[opensuse] Installed 3Com 3c595 rate dropped from 11.59MB/s to 3.27MB/s -- howto fix?
I'll post here as well where the smart folks are (original post on arch list): Guys, This is a new one on me. I have been using the onboard LAN (RTL-8139/8139C/8139C+ (rev 10)) for several years and always got great transfer rates: 01:0b.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10) Subsystem: Giga-byte Technology GA-7VM400M/7VT600 Motherboard Flags: bus master, medium devsel, latency 32, IRQ 16 I/O ports at 8000 [size=256] Memory at e8000000 (32-bit, non-prefetchable) [size=256] Capabilities: [50] Power Management version 2 Kernel driver in use: 8139too Kernel modules: 8139too, 8139cp 10:02 providence:~/arch/pkg/new-110718/virtualbox_bin> rsync -uav --progress virtualbox_bin-4.0.12-1-i686.pkg.tar.xz phoenix:/srv/http/dl/arch/i686 sending incremental file list virtualbox_bin-4.0.12-1-i686.pkg.tar.xz 55685508 100% 11.59MB/s 0:00:04 (xfer#1, to-check=0/1) sent 55692427 bytes received 31 bytes 10125901.45 bytes/sec total size is 55685508 speedup is 1.00 I was having intermittent lockups during large file transfers (memtest is fine), so I thought I would change the NIC. I installed a 3Com Corporation 3c595 100BaseTX [Vortex] card and tested the throughput. The throughput is 1/3 what I was getting with the Realtek card: 01:07.0 Ethernet controller: 3Com Corporation 3c595 100BaseTX [Vortex] Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx- Status: Cap- 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx- Latency: 248 (750ns min, 2000ns max) Interrupt: pin A routed to IRQ 19 Region 0: I/O ports at 8000 [size=32] [virtual] Expansion ROM at 80000000 [disabled] [size=64K] Kernel driver in use: 3c59x Kernel modules: 3c59x 12:47 providence:~/arch/pkg/new-110718/virtualbox_bin> rsync -uav --progress virtualbox_bin-4.0.12-1-i686.pkg.tar.xz phoenix:~/tmp sending incremental file list virtualbox_bin-4.0.12-1-i686.pkg.tar.xz 55685508 100% 3.27MB/s 0:00:16 (xfer#1, to-check=0/1) sent 55692427 bytes received 31 bytes 3182426.17 bytes/sec total size is 55685508 speedup is 1.00 I searched the forums and I've tried ethtool in setting speed, duplex and autoneg, but it makes no difference. i.e.: ethtool -s eth0 speed 100 duplex full autoneg off What else can I try to get this card working correctly? All light indications on the card and the switch it is connected to show a 100TX connection is being made. What say the gurus? -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2011/07/26 13:20 (GMT-0500) David C. Rankin composed:
What else can I try to get this card working correctly? All light indications on the card and the switch it is connected to show a 100TX connection is being made. What say the gurus?
I'm no guru, but I do have a suggestion: don't try. I don't remember ever having success with a 10/100 3Com PCI (unlike onboard) NIC, while Realtek 8139 and Intel e100 have always just worked. Especially in Windows, matching correct driver to any 3Com PCI NIC seems always an exercise in futility, as if they intentionally misidentify themselves so as to work only for the OEMs who originally bought them and know particular magic incantations to make them work. It wouldn't surprise me if misidentification makes them fall back to a lowest common denominator configuration in Linux. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 7/26/2011 12:11 PM, Felix Miata wrote:
On 2011/07/26 13:20 (GMT-0500) David C. Rankin composed:
What else can I try to get this card working correctly? All light indications on the card and the switch it is connected to show a 100TX connection is being made. What say the gurus?
I'm no guru, but I do have a suggestion: don't try. I don't remember ever having success with a 10/100 3Com PCI (unlike onboard) NIC, while Realtek 8139 and Intel e100 have always just worked. Especially in Windows, matching correct driver to any 3Com PCI NIC seems always an exercise in futility, as if they intentionally misidentify themselves so as to work only for the OEMs who originally bought them and know particular magic incantations to make them work. It wouldn't surprise me if misidentification makes them fall back to a lowest common denominator configuration in Linux.
I concur. Run away from the entire 3Com line but especially the 3c5xx line like your hair is on fire. I've had so many problems with this nic over the years. If you don't find a way to lock it to 100 or 1000 and duplex mode at each boot you will find them doing their own thing in no time flat, and once they retreat to slower speeds they never step up till you power them off. I finally just trashed them all. Almost any other chipset works better. 10/100 nics are a dime a dozen these days and even 10/100/1000 cost less than David has spent in time trying to make it work. Don't waste an hour trying to avoid paying 7 bucks. -- _____________________________________ ---This space for rent--- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 07/28/2011 01:42 PM, John Andersen wrote:
On 7/26/2011 12:11 PM, Felix Miata wrote:
On 2011/07/26 13:20 (GMT-0500) David C. Rankin composed:
What else can I try to get this card working correctly? All light indications on the card and the switch it is connected to show a 100TX connection is being made. What say the gurus?
