*start* 00453 00024 US Date: 2 JAN 1978 1502-PST From: MCDANIEL Subject: [ifs] diagnostic files To: D1Users: There is a new version of the memory diagnostics on ifs. this version represents a change in the file organization of the diagnostics. Labels are unchanged (program labels are unchanged, file names are different in many cases). You should get a new version of the diagnostics, and it is reasonable to not get new listings. Gene ------- *start* 00481 00024 US Date: 6 JAN 1978 0325-PST From: MCDANIEL Subject: new diagnostics on ifs To: D1Users: There are new versions of kernel and mem*. diagnostics on ifs. This release incorporates a "midas subroutine" for turning off the interim display: displayOff;G will cause the display to be turned off. the code has a breakpoint at the end. This release includes requested changes to MemD.I will send another message after I have checked it out on the d1. ------- *start* 00491 00024 US Date: 10 JAN 1978 0228-PST From: FIALA Subject: D1Lang To: D1Users: On January 3, I dumped a new D1Lang with the following changes: 1. MCP[i] deimplemented 2. MOS←, B←Junk, GenOut←B added. 3. TestSyndrome← renamed LoadTestSyndrome 4. Renumbering of some IFU section functions. These changes are in response to some decoding changes being made on the IFU board. See Severo O. for details about IFU changes, Doug C. for details about LoadTestSyndrome. ------- *start* 00316 00024 US Date: 12 JAN 1978 1841-PST From: MCDANIEL Subject: new diagnostics on ifs To: D1Users: There are new diagnostics (all) on ifs. They reflect a change in preamble that prevent the "interations" word in IM from being incremented properly. There is a fix to a minor bug in memc. Gene ------- *start* 01060 00024 US Date: 13 JAN 1978 0859-PST From: FIALA Subject: <D1>D1MidasRun.dm;53 To: D1Users: This release of Midas fixes all of the printout problems reported to me and also has the following changes: 1. Some bugs in the DMux simulator related to Hold are fixed; 2. The special Alto microcode replacing DoStrobe and ReadDmux has been added. 3. The rest of the multi-radix I/O stuff is commissioned (probably uninteresting for D1 Midas) and features for symbolic and address printout in the main menu areas. It is now possible to display memories which hold instruction addresses (TPC, LINK, and OLINK but not CIA, BNPC, etc.) as symbol+/-offset by means of a command file command (e.g., "L A10 Search " will cause the value of the item at A10 to be searched for in the symbol table). 4. MAP and VM are now more easily testable (not in TestAll, however), although you will have to setup the map and address ranges carefully before testing VM. The new Alto microcode results in about a factor of 10 speedup in testing DMux stuff. ------- *start* 00263 00024 US Date: 13 JAN 1978 2254-PST From: FIALA Subject: Admin To: D1Users: I have renamed the <D1DOCS>D1users.msg file into the <SECRETARY> directory where all the other message files are kept. I think that J. Jenkins maintains these. ------- *start* 00231 00024 US Date: 16 JAN 1978 0415-PST From: FIALA Subject: <D1>D1midasrun.dm;54 To: D1Users: This release includes changes to the Set-Clock, Reset, and DMux printout code for the clock board design changes. ------- *start* 00685 00024 US Date: 16 JAN 1978 1805-PST From: MCDANIEL Subject: new kernel on ifs To: D1Users: There is a new kernel diagnostic on ifs. All bugs known to me are fixed. There are three new "midas" subroutines: xorTaskCirc xors the flags.testTasks bit in FLAGS xorHoldSim xors the flags.holdSim bit in FLAGS xorTaskSim xors the flags.taskSim bit in Flags These subroutines remove the necessity of knowing which bit to set in the FLAGS control word in Kernel (FLAGS controls the two simulator functions and it controls whether or not Kernel tries to execute the diagnostics at different task levels (flags.testTasks). Kernel runs tasks [0..7] without error. Gene ------- *start* 00283 00024 US Date: 17 JAN 1978 1227-PST From: FIALA Subject: <D1>D1midasrun.dm Release To: D1Users: New Midas release fixing some problems which caused Midas to swat following command file errors. Several other changes also. Please report problems by Sndmsg. ------- *start* 00306 00024 US Date: 18 JAN 1978 1525-PST From: WILLIE-SUE Subject: D1 alu definitions file To: D1Users: [ifs]<d1source>d1-aludefs.mc claims to be the truth about the alu, for code that wishes to run with the alto emulator. Please read the comments at the beginning of the file. WS ------- *start* 00325 00024 US Date: 19 JAN 1978 0922-PST From: WILLIE-SUE Subject: more on alu defs To: D1Users: New version of [ifs]<d1source>d1-aludefs.mc. Again, please read the note contained in the file. ALU function A0 has been restored; aluf[15] gets used by BITBLT and then restored to a standard value. WS ------- *start* 00189 00024 US Date: 27 JAN 1978 0953-PST From: FIALA Subject: <d1>D1midasrun.dm release To: D1Users: New Midas makes latest requested changes for read/write of CACHEA. ------- *start* 00201 00024 US Date: 27 JAN 1978 1733-PST From: FIALA Subject: <D1>d1midasrun.dm;61 To: D1Users: This release fixes bug in CACHEA read-write and adds more speeds to Set-Clock menu. ------- *start* 00488 00024 US Date: 1 FEB 1978 2003-PST From: TAFT Subject: Dorado I/O Microprogramming To: D1Users: I have prepared a brief memo (2 pages) describing my experiences in writing the microcode for the Dorado Ethernet. My main point is that the Dorado's branching scheme appears not to be very well-suited to the needs of I/O microcode. This memo is an exercise in hindsight, and obviously I don't expect to influence the design of the machine at this late date. Ed ------- *start* 00157 00024 US Date: 1 FEB 1978 2004-PST From: TAFT Subject: That memo... To: D1Users: is available as <Taft>D1IOProg.bravo (or .ears). ------- *start* 00446 00024 US Date: 14 FEB 1978 0944-PST From: TAFT Subject: LdTPC restriction To: D1Users: Ken has informed me of the following restriction on the use of the LdTPC operation: it may not be used if the task whose PC is being loaded can wake up simultaneously. If control were to switch to that task at the end of the LdTPC instruction, it would start running at the old PC and the effect of the LdTPC would be ignored. Ed ------- *start* 00337 00024 US Date: 15 FEB 1978 1501-PST From: FIALA Subject: <D1>D1midasrun.dm release To: D1Users: New release fixes bug in LoadDMD routine so that CACHEA write might work. Large clock speed values in menu that are not supported by the hardware are eliminated. Printout bug in Config is fixed (now "notConfig"). ------- *start* 01322 00024 US Date: 15 FEB 1978 1716-PST From: PIER Subject: Model 1 changes To: Dorado Users Group:, boggs I am soliciting immediate comments on the following two intentions for the Model 1 control section: 1. All the control ram chi[ps will be moved to Control B. All the randonm logic will live on control A. Control B will be physically below Control A in the