Re: packet drop with intel gigabit / marwell gigabit

Gary Thorpe wrote:

[No subject in first one, sorry for repost]
...
1.6 Gb/s = system bus bandwidth. Cache won't affect this bandwidth. DDR400 has 400 MB/s: only attainable for long sequential accesses of either read or write but not a mix of both. DMA should be able to get near this limit (long and sequential, read or write only per transfer). A NIC with bus mastering DMA should be able to effectively use the memory bandwidth.


This is not such simple thing and it is hard to explain in email.
Two things to study:
(1)
DMA does not directly affect memory bandwidth. It directly sticks with I/O bandwidth,
which is indirectly occupy the memory bandwidth. Slower I/O bus takes more memory
bandwidth. DMA burst size also affects bandwidth efficiency. Smaller size is good for CPU,
while large burst size is good for I/O. Does this make sense?
(2)
Try to analyze Intel 860 chipset -- 82806 AA PCI Hub (P64H) and 82860
memory controller Hub (MCH). This is well know problem, so you may find
discussion on the Internet.
This chipset had only 90 MB/s system bandwidth. When changing DT registers setting in
P64H (see below), you may get 105MB/s in average (or 117 MB/s maximum) bandwidth
by setting
Dev31, Fun0, Reg50h[2] = 1
Dev31, Fun0, Reg80h[1:0]=2

It uses DDR400. Why is its system bandwidth not 1.6GB/s, not even 0.16GB,
but 90MB/s? or 117 MB/s after modifying register setting? (This has nothing
to do with cache or CPU. Any speed CPU has the similar system bandwidth
when used with motherboard having such chipset.)

(That is)
The system bandwidth is not equal to memory speed or bus_speed times bus_width
in the real world.
(Done)

-Jin


_______________________________________________
freebsd-performance@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-performance
To unsubscribe, send any mail to "freebsd-performance-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: Macs vs Dell, again from the AP.
    ... bothering to supply enough memory or memory bandwidth. ... GB/sec. ... PCI Express graphics slot, ...
    (comp.sys.mac.advocacy)
  • Re: Macs vs Dell, again from the AP.
    ... memory and 16 TB of virtual memory." ... bothering to supply enough memory or memory bandwidth. ... GB/sec. ... That 16% faster is based on 533 MHz memory in the Dell. ...
    (comp.sys.mac.advocacy)
  • Re: GeForce 6800 Go bus bandwith problem?
    ... I also thought about the memory bandwidth (though admittedly only after ... but AFAICS it uploads the same texture in every pass so various caches ... For a long time, and in many cases even now, the limit is the driver ...
    (comp.graphics.api.opengl)
  • Re: GeForce 6800 Go bus bandwith problem?
    ... Our code does not use PBOs, but if time permits we'll change that. ... I also thought about the memory bandwidth (though admittedly only after ... but AFAICS it uploads the same texture in every pass so various caches ...
    (comp.graphics.api.opengl)
  • Re: 4-way Opteron vs. Xeon-IBM X3 architecture
    ... >> is how you get to 128GB and AMD's specs talk of number of ranks of memory. ... >> clock speed has to be reduced with an obvious loss in bandwidth... ... they won't have as many problems with FSB speeds. ... most of the DB2 submissions are for AIX ...
    (comp.sys.ibm.pc.hardware.chips)