Note the packaging does NOT claim Linux compatibility. I hard to surf on the web to find a case of a user who had used this card successfully with Linux, and eventually I did find such a documented case, and hence only then (after reading it worked with Linux) did I feel comfortable in ordering this card.
I got around to installing the card in my Intel Core i920 PC (w/Asus P6T Deluxe V2 motherboard) after lunch today.
I then booted to a 64-bit openSUSE-11.3 w/KDE-4.4.4 and the nominal 2.6.34.7-0.7-default kernel.
/sbin/lspci -nnk #gives:
03:00.0 USB Controller [0c03]: NEC Corporation uPD720200 USB 3.0 Host Controller [1033:0194] (rev 03)
Subsystem: ASUSTeK Computer Inc. Device [1043:8413]
Kernel driver in use: xhci_hcd
I plugged in my External hard drive (that supports USB-3.0) and it was recognized:
Bus 001 Device 002: ID 0409:005a NEC Corp. HighSpeed Hub
Bus 002 Device 003: ID 0409:005a NEC Corp. HighSpeed Hub
Bus 009 Device 002: ID 152d:0539 JMicron Technology Corp. / JMicron USA Technology Corp.
I copied a 7GB file to the external drive from my PC. It was fast. Much faster than the Firewire 800 I have, and of course faster than the older USB-2.0.
The USB 3.0 controller gets well recognized, just like in your case above. However, when I attach my Kingston HyperX Max USB 3.0 Flash drive to it, the drive only works in USB 2.0 speed. This is shown by a green light on the drive (as opposed to a blue light for USB 3.0), and also tested by copying speed (I get about 33 MB/s, so USB 2.0 type speed). I have a double boot system, and under Windows XP SP3 the same drive works fine with USB 3.0 (also blue light shown), so there is no hardware error.
**
So my question is: Is there anything I could do to make this drive work with USB 3.0 (instead of USB 2.0) under SuSE Linux 11.3?**
Can you test the transfer speed with another operating system (such as an MS-Windows variant) to confirm that this is not a hardware incompatibility between the two USB-3.0 devices ?
Tested, with copying exactly the same file from and to exactly the same place. Then I got with Windows XP SP3 about 75 MB/s (and the drive lamp shows a blue light, as it should for USB 3.0), instead of 33 MB/s with SuSE 11.3 (and the drive lamp shows a green light, as it does for USB 2.0). The 75 MB/s is not the full theoretical speed of USB 3.0 (which may have to do with my hard disk from which the file comes being quite slow), but anyway clearly above USB 2.0, and also in the range of other real-world tests (e.g., Kingston HyperX Max 3.0 128GB Drive Review | KitGuru).
So I guess my question is still the same as above…
Hmmm … It might be worth testing on openSUSE-11.4 RC2 and then writing a bug report if appropriate speeds are not reached. Could it be different Linux file manager programs yield different speeds?
polarhki, I HIGHLY suggest you consider a kernel upgrade to 2.6.37. USB 3 drives have issues when using the default 2.6.34 that came with openSUSE 11.3. I have a bash script file that can make the process very painless. Do consider that if you install an external AMD or nVIDIA video driver, it must be reloaded if you change your kernel version. Here is the link:
Message #17 has the most recent version of SAKC. There is a companion script file I called KLIST in the same message thread at #22 that can help in managing your kernel source file compiles. Good luck.
Here comes the relevant boot messages (I think):
1.489033] usb 2-1: new high speed USB device using ehci_hcd and address 2
1.604250] usb 2-1: New USB device found, idVendor=05e3, idProduct=0608
1.604359] usb 2-1: New USB device strings: Mfr=0, Product=1, SerialNumber=0
1.604465] usb 2-1: Product: USB2.0 Hub
1.604905] hub 2-1:1.0: USB hub found
1.605249] hub 2-1:1.0: 4 ports detected
1.708137] usb 2-4: new high speed USB device using ehci_hcd and address 3
I’m not quite sure what it means though, but anyway the result is that the USB 3.0 flash drive is only detected as USB 2.0.
Fast was subjective. I could notice the difference.
Just how reliable are benchmarks? Over the years I’ve seen so many claim benchmarks are good/gospel MUST to be believed, only to find out months or years later that the bench marks were not very good, and so I confess I am very skeptical now.
But since you asked with “hdparm -t /dev/sdb” , with my USB-3.0 drive (identified as sdb) connected to a USB-2.0 controller (and a very short cable), using its USB2.0 compat mode I note 102 MB in 3.02 seconds = 33.82 MB/sec
With my USB-3.0 device connected to the USB-3.0 controller (and a much longer cable), I note 118 MB in 3.05 seconds = 38.70 MB/sec.
The same test for Firewire-800 on a different drive gave 39.55 MB/sec (and my subjective feel was the USB-3.0 transfers were faster than the Firewire-800) so I’m skeptical as to the results of that method of measuring speed.
But I don’t put much faith in those bench mark tests. Subjectively I note large file transfers are significantly shorter in time. And I would prefer to do my own timing test if I thought it suitable to do such a test (ie copy to/from the external drive a large file, say 5 GB in size).
On the product page of the DeLock PCI-E USB 3.0 card (see link in my earlier message above) is states that the card would support Linux with Kernel 2.6.31 or higher. As I wrote before, the USB 3.0 controller gets well recognized, just like in your case above. I was under the (maybe wrong?) impression that the communication between the USB 3.0 controller and the USB 3.0 device would be independent of the operating system, that is, if the USB controller (i.e., the PCI-E USB 3.0 card) is properly recognized then it would work with any USB 3.0 device…
It should be independant, but it depends (IMHO) on the quality of the operating system driver, and also depends (IMHO) on how close the USB-3.0 hardware on both ends are to meeting the USB-3.0 specification.