cage. Control B will have a manifold/muffler interface as well as the normal interface, so it can be checked out without control A from Midas. Microcode impact: none Midas impact: new code to use and reset extra checkout interface Backpanel impact: reverses positions of the two boards 2. Task specific NoTask is deleted and replaced by a counter which is incremented by execution of TaskingOOff and decremented by the execution of TaskingOn. When the counter reaches -1 (17b, 4 bit counter), tasking is enabled and decrementing is disabled. Overflowing by incrementing brings disaster. Microcode impact: none, I think. Need input from Taft, Fiala. Midas impact: replaces whatever code manages task specific tasking with management of the counter. We may have to think about this a little. Item 1 will be implemented immediately for the new backpanel unless someone supplies a good reason not to. Item 2 is not urgent. Thank You Ken ------- *start* 00430 00024 US Date: 15 FEB 1978 1721-PST From: PIER Subject: Addition to previous message, Model 1 control To: Dorado Users Group:, boggs I neglected to add that the new NoTask scheme will not prevent blocking, but will prevent pre-emption. Thus, a task would normally increment the register, do its thing, then decrement the register, then BLOCK. The block is effective even though the register is # 17b. K ------- *start* 00381 00024 US Date: 17 FEB 1978 1517-PST From: MCDANIEL Subject: new diagnostics To: Dorado Users Group: There are new diagnostics on the d1 debugging disk and on ifs. All the diagnostics appear to work. The problem w/ kernel bombing during the link test has been fixed. The problem w/ not being able to start memC at begin after an error has been fixed. Gene ------- *start* 00774 00024 US Date: 9 MAR 1978 1900-PST From: FIALA Subject: <D1>D1MidasRun.dm release To: Dorado Users Group: The new release has primarily the following changes: 1. CACHED is implemented. 2. "DisHold" is removed for VM, VMX, and MAIN, so that these memories should work correctly on cache misses if RunRefresh is true (? let me know if this is so ?) 3. Prettyprint for CACHEA will now print the contents of the 16 CACHED entries associated with that CACHEA entry as well as the flags and VA. 4. Junk IO reset is incorporated in the Reset and Go at new address code. The addresses for CACHED are in the range 0 to 3777 where the top five bits of address refer to the row, next two bits to the column, and low four bits to the word in the cache. ------- *start* 00281 00024 US Date: 14 MAR 1978 1707-PST From: DEUTSCH Subject: New MicroD To: Micro Users: This release passes all memory and symbol definitions for unknown memories to Midas unchanged, at Ed Fiala's request. No changes from the microprogrammer's viewpoint. ------- *start* 00591 00024 US Date: 15 MAR 1978 1624-PST From: ORNSTEIN Subject: A Small Step for the Dorado Eleven To: Dorado Users Group: cc: Taylor Today we passed a modest milestone in that the Dorado displayed a mouse-controlled cursor whose pattern registered keystrokes, mouse button pushes, etc. This means that the junk and display tasks worked in conjunction with the Emulator task executing BCPL Alto code from the cache. The next step is to get the storage working with Midas so that we can load some non disking program (SIL probably) into the main memory and run it. ------- *start* 03334 00024 US Date: 16 MAR 1978 0306-PST From: FIALA Subject: <D1>D1MidasRun.dm;68 and new Microassembler To: Dorado Users Group: There are new releases of Midas and the Dorado microassembler. You should be sure to get the new Midas and new MicroD before getting the new microassembler since the new assembler produces some stuff that previous versions of Midas won't load. To get the new Midas, first delete the obsolete command files mx.midas, mc.midas, and md.midas, then load <D1>d1midasrun.dm;68 as usual from Maxc1. The changes in the new Midas are as follows: 1. VM, CACHEA, CACHED work correctly finally. 2. Reset code has been modified to fix all known bugs. 3. Reset before Go at new address resets all the peripherals. 4. A new read-only memory called MD is available that will allow you to look at the task-specific MD register, and a command file will show all 16 MD registers; the former MC, MD, and MX command files were renamed to MMC, MMD, and MMX to be consistent while not conflicting with the MD command file for showing MD. 5. Items which change on the display are now "*"ed to indicate what has changed state; this will probably be useful after breakpoints. 6. DMux pretty-printing for the disk controller is implemented and a command file DSK is provided to show the disk DMux signals. 7. A new fake memory TASKN is provided to show the contents of registers containing task numbers symbolically. The new assembler outputs symbols in this memory for convenience during debugging. 8. A new fake memory DEVICE is provided to show the contents of TIOA symbolically; the new assembler allows you to define device numbers symbolically for convenience during debugging. 9. The Midas pretty-printing procedures now use the TASKN, DEVICE, and BR address symbols during printout. Several bugs were fixed in the microinstruction pretty-printer. The new assembler consists of a new <D1Docs>D1Assem.Ears file describing the changes, new <D1SOURCE>D1lang.mc and <D1SOURCE>D1Alu.mc. You may need a new alu defs file from Willie-Sue also. The new assembler outputs additional memory address symbols for use with Midas, as discussed above, and you need LPD's new MicroD to pass these safely through. The changes in the assembler are as follows: 1. A new compact form for specifying EF/IF format shift instructions is available. Although the old form is still defined, please convert your microcode to the new format so that I can delete the obsolete form. The new shift format is discussed in the documentation. 2. BR[name,n] defines a base-register address for use in MEMBASE←name IFUM assembly, and symbolic printout by Midas (see documentation). 3. DEVICE[name,n] defines a device number. 4. TASKN[name,n] defines a task number (EMU and FLT predefined). 5. EQV defined as a synonym ALU operation for XNOR and #. 