**** BEGIN LOGGING AT Tue Nov 17 02:59:58 2015 Nov 17 03:52:45 so make image to 16 G SD card expand image boot use 2T USB drive for SVN "storage" I suppose adding FTP would work too ... I wonder if I could toss a media server on it even. Nov 17 03:52:57 Oh well sleep time for me Nov 17 12:59:36 Those 2 scripts in the scripts folder on the bbb. Do I need to run them on my host in order to connect? Nov 17 12:59:48 The getting started page doesn't seem to work for me Nov 17 13:59:19 does kerenl 4.4 offer any new features for us? Nov 17 17:39:28 I've updated to 4.2.6 & I'm seeing lots more usb load (pulseaudio with loopback to a usb audio stick is taking up 2x as much cpu) Nov 17 17:39:31 any ideas? Nov 17 17:40:50 was running 3.19.8 previously Nov 17 18:45:44 bwar Nov 17 19:57:07 Hey everyone. Need some help with PRUs. How does one send/receive data from PRUs using C program? Nov 17 19:58:18 Supposing, I have an array created in my C program which i want to send to the PRUs to do some fast calculations. Nov 17 19:58:35 Is it possible to achieve this? Nov 17 20:03:08 PRU for fast calculations? ehh... what? Nov 17 20:05:01 the cortex-a8 (especially the Neon unit) should be a lot faster for doing math of any kind Nov 17 20:07:09 there is a MAC... Nov 17 20:07:32 ds2: with fixed registers, so you need to add the cost of moving data into / out of those Nov 17 20:07:58 ds2: Neon can do two 32×32->64 bit multiply-accumulates per two cycles Nov 17 20:09:01 (and data movement in parallel) Nov 17 20:10:24 the MAC is good to have in PRU, but offloading a computation from the A8 to PRU still makes little sense Nov 17 20:15:33 I'd look at it differently Nov 17 20:15:53 if the data is sourced on the PRU and doing the math will reduce the size of the xfer AND the NEON is busy... Nov 17 20:16:45 that's a lot of "if"s, and doesn't relate at all to the original question Nov 17 20:18:07 I read it differently Nov 17 20:19:09 i wonder if the PRU's registers can be exploited for the buterflying in an FFT Nov 17 20:19:23 bbl Nov 17 22:27:18 rcn-ee: Should there be install-me.sh scripts on your site, or have you moved away from that method of upgrading kernels completely? I'm not seeing any. Nov 17 22:27:45 MathOnNapkins, everything is builtin: sudo apt-get install linux-image- Nov 17 22:28:03 then an initramfs script updates the boot location Nov 17 22:28:24 I have been using that method, I was just wondering if the scripts were completely out. Nov 17 22:29:18 yeap, the install-me's are dead.. things like headers and linked modules are way easier when they are all in the repo. ;) Nov 17 22:30:30 I hate to bother you too much, but I saw a post from you describing that the MLO was contained in the MBR these days Nov 17 22:30:48 Can you explain how exactly that works and how to poke around and find that? Nov 17 22:31:14 * MathOnNapkins goes fishing around to find that post... Nov 17 22:31:47 MathOnNapkins, this was a new bootrom featured in the dm81xx family (post omap37xx), called mmc raw boot in the datasheet... Nov 17 22:31:55 "No, that's incorrect, they are stored in the master boot record (MBR), 1MB hole at the front of the drive mmcbklX." Nov 17 22:32:18 the push came as users started deleting "mlo/u-boot.img" files in the fat partition of the bbb and wondered why it stopped booting.. Nov 17 22:32:27 so we moved it out of their way. ;) Nov 17 22:33:10 so older devices beagle classic/xm can't boot this new way... Nov 17 22:35:09 So if I wanted to 'find' this file, how would I do so? I apologize as I am not a whiz with partitions. Nov 17 22:35:38 I take it it's not mountable, but is actually in the front of the MMC or SD card? (e.g. some number of sectors at the front) Nov 17 22:38:50 MathOnNapkins, http://www.ti.com/lit/ug/spruh73l/spruh73l.pdf page 4939/4940 Nov 17 22:39:19 you'll notice raw mode is before file mode, so it's a little faster too. ;) Nov 17 22:39:34 Thanks much, sir. Nov 17 22:39:51 The raw addresss are: https://eewiki.net/display/linuxonarm/BeagleBone+Black#BeagleBoneBlack-SetupmicroSDcard Nov 17 22:40:07 you could dd them back... Nov 17 22:41:11 Sounds like a plan. Thanks again! Nov 17 23:50:40 so it looks like my issue (new kernel being slow in usb xfer) was due to turning USB DMA _on_. Apparently the musb dma is extremely cpu hungry, which is a bit supprising considering that DMA is supposed to free up the cpu to do other work. I guess the driver still needs some work. **** ENDING LOGGING AT Wed Nov 18 02:59:59 2015