Dave Caroline wrote:
On Tue, Oct 25,
2011 at 5:10 PM, Chuck Guzis <cclist at sydex.com> wrote:
On 25 Oct
2011 at 10:50, Mouse wrote:
"The most amazing achievement of the computer
software industry is its
continuing cancellation of the steady and staggering gains made by the
computer hardware industry." - Henry Petroski
I remember an Intel engineer summarizing this more succinctly
somewhat like this:
"We work hard to build chips with amazing performance and the
software people piss it all away."
An easy cure is give developers slow machines with limited memory and
slow disk subsystems.
I do that to myself. I find it a good way to discover the "right way" TM
At one point, I enhanced the device driver for the DEC RX02
(double density 8" floppy) disk drive on a Qbus system. DEC
never bothered to upgrade the RX02 Qbus controller in the
same manner as the RL02 controller which was upgraded
to support 4 MB of memory. The DEC RX02 controller
supports only 256 KB of memory. If the user buffer is located
above 256 KB (in extended memory in RT-11), then the
RX02 Qbus controller can't transfer the silo directly to the
user buffer.
I figured that it could not be that hard to use a bounce buffer,
so I wrote and tested the code on a PDP-11/73. No problems
with the concept or implementation. For some reason that
I can't remember, I then used that new feature in the RX02 device
driver on a PDP-11/23 - it slowed down by a factor of about 20
since it was missing the interleave window. The PDP-11/73 was
able to transfer the silo to the bounce buffer and then to the user
buffer within the sector interleave time. On the PDP-11/23, the
transfer of the bounce buffer was not fast enough.
The simple solution (which required a few more temporary variables)
was to transfer the silo to the bounce buffer, then initiate the next
I/O request in the case of a read (and the other appropriate things
in the case of a write) followed by a transfer of the bounce buffer
to the user buffer WHILE the next sector was being read into the
silo. The additional code took about as long to figure out as the
original extra code to use the bounce buffer. But there was no
other choice when a PDP-11/23 was being used.
Just confirming that I AGREE WITH YOU!!!!!!!!!!!!!!
Jerome Fine
