Please reply to the mailing list as well as to me. On Sat, 6 Aug 2005, Mark Rutherford wrote:
> Alan, > > Some good news firstly (well kinda), after removing EHCI_HCD I was able > to burn cd via pcmcia usb2 :) which was the same with winXP.... > > Ok, so lets go through what I did today :), > > Two methods I used for testing / monitoring speeds while both OHCI and > EHCI modules loaded : > > 1. Via KDE copy the directory of 100 meg of stuff to HDD and as the GUI > is copying, it comes up averaging 2.7 - 3.1 MB/s > > 2. using cmd line: hdparm -tT /dev/scd0 running it several times > > [EMAIL PROTECTED] ~]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 996 MB in 2.01 seconds = 495.84 MB/sec > Timing buffered disk reads: 10 MB in 3.34 seconds = 2.99 MB/sec > [EMAIL PROTECTED] ~]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 984 MB in 2.00 seconds = 491.83 MB/sec > Timing buffered disk reads: 8 MB in 3.18 seconds = 2.52 MB/sec > [EMAIL PROTECTED] ~]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 1000 MB in 2.01 seconds = 498.58 MB/sec > Timing buffered disk reads: 6 MB in 3.07 seconds = 1.95 MB/sec > [EMAIL PROTECTED] ~]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 956 MB in 2.00 seconds = 477.83 MB/sec > Timing buffered disk reads: 8 MB in 3.19 seconds = 2.51 MB/sec Okay, maybe not great rates but appropriate for an optical device. You're definitely getting high-speed transfers. > You asked if I could unload the EHCI driver, so I quickly done a lsmod > to determine that module odci_hcd and ehci_hcd exist. > > I firstly rmmod ohci_hcd, remount the CD/DVD and ran the tests, get this > it ran as USB 1.1 steady unlike before I was getting 3mb/s: Wrong. ohci-hcd is the USB 1.1 driver. By removing it and leaving ehci-hcd present, you are still using high-speed transfers. > [EMAIL PROTECTED] cdrecorder]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 1020 MB in 2.00 seconds = 509.06 MB/sec > Timing buffered disk reads: 4 MB in 3.57 seconds = 1.12 MB/sec > [EMAIL PROTECTED] cdrecorder]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 1020 MB in 2.00 seconds = 508.81 MB/sec > Timing buffered disk reads: 4 MB in 3.58 seconds = 1.12 MB/sec > [EMAIL PROTECTED] cdrecorder]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 1016 MB in 2.01 seconds = 506.56 MB/sec > Timing buffered disk reads: 4 MB in 3.58 seconds = 1.12 MB/sec I don't know why the buffered rates here are lower than the previous ones. Maybe something was different about the conditions for your tests. However 1.12 MB/sec is still faster than USB 1.1's full speed, although not by much. > * > Special note*: when I unloaded the OHCI_HCD driver my usb mouse stopped > to work ??? That's because the mouse is a USB 1.1 device, so it can't be used without the ohci-hcd (the USB 1.1 driver). > So Next I tried Unloading the EHCI_HCD and reloaded OHCI_HD here is the > results of speed test: > > [EMAIL PROTECTED] media]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 992 MB in 2.00 seconds = 495.83 MB/sec > Timing buffered disk reads: 4 MB in 5.60 seconds = 731.02 kB/sec > [EMAIL PROTECTED] media]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 1000 MB in 2.00 seconds = 499.08 MB/sec > Timing buffered disk reads: 4 MB in 5.62 seconds = 728.68 kB/sec > [EMAIL PROTECTED] media]# hdparm -tT /dev/scd0 > > /dev/scd0: > Timing cached reads: 996 MB in 2.00 seconds = 497.58 MB/sec > Timing buffered disk reads: 4 MB in 5.63 seconds = 727.25 kB/sec > > > Bugger me, whats going on here the speeds are less that USB 1.1 :( No they aren't. They are exactly what a typical full-speed USB 1.1 transfer gives. Although the bus transfer rate is 12 Mb/s = 1.5 MB/s, overhead in the protocol limits bulk transfers to 19 64-byte packets per millisecond. That's 1216 B/ms = 1187.5 KB/s. Delays caused by driver processing and disc activity reduce the actual figure to 700-800 KB/s. > So I then used K3b and attempted to burn some data to a CD and it worked > while the EHCI driver was not loaded. > > I found a website for rebuilding the kernel, and my god how many > questions are there during "make config", when I last built a linux > kernel during my uni days 7 years back, I remember that I could just > press Enter to assume last known / default value, which I did except for > the debug PCI, PCMCIA and USB and USB STORAGE which I set to "Y". You're not supposed to answer all those questions yourself. As part of your Linux distribution, included somewhere with the kernel's source code, there should be a copy of the config file used for building the distribution's kernel. Just copy that file to .config and run "make oldconfig" followed by "make menuconfig". > After running the make command it appearred to build correctly, but > after attemptting "make install" it failed. Were you logged in as root at the time? > So I'm surfing the web for another install procedure / direction. > > Yeah I was thinking of contacting ENE Technology, and the PCMCIA guys, > but I don't want to until I get the detailed debug info, so If you dont' > mind me bothering you a bit more, I'd appreciate it :) > > Regards > Mark > > Alan Stern wrote: > > >Please reply to the mailing list as well as to me. You didn't follow this advice last time. Maybe you will for your next message?... We're still not going to be able to accomplish much for your writing problems without some debugging information. Alan Stern ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ [email protected] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
