Re: [PARPORT] Speed of ppa 1.40 driver


David Campbell (campbell@gear.torque.net)
Fri, 6 Feb 1998 08:28:13 +0000


> Date: Fri, 6 Feb 1998 00:19:34 +0100 (CET)
> From: Andrea Arcangeli <arcangeli@mbox.queen.it>
> To: Bas Mevissen <sgm@stack.nl>
> Cc: Riley Williams <rhw@bigfoot.com>,
> David Campbell <campbell@gear.torque.net>,
> Linux Parallel Port Development Mailing List <linux-parport@torque.net>
> Subject: Re: [PARPORT] Speed of ppa 1.40 driver

> On Thu, 5 Feb 1998, Bas Mevissen wrote:
>
> >That's a good offer. CPU consumption reported by 'time' is about 4% now.
>
> Wrong. CPU load is very more high than 4%. Only /proc/loadavg report the
> right load of the CPU because the ppa_engine() is interrupt timer driven
> and is not checked by usual user level programs such us ps or top (or
> time). Maybe with the old driver in 2.0.33 this will not happen because
> ppa was tq_scheduler and not tq_timer driven but I am not sure of this.

Agree on high CPU loading.

Load average is defined as the "average number of processes in the run
queue".

With EPP protocol the processor is in "halt" mode a large amount of the
time while the IO chipset is handshaking/data transfer with the ZIP drive.
The 4% kernel time is calculated by dividing clock ticks while in the
kernel by the wall clock time.

We are talking about worse case senario here of continous data transfer.
In reality there would be disk head sweep time, during this time the
ppa_engine() drops out allowing other processes to run.

If you want better performance - use SCSI.
(expecting to be flamed by laptop owners here)

I will NOT support any drivers where tq_timer is changed to tq_scheduler.
(*putting my foot down before anyone gets some bright idea*)

It causes all sorts of problems when the loadavg exceeds 0.9. Compiling a
kernel or GCC on a ZIP drive REQUIRES the driver to be using tq_timer (to
ensure constant data transfer) otherwise the kernel panics in the SCSI
mid-level driver.

David Campbell
=======================================================
campbell@torque.net (Parallel port device related mail)
dcampbel@p01.as17.honeywell.com.au (For all other mail)

"All parallel ports are equal - Some are more equal than others"

-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --



This archive was generated by hypermail 2.0b3 on Wed 30 Dec 1998 - 10:17:25 EST