6. H3 literals allow you to write statements like Store←Rtemp, Md←T, T←0H; which read the contents of ALUFM as a literal value. This may be useful occasionally. As you start defining task numbers symbolically, please limit the names to 3 characters since a future change in Midas will print the task-specific register addresses symbolically, and long task names won't fit in the display area. (e.g., TIOA 17 will be TIOA FLT). Please report any problems. ------- *start* 00301 00024 US Date: 16 MAR 1978 2034-PST From: DEUTSCH Subject: New release of Micro To: Micro Users: cc: ludolph Only minor changes. See the documentation. The purpose of the release was to alleviate a space constraint which prevented the Dorado diagnostics from assembling. ------- *start* 00249 00024 US Date: 17 MAR 1978 0848-PST From: MCDANIEL Subject: new [ivy]<d1>memA.dm To: Dorado Users Group: There is a new version of memA(ll) on ivy. It is compatible with the new release of midas and the new assembler. Gene ------- *start* 00287 00024 US Date: 17 MAR 1978 1550-PST From: DEUTSCH Subject: New release of Micro -- apologies To: Micro Users: cc: ludolph The bit table primitives (other than MEMBT, which was never implemented anyway) have been restored by popular demand. No other changes. ------- *start* 01518 00024 US Date: 20 MAR 1978 1745-PST From: FIALA Subject: <d1>d1midasrun.dm;69 To: Dorado Users Group: This release adds the subroutine call feature requested by G. McD and fixes various signal inversion hardware changes in the Pipe readout. The CALL feature allows a microprogrammed subroutine to be called with 0 or 1 arguments. To do this, type: SUBR(arg) where SUBR is an IM or IMV address such as 141 or FOO+3, and where arg evaluates to a 16-bit number. If arg is passed, it will be loaded into T before the subroutine is called. The call feature is implemented by planting a breakpoint instruction at absolute location 7776 in the microstore. This location is loaded with an appropriate value by Reset; you must not use 7776 in any microprograms you write. The next release of D1LANG will output an appropriate instruction for 7776 when it assembles the END statement. In the meantime, microprograms will still work unless 7776 is unluckily used by MicroD. The prettyprint routine and readout of the Pipe stuff for Syndrome and for several map bits has been changed in accordance with hardware changes reported by KP. I am proposing to change the following names in D1LANG in the next release because of the various signal inversions which are in the memory system hardware; if this is inconvenient, please come discuss these with me: B←PIPE2 to B←NOTPIPE2 B←PIPE3 to B←NOTPIPE3 B←PIPE4 to B←NOTPIPE4 B←FAULTINFO to B←NOTFAULTINFO B←ERRORS to B←NOTERRORS B←MAP to B←NOTMAP. ------- *start* 00236 00024 US Date: 28 MAR 1978 2314-PST From: FIALA Subject: <D1Source>D1Lang.mc release To: Dorado Users Group: I have just released a new D1Lang.mc which fixes a bug in DEVICE reported by EAT. No other changes. ------- *start* 00240 00024 US Date: 29 MAR 1978 0953-PST From: MCDANIEL Subject: minutes of the Dorado working retreat are available To: Dorado Users Group: see <mcdaniel>dorado-yosemite.minutes or <mcdaniel>dorado-yosemite.ears Gene ------- *start* 00360 00024 US Date: 31 MAR 1978 1045-PST From: DEUTSCH Subject: New Micro and MicroD To: Micro(D) Users: cc: ludolph This release runs only under OS 14 and later -- now is the time to update your disks if you haven't already done so. MicroD has more space, and always does a Junta -- the /J switch has been abolished. No other changes. ------- *start* 01261 00024 US Date: 5 APR 1978 1217-PST From: FIALA Subject: <d1>d1midasrun.dm;71 To: Dorado Users Group: This release of Midas contains the following changes: 1. DMux signal changes on MemX 2. SkipVEql, SkipVGr, and SkipVLs are replaced by SkipE, SkipNE, SkipG, SkipL, SkipLE, and SkipGE in Midas command files. (i.e., the three existing actions are renamed and three more actions added) 3. The Test and Test-All actions are modified to report some additional error summary information and the right-column of the standard display is modified to show the new information. Now, typing <cr> after a test has failed will continue running the test indefinitely while accumulating information about the errors that occur. When the test is stopped by typing control-C (or bugging Abort), the summary information about the errors will be displayed. 4. A WriteDT action has been added for command files that will append the date and time to the current output file. 5. The WriteMessage action for command files has been modified so that the "\" character is replaced by a blank. This is provided so that blanks can be inserted at the beginning of a message. Note: you should upgrade to OS version 14 or later for this release of Midas. ------- *start* 00523 00024 US Date: 10 APR 1978 1249-PST From: MCDANIEL Subject: [ivy]<d1>memA.dm To: Dorado Users Group: There is a new version of the memory diagnostic. It supports a facility that causes the memory storage diagnostics to run with MCR set up to force all misses to select a particular column. This has the effect of making the cache appear to be one column wide. This feature can be used in cnjunction with the memory tasking code as well. The listing and documentation in the d1 lab are new. Gene ------- *start* 00470 00024 US Date: 12 APR 1978 2251-PST From: MCDANIEL Subject: new documentation for diagnostics To: Dorado Users Group: I've prepared a list of some of the most common subroutines called from Midas (or seen frequently in listings). The list is alphabetically ordered and contains a short summary of what the routines do. Hopefully this will make it easier to use/understand the diagnostics. Please mention to me any routines I've forgotten. Gene ------- *start* 00246 00024 US Date: 13 APR 1978 1435-PST From: FIALA Subject: <d1>D1MidasRun.dm release To: Dorado Users Group: This new release fixes several problems in prettyprinting the STKRB and PJUNK DMux words and the D1OUT register. ------- *start* 00222 00024 US Date: 24 APR 1978 1409-PST From: FIALA Subject: <D1SOURCE>D1LANG.MC release To: Dorado Users Group: This release allows KNOWRBASE to be used with RM addresses as well as RM region names. ------- *start* 00210 00024 US Date: 25 APR 1978 1826-PST From: FIALA Subject: <D1>D1MidasRun.dm;75 release To: Dorado Users Group: This release adds the prettyprinting stuff for the ethernet DMux signals. ------- *start* 00745 00024 US Date: 26 APR 1978 1251-PST From: ORNSTEIN Subject: Friday Meeting and IFU To: Dorado Users Group: 1. Due to the Friday picnic, let us postpone our regular meeting to 1:30PM 2. We are considering two reductions in IFU function: a) Eliminating the whole business of switching instruction set for the ensuing instruction b) Reducing the number of bits for helping to initialize the emulator's membase from 3 to 2 (the bits from the decoder RAM) If you have strong feelings about either or both of these, come see me. And be prepared to defend assertions about the need for the function with illustrative examples of microcode showing how and where you will be hurt if we take it out. Ta, ta... Severo ------- *start* 00629 00024 US Date: 9 MAY 1978 1413-PDT From: FIALA Subject: <D1>D1MidasRun.dm release To: Dorado Users Group: A new release of midas is on <D1>D1MidasRun.dm. This release adds an IOReset during a step or go at a new address. This is to fix a problem with the ethernet controller that results when Midas executes instructions on behalf of the ethernet controller's tasks. A side effect of the change is that you may inadvertently change the run-enables (RunRefresh and EnRefreshPeriod') if you have failed to execute