What could be a better test to check the speed of a connection to an external hard disk than copying a (large enough) file? Do you really think your subjective feeling is better than that?
The test I carried out was made with a single 30GB file, which should give good statistics of the average speed. 118 MB may be a bit too small to really get good statistics, but the speeds you get from that do sound like far below of what one should expect from USB 3.0, and in fact very close to USB 2.0 speed… So are you sure that your connection actually really works in USB 3.0 mode? (The log messages alone might not be enough of a proof…)
There are a lot of variables to consider. In my example computer one is running openSUSE 11.3 but with kernel 2.6.38-rc6 and KDE 4.4.4 while computer two is running openSUSE 11.4 and kernel 2.6.37 and KDE 4.6. It is hard to know what makes a difference or not though I would never guess the USB3 cable length would make any difference but perhaps the cable quality does. I did have issues a while back with optical cables and came to realize that while one was often hard pressed to tell a difference between cheap and expensive wired cables, one could see a real difference with crappy optical cables. Surely a longer USB3 cable would slow you down unless its quality was much better. Such issues often relate to a cable’s capacitance (often grouped with inductance and thus referred to its impedance) which tends to shunt out higher frequencies or its resistance which reduces the overall signal strength.
Thank you for this advice. I am unfortunately anything but a Linux shaman, and have never compiled any kernel myself. So it sounds a bit scary… Since I mostly use the external USB 3.0 hard disk from the Windows side anyway (where it works without problem), and as it anyway works under SuSE Linux, even though only in USB 2.0 speed, I might rather wait for SuSE Linux 11.4…
I read in another message that installing the 2.6.37 kernel to SuSE Linux 11.3 drastically reduced the speed, almost by a factor of two. Do you have any comments on that?
Thank you for this advice. I am unfortunately anything but a Linux shaman, and have never compiled any kernel myself. So it sounds a bit scary… Since I mostly use the external USB 3.0 hard disk from the Windows side anyway (where it works without problem), and as it anyway works under SuSE Linux, even though only in USB 2.0 speed, I might rather wait for SuSE Linux 11.4…
I read in another message that installing the 2.6.37 kernel to SuSE Linux 11.3 drastically reduced the speed, almost by a factor of two. Do you have any comments on that?
I have not seen any slow downs using kernel 2.6.37 AND it is the basis for the new openSUSE 11.4. As you can see from my previous message’s there is a lot more to the speed measurement than raw speed. What was the desktop used, exact hard disk hardware and how did one determine that a second setup was slower? One thing is for sure and that is newer is not always better. You need some reason to attempt a kernel update over the default. I will say however, that using the SAKC bash script is very painless. It maintains your old kernel which you can drop back to. If you do load the nVIDIA or AMD proprietary video drivers, these must be reloaded if you update your kernel version and there is no automatic way to remove a kernel that you compiled. You can manually remove an unwanted kernel and the latter is only an issue if you are very low on disk space as having more than one kernel to use can be a good thing when problems come up in one you can switch to the other.
Looking more closely to the drive light when SuSE Linux 11.3 boots, it shows blinking green (USB2.0), and then shortly blue (USB3.0), and this repeats until I log in to an account. Then the light remains green (USB2.0). So somehow it seems that SuSE Linux 11.3 tries to connect in USB3.0 mode, but does not succeed.
Also I found the boot messages when SuSE Linux 11.3 tries to connect to the 128GB USB 3.0 flash drive. Looks like some warnings/ errors appear:
<6> 17.581100] usb 8-4: new high speed USB device using xhci_hcd and address 0
<4> 17.657892] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<4> 17.658256] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<4> 17.658631] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<4> 17.659132] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<6> 17.659272] usb 8-4: New USB device found, idVendor=11b0, idProduct=6398
<6> 17.659377] usb 8-4: New USB device strings: Mfr=3, Product=4, SerialNumber=5
<6> 17.659470] usb 8-4: Product: HyperX Max 3.0
<6> 17.659561] usb 8-4: Manufacturer: Kingston
<6> 17.659651] usb 8-4: SerialNumber: 0026182C72650AC10FFF0080
<4> 17.660756] xhci_hcd 0000:04:00.0: WARN: transfer error on endpoint
<4> 17.661382] xhci_hcd 0000:04:00.0: WARN: transfer error on endpoint
<6> 17.662068] scsi8 : usb-storage 8-4:1.0
<6> 17.662286] usb 8-4: USB disconnect, address 2
<6> 17.998096] usb 8-4: new high speed USB device using xhci_hcd and address 0
<4> 18.075469] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<4> 18.075967] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<4> 18.076467] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<4> 18.077093] xhci_hcd 0000:04:00.0: WARN: short transfer on control ep
<6> 18.077230] usb 8-4: New USB device found, idVendor=11b0, idProduct=6398
<6> 18.077348] usb 8-4: New USB device strings: Mfr=3, Product=4, SerialNumber=5
<6> 18.077466] usb 8-4: Product: HyperX Max 3.0
<6> 18.077553] usb 8-4: Manufacturer: Kingston
<6> 18.077640] usb 8-4: SerialNumber: 0026182C72650AC10FFF0080
<