I'm no guru, but I do have a suggestion: don't try. I don't remember ever having success with a 10/100 3Com PCI (unlike onboard) NIC, while Realtek 8139 and Intel e100 have always just worked. Especially in Windows, matching correct driver to any 3Com PCI NIC seems always an exercise in futility, as if they intentionally misidentify themselves so as to work only for the OEMs who originally bought them and know particular magic incantations to make them work. It wouldn't surprise me if misidentification makes them fall back to a lowest common denominator configuration in Linux.
I concur. Run away from the entire 3Com line but especially the 3c5xx line like your hair is on fire. I've had so many problems with this nic over the years. If you don't find a way to lock it to 100 or 1000 and duplex mode at each boot you will find them doing their own thing in no time flat, and once they retreat to slower speeds they never step up till you power them off. I finally just trashed them all.
Almost any other chipset works better. 10/100 nics are a dime a dozen these days and even 10/100/1000 cost less than David has spent in time trying to make it work. Don't waste an hour trying to avoid paying 7 bucks.
Bummer! I bought 2 a year or so ago for $8.99. Should have known there would be problems :) On the plus side - no lockups since the NIC change. Looks like I'm shopping for another NIC. Thanks for the help! -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 07/28/2011 01:51 PM, David C. Rankin wrote:
Bummer! I bought 2 a year or so ago for $8.99. Should have known there would be problems :) On the plus side - no lockups since the NIC change. Looks like I'm shopping for another NIC. Thanks for the help!
We are going to try a NETGEAR GA311. Any negative? -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2011/07/28 16:42 (GMT-0500) David C. Rankin composed:
We are going to try a NETGEAR GA311. Any negative?
http://forum1.netgear.com/showthread.php?p=68776 -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Tue, Jul 26, 2011 at 2:20 PM, David C. Rankin <drankinatty@suddenlinkmail.com> wrote:
I'll post here as well where the smart folks are (original post on arch list):
Guys,
This is a new one on me. I have been using the onboard LAN (RTL-8139/8139C/8139C+ (rev 10)) for several years and always got great transfer rates:
01:0b.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10) Subsystem: Giga-byte Technology GA-7VM400M/7VT600 Motherboard Flags: bus master, medium devsel, latency 32, IRQ 16 I/O ports at 8000 [size=256] Memory at e8000000 (32-bit, non-prefetchable) [size=256] Capabilities: [50] Power Management version 2 Kernel driver in use: 8139too Kernel modules: 8139too, 8139cp
10:02 providence:~/arch/pkg/new-110718/virtualbox_bin> rsync -uav --progress virtualbox_bin-4.0.12-1-i686.pkg.tar.xz phoenix:/srv/http/dl/arch/i686 sending incremental file list virtualbox_bin-4.0.12-1-i686.pkg.tar.xz 55685508 100% 11.59MB/s 0:00:04 (xfer#1, to-check=0/1)
sent 55692427 bytes received 31 bytes 10125901.45 bytes/sec total size is 55685508 speedup is 1.00
I was having intermittent lockups during large file transfers (memtest is fine), so I thought I would change the NIC. I installed a 3Com Corporation 3c595 100BaseTX [Vortex] card and tested the throughput. The throughput is 1/3 what I was getting with the Realtek card:
01:07.0 Ethernet controller: 3Com Corporation 3c595 100BaseTX [Vortex] Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx- Status: Cap- 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx- Latency: 248 (750ns min, 2000ns max) Interrupt: pin A routed to IRQ 19 Region 0: I/O ports at 8000 [size=32] [virtual] Expansion ROM at 80000000 [disabled] [size=64K] Kernel driver in use: 3c59x Kernel modules: 3c59x
12:47 providence:~/arch/pkg/new-110718/virtualbox_bin> rsync -uav --progress virtualbox_bin-4.0.12-1-i686.pkg.tar.xz phoenix:~/tmp sending incremental file list virtualbox_bin-4.0.12-1-i686.pkg.tar.xz 55685508 100% 3.27MB/s 0:00:16 (xfer#1, to-check=0/1)
sent 55692427 bytes received 31 bytes 3182426.17 bytes/sec total size is 55685508 speedup is 1.00
I searched the forums and I've tried ethtool in setting speed, duplex and autoneg, but it makes no difference. i.e.:
ethtool -s eth0 speed 100 duplex full autoneg off
What else can I try to get this card working correctly? All light indications on the card and the switch it is connected to show a 100TX connection is being made. What say the gurus?
-- David C. Rankin, J.D.,P.E.
David, Just food for thought. About a year ago I started seeing speed issues on my office network. I did the usual troubleshooting like you are. I could not figure out the issue. Finally as a troubleshooting effort a put a small dedicated 5-port 1 GB switch between two of the machines that had speed issues. My speed issue went away between those 2 machines. Turned out I had a failing switch. And somehow it was confusing even other switches on the LAN. My further troubleshooting process was to use that small 5 port as my foundation and move my other switches to it one at a time. All was good until I connected up one specific switch to the LAN. Then I lost speed. I don't know what kind network protocol commands the switches send between themselves, but somehow that one switch was polluting my entire network. I replaced that switch and all has been good since then. Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Greg Freemyer wrote:
I don't know what kind network protocol commands the switches send between themselves
Spanning Tree Protocol https://secure.wikimedia.org/wikipedia/en/wiki/Spanning_tree_protocol Switches use it to map out the best path and also avoid loops. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (5)
-
David C. Rankin -
Felix Miata -
Greg Freemyer -
James Knott -
John Andersen