a Reset command after firing up Midas. Please let me know if this causes any trouble. ------- *start* 00229 00024 US Date: 10 MAY 1978 0044-PDT From: BOGGS Subject: Model 0 is on the net To: Dorado Users Group: I am logged into Maxc from the Model 0 Dorado using a Chat booted from the Ethernet. /David and Ed ------- *start* 00626 00024 US Date: 30 MAY 1978 1129-PDT From: DEUTSCH Subject: New MicroD To: Micro(D) Users: cc: kierr This release has approximately 1000 words more free space. No other changes (in particular, still places CALL instructions at even locations on D0). If you encounter difficulties with MicroD at any time, first, please take the time to fetch the latest version from [Iris]<D0> or [Maxc1]<Alto>. If you still have problems, please put the .DIB file on some file server before you send me a message, and tell me in your message where I can find this file. This will save everyone time and trouble. ------- *start* 00255 00024 US Date: 13 JUN 1978 1248-PDT From: DEUTSCH Subject: New release of Micro To: Micro(D) Users: This release has roughly 1800 words more free space and can assemble the complete Dorado microcode. Please report any problems. ------- *start* 01145 00024 US Date: 19 JUN 1978 1752-PDT From: FIALA Subject: Dorado Model 0 Hardware Manual To: Dorado Users Group: The final release of the Dorado Model 0 Hardware Manual is now available as <D1Docs>D1Man-Mod0-A.press and <D1Docs>D1Man-Mod0-B.press. These two files are a revision of the 17 November 1977 release of the hardware manual which was D1manual.ears. This is intended to be the final release of the manual describing the model 0 Dorado, and none of the changes planned for Model 1 are (intentionally) documented in it. The main changes since the 17 November release are as follows: 1. Addition of Junk IO section. 2. Addition of DMux signals and other revisions to Alto Debugging Interface 3. Many changes in Figure 9 (Pipe, Config, etc.) 4. Some changes in the Control and Slow IO chapters mostly related to task switching stuff, including model 0 bugs. 5. The shifter change for D0 compatibility (not yet changed in the model 0 hardware, but will be changed soon). My intention is to start revising the manual to describe the Model 1, and the next release of the manual will be a first cut at that. ------- *start* 00260 00024 US Date: 22 JUN 1978 2307-PDT From: DEUTSCH Subject: New MicroD To: Micro(D) Users: On [Maxc1]<Alto> and [Iris]<d0>. Fixes a bug which prevented placement of some very large programs. Major internal (but no external) changes. ------- *start* 00320 00024 US Date: 25 JUN 1978 1351-PDT From: MCDANIEL Subject: new kernel on ivy, d1debugging disk To: Dorado Users Group: There is a new version of the kernel diagnostics that understands the new shifter implementation and that fixes a bug associated with using the hold and task simulator. Gene ------- *start* 01249 00024 US Date: 5 JUL 1978 1733-PDT From: PIER Subject: mea culpa plus new features To: Dorado Users Group: 1. Page 34 of the Dorado Hardware Manual - Model 0, should be amended in paragraph 1 to read: 4. Do not do H3←Input, BLOCK, BDispatch, BigBDispatch, Multiply, TaskingOn, or Tasking Off in an instruction that might cause HOLD. Do not do the branch FOLLOWING a BDispatch or BigBDispatch in an instruction that might cause HOLD. That is, do not do memory operations in the instruction with a dispatch OR in the next instruction. 2. A proposal to reduce the number of CALL locations in Model 1 has been made and seconded. The number of CALLs is halved, from one-eitghth of the total number of locations to one-sixteenth. Only instructions with the bottom four bits of TNIA equal to 0 will be detected as CALLs. This modification aids in placement of subroutines and dispatches. It reduces the total number of subroutine locations in the IM to 256 and the number of "local" subroutines to four per page. An informal survey of emulator authors felt that this was not a real restriction. This is a revokable feature if for some unanticipated reason we run out of CALL locations. It applies to Model 1 ONLY. K ------- *start* 01082 00024 US Date: 5 Jul 1978 6:17 pm (Wednesday) From: Taft Subject: "Standard" Alto Emulator To: D1Users↑ cc: Taft The sources and command files for building a basic Alto Emulator are now contained in [Ivy]<D1Sources>Aemu.dm. This includes Alto emulator, BitBlt, and I/O microcode, but not Mesa or other microcode. (D1lang and D1-Aludefs are not contained in the dump file.) File Aemu.mb is the result of assembling this, and Aemu.midas is the command file for loading it into the Dorado. The organization that Willie-Sue and I have in mind is that other emulators (Mesa, Smalltalk, Lisp) be built using the basic Alto emulator as a base. The basic emulator will contain hooks for getting into the other emulators, and these hooks will have stubs hung on them in the basic emulator. For example, the JMPRAM instruction now has a test for Ram address = 420 (the Mesa emulator entry point in the Alto), and branches to MGO if so. Aemu.dm contains a file called MesaStub.mc that diverts MGO to Trap. Presumably other emulators can receive similar treatment. Ed *start* 00241 00024 US Date: 7 Jul 1978 9:51 am (Friday) From: Sosinski Subject: Temperature Sensing To: D1users↑ There is a sample Temperature Sensing circuit for the Dorado Model 1 Boards located on ivy <D1Logic>tempSense.sil. Charlie *start* 00476 00024 US Date: 11 JUL 1978 1220-PDT From: DEUTSCH Subject: Release of MicroD 2.1 To: Micro(D) Users: Internal changes have resulted in an increase of about 4000 words in free space. No one should ever run out of space in MicroD again. A version of MicroD is in the works that will allow you to assemble parts of your microprogram separately and then run the .MB files through MicroD together to produce a single .DIB file. ETA is end of next week. ------- *start* 00248 00024 US Date: 11 JUL 1978 1222-PDT From: DEUTSCH Subject: oops To: Micro(D) Users: What I meant, of course, is that MicroD 3 will allow you to run several .DIB files together to produce a single .MB file, not vice versa. ------- *start* 00374 00024 US Date: 11 JUL 1978 1755-PDT From: WILLIE-SUE Subject: emulator sources To: Dorado Users Group: New versions of many emulator sources files. You should load from [ivy]<d1source>aemu.dm. WS changes: AT declarations for emulator at 2000+ addresses display code and BitBlt are now happy with each other disk task should not hang anymore ------- *start* 00218 00024 US Date: 11 Jul 1978 9:09 pm (Tuesday) From: Taft Subject: D1Lang To: D1Users↑ cc: Taft A new version of [Maxc]<D1Source>D1Lang.mc is released. The only change is to fix a bug in the DPF macro. *start* 01544 00024 US Date: 20 JUL 1978 1515-PDT From: DEUTSCH Subject: MicroD 4.0 To: Micro(D) Users: This new release of MicroD has one major new feature: you can now assemble parts of your program separately with Micro, and run the .DIB files through