is definitely because of larger buffers. If you want to get a real figures you have to measure performance for a set of buffer sizes like I do with iozone tool: root@egpr:~# uname -a Linux egpr 4.6.0-sunxi #1 SMP Tue May 10 20:50:21 EEST 2016 armv7l GNU/Linux root@egpr:/mnt# iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -i 0 -i 1 -i 2 Iozone: Performance Test of File I/O Version $Revision: 3.429 $ Compiled for 32 bit mode. Build: linux Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins Al Slater, Scott Rhine, Mike Wisner, Ken Goss Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR, Randy Dunlap, Mark Montague, Dan Million, Gavin Brebner , Jean-Marc Zucconi, Jeff Blomberg, Benny Halevy, Dave Boone, Erik Habbinga, Kris Strecker, Walter Wong, Joshua Root, Fabrice Bacchella, Zhenghua Xue, Qin Li, Darren Sawyer, Vangel Bojaxhi, Ben England, Vikentsi Lapa. Run began: Tue May 10 23:28:08 2016 Include fsync in write timing O_DIRECT feature enabled Auto Mode File size set to 102400 kB Record Size 4 kB Record Size 16 kB Record Size 512 kB Record Size 1024 kB Record Size 16384 kB Command line used: iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -i 0 -i 1 -i 2 Output is in kBytes/sec Time Resolution = 0.000001 seconds. Processor cache size set to 1024 kBytes. Processor cache line size set to 32 bytes. File stride size set to 17 * record size. random random kB reclen write rewrite read reread read write 102400 4 3653 4003 12637 12521 9501 3997 102400 16 7277 7531 16300 16413 14910 7392 102400 512 11733 11739 39496 39444 39254 11718 102400 1024 11793 11843 40015 40086 39951 11887 102400 16384 11977 11895 40680 40781 40656 11952 random random kB reclen write rewrite read reread read write 102400 4 3768 4008 12458 12432 9565 3895 102400 16 7231 7526 16488 16634 15085 7432 102400 512 11608 11700 38424 38358 38172 11655 102400 1024 11841 11853 39543 39736 39573 10977 102400 16384 11923 11950 40756 40823 40824 11994 iozone test complete. It is strange that read speed of ~40 MB/s is reached at 512+ kB buffers but write speed of 17MB/s (as you mention) is never reached at my tests. For reference iozone test gives eMMC speeds at kernel 4.5.3: random random kB reclen write rewrite read reread read write 102400 4 3444 3677 9663 9626 7801 3577 102400 16 6290 6514 13084 12740 11873 6409 102400 512 11623 11653 21592 21567 21485 10617 102400 1024 11787 11859 21862 21907 21811 11805 102400 16384 11944 11914 22131 22125 22147 11947 4GB SD card class 10 speeds at kernel 4.6.0 are: random random kB reclen write rewrite read reread read write 102400 4 630 736 5242 5149 4683 171 102400 16 1140 1507 11630 11651 11517 331 102400 512 6247 6465 21572 20748 21567 945 102400 1024 6297 6521 20707 21943 21901 1945 102400 16384 6363 6416 22209 21001 22208 6459 4GB SD card class 10 speeds at kernel 4.5.3 are: random random kB reclen write rewrite read reread read write 102400 4 222 227 5790 5821 4180 203 102400 16 5681 6149 9470 9661 8340 62 102400 512 12033 11656 21194 21194 20982 2199 102400 1024 12548 11792 21606 21616 20760 4737 102400 16384 12755 11831 21929 21964 21214 10531 Strange is that SD card write speed is better at kernel 4.5.3 than at kernel 4.6.0 while read speed is almost the same. About using your patch in kernel 4.6 at Armbian build I did not see any errors at patching phase. Maybe I have to try patched with my patches u-boot and kernel 4.6 patched with your patch. But I am not sure that u-boot has to do something for kernel to work with eMMC. I am ready to test if you have some ideas because your patch is probably better than one used by me at the moment. Best regards Chris -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to linux-sunxi+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.