**** BEGIN LOGGING AT Wed Aug 26 02:59:58 2009 Aug 26 03:00:19 I have read the instructions a few times, there are only 94 lines in the readme file :-p Aug 26 03:14:15 well you missed something Aug 26 03:39:27 necronian: http://www.plugcomputer.org/plugwiki/index.php/SheevaPlug_Installer Aug 26 03:50:57 yea, I read all that. I wouldn't even say my problem is with the sheevaplug installer Aug 26 03:51:40 it's with openocd not being able to flash uboot and uboot-env Aug 26 03:53:41 hmmmm and you have usbTTY0 and usbTTY1 active on the host machine ? Aug 26 03:54:09 and the cable is plugged in securely into the sheevaplug Aug 26 03:54:31 no I just have a usbTTY0, since about a month ago when I updated my kernel Aug 26 03:54:47 wrong!!! Aug 26 03:55:23 unload the module and then load it again Aug 26 03:56:29 necronian: http://plugcomputer.org/plugforum/index.php?topic=2.0 Aug 26 03:57:00 I think that it's udev doing it, it ignors the jtag interface and puts the serial console on ttyusb0 Aug 26 03:58:44 hmmmm what did you do? Aug 26 03:59:07 you should have 2 Aug 26 04:01:06 "/dev/ttyUSB0" and "/dev/ttyUSB1" Aug 26 04:01:56 I don't think that's the problem because of this http://plugcomputer.org/plugforum/index.php?topic=501.0 Aug 26 04:02:01 do a rmmod ftdi_sio Aug 26 04:02:20 and then modprobe blah blah Aug 26 04:03:01 see http://www.openplug.org/plugwiki/index.php/Serial_terminal_program#Getting_the_USB_serial_device_working Aug 26 04:03:18 but if nothing else works maybe I'll track down a livecd to boot off and see if that works Aug 26 04:04:49 hmmmm I am using the latest kernel on my desktop and all is fine Aug 26 04:06:34 Well I don't know if it was the kernel update or something else that got updated Aug 26 04:06:59 which distro and kernel? Aug 26 04:07:40 archlinux 2.6.30.4 Aug 26 04:08:07 hmmmm Aug 26 04:08:33 it was never a problem before, I just said that's weird and pointed minicom at ttyUSB0 instead Aug 26 04:08:39 well you have the source....did you check the config file? Aug 26 04:09:39 no, well I know that the ftdi driver is working fine Aug 26 04:10:40 because I was using the serial console and broke into uboot right before I tried running the sheevaplug installer Aug 26 04:10:43 might have something commented out that should not be Aug 26 04:11:13 my understanding the jtag is on the other port Aug 26 04:11:21 getting tired... Aug 26 04:13:03 well thanks for the help anyway Aug 26 04:14:12 he first serial device ("/dev/ttyUSB0" is supposed to be the JTAG interface; the second serial device ("/dev/ttyUSB1) is supposed to be the serial console. well good luck...when you solve the mystery you should update http://www.openplug.org/plugwiki/index.php/Serial_terminal_program#Getting_the_USB_serial_device_working Aug 26 04:18:02 necronian: I did this on my desktop http://www.openplug.org/plugwiki/index.php/Load_Serial_Drivers_Automatically_Using_udev Aug 26 04:21:55 I'm probably just going to throw ubuntu or something on another partition, something more mainstream and less bleeding edge Aug 26 04:23:22 yeah well ubuntu tneds to be trailing edge Aug 26 04:23:27 tends Aug 26 04:24:24 exactly, I'll have an older kernel and I wont need to start messing around with anything on my archlinux install Aug 26 04:24:41 I just got it set up perfectly the way I want it Aug 26 04:26:29 I am using 2.6.29.6 Aug 26 04:26:55 which is from a few days ago with the null handling fix Aug 26 07:59:40 how would a usb flash drive go for downloading torrents? Aug 26 08:00:07 one of the 16/32gb ones would suffice in size, im just wondering how quickly the memory will go bad Aug 26 13:00:45 guys ive just found some conflicting results Aug 26 13:01:06 http://computingplugs.com/index.php/SheevaPlug_Performance Aug 26 13:01:13 this page shows very poor samba performance Aug 26 13:01:33 but ive seen people in the forums claiming 25-30MBps Aug 26 13:01:42 ymmv :) Aug 26 13:01:58 more than twice as much? Aug 26 13:02:00 lol Aug 26 13:02:01 more seriously though, test along the lines of "I copied a file and got this number" is