EDSI Problem

From: Jerome Fine <jhfine_at_idirect.com>
Date: Sat Jul 21 14:56:54 2001

Hi PDP-11 users,

I need some really expert advice. I am running an Hitachi DK515-78 ESDI
600 MByte hard disk drive with a Sigma RQD11-EC ESDI quad Qbus controller.
The box is a good working BA123 system (not the one with the bad power supply).

The current configuration is:
2 MBytes quad PMI Memory
2 MBytes quad PMI Memory
11/83 quad CPU
Sigma RQD11-EC quad controller
DHV11 quad serial card
Sigma dual EDSI controller / DLV11-J
TK50 dual controller / TK70 dual controller

At one time, I also had a quad TQK25 controller,
a DZV11 quad serial card
RQDX2 quad MFM controller
DSD dual controller for DSD 880/30 (RL02/RX02)

I think the system functioned OK for a while with the
extra boards, but I can't be sure. In any case, about
4 years ago, I suddenly realized that the DK515-78
EDSI hard disk drives were not running at full speed
during many multiple READ operations - like at least
5 MBytes - although it became really noticeable when
I compared a full 32 MByte partition on one of the
DK515-78 drives with another 32 MByte partition on
a second DK515-78 drive. It is possible that this happened
right from the start, but it is now so long ago that I can't
remember. In any case, somehow I found that if I limited
the total number of boards in the backplane to 7 quad
boards or less (or their equivalent in dual/quad combination -
i.e. a total of 7 quad slots), then the drives could still be READ
at full speed. For example, right now I can compare a 32 MByte
partition on one drive with an identical 32 MByte partition
on a second drive in 4 minutes - when there are 7 quad boards.
But if I add any more boards (RQDX2, TQK25, DZV11, etc. -
not all, just one more), then the total compare time is more than
10 minutes. I don't have any drives connected to the dual ESDI
controller right now, but when I did have one, a compare with
more than 7 quad Qbus boards using just one DK515-78 drive
plus an ST8760E on the dual ESDI controller still took about
10 minutes. When there were only 7 quad boards or less, the
compare time dropped back to 4 minutes.

HOWEVER, if I did a copy from the ST8760E on the dual controller
to a DK515-78 on the Sigma quad controller, the time was only about
3 minutes regardless of how many boards were in the backplane.

I suspect that the actual problem is that the sector size is too small
on the Hitachi DK515-78 drives since there are only two options
for sectors which end up as 512 bytes per block. WRITE operations
seem to get through OK, but if the backplane is more loaded, then
a READ operation often misses with the 1:1 interleave - sufficient
to slow throughput by a factor of 2 or 3.

If I have not provided enough information, please ask some more questions.
Does anyone have a clue?

Sincerely yours,

Jerome Fine
Received on Sat Jul 21 2001 - 14:56:54 BST