On 23 December 2015 at 13:44, CLASystems <clasystems at gmail.com> wrote: Hmm, I just tried that in SIMH, and that doesn't XOR at all. I haven't a clue what it does. What I have entered: sim> ie -m 100-105 100: TAD 76 101: AND 77 102: CLL RAL 103: CIA 104: TAD 76 105: DCA 77 Locations 076 and 077 being ARGONE and ARGTWO respectively, at the start: sim> ie 076-077 76: 1234 77: 4321 After running the above code sample: sim> ie 076-077 76: 1234 77: 0574 If we "flip" ARGONE and ARGTWO's values (to 4321 and 1234 respectively): sim> ie 076-077 76: 4321 77: 3661 Neither of those is the expected 5115 of an XOR operation. Am I missing something blindingly obvious? Cheers, Christian -- Christian M. Gauger-Cosgrove STCKON08DS0 Contact information available upon request.

One more factoid: Many who have DECtape drives do not run them at the final spec of relevant ECOs. The TD8E will work quite as well as it can on the hardware side, but only if you perform all lf the ECOs [or equivalent]. Some are volunteering their time to help document [and in some instances implemente!] some of the more common missing elements towards that goal.] DECtape started out so reliable, no one realized that it could get just enough unreliable to eventually become a "gotcha" unless you have the final spec. All of the following MIGHT apply to any particular TU55 or TU56 drive. ALL of these are fixable with some effort [that may require some cooperation to keep the community cost down to quite affordable: 1) There are three types of front wear plates:. The final is a ceramic in the exact shape as the oldest one made from soft steel. Either is fine; arguably the steel one is actually better in terms of maintenance because it can be polished nicely; the ceramic one can be somewhat cleaned by bathing in certain chemicals to remove the iron oxide buildup that gets INTO the pores of the ceramic, but that's mostly cosmetic, etc. What MUST be avoided is the dreaded mistake that was a major-step backwards: DUAL-SIDED ceramic wear plates. The smooth edge of the other two is replaced by a sharp chisel! It is known tp SHRED the tapes [and sometimes not so slowly!]. I doubt if it can be "blunted" enough to become smotth [it will likely crack apart] but soft steel can be bought; a good machinist can repro he old one, and perhaps by using slightly harder metal, can come up with one arguably better than either DEC original, etc. 2) Inside of the cover on the front of the guides is a small hole with a very coarse spring. The spring is to be OUTLAWED as it is the chief culprit in why everything else is screwed up [including the need to clean/polish the front wear plates as described in 1) above. Most newer TD8E-worthy TU56 fortunately have this changed: The backside of the front cover needs a flat-bottom larger but less-deep hole in a different place [the other hole can be ignored]. This is also a pretty straight-forward machine-shop job. Once that is established, there is a very "floppy" gentle-pushing spring that lowers the force pushing the tape rearward at the rear "cheek" plates [which can score them, see below]. This way, you avoid NEW damage to the tapes, the wear plates [and the rear cheek plates, but you cannot fix present damage this way, so this is still not complete]. 3) The wear plate is defined to have some definite number of thousandths of an inch it can push the tape rearwards. DEC wisely changed it to a bit more [maybe 12 thousandths more total travel, I am not a machinist to measure mill-smidgens, but some people can come up with this by comparing the before and after to get that exact to a macihinist's rule of thumb, etc. In any case, it's a straight-forward milling job and not all that much. [Note: Some guides are made of aluminum, while the spec is for steel, it's not as important to change the guides as it is to change this milling job being applied. Many older TU55 drives are all aluminum and the LINC/LINC-8 drives are all Aluminum. Steel if you have it is preferred, but is not on the list of things to be concerned about; all I mention here is [if your drive needs upgrade]. This allows the tapes to be "thinner" [slightly less than the nominal 3/4 inch wide] because DEC wanted SLIGHTLY damaged tapes to still be useable. [They realized they greatly screwed up from the beginning.] 4) The rear-cheek plates issue: All TU55 drives [and some very early TU56 drives have one-sided rear plates that cannot be swapped left to right and vice verse because it lacks a bevel. This is actually an opportunity! The newer plates were fflipable. the problem is that often field circus did just that, so you can have ruined on both sides guides. If you have the older plates, then get machining to create a virgin new surface. All dual-sided ones need to be evaluated. If you are lucky enough to have only damage on one sided, then just swap them after you do all of the above. They will score very, very slowly, easily 2-3 orders of magnitude lower with the floppy spring than with the old gross one, etc. Each damaged side of any plate has to be a one-off judgement call [actually both sides of any drive are likely in the same shape, good or bad]. It is possible there is the need to polish out the score line, but some are so bad, you may not be able to. Hopefully many will at least pass on one side,etc. We now have hard-anodize capability that didn't exist back then. So, after score-line issues and bevelling if necessary to create a new reverse side for the older ones, they need to be made "naked" and be placed into a modern anodizing bath. If there is no score line [such as on older drives' "back side" you have the best possible outcome. [But you cannott undo deep scores, agtan hopefully no one has deep scoring on both sides. Scoring may not be as bad as it looks. That's because the anodize scores first. Once it is naked metal, it might be polished out, YMMV 5) Roll pins versus dowel pins. If you encounter roll pins, throw them way [targets optional]. Replace with dowel pins [which are apparently not only available, they are cheap in boxes of 100, but any one drive only needs 8.] Share them with your drive-possessing friends. Past some point, they all have these. It makes taking the guides off and the ability to inspect the rear plates [and clean thoroughly] easy instead of a true nightmare. [You'll have to learn to curse at something else :-) ] All TU55 and TU56 are the same on all of the above issues [except where noted]. The heads are not involved [air bearings so the head never touches, but the tape does come close]. DO NOT take the heads off ever [unless you know and are prepared to deskew, an entirely different issue. Restoring things to what they were should mean no need to be concerned here, etc. For purists who get past all of the above,contact me for info about that more advanced subject.] Warning, there are drives in such poor condition out there that the knowledgable will NOT allow their tapes to be placed on them. But all drives can be fixed up with some effort. [I would say taking up a collection would get all happy if we organized. $25/head [that is a DECtape head :-)] is the figure I suggest; the more "generous" collectors can foot the rest of the bill. [Of course only a few are concerned with such as repro front wear plates, etc. Dowel pins are so cheap it doesn't matter much either way.] Tapes from a 1962 LINC can be read today on LINC/LINC-8 drives just fine. The all-Aluminum design has no front nor rear stuff, just a hand-made channel in an Aluminum block. [Note: The speed of the newer drives is 133 microseconds a word there. On the LINC/LINC-8 drives it's 160 microseconds and it is just effortlessly working [but you do need to buy new belts!]. DECtape/LINCtape is a great concept, but there were blunders along the way; it has held up surprisingly well since some tapes are nearly 54 years old now! [And ALL of the potential problems can be remedied.] [Now imagine running fully ECO'd TD8E and TU56 AND software beefed up all at the same time!] cjl On Thu, Dec 24, 2015 at 12:39 AM, CLASystems <clasystems at gmail.com> wrote: -- "In the future, OS/2 will be on everyone's desktop" Bill Gates, 1992

