Re: [cobalt-developers] 2.2.14 kernel

[Dennis <dennis@xxxxxxxxx>]

At 11:12 AM 5/30/00 -0700, you wrote:
>Dennis wrote:
>
>> Maybe Cobalt's market is selling to the brain-dead, non linux savvy
>> community. But there is a whole (huge) market you are missing by
>> (apparently) being 1 or 2 engineers short. My point, if you care, is that
>> for the amount of money you raised you could sell a LOT more units by
>> having a couple more people.
>
>We are always ready to hire available talent - you'd be surprised (given
>how popular Linux has become) how hard it is to hire anyone who has the
>level of experience and enthusiasm we want.  There are a lot of linux
>hackers out there, but most folks who are hardcore OS guys don't want to
>work at Cobalt, they want to be designing a new OS :)  Unfortunately,
>that is not the business we are in.

Believe me, this I know.....

>
>> But .14 is not yet released, and .16 is almost out....so really .12 is
>> being shipped and 16pre is available. So you are "almost" only 2 revisions
>> back. The ethernet contoller fixes for the intel processor was not released
>> until .15, so you may be short on that also.
>
>the .15 eepro100 update also has a MAJOR performance impact, as well as
>a slew of bugs of it's own.  It may fix YOUR problems, but we sell
>thousands of systems, and I need to make sure it works EVERYWHERE.  We
>have a 2.2.14 available on teh ftp site, into which you can easily drop
>the 2.2.15 eepro100.c and recompile - watch your bandwidth capabilities
>drop by 10 fold.  I am working with the driver maintainers to understand
>teh problem better.

the 1.06 driver locks up under heavy load, so it is unusable in a heavily
loaded environment. We havent seen significant perfornance problems with
the 1.2 driver, and it doesnt fail under exceptionally heavy load. There
are no changes in the 1.2 driver that would cause a "10 fold" decrease in
perfornance, so I dont know how you came to that conclusion.  I've got a
test running right now pumping 35000pps through a 200Mhz pentium and its
not even breathing hard running that driver. So I think that you are badly
mistaken if you think you can get 10 times that performance on a 1.06
driver (which chokes on this test, btw). 

I've also been trying to tell you that compiling eepro100 drivers does NOT
WORK "out of the box" on the RaQ3. I get continuous mdio_read() errors and
the driver doesnt function properly.

DB