22 Apr
2020
22 Apr
'20
8:59 a.m.
From: Mark Matlock
Are you able to use a Qbus MTI controller in the
11/84's Qbus section
of the backplane? This is something I've often wondered about but never
tried.
I looked into this in some detail, but I don't know:
https://gunkies.org/wiki/KTJ11-B_UNIBUS_adapter#QBUS_slots
Amswering it definitively would probably require looking at DMA and interrupt cycles
with a logic analyzer on the bus, to see what the CPU does with them. And the KDJ11-B
and KDJ11-E (the 11/94 uses the same bacplane with a different CPU card) might act
differently.
For anyone who is curious about what Happens when a
M8637-C version
board is used as PMI memory in an 11/83 I can speak From experience.
...
After this the disk is corrupted and you will need to restore the
system disk from backups
Ah, very informative; thanks for reporting.
So whatever the fault is (perhaps the QBUS block transfer issue reported
up-thread), it must _seem_ to work, but fail in actuality. It would be
interesting to appply a logic analyzer, and see what the bus transaction
looks like, if it looks OK on the bus (in which case it's an internal
failure).
Noel
22 Apr
22 Apr
5:09 p.m.
So whatever the fault is (perhaps the QBUS block
transfer issue reported
up-thread), it must _seem_ to work, but fail in actuality. It would be
interesting to appply a logic analyzer, and see what the bus transaction
looks like, if it looks OK on the bus (in which case it's an internal
failure).
One possibility is that it's not handling multiple DMA devices talking
at the same time or calling to arbitrate the bus. Right now I have an
RL02 controller and the MTI controller but the RL02's are powered down.
I'll see if I can run it for a bit on RSX11M and get it to fault (and
backup my disk first as it's nice to have it working again).
C
23 Apr
23 Apr
3:30 p.m.
New subject: PMI memory error
So the memory board might be bad. When the 11/83 runs memory tests the
1mb board (of course the one that's fixed to run on Q bus systems) fails
out with:
Testing in progress - Please wait
Memory Size is 3072 K Bytes
9 Step memory test
Step
Error 47
Memory CSR Error
See troubleshooting documentation
Error PC = 173354 PCR page = 14 Program listing address = 014354
R0 = 060002 R1 = 100000 R2 = 172136 R3 = 172346
R4 = 100000 R5 = 000000 R6 = 172300 Par3 = 102000
It's there and I can see it in the memory map:
Memory Map
Starting Ending Size in CSR CSR Bus
Address address K Bytes address type type
00000000 - 07777776 2048 17772100 ECC PMI
10000000 - 13777776 1024 17772136 ECC PMI
But setting any CSR seems to fault it. Drat.
Any thoughts?
3:39 p.m.
New subject: PMI memory error
On Thu, Apr 23, 2020 at 3:30 PM Chris Zach via cctalk
<cctalk at classiccmp.org> wrote:
Error 47
Memory CSR Error
Memory Map
Starting Ending Size in CSR CSR Bus
Address address K Bytes address type type
00000000 - 07777776 2048 17772100 ECC PMI
10000000 - 13777776 1024 17772136 ECC PMI
But setting any CSR seems to fault it. Drat.
Are the CSR dip switches 1-4 all set to 0 on the first MSV11-J and all
set to 1 on the second one? Did you already try setting CSR dip
switches 1-4 to 0,0,0,1 on the second one so that the CSR address gets
mapped to 17772102? That would be the expected CSR address for the
second one with the first one at 17772100.
Without trying it on a system myself with two MSV11-J boards I'm not
sure what happens if the CSR addresses are not at the expected
locations.
3:45 p.m.
New subject: PMI memory error
Yes, same failure. Likewise if I put board 1 in the same dip settings as
board 0 I get the same failure (and pulling out board 0 the 2mb board)
Running the 2mb memory board here with RT11, seems fine.
C
On 4/23/2020 6:39 PM, Glen Slick wrote:
On Thu, Apr 23, 2020 at 3:30 PM Chris Zach via cctalk
<cctalk at classiccmp.org> wrote:
Error 47
Memory CSR Error
Memory Map
Starting Ending Size in CSR CSR Bus
Address address K Bytes address type type
00000000 - 07777776 2048 17772100 ECC PMI
10000000 - 13777776 1024 17772136 ECC PMI
But setting any CSR seems to fault it. Drat.
Are the CSR dip switches 1-4 all set to 0 on the first MSV11-J and all
set to 1 on the second one? Did you already try setting CSR dip
switches 1-4 to 0,0,0,1 on the second one so that the CSR address gets
mapped to 17772102? That would be the expected CSR address for the
second one with the first one at 17772100.
Without trying it on a system myself with two MSV11-J boards I'm not
sure what happens if the CSR addresses are not at the expected
locations.
7:20 p.m.
New subject: PMI memory error
I have a stack of them here if you need them. I think they are $175 each,
$150
each for 2 or more, + $12.50 shipping for any qty.
Paul
On Thu, Apr 23, 2020 at 5:40 PM Glen Slick via cctalk <cctalk at classiccmp.org>
wrote:
On Thu, Apr 23, 2020 at 3:30 PM Chris Zach via cctalk
<cctalk at classiccmp.org> wrote:
Error 47
Memory CSR Error
Memory Map
Starting Ending Size in CSR CSR Bus
Address address K Bytes address type type
00000000 - 07777776 2048 17772100 ECC PMI
10000000 - 13777776 1024 17772136 ECC PMI
But setting any CSR seems to fault it. Drat.
Are the CSR dip switches 1-4 all set to 0 on the first MSV11-J and all
set to 1 on the second one? Did you already try setting CSR dip
switches 1-4 to 0,0,0,1 on the second one so that the CSR address gets
mapped to 17772102? That would be the expected CSR address for the
second one with the first one at 17772100.
Without trying it on a system myself with two MSV11-J boards I'm not
sure what happens if the CSR addresses are not at the expected
locations.
8:28 p.m.
New subject: PMI memory error
Thanks! I might take you up on the offer if this 11/84 memory is
unstable. Meantime I'm trying to figure out just what kind of
performance improvement PMI memory really gives. So far it's looking
like zero when it comes to disk or I/O performance.
Are there benchmarks one can compile for pdp11's?
C
(Hm. I wonder if 11/84 PMI memory will exhibit errors when run with a
pdp11/23+ CPU. Or more to the point if it is put in with an 11/83 with a
space between it and the CPU so the CPU accesses it only via Q-Bus)
(Hm. I wonder if you can replace a pdp11/84 CPU with an 11/23+ cpu. It
could be a way to get CIS on Unibus (although you could always go with
an 11/24 or 44)
24 Apr
24 Apr
2:32 p.m.
New subject: PMI memory error
On Thu, Apr 23, 2020 at 3:30 PM Chris Zach via cctalk
<cctalk at classiccmp.org> wrote:
So the memory board might be bad. When the 11/83 runs memory tests the
1mb board (of course the one that's fixed to run on Q bus systems) fails
out with:
Testing in progress - Please wait
Memory Size is 3072 K Bytes
9 Step memory test
Step
Error 47
Memory CSR Error
Is it possible to continue past that error and then try to boot?
If you can boot XXDP and run the VMJAB0 MSV11-J Memory diagnostic
maybe it would provide more detailed error information.
1473
days inactive
1475
days old
7 comments
4 participants
participants (4)
-
cz＠alembic.crystel.com -
glen.slick＠gmail.com -
jnc＠mercury.lcs.mit.edu -
useddec＠gmail.com