Ok, I think I get it The acc is one value. Both arg1 and agr0 are init to the second value ( start the same ) I'm not sure how to save a value to the acc on a pdp-8. I know how on my Nicolet 1080 that has a similar ALU. On my Nicolet: MEMA ARG0 A+MM ARG0 ANDAM ARG1 A+MA ARG1 ANGA A+MAM ARG0 ( acc and ARG0 have the xor of the original ARG0 and ARG1 ) I suspect it is similar on a pdp-8 as they have similar ALUs Dwight ________________________________________ From: cctalk <cctalk-bounces at classiccmp.org> on behalf of dwight <dkelvey at hotmail.com> Sent: Wednesday, December 23, 2015 10:38 PM To: General Discussion: On-Topic and Off-Topic Posts Subject: Re: Has Anyone Written PDP-8 .XOR. Code Using the MQ Register (Without the EAE)? I'm not sure how to code it on the machine but I know how to do it. Take A AND B = C 2* C = D ( ignore carry ) negate D ( complement and add 1 ) A + B = E ( this could have been done first ) E + D = A XOR B I think it needs 3 variables? It works because the carries are the result of the AND of the times 2 ( his RAL ) negating and subtracting is what he want to do, leaving the XOR. Dwight ________________________________________ From: cctalk <cctalk-bounces at classiccmp.org> on behalf of Christian Gauger-Cosgrove <captainkirk359 at gmail.com> Sent: Wednesday, December 23, 2015 7:14 PM To: CLASystems at gmail.com; General Discussion: On-Topic and Off-Topic Posts Subject: Re: Has Anyone Written PDP-8 .XOR. Code Using the MQ Register (Without the EAE)? On 23 December 2015 at 13:44, CLASystems <clasystems at gmail.com> wrote: Hmm, I just tried that in SIMH, and that doesn't XOR at all. I haven't a clue what it does. What I have entered: sim> ie -m 100-105 100: TAD 76 101: AND 77 102: CLL RAL 103: CIA 104: TAD 76 105: DCA 77 Locations 076 and 077 being ARGONE and ARGTWO respectively, at the start: sim> ie 076-077 76: 1234 77: 4321 After running the above code sample: sim> ie 076-077 76: 1234 77: 0574 If we "flip" ARGONE and ARGTWO's values (to 4321 and 1234 respectively): sim> ie 076-077 76: 4321 77: 3661 Neither of those is the expected 5115 of an XOR operation. Am I missing something blindingly obvious? Cheers, Christian -- Christian M. Gauger-Cosgrove STCKON08DS0 Contact information available upon request.