MicroD together to produce a single .MB file. For example, suppose your program is made up of 4 files called prog1.mc through prog4.mc, plus d1lang.mc, aludefs.mc, and mydefs.mc (your own definitions). One way to assemble your program might be: micro d1lang aludefs mydefs prog1 micro d1lang aludefs mydefs prog2 ... microd prog/o prog1 prog2 prog3 prog4 where the output goes on prog.MB and prog.DLS. A better way that saves more assembly time is to pre-assemble the definitions: micro savedefs/s d1lang aludefs mydefs micro savedefs/r prog1 micro savedefs/r prog2 ... which essentially saves a snapshot of Micro on savedefs.ST after assembling your definition files. (If you do do this, you must be careful to reassemble the definitions and remake the .ST file whenever you change your defs file or whenever Micro or the standard definition files change.) There is no requirement that you do things in this way: if you only have a single .DIB file, everything will work as before. I am leaving for a week's vacation tomorrow (Friday 7/21) night, so you may want to keep an older version of MicroD somewhere. The new MicroD has successfully processed the Dorado Alto Emulator, so it is unlikely that any problems will arise. Please report all problems to me. ------- *start* 00823 00024 US Date: 20 JUL 1978 2237-PDT From: DEUTSCH Subject: P.S. on MicroD 4.0 release To: Micro(D) Users: There is a known glitch in MicroD 4.0: the amount of free space it prints out at the end is always tiny (usually 1 word). Be assured that MicroD is nowhere near running out of room -- the next release will fix the message. I was asked how you declare exported (defined here, used in another file) and imported (used here, defined in another file) names. You don't! Any symbol used as a branch address and not defined in the group of files you assemble together is assumed imported: Micro gives you a list of them at the end of assembly. All labels are assumed exported. MicroD will complain if you import (use) a name that isn't exported (defined) in any of the .DIB files you give it. ------- *start* 00437 00024 US Date: 21 JUL 1978 1640-PDT From: DEUTSCH Subject: MicroD 4.0 withdrawn To: Micro(D) Users: Numerous users have encountered difficulties with this new release; hence I have withdrawn it, and deleted it from Maxc and Iris. You should retrieve MicroD from your favorite server again (you will get the previous release, 3.0). A repaired version with multi-file capability should appear in about 2 weeks. ------- *start* 00259 00024 US Date: 25 JUL 1978 1439-PDT From: SUZUKI Subject: Mesa emulator source and object files are on IVY To: Dorado Users Group: Mesa emulator source file is in MESAMC.DM at [IVY]<D!SOURCE> and object file is [IVY]<D1>EMU.MB. Nori ------- *start* 00264 00024 US Date: 1 AUG 1978 1543-PDT From: DEUTSCH Subject: Release of MicroD 4.1 To: Micro(D) Users: The new MicroD contains the promised multiple-file feature, and appears to be debugged. Documentation is on Maxc1 <ALTODOCS>MICROD.TTY. ------- *start* 00235 00024 US Date: 1 AUG 1978 1637-PDT From: DEUTSCH Subject: Re-release of MicroD 4.1 To: Micro(D) Users: A serious bug has been found and fixed. New files are on the usual places (Maxc1 <Alto> and Iris <D0>). ------- *start* 01126 00024 US Date: 2 Aug 1978 9:56 am (Wednesday) From: Suzuki Subject: Common D1 microprogramming errors To: d1users↑ cc: Suzuki This is the summary of the talk I gave at the last D1 meeting. The experience of D1 microprogramming showed that subtle bugs are mainly caused by simple mistakes not by conceptual mistakes. These bugs are very hard to find, sometimes it takes a couple of days to find, and can be mechanically picked up at source code level. Probably it is a good idea to locate anomalies which might lead to these errors by MicroD or some other software and give warnings. 1. Rbase incompatibilities. Almost all the time, if a location is the target of several jumps, or calls, then Rbase value is expected to be the same. 2. Membase incompatibilities. Similarly when a control comes to a location, Membase value is expected to be the same almost all the case. 3. Timing Shifter or R register condition branch expects the value of R register to be frozen two instructions before. If these errors are picked up by software, the debugging time would have been two weeks shorter. Nori *start* 00223 00024 US Date: 2 AUG 1978 1525-PDT From: DEUTSCH Subject: Release of MicroD 4.2 To: Micro(D) Users: 4.2 has about 4000 more words of free space than 4.1 but is supposed to be otherwise identical. ------- *start* 01676 00024 US Date: 4 AUG 1978 0942-PDT From: FIALA Subject: Model 0 Midas and Model 1 HW Manual To: Dorado Users Group: There is a new release of <D1>D1Midasrun.dm;77 on Maxc1 which fixes a bug in printout of syndrome. The first release of the model 1 hardware manual is on <D1Docs> on Maxc1. It incorporates the changes in the control and processor sections. The material for interfacing io devices formerly in the Slow IO and Fast IO chapters has been removed. It will be put in some other document eventually. The chapter on Alto Debugging Interface has also been removed, and is now in a document called "Debugging Interface" also stored on <D1Docs>. The muffler signals formerly in this chapter are being moved to the Midas documentation which isn't ready yet. Some of the hardware details in "Control Section" were also removed. Some of my editorial comments and opinions in the "Memory Section" chapter have been moved to a new chapter called "Performance Issues" and a little of the discussion on hardware errors formerly in the "Alto Debugging Interface" chapter have been moved to a new chapter called "Error Handling" which also discusses the new error handling stuff by the processor. Still missing from the hardware manual are the following: 1. Chapters on "Disk", "Display Controller", and "Ethernet" 2. TIOA device numbers assigned in the "Slow IO" chapter. 3. Task numbers used by standard peripheral device handlers in the "Slow IO" chapter. 