Welcome, Guest

Author Topic: Underclocked to 912MHz...  (Read 2036 times)

aw

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Underclocked to 912MHz...
« on: April 08, 2016, 06:27:09 PM »
I'm a noob to this space, so I apologize in advance for any dumb questions, however I thought I was buying a 1GHz part with the A10-LIME, only to see these:

http://git.denx.de/?p=u-boot.git;a=commit;h=961e77712bbf14a6797bc0d725aed956cd06bce1
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/arch/arm/boot/dts/sun4i-a10-olinuxino-lime.dts?id=977104e5606a6df8fe22c0dacd3620fc00b58d61

Summary, both uboot and the Linux kernel cap the device at 912MHz due to reports of instability at 1.008GHz.

Is this a heat issue?  Can I resolve it with a small heatsink?  If I do, can I ever run a mainline kernel/uboot?

I'm kind of disappointed that this board apparently isn't capable of the rather meager specs it advertises.  Buyer beware.  Thanks

aw

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Re: Underclocked to 912MHz...
« Reply #1 on: April 08, 2016, 10:09:34 PM »
I think the way to work around this is to build a custom u-boot and dtb image with the above mentioned commits removed.  Take the new A10-OLinuXino-Lime/u-boot-sunxi-with-spl.bin image and dd it onto the boot device, just as done during install (https://fedoraproject.org/wiki/Architectures/ARM/F23/Installation#Write_U-Boot_to_Media).  For the dtb, I created /boot/dtb/ and copied my modified sun4i-a10-olinuxino-lime.dtb there.  My fdtdir is then set as /dtb/ in my extlinux.conf.  Hopefully all of that will stick across kernel upgrades.  If there are better ways to do this, please let me know.

Now to see if I have any of the claimed stability issues at 1GHz...

aw

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Re: Underclocked to 912MHz...
« Reply #2 on: April 09, 2016, 07:20:24 PM »
Well, I find that my particular board seems to run stable up to 1032MHz.  At 1056MHz the djpeg tests that were originally used to justify the downclock start to fail.  I also added a couple more frequency points on the low end, 384MHz and 168MHz at 1.1V and 1.0V respectively, though I'm not convinced the voltage specification in the dtb actually does anything.  These take my idle power draw down from 1.4W to 1.1W.  Lower than 168MHz I seem to get hangs.  Running compiles concurrent to cpufreq-ljt-stress-test to validate the stability at all these points.