You are completely right, and of course you can configure the 68EN360 for 3
No doubt on my feelings, this discussion is only to help you in your design
and to benefit from my experience.
What i mentionned is to better use a MC68EN360 at 33MHz due to the 30Mbits/s
bandwidth between the CPM and the CPU core. Thereby, you should have the
maximum performance of your 3 Ethernet channels, whatever you product is
dedicated for (router, data injection ...etc). However, i see no interest to
offer Ethernet channels with a bandwidth divided by a factor of 3 or 5!
For example, I successfuly designed CPUs with MC68360 at 25MHz and recently
But for 3 Ethernet ports at the maximum bandwidth (10 Mbits/s), i
successfuly designed a CPU that integrate a MC68EN360 at 33MHz in companion
mode with a MC68EC060 at 66MHz.
> -----Original Message-----
> From: Michael Leslie [SMTP:firstname.lastname@example.org]
> Sent: Wednesday, November 22, 2000 7:14 PM
> To: Paugam Luc
> Cc: 'uClinux Mailing-List'; 'uClinux-dev Mailing-List'; 'uCsimm
> Subject: RE: [uClinux-dev] BEWARE: Lineo 68360 board
> It's not quite true that you cannot configure the 68en360 for more than
> two Ethernet ports. You can do so, but you are limited, as you observe, to
> an aggregate throughput of about 20Mbps.
> I would refer anyone interested in the gory details to the 360 manual,
> appendix A: "Serial Performance":
> and also the "CPM Performance Spreadsheet" and accompanying documentation
> As an example, a router with multiple ethernet connections to 2Mbps DSL
> modems, would be well within the limits. From Motorola's worksheet, the
> following configuration yields ~73% CPM usage for a 25MHz part:
> Ethernet on SCC1, 10Mbps throughput
> Ethernet on SCC2, 3, and 4, 2Mbps throughput
> The perormance of the CPU core should, in an application like a router,
> have little to no effect on the throughput of the communications
> engine. As long as CPM interrupts are serviced quickly, and the amount of
> buffer copying is kept to a minimum, fully loading the CPU should have no
> effect on data transfers. This is the basic idea behind the '360, of
> course. If the CPU core is performing significant manipulation of the
> rx/tx data, then of course it will become a bottleneck. But then a 25MHz
> 68k core is no DSP...
> Michael Leslie,
> Lineo Inc.
> On Wed, 22 Nov 2000, Paugam Luc wrote:
> > BEWARE If you use a 25MHz Motorola MC68EN360, you cannot have 3 Ethernet
> > ports due to the bandwidth between the CPU core and its CPM
> > Processor Module). It is about 22 Mbits/s.
> > You better should use a 33MHz chip: the bandwidth between the CPU core
> > its CPM is 30 Mbits/s (OK for 3 LAN/WAN ports).
> > Generally it is not a good idea to design a CPU with more than 1
> > port, with only a MC68EN360.
> > This uP is not very fast: at 33MHz, 6 MIPS, 10500 dhrystones/second (5.5
> > BogoMips on uClinux). You have not enough power to fill up 2 or 3
> > ports and to execute apllications.
> > In order to remedy this fact and to really use 2 or 3 LAN/WAN ports, you
> > need to use your MC68EN360 at 33MHz in companion mode with a MC68EC060
> > 66MHz.
> > Best Regards
> > Luc
> > >Michael Leslie [SMTP:email@example.com] wrote:
> > >
> > >For those interested in Lineo's MC68360 board.
> > >
> > >The device is designed to be a high-end T1/E1 router platform.
> > >For availability and pricing, please contact Jim Stewart
> > <firstname.lastname@example.org>
> > >
> > >Again, this board's vital statistics are:
> > >
> > > - 25MHz Motorola MC68EN360
> > > - 32M DRAM
> > > - 8M FLASH
> > > - 10-base-T Ethernet
> > > - Serial console
> > > - 3 highly configurable WAN ports
> > > - IDE interface
> > > - Nice and small; designed to mount on a laptop hard disk
> > >
> > >
> > >Cheers,
> > >Michael Leslie,
> > >Lineo Inc.
This message resent by the email@example.com list server http://www.uClinux.com/
This archive was generated by hypermail 2b30 : Sun Apr 07 2002 - 00:01:39 EST