jefflunt
gpu_mem not set to 128 in latest demo build
cat /boot/config.txt gpu_mem=64 arm_freq=800
again those other elf files don't exist in the latest build:
pi@raspberrypi:/boot$ ls bootcode.bin config.txt fixup_cd.dat issue.txt kernel_emergency.img start_cd.elf cmdline.txt config.txt.backup fixup.dat kernel_cutdown.img kernel.img start.elf
so we need to update config.txt:
scripted
sed -ie "s/gpu_mem.*/gpu_mem=128/g" /boot/config.txt
Is there a performance difference? In testing on the 256 board some weeks ago 64 seemed to be the best overall setting. I can try it again tonight, but I'm curious what impact there is on the speed/quality of emulation.
Haven't tested to that extent...still working on initial development of some missing features...
I never did get around to testing 11 days ago when I mentioned. However, I am going through a round of testing tonight in an effort to get everything working in preparation for a release. I will test the 64 vs. 128 split on both the 256 and 512 boards and let you know if there is any noticeable difference.
Doesn't seem to make any difference in performance. I know that the petrockblog recommended a 128/128 split, but early on I found this to be sub-optimal. With the current state of things I can't see that it actually makes a positive performance difference.
Sure, but only if we find that we need it to get performance for some games. I don't want to shorten the life of anyone's Pi, even though I know they've announced that it no longer voids your warranty if you do it via the raspi-config menu.
If we find it's necessary for some games, we'll need to open a separate case for overclocking, or just add it to a "hacking guide" as part of the project documentation so people can tweak it themselves.
-Jeff
On Sun, Jan 6, 2013 at 12:49 AM, Brandt Daniels [email protected]:
also overclocking of cpu and/or gpu is another option.
— Reply to this email directly or view it on GitHubhttps://github.com/normalocity/glint-nes/issues/27#issuecomment-11925207.
Dynamic memory splitting is now available:
http://elinux.org/RPiconfig#CMA_-_Dynamic_Memory_Split
Okay, but does it actually make any difference in performance or other important things, and how/why would we go about managing the memory at run time?