196,644 users have contributed to 43,021 threads and 258,416 posts.
In the past 24 hours, we have 0 new thread(s), 0 new post(s) and 92 new user(s).
Hi Guys,
I recently spoke to a PC specalist who builds DAW's for a living, and he mentioned that for his VEP clients he tends to recommend dedicated Intel PCIE ethernet cards. According to him they lower the cpu costs and posibillities of crackling on big sessions, as they're more efficient and capable than most onboard lan connections.
Is it worth the trouble, has anyone got any experience?
Cheers,
Joe
often a separate NIC (ethernet PCIe card) is a good idea since many manufacturers are using *low budget* ethernet controllers on board.
such controllers leave much of the work (buffering, TCP checksums, ect) to the driver, in fact the operating system.
more advanced (and more expensive) controllers do this work in hardware on the PCIe card so leaving more CPU resources (and I/O operations) available to the operating system.
dual port cards (often used with servers) _can_ allow doubling the bandwidth - this is a feature called teaming which is provided by the respective driver (synchronizing the traffic from both lines).
however there are different configurations for teaming, some do only provide failover (if one line fails the second still works) and the adjecent NIC as well as the switch between both dual-NICs also has to support such a teaming - more precisely call it link aggregation.
so i would recomment to leave selecting all involved devices to a specialist or reading the respective product descriptions very attentively.
What about the built-in dual Intel I210 Ethernet controllers on the ASUS P9X79-E WS? My new computer is using this mobo. I've dedicated one port for Remote Desktop and internet sharing and the other for VEPro.
https://www.asus.com/Motherboards/P9X79E_WS/
For years I have had a problem where if I leave a session connected for a while (8+ hours) eventually I get a weird breakup that requires a restart. I thought my new mobo would fix this problem but it did not. The problem most frequently happens when I leave a session up overnight and come back to it in the morning.
the Intel I210 is one of the *better ones* (eg. supports checksum offloading, ECC error correction in buffer memory, ect).
regarding your breakup issue: go to properties of your network controller, watch out for a TAB labeled *Power Management* and uncheck the *Allow the computer to turn off this device to save power* option - this may help.
Wow, thanks! I dont' see that particular setting. I have:
- Respond to ARP requests without waking system (checked)
- Respond to NS requests without waking system (checked)
- Energy Efficient Ethernet (checked)
- Reduce Link speed during standby (checked)
- Wake on Magic Packet (checked)
- Wake on Pattern Match (checked)
- Wake on Magic Packet from power off state (unchecked)
- Wake on Link Settings (uncheked)
unfortunately i'm not familiar with the details of this driver dialog.
i would try to uncheck
@mohurwitzmusic said:
- Energy Efficient Ethernet (checked)- Reduce Link speed during standby (checked)
these two apear to be related to energy saving and a wakeup might fail (eg.. the adjecent NIC doesn't properly recognize the wakeup)
hth, cm
edit: uncheck on involved NICs on both computers ...
Thanks you are very helpful! I am going to try this tonight. This has been a problem for years.
Hey CM,
Thanks for your reply.
Is there any issue with differing ip's on team bonded network ports?
I understand that VE Pro connects to one network port/address at a time; but I assume that a bonded port has a shared ip address?
Also, though I've read that bandwith saturation isn't an issue, I've seen cpu increases and spikes on heavy sessions, eventhough cpu use on the slave isn't particularly high...(but higher on the slave).
So, would linking/bonding on each machine be worth it?
Cheers,
joe
correct - the driver applies a single IP to both NICs. however each network link will still have its own MAC-address, that's why i mentioned dependency on driver and switch because the ARP (address resolution protocol) comes into the game.
one link has to send the first request package and the switch *remembers* the MAC address from which it has been sent.
now if link aggregation is not supported by all involved devices the following responses goes only to this MAC address ... the result is that again only one link is used for communication.
also correct: bandwidth saturation is usually not an issue, a single gigabit network link should be able to transport up to 350 channels (44.1 kHz 32bit), so if nothing else but a VE PRO connect has to be *shipped* across this link it isn't worth the efforts required.
What about the built-in dual Intel I210 Ethernet controllers on the ASUS P9X79-E WS? My new computer is using this mobo. I've dedicated one port for Remote Desktop and internet sharing and the other for VEPro.
https://www.asus.com/Motherboards/P9X79E_WS/
For years I have had a problem where if I leave a session connected for a while (8+ hours) eventually I get a weird breakup that requires a restart. I thought my new mobo would fix this problem but it did not. The problem most frequently happens when I leave a session up overnight and come back to it in the morning.
Hello mohurwitzmusic!
Can you please explain to me how exactly you dedicated one port for Remote Desktop and internet sharing and the other for VEPro? I dont know how to do this with my p9x79-ws. Thank you very much.
correct - the driver applies a single IP to both NICs. however each network link will still have its own MAC-address, that's why i mentioned dependency on driver and switch because the ARP (address resolution protocol) comes into the game. one link has to send the first request package and the switch *remembers* the MAC address from which it has been sent. now if link aggregation is not supported by all involved devices the following responses goes only to this MAC address ... the result is that again only one link is used for communication. also correct: bandwidth saturation is usually not an issue, a single gigabit network link should be able to transport up to 350 channels (44.1 kHz 32bit), so if nothing else but a VE PRO connect has to be *shipped* across this link it isn't worth the efforts required.True. Server teaming can send on both links, but most teaming modes can't only receive on one. For true load-balancing over nics you would need a link aggregation which you have to configure on both the server and a managed switch. The offload features on a NIC are hard to understand and can easily do more harm. Quality drivers always helps. A more serious ethernet switch also. As stated, you can do a lot of i/o over gigabit.
What about the built-in dual Intel I210 Ethernet controllers on the ASUS P9X79-E WS? My new computer is using this mobo. I've dedicated one port for Remote Desktop and internet sharing and the other for VEPro. https://www.asus.com/Motherboards/P9X79E_WS/ For years I have had a problem where if I leave a session connected for a while (8+ hours) eventually I get a weird breakup that requires a restart. I thought my new mobo would fix this problem but it did not. The problem most frequently happens when I leave a session up overnight and come back to it in the morning.Hello mohurwitzmusic! Can you please explain to me how exactly you dedicated one port for Remote Desktop and internet sharing and the other for VEPro? I dont know how to do this with my p9x79-ws. Thank you very much. The only simpel way would be cross-connecting the extra interfaces to eachother. Use a cross-cable and configure a different IP subnet on these interfaces. But also, there is no need to.