useless Aug 26 13:02:29 echosystm, what if one person was using 100Mbps ethernet, and the other 1000Mbps one? Aug 26 13:02:57 if you look on the benchmark page you'll see he's using gb ethernet Aug 26 13:03:09 bummer Aug 26 13:03:19 im not sure who to beleive Aug 26 13:03:28 but im leaning towards the forum users Aug 26 13:03:34 you'd have to test it yourself Aug 26 13:03:50 well, i dont want to buy one if its not going to work well haha Aug 26 13:04:37 :) Aug 26 13:04:57 that's kind of the reason im looking at these benchmarks in the first place Aug 26 13:08:04 are there any howtos about possible gcc optimizations? Aug 26 13:08:46 i would love to get it to transcode properly, but am i asking too much? Aug 26 13:09:09 this marvel chip doesn't have fpu, does it? Aug 26 13:09:17 got vdr working yesterday btw, that was my main goal so i'm happy Aug 26 13:09:20 yeah :( Aug 26 13:09:40 i only read about it after i ordered Aug 26 13:10:00 i guess it would have helped a lot Aug 26 13:10:16 usb-based fpga array... :) Aug 26 13:12:07 but there must be some gcc switches i could try Aug 26 13:12:18 it wasn't totally hopeless Aug 26 13:13:54 and vlc seems to be little heavier than other solutions Aug 26 14:30:03 audio transcoding would suffer from the lack of fpu Aug 26 14:30:22 but video transcoding doesn't use much floating point Aug 26 14:35:49 ukki, gcc -O2 and -O3 are supposed to mean "enable optimizations that give you best average speedup". Aug 26 14:36:36 you can eke out marginally more performance by experimenting with the detailed knobs but that won't make your sheevaplug a video transcoding monster Aug 26 14:37:11 acovea can automate the knob finding Aug 26 14:37:26 I'd set my expectations somewhere near the performance of a 300MHz Pentium2 :-) Aug 26 14:38:06 (P2 has ooo execution and MMX, giving it much better performance per clock) Aug 26 14:39:59 the kirkwood core is actually ooo Aug 26 14:54:23 ah, didn't know that Aug 26 15:09:15 ShadowJK: can you run a test and provide numbers? Aug 26 15:09:39 ShadowJK: I hate to see people giving random numbers without backing them up with lab benchmarks Aug 26 15:12:49 http://www.linleygroup.com/pdf/WP_MRVL.pdf seems to have slightly more tidbits on the microarchitecture than the marvell pr stuff Aug 26 15:15:44 the armv5te instruction set seems to include some codec acceleration instructions like saturating arithmetic .. no simd though Aug 26 15:46:23 Why don't you look at the Kirkwood functional specs? Aug 26 15:48:58 i don't remember it having any implementation details, just stuff on how to program it Aug 26 15:49:54 i'll have another look. Aug 26 15:50:38 it only has this on the core Aug 26 15:50:44 For full details and specifications about the Sheeva CPU core refer to the following documents: SheevaTM 88SV131 ARM v5TE Processor Core with MMU and L1/L2 Cache Datasheet (Doc No. MV-S104950-00) Aug 26 15:50:53 but that's apparently a secret document Aug 26 15:54:19 sounds like you did not read the document Aug 26 15:55:03 well, it's 700+ pages, pointers gladly accepted :) Aug 26 15:55:23 http://www.mail-archive.com/linux-crypto@vger.kernel.org/msg03695.html <- seems crypto engine support is coming along Aug 26 15:56:50 Now these look interesting as they have VFP http://www.marvell.com/products/communication/discovery_innovation/index.jsp Aug 26 15:57:50 searching through the functional spec for uarch terms like "branch prediction" doesn't yield anything either (besides the pr summary and reference to the secret document) Aug 26 16:01:01 your assuming it is fully indexed Aug 26 16:01:33 http://www.marvell.com/products/embedded_processors/kirkwood/index.jsp Aug 26 16:05:21 hmmm Aug 26 17:40:04 Hello Aug 26 17:55:43 hi Aug 26 22:56:51 Hello again Aug 26 22:57:53 I would like to know if someone has got unionfs/aufs or mini_fo running under debian (armel/kirkwood) ? Aug 27 02:10:21 would a usb hard drive use significantly more cpu than sata? Aug 27 02:10:55 ie. could i get better samba performance with an openrd (sata) ? **** ENDING LOGGING AT Thu Aug 27 02:59:57 2009