On 12/23/2015 7:14 PM, Christian Gauger-Cosgrove wrote: I think there was just a bit of brain fade in the original listing. From DEC-08-FFAA-D PDP-8 Math Library circa 1968 the XOR routine is as follows, transcribed to SIMH: sim> de 76 1234 sim> de 77 4321 sim> ie -m 100-107 100: CLA CLL 101: TAD 76 102: AND 77 103: CIA 104: CLL RAL 105: TAD 76 106: TAD 77 107: HLT sim> go 100 HALT instruction, PC: 00110 (AND 0) sim> ex ac AC: 5115 Which produces the correct result in the AC. Locations 103/104 were swapped in order, and location 106 was left out. And for completeness location 100 was added to initially clear the AC/Link. Don

Without the TMP on the Nicolet MEMA ARGONE ANDA ARGTWO LASH 1 ANGA A+MA ARGONE A+MA ARGTWO ( or A+MM ARGTWO to save in ARGTWO ) The link bit isn't used by LASH or adds on the Nicolet. It is only the carry out of an add. It is a little more clumsy to use as a carry. It can only be used as a condition of a skip or execute. ________________________________________ From: cctalk <cctalk-bounces at classiccmp.org> on behalf of Don North <north at alum.mit.edu> Sent: Wednesday, December 23, 2015 11:21 PM To: cctalk at classiccmp.org Subject: Re: Has Anyone Written PDP-8 .XOR. Code Using the MQ Register (Without the EAE)? On 12/23/2015 10:50 PM, Don North wrote: FYI here is a link to find the document: http://manx.classiccmp.org/details.php/1,3191 I can't find it under dec/pdp8 on bitsavers, maybe I'm just not seeing it. Don

Steve Gibson talks about XOR on the PDP-8 here: https://www.grc.com/pdp-8/isp-musings.htm In summary, Steve uses this condensed code using "carryless addition" for XOR: CLA / clear accumulator (AC) since all we have is add! TAD ArgOne / add (TAD) ArgOne to the just-zeroed AC AND ArgTwo / AND ArgTwo to determine where the carrys will be CLL RAL / clear the LINK (CLL) and rotate the accumulator left (RAL) CMA IAC / compliment (CMA) & increment (IAC) the accumulator (negate) TAD ArgOne / add the first argument to the negated accumulator TAD ArgTwo / and add the second argument as well / the accumulator now contains the XOR of ArgOne & ArgTwo Best I can tell, Charles proposed the same thing, but for whatever reason changed the last instruction to DCA instead of TAD. Kyle