[Openocd-development] What's up? Why so slow?
EK-LM3S9B90 ICDI JTAG openocd git jtag_khz 6000 (gdb) load Loading section .text, size 0x1e5f8 lma 0x0 Loading section .data, size 0xe0 lma 0x1e5f8 Start address 0xc63d, load size 124632 Transfer rate: 3 KB/sec, 13848 bytes/write. Seriously, in 2011? //Peter ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
USB roundtrips? On Sun, Oct 9, 2011 at 3:40 AM, Peter Stuge wrote: > EK-LM3S9B90 ICDI JTAG openocd git jtag_khz 6000 > > (gdb) load > Loading section .text, size 0x1e5f8 lma 0x0 > Loading section .data, size 0xe0 lma 0x1e5f8 > Start address 0xc63d, load size 124632 > Transfer rate: 3 KB/sec, 13848 bytes/write. > > > Seriously, in 2011? > > > //Peter > ___ > Openocd-development mailing list > Openocd-development@lists.berlios.de > https://lists.berlios.de/mailman/listinfo/openocd-development > -- Øyvind Harboe - Can Zylin Consulting help on your project? US toll free 1-866-980-3434 http://www.zylin.com/ ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
(At least) in SWD mode with an FTDI programmer openocd is sending data bit by bit. I am planning to send in a patch to fix that, but I am currently unable to test it, because I am waiting for my SWD-capable programmer. Regards, Ákos Vandra On 9 October 2011 08:21, Øyvind Harboe wrote: > USB roundtrips? > > On Sun, Oct 9, 2011 at 3:40 AM, Peter Stuge wrote: >> EK-LM3S9B90 ICDI JTAG openocd git jtag_khz 6000 >> >> (gdb) load >> Loading section .text, size 0x1e5f8 lma 0x0 >> Loading section .data, size 0xe0 lma 0x1e5f8 >> Start address 0xc63d, load size 124632 >> Transfer rate: 3 KB/sec, 13848 bytes/write. >> >> >> Seriously, in 2011? >> >> >> //Peter >> ___ >> Openocd-development mailing list >> Openocd-development@lists.berlios.de >> https://lists.berlios.de/mailman/listinfo/openocd-development >> > > > > -- > Øyvind Harboe - Can Zylin Consulting help on your project? > US toll free 1-866-980-3434 > http://www.zylin.com/ > ___ > Openocd-development mailing list > Openocd-development@lists.berlios.de > https://lists.berlios.de/mailman/listinfo/openocd-development > ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
Øyvind Harboe wrote: > On Sun, Oct 9, 2011 at 3:40 AM, Peter Stuge wrote: > > Transfer rate: 3 KB/sec, 13848 bytes/write. > > USB roundtrips? Ḯ'll have a look with usbmon soon. Akos Vandra wrote: > (At least) in SWD mode with an FTDI programmer openocd is sending > data bit by bit. I am planning to send in a patch to fix that, but > I am currently unable to test it, because I am waiting for my > SWD-capable programmer. LM3S knows JTAG and SWD. ICDI is FT2232 based and knows JTAG and SWD, and it's what the late David Brownell was doing his SWD work on, but my setup is using JTAG. //Peter ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
On Sun, Oct 9, 2011 at 10:43 PM, Peter Stuge wrote: > Øyvind Harboe wrote: > > On Sun, Oct 9, 2011 at 3:40 AM, Peter Stuge wrote: > > > Transfer rate: 3 KB/sec, 13848 bytes/write. > > > > USB roundtrips? > > Ḯ'll have a look with usbmon soon. > U+0618 LATIN CAPITAL LETTER I WITH DIRT ON TOP? I tried to scrape it off my monitor, dammit... :) It sounds like the LM3S flash algorithm could use a rewrite like the one I merged for STM32F10x yesterday. Flash speed went up from 9KB/s to 29 KB/s. I don't know why "before"-speed was so low when I tested yesterday, I normally got around 14 KB/s. Maybe the bus/computer was extra heavily loaded. I've never seen it as far down as your figures though. Almost as if the LM3S driver lacks block write completely. Examining the timestamps in the full debug log is a good complement to usbmon when debugging performance problems, btw. > Akos Vandra wrote: > > (At least) in SWD mode with an FTDI programmer openocd is sending > > data bit by bit. I am planning to send in a patch to fix that, but > > I am currently unable to test it, because I am waiting for my > > SWD-capable programmer. > > LM3S knows JTAG and SWD. ICDI is FT2232 based and knows JTAG and SWD, > and it's what the late David Brownell was doing his SWD work on, but > my setup is using JTAG. I'm pretty sure that mainline ft2232 uses byte writes whenever possible. Akos: Is SWD even remotely up and running in mainline or are you using a fork? /Andreas ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