4. All the new stuff for the "IFU section" chapter. I expect to have the Model 1 Midas manual ready later today and the Model 1 Midas and Assembler ready next week or the week after. ------- *start* 00301 00024 US Date: 10 AUG 1978 1547-PDT From: PIER Subject: EclDict.analyze To: Dorado Users Group: Latest version on Maxc<sil>EclDict.Analyze. Minor corrections to timing analyzer section only. Thanks due to Will Crowther for reviewing the analyzer section for correctness. Ken ------- *start* 00560 00024 US Date: 11 AUG 1978 2003-PDT From: FIALA Subject: [Maxc1]<D1Docs>D1Assem-model1.press;1 To: Dorado Users Group: This document describes the Dorado model 1 microassembler which is now released. MicroD will require some changes before the new assembler's output can be loaded. You may wish to read the appendix that describes differences between the model 0 and model 1 hardware/assembly language before doing anything else. This assembler is in approximate agreement with the model 1 hardware manual released on 1 August 78. ------- *start* 00342 00024 US Date: 14 AUG 1978 0249-PDT From: FIALA Subject: <Altodocs>Micro.press To: Dorado Users Group:, hankins, charnley I have resurrected and revised the ancient Micro document formerly called Maxc 9.2. It is now considerably cleaned up and online. Sources for this are <D1source>micro1.bravo and micro2.bravo. ------- *start* 01871 00024 US Date: 16 AUG 1978 1712-PDT From: WILLIE-SUE Subject: Alto and Mesa emulators To: Dorado Users Group: The 'standard' emulator for the Dorado is now released. It comes equipped with Mesa and a reasonably fast Alto emulator. Retrieve [ivy]<d1>emu.mb and [ivy]<d1>emu.midas. Fire up Midas and type EMU, then bug READ-CMDS. When Midas has finished loading the code, type INITMAP;G. (If the code breaks at NOMSA, it thinks there is no main storage - call me or try again.) When the code breaks at DONEMSA, type ;P to initiate the boot sequence. As on the Alto, the Dorado will boot from the disk, or, if the same various keys are held down, etherbooting will occur instead. To restart the code after a breakpoint or ↑C interrupt, set TASK to 16b and type BOOT;G (keys comment above applies here too). NOTE: the necessary parts of the command file POWERON have been incorporated into EMU.midas. New feature: if ALL 3 mouse buttons are held down, the Dorado ucode will go into its boot sequence, just as if you had stopped the Dorado and typed BOOT;G to Midas (with TASK having been set to 16b). Note: the code does NOT check if the disk is idle before booting, SO BEWARE. The rest of this message is of interest to ucode writers only. Sources for Emu.mb are on [ivy]<d1source>EmuSources.dm, along with command files for assembling various pieces of the total system (to take advantage of the new MircroD feature). The .ST and .DIB files are on [ivy]<d1source>Emu-dibs.dm. These were assembled with [ivy]<deutsch>Micro.run, which makes a .ST file incompatible with [maxc]<alto>Micro.run; you will get an error message if you have the wrong Micro. A detailed exposition of how the sources and command files are organized is on [ivy]<d1source>MakingEmu.txt. If you have any questions or complaints, please see me or sndmsg. WS ------- *start* 00565 00024 US Date: 18 AUG 1978 1302-PDT From: DEUTSCH Subject: New release of Micro To: Micro(D) Users: cc: ludolph This long-awaited release fixes all known glitches and offers a speed improvement of 20-25%. There is also an on-line reference manual now, on Maxc1 <Altodocs>Micro.Press. The release is on Maxc1 <Alto> and Iris <d0> as usual. Please report problems. There are some new features in this release which will be used by the D1Lang for the Dorado model 1, and possibly by future versions of D0Lang, so you should get it soon. ------ *start* 00373 00024 US Date: 19 AUG 1978 0247-PDT From: MCDANIEL Subject: new version of memA To: Dorado Users Group: There's a new version of the memory diagnostic on the debugging disk and in [ivy]<d1>memA.dm. The address test bug and the fault task bug have been fixed. The diagnostic runs without error on model 0 when error correction is turned on. Gene ------- *start* 00263 00024 US Date: 22 AUG 1978 0921-PDT From: FIALA Subject: [Maxc1]<D1>D1midasRun/model0.dm;78 To: Dorado Users Group: This new release expands the symbol table on the disk for Gene and adds the changes to Midas.Programs which Ken gave me. ------- *start* 00181 00024 US Date: 23 AUG 1978 1518-PDT From: DEUTSCH Subject: New release of Micro To: Micro(D) Users: This release (dated 8-23-78) just fixes two minor bugs. ------ *start* 00236 00024 US Date: 29 AUG 1978 1708-PDT From: JENKINS Subject: Micro.dl To: Micro & Micro(D) Users: cc: Jenkins The name of the above list has been changed, at Peter Deutsch's request, to Micro&MicroDUsers.dl ------- *start* 00284 00024 US Date: 30 AUG 1978 1133-PDT From: JENKINS Subject: The ampersand &&&&&&& To: Micro-Micro(D) Users: cc: Jenkins Please note that I have replaced the "&" in your distribution list name with a "-" and it now reads: Micro-MicroDusers.dl Jeannette ------- *start* 00225 00024 US Date: 8 SEP 1978 1011-PDT From: OVERTON Subject: Dorado Backpanel Drawings To: Dorado Users Group: The current Dorado Backpanel Drawings are stored on [ivy]<D1Drawings>DoradoBPDwgs.press. ------- *start* 00345 00024 US Date: 14 SEP 1978 1906-PDT From: JENKINS Subject: The name of the list has been changed again .... To: Micro-Micro(D) Users: cc: Jenkins Micro-MicroDUsers.dl has been changed again to Micro.dl. All these name changes have a purpose. Please believe me even if you don't understand it. Thanks. Jeannette ------- *start* 00299 00024 US Date: 18 SEP 1978 0952-PDT From: DEUTSCH Subject: Release of MicroD 6.0 To: Micro-Micro(D) Users: This version has been on my Ivy directory for 3 weeks and seems solid. It fixes a couple of bugs related to multiple .DIB files, and also supports the Dorado model 1. ------ *start* 00328 00024 US Date: 19 SEP 1978 1729-PDT From: DEUTSCH Subject: Release of MicroD 6.1 To: Micro-Micro(D) Users: Now on Iris and Maxc. Removes the previous limit of 10 .DIB files, and also a bug which caused a Swat trap if there were 8 or more .DIB files. There is now no limit on the number of .DIB files. ------- *start* 00190 00024 US Date: 22 SEP 1978 1030-PDT From: SUZUKI Subject: Mesa 4.1 on Dorado To: Dorado Users Group:, geschke, levin Mesa 4.1 now runs on Dorado and tested. Nori ------- *start* 00609 00024 US Date: 30 SEP 1978 2229-PDT From: DEUTSCH Subject: Release of MicroD 7.3 To: Micro-Micro(D) Users: This release incorporates the ability to distinguish 3 different versions of the D0 hardware (with "OR 1" vs. "+1" returns from CALLs, and with slow vs. fast page switching). Be sure you assembled your program with the right version of D0LANG! There are also minor cosmetic improvements, and some changes in the placement algorithm to allow fuller packing of the Dorado memory. I will be gone from 10/1 through 10/8 and will not be reading my on-line mail during that time. ------ *start* 01192 00024 US Date: 12 OCT 1978 1227-PDT From: FIALA Subject: Dorado Model 1 HW Manual Release To: Dorado Users Group: A new release of the model 1 Dorado Hardware Manual is available on [Maxc1]<D1Docs>D1Man-mod1-a.press and D1Man-mod1-b.press. The major changes are as follows: 1. Branch new writeup on the IFU, believed to reflect the current design pretty accurately. 2. Some renumbering of FF decodes for the processor section. 3. Changes to FF-controlled shifts (source for B on FF-controlled shift is now Q; single-register cycles of T and R are now possible). 4. Addition of the IFetch← and LongFetch← operations in the memory section and rearrangement of the ASEL-FF[0:1] decoding for memory references. 5. Some revisions to the Control and Memory section chapters. 