[Beowulf] Help with inconsistent network performance
Patrick Geoffray
patrick at myri.com
Tue Dec 18 18:05:41 PST 2007
Hi Greg,
Greg Lindahl wrote:
> ethtool -a eth0
>
> and it says RX/TX pause are on, doesn't that mean that the switch
> supports it?
No, it just means the NIC supports it. RX means that the NIC will send
PAUSE packets if the host does not consume fast enough (rare) and TX
means that the NIC will stop sending when receiving a PAUSE packet (more
likely). It's independent of the switch flow control settings.
> My dumb Netgear 24-port 1-gig switch supports hw flow control. Sounds
> like things are a bit more difficult with low-end 10gigE ports.
For RX hardware flow-control, you need enough buffer space to keep one
full frame plus the latency on the longest wire, for every port. It is a
bit more expensive to do with 10GigE, because you need faster memory and
more of it. Some recent 10GigE chips use a shared SRAM buffer that is
not big enough for the worst case with 9K packets: it works fine as long
as a few ports are blocked, then it happily collapses and drops packets.
Flow-control is not for everyone, and that's why it is often turned off
by default. When a sender is paused, it will stop sending anything,
including packets for different destinations. Dropping packets is
expensive to recover but it keeps things moving.
Patrick
More information about the Beowulf
mailing list