P.S. Is it just me, or the default reply address is the sender himself, and not the list? Regards, Ákos On 9 October 2011 23:28, Akos Vandra wrote: > I am using the libswd fork, but I don't know how functional it is as > of yet, because I don't have any hardware to test it. > I just joined the list not so long ago, as my thesis is about the SWO > trace output (both analyzing and configuring) so it is closely tied to > SWD. I'm thinking about using openocd to send the SWD commands, and > maybe integrating this support into openocd later, if it works out, > and you guys are interested in it. > > Regards, > Ákos Vandra > > On 9 October 2011 23:13, Andreas Fritiofson > wrote: >> >> On Sun, Oct 9, 2011 at 10:43 PM, Peter Stuge wrote: >>> >>> Øyvind Harboe wrote: >>> > On Sun, Oct 9, 2011 at 3:40 AM, Peter Stuge wrote: >>> > > Transfer rate: 3 KB/sec, 13848 bytes/write. >>> > >>> > USB roundtrips? >>> >>> Ḯ'll have a look with usbmon soon. >> >> >> U+0618 LATIN CAPITAL LETTER I WITH DIRT ON TOP? I tried to scrape it off my >> monitor, dammit... :) >> It sounds like the LM3S flash algorithm could use a rewrite like the one I >> merged for STM32F10x yesterday. Flash speed went up from 9KB/s to 29 KB/s. >> I don't know why "before"-speed was so low when I tested yesterday, I >> normally got around 14 KB/s. Maybe the bus/computer was extra heavily >> loaded. I've never seen it as far down as your figures though. Almost as if >> the LM3S driver lacks block write completely. >> Examining the timestamps in the full debug log is a good complement to >> usbmon when debugging performance problems, btw. >> >>> >>> Akos Vandra wrote: >>> > (At least) in SWD mode with an FTDI programmer openocd is sending >>> > data bit by bit. I am planning to send in a patch to fix that, but >>> > I am currently unable to test it, because I am waiting for my >>> > SWD-capable programmer. >>> >>> LM3S knows JTAG and SWD. ICDI is FT2232 based and knows JTAG and SWD, >>> and it's what the late David Brownell was doing his SWD work on, but >>> my setup is using JTAG. >> >> I'm pretty sure that mainline ft2232 uses byte writes whenever possible. >> Akos: Is SWD even remotely up and running in mainline or are you using a >> fork? >> /Andreas >> >> ___ >> Openocd-development mailing list >> Openocd-development@lists.berlios.de >> https://lists.berlios.de/mailman/listinfo/openocd-development >> >> > ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
Akos Vandra wrote: > default reply address is the sender himself, and not the list? Please see http://woozle.org/~neale/papers/reply-to-still-harmful.html Mailing lists are a third form of email communication pattern. (The first two are person-to-person, and person-to-many-persons.) Using an email software which does not have explicit support for mailing lists makes life with lists very inconvenient. :\ //Peter ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] What's up? Why so slow?
On Sun, Oct 9, 2011 at 11:28 PM, Akos Vandra wrote: > P.S. Is it just me, or the default reply address is the sender > himself, and not the list? > The list server doesn't set Reply-To to the list address, for some very good reason which I have forgotten now. Easiest fix is to set reply-to-all as the default reply action in your client. /Andreas ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development