6. Nomenclature changes: "Pd" replaces "H3" everywhere; "Id" replaces "IFUData"; "opcode" replaces "macroinstruction"; obsolete uses of "H1" and "H2" removed; "CPReg" replaces "ABMux"; registers with inverted readout renamed with "'" (e.g., "NotPipe2" renamed "Pipe2'"); IFU section names different: "PCF←B" replaces "NewPC←B"; "IFUMLH" replaces "DecHi"; "IFUMRH" replaces "DecLo". ------- *start* 00325 00024 US Date: 12 OCT 1978 1230-PDT From: FIALA Subject: [Maxc1]<D1Docs>D1midas-model1.press To: Dorado Users Group: A new release of the Dorado model 1 Midas document is available. Only significant change is the revision of the DMux signal table to incorporate changes in the ProcH/L DMux signals. ------- *start* 00270 00024 US Date: 12 OCT 1978 1625-PDT From: DEUTSCH Subject: Release of MicroD 7.4 To: Micro-Micro(D) Users: This release has a better and slightly faster placement algorithm. No other changes. Available on [Iris]<d0> and [Maxc1]<Alto> as usual. ------ *start* 01344 00024 US Date: 12 Oct 1978 5:21 pm (Thursday) From: Taft Subject: Dorado BitBlt interface To: D1Users↑ I am considering ways of modifying the BitBlt subroutine to permit accessing extended memory. At present, BitBlt uses the base register in effect at the time of the call as the base for the BBTable and the source and destination blocks. I would like to make a change that doesn't perturb the current interface too much and yet is flexible enough to be used by all the language interpreters, which have different ideas about how to manage base registers. Here is my proposal. BitBlt will continue to assume that the BBTable lives in the space defined by the base register selected by MemBase at the time of the call, as before. There will be two new base registers, to be used by BitBlt only, that define the spaces within which the source and destination blocks live. These base registers must be set up by the caller. The source and destination bit map addresses in the BBTable will be interpreted relative to the respective base registers. Upon return from BitBlt, the caller must not assume anything about MemBase. This means that the caller must interpret the extended memory flags in the BBTable and read the bank bits (or whatever) in order to set up the source and destination base registers for BitBlt. Ed *start* 00499 00024 US Date: 14 OCT 1978 1551-PDT From: MCDANIEL Subject: new memA on debugging disk To: Dorado Users Group: By using Peter's new MicroD I'm able to generate a version of the memory diagnostics that includes all the board tests. That diagnostic is the one on the debugging disk and the one that can be found on [ivy]<d1>memA.dm. It includes the fio test, which does not run unless specifically enabled by the operator. The documenttion in the d1 lab discusses this. Gene ------- *start* 00392 00024 US Date: 17 OCT 1978 1321-PDT From: MCDANIEL Subject: new kernel on debugging disk To: Dorado Users Group: Roger has fixed the task and hold simulator logic in Model 0. There is a new kernel on the debugging disk (and on <d1>) that will use those simulators. The default situation is to not use them, however. See the documentation for more instructions Gene ------- *start* 00746 00024 US Date: 20 OCT 1978 1513-PDT From: FIALA Subject: <D1Source>D1Lang-model1.mc;10 and <D1Source>D1alu-model1.mc;5 To: Dorado Users Group: This new release of the model1 microassembler includes revisions to handle the hardware as described in the October 11 HW manual plus several other changes that have occurred since then. The new assembler is documented on <D1Docs>D1Assem-model1.press. However, note that the dispatch table revisions to the assembler are not implemented yet; continue to use AT and GLOBAL for placement declarations until further notice. A list of 50 or so hardware, assembler, and other differences between the model 0 and model 1 Doradoes is listed in an appendix to the assembler manual. ------- *start* 00346 00024 US Date: 23 OCT 1978 1616-PDT From: MCDANIEL Subject: please modify your buildbackuptemplate.cm To: Dorado Core Group: We agreed a while ago to keep prescan listings on [ivy]<doradodrawings> for each board. This will only happen if you modify your buildBackupTemplate accordingly. So, please modify ....etc. Gene ------- *start* 00222 00024 US Date: 23 OCT 1978 1717-PDT From: FIALA Subject: <D1>D1MidasRun.dm;6 is released To: Dorado Users Group: Fixes bugs on symbolic instruction type-in and hexadecimal io for baseboard stuff. ------- *start* 00224 00024 US Date: 24 OCT 1978 1355-PDT From: DEUTSCH Subject: Release of MicroD 7.5 To: Micro-Micro(D) Users: This release fixes a bug which garbled the JCN field for Dorado Model 1. No other changes. ------- *start* 01180 00024 US Date: 24 Oct 1978 3:23 pm (Tuesday) From: Pier Subject: Model 0 environmental upgrade To: DoradoCore↑ cc: Horn, Deutsch, Yeary, Maleson We now have the terminal hardware multiplexers and cables to allow us to share terminals among Altos and Dorados. I have asked Mike Overton to install the appropriate hardware to give us two sets of Dorado/Alto terminals. One set lives inside the Model 0 room, and looks just like the current set of terminals. The other is on a table where Viola's terminal is currently. Viola's printer will be moved to near the center of the graphics lab and its Alto/Trident will be moved to the north wall (currently unused). I would like to erect a partition between the Viola area and the new Dorado programmers area. I think it is important that we do this now, in order to test out the feasibility of the multiplexer idea for the future and in order to make programming more reasonable and pleasant (not perfect, but better than the currently terrible noisy environment). If programmers prefer to work in the lab, this arrangement allows them to do so. Unless objections are promptly forthcoming, we will plunge ahead. Ken *start* 00314 00024 US Date: 30 OCT 1978 0815-PST From: PIER Subject: Proposed new Dorado clock scheme To: Dorado Core Group: Error correction: the gate at the bottom of BoardClocks.sil which generates IgnoreProcessorClocks should be an MC210 type non inverting gate. An inverting gate is shown. Ken ------- *start* 00386 00024 US Date: 30 Oct 1978 2:51 pm (Monday) From: pier Subject: Proposed new Dorado clock scheme To: DoradoCore↑ Error correction: the edge pin for MemClkEn' for boards I C X D shown on CharClocks.sil should be 12, rather than 8. Clarification: boards I C X D will receive MemSH on pin 5. All IO boards will receive MemSH'. Keep those cards and letters coming. Ken *start* 00165 00024 US Date: 1 NOV 1978 0933-PST From: FIALA Subject: <D1Source>D1Lang-model1.mc;12 To: Dorado Users Group: Is released. Fixes 3 bugs. ------- *start* 00387 00024 US Date: 3 Nov 1978 2:20 pm (Friday) From: Pier Subject: New Dorado clock scheme To: DoradoCore↑ Doug Clark has pointed out that we should standardize the edge chip locations wherever possible. Consequently, the SE231 should place at location l2 and the SE210 receiving CLK' at location l1. We are