Question about DEC PMI Processors & Gear Acquisition

Been there - Done that! However, there are many different non-Q22/CD backplanes. There are also many Q22 ABAB backplanes. If you are starting with a 16 bit backplane and a PDP-11/03, then DEFINITELY!!!!!!!!!!!!! DO NOT attempt to use a PDP-11/23 or a PDP-11/73. The PDP-11/03 has (if I remember correctly) extra power points in the backplane. However, if you are starting with a system which is already running even a dual PDP-11/23 or a dual PDP-11/73 such as the VT103, then my actual experience is that there will not be a problem and the VT103 backplane can easily be upgrade from 18 bits to 22 bits with the required documentation and someone with almost no experience with a soldering iron. It is not trivial, but Tony Duell could probably do it while sleeping. I have also heard of someone upgrading the first two slots of the VT103 to be ABCD and installing a MVII. I do not have the details and have no idea what needs to be done or where to find the details. With regard to the DEC manual, I agree that as stated, it is CORRECT!!!! What is missing is a discussion of other configurations. To make sure we are all discussing the same thing: M8190-AA is a KDJ11-BA with a 15 MHz crystal M8190-AB is a KDJ11-BB with a 15 MHz crystal M8190-AC is a KDJ11-BC with a 15 MHz crystal and the FPU chip M8190-AE is a KDJ11-BF with an 18 Mhz crystal and the FPU chip I will state again: I have placed an M8190-AA board in the VT103 backplane in the first slot and a 4 MB memory board in the second quad slot. The third quad slot was a quad EDSI disk controller and the fourth quad slot was a DHV11. The backplane was originally 18 bit, but was upgraded to 22 bit by soldering 4 wires to 8 points each on the backplane, one for each of the dual AB locations which are available via a solder connection which was previously left available but hanging. SORRY - lead based solder is required since the original backplane used, at the time, lead based solder. Wire wrap wire was used and the insulation was left on between the solder joints. The VT103 has a 4 slot ABAB serpentine backplane since I have also connected 8 dual boards to that same backplane such as an M8192 and an M8043. The manual that you have copied a portion of is CORRECT. If you place a quad M8190 board (all M8190 boards as far as I know including the M8190-AA) in the third slot of a BA23 and place 2 * 2 MB PMI quad memory boards in slots one and two (the first three slots of a BA23 are ABCD), then RT-11 will report that the CPU is a PDP-11/83 with 4 MB of PMI memory. If you place the M8190 in slot one and the PMI boards in slots two and three, then RT-11 will say NOT say PMI memory is present. What I did not do (since I did not have an M8190-AE to test at the time) was place an M8190-AE into the VT103 backplane. However, I would have at the time since I did not know any better and I am confident that if I had access to all of the stuff needed to test the VT103 that I could do so now. But, since I have not done it yet, I can't state with 100% confidence that the magic smoke would not be released. By the way, some benchmarks I did might be of interest. The M8190-AE boards configured with PMI activated memory is about 33% faster than an M8190-AA board without PMI activated memory. An M8190-AE board without the PMI memory activated is about 13% faster than an M8190-AA board without PMI activated memory. An M8190-AA board (with the 15 Mhz crystal) with PMI activated memory is about 20% faster than an M8190-AA without PMI activated memory. This means that about 60% of the speed improvement of the M8190-AE with PMI activated memory is due to the use of PMI activated memory. The other 40% of the speed increase is due to the 18 MHz crystal vs the 15 MHz crystal. I caution that this was RT-11 running only a single job (specifically BINCOM which compares blocks of a file). For the file, I substituted the VM: device and compared about 7000 blocks with itself, thereby eliminating any disk I/O. Since ONLY the CPU was involved, I felt that the test was a very good rough approximation to the overall CPU speed. The time was measured to the nearest 1/60 of a second, so anyone can easily repeat the test. I know that the above information can be verified. If anyone has access to these boards and can check the electrical connections and the difference between ABAB and ABCD, I suspect that it will be possible to verify that the M8190-AE is different from the M8190-AB only in the 18 Mhz crystal (and the different revs of the CPU plus the extra FPU chip). Jerome Fine