Re: [nslu2-linux] Re: possible to mod the slug for gigabit ethernet?
- Trevor wrote:
> --- In email@example.com, Manu Abraham <manu@k...> wrote:
>>If the Gig-E chips can be interfaced through the 34 pin PIO
> connector as
>>described, probably a Gig-E be made to interface to the 34 pin
> connector ?
> What 34 pin PIO connector? Are you talking about Rod's 34 pin
> universal connector mod in the wiki at
> http://www.nslu2-linux.org/wiki/HowTo/AddAThirtyFourPinUniversalConnectorI did not know the exact details..
> ? If so then no, all that does is bring out the internal USB client
> port, JTAG, I2C and the two serial ports.
>Work on Hardware is not that easy. Usually, would result in breaking it
> Adding GigE to the NSLU2 is basically not going to happen unless
> Linksys decide to make a variant which has it built in.
> The NPEs inside the IXP420 max out at 100Mbps so you'd have to use an
> external chip to handle it. The PCI bus is connected directly to the
> NEC USB 2.0 controller which is a BGA so tapping those lines isn't
> going to be easy. You could do a hack involving the expansion bus I
> guess which is slightly easier to tap but that'd involve a lot of work
> creating the necessary glue logic to go between it and a NIC chipset.
in the first case, trying to do modifications..
> In short, adding GigE would not be worth the effort and as noted byI was not suggesting exactly that, a hack would be required, that was
> Inge, you won't see any performance increases.
what i meant by "info".
Most of the GigE chips have a MIPS core, which would help to a certain
extent in reducing the overhead..
And the hack was the "interface" i was talking about.(Something similar
to a small host bus.) The "glue logic" could vary.. If somebody could
work with a FPGA as a glue logic to the ASIC based GigE, interfaced to
the PCI bus then all that would make sense.. But it is quite some work,
require to do some work on an external prototyping board also. As you
explained, getting the connections out would be a major task, but i have
an idea here.. but quite difficult, if you can get the processor, out
from the PCB, extend it vertically upwards (it would be a real gimmick).
Then the PCI expansion connections could be tapped out.
But anyway, it would be quite a lot of work, and it would not be an easy
hobbyist work on BGA's.
- --- In firstname.lastname@example.org, Manu Abraham <manu@k...> wrote:
> Work on Hardware is not that easy. Usually, would result inbreaking it
> in the first case, trying to do modifications..I never said it would be easy. Just that it's technically possible
but for various reasons it's not worth doing it.
> to a small host bus.) The "glue logic" could vary.. If somebodycould
> work with a FPGA as a glue logic to the ASIC based GigE, interfacedto
> the PCI bus then all that would make sense.. But it is quite someA few of us do actually have the necessary experience to do this.
Time, money and motivation being the main limiting factors...
> an idea here.. but quite difficult, if you can get the processor,out
> from the PCB, extend it vertically upwards (it would be a realgimmick).
> Then the PCI expansion connections could be tapped out.It'd be easier and cheaper to just buy a development board like the
IXDP425 kit if you're doing all that. The board has a IXP425 instead
of the IXP420 we've got in the NSLU2 so you can use the crypto
features as well if you wish and it has the full complement of
interfaces. The PCI bus is brought out to a slot as well.