simply following the existing Proc boards in this selection. Ken *start* 00214 00024 US Date: 4 NOV 1978 2041-PST From: FIALA Subject: <D1>D1midasrun-model0.dm;80 To: Dorado Users Group: is released. Fixes several bugs in the DMux simulator and in the ProcVA test. ------- *start* 00556 00024 US Date: 5 NOV 1978 1210-PST From: DEUTSCH Subject: MicroD 7.6 To: Micro-Micro(D) Users: The changes in this release only affect the Dorado model 1. PREVIOUS MICROD RELEASES DO NOT WORK CORRECTLY FOR THE DORADO MODEL 1. Furthermore, a minor hardware change is required for release 7.6 to work correctly. DO NOT ATTEMPT TO RUN DORADO MODEL 1 PROGRAMS OUTPUT BY MICROD UNTIL THIS CHANGE IS MADE. This is a matter of at most a few days, according to Ken Pier. Another message will inform you when the change has been made. ------ *start* 00307 00024 US Date: 8 NOV 1978 1613-PST From: DEUTSCH Subject: MicroD 7.7 To: Micro-Micro(D) Users: This release is of interest only to Dorado users. It fixes some bugs relating to the IFU memory. The hardware change to the Model 1 is now complete: MicroD and the hardware now agree. ------ *start* 00337 00024 US Date: 13 NOV 1978 1516-PST From: CHANG Subject: ECLDict.Analyze To: Dorado Users Group: cc: Chang Changed the number of pins for MB071 from 24 to 26 (pin #25 and #26 are GNDs'). Also, changed IC type of F181K from "E45W" to "F100K". All changes are made in the HEADER portion of the dictionary. Tom ------- *start* 00268 00024 US Date: 14 NOV 1978 1649-PST From: DEUTSCH Subject: MicroD 7.8 To: Micro-Micro(D) Users: Now released. Fixes a bug on the Dorado model 1 -- fast conditional branches were sometimes getting changed into long unconditional branches (!). ------ *start* 00306 00024 US Date: 21 NOV 1978 1808-PST From: FIALA Subject: Midas release To: Dorado Users Group: A new release of the Dorado model1 Midas is available on Maxc <d1>. This release fixes several IFU bugs and has prettyprinting changes for some muffler signals that have been modified. ------- *start* 00410 00024 US Date: 22 NOV 1978 1537-PST From: PIER Subject: Buildbackuptemplate To: Dorado Core Group: The latest Maxc<sil>DoradoBuildBackuptemplate.cm illustrates use of several new features, including the loading chart file BOARDNAME01.lc now issued by ROUTE and the generation and storage of prescan files along with press files. Please update your backuptemplates appropriately. Ken ------- *start* 00358 00024 US Date: 27 NOV 1978 1827-PST From: FIALA Subject: Midas Release To: Dorado Users Group: A new release of Dorado model1 Midas is available on Maxc1 <D1>. This release has almost all the changes needed to support MemC, MemD, and MemX and incorporates all decoding changes that I am aware of. Several other bugs are also fixed. ------- *start* 00311 00024 US Date: 5 DEC 1978 1336-PST From: DEUTSCH Subject: New Micro To: Micro-Micro(D) Users: Fixes a bug which caused infinite looping or other strange behavior at the end of assembly when storage was nearly full. No other intended changes. On [Maxc1]<Alto> and [Iris]<d0> as usual. ------ *start* 01335 00024 US Date: 16 DEC 1978 1244-PST From: MCDANIEL Subject: you'll love it To: Dorado Core Group: Date: 15 DEC 1978 1952-PST From: STROLLO Subject: A Dorado Smalltalk experience To: Smalltalk Users Group: cc: goldstein I had the opportunity tonight to try out Dorado Smalltalk on my Smalltalk Calendar system. It is impressively speedy. A loop which I have timed pretty carefully on my Alto at 15 seconds takes a shade over 3 seconds on Dorado. This is more impressive when you consider I am using XM Smalltalk on my Alto and had to use the 64K AI/AII (latest WFS release) version on Dorado whose emulator does not yet support XM Altos. On other things which I have timed the speed up factors range between 2 and 5. To those not familiar with the way Smalltalk is being emulated on Dorado, realize that the Alto and the Nova code and all that are being emulated. The raw Dorado horsepwer is of course much more than the trivial factors quoted above. Most important it feels acceptably fast, and, since the average number I am getting out of people for Smalltalk on NoteTaker is in the 2 to 5X Alto Smalltalk range, Smalltalk Calendar will be a useable tool on the NoteTaker. I left the file Calendar.Boot on the simulated 44 disk of the Dorado in case anyone would like to play with it. Ted ------- ------- *start* 00327 00024 US Date: 16 DEC 1978 1330-PST From: MCDANIEL Subject: don't acquire a new midas for model1 To: Dorado Core Group: [maxc]<d1>d1midasrun-model1.dm;17 written on 11 dec 78 will not initialize -- it goes into swat. Keep your current version of midas until Ed gets a channce to fix the problem. Gene ------- *start* 00664 00024 US Date: 18 DEC 1978 0859-PST From: MCDANIEL Subject: testmodel0.midas To: Dorado Users Group: I've updated the midas.run on the model 0 debugging disk to a new version so it can run the command file, "testModel0". This command file sequences the machine through the microdiagnostics that are currently available. Unfortunately, any error that gets detected will cause the command file to halt. TestModel0.midas is available on the model 0 debugging disk. This file causes the diagnostics to execute in an infinite loop. It may be useful if you suspect a hardware problem. Try ruknning it when you are done with the machine. Gene ------- *start* 01547 00024 US Date: 19 Dec 1978 2:33 pm (Tuesday) From: Taft Subject: Model 1 instruction placement To: D1Users↑ This message is to warn you of a problem in Dorado model 1 instruction placement which you may encounter while converting model 0 microcode to run on model 1. I hope this warning will save you some time. In the hardware, there are potentially 14 pairs of targets for JCN-encoded conditional branches in each page. However, MicroD is presently able to make use of only 7 of them. Furthermore, if you have a CALL at the true target of a JCN-encoded conditional branch, the .+1 constraint on the next instruction may make another conditional branch target pair unusable. Unfortunately, the changes required to fix this are not trivial. This may eventually be fixed when the dispatch table mechanism is implemented, but there are no immediate plans to do that. In the meantime, you may have to leave FF free in more instructions than you did in model 0. The symptom of having too many JCN-branch targets in one page is that MicroD gives the error message "ran out of pages attempting to assign ring:" followed by a bunch of instruction addresses. Among those addresses will be all the targets of JCN-encoded conditional branches that have to be on the same page. Note that if you leave FF free in an instruction containing a conditional branch, the branch will be encoded in FF rather than JCN, so it won't consume a JCN-branch target pair. My thanks to Peter for figuring this all out and explaining it to me. Ed *start* 00250 00024 US Date: 22 DEC 1978 1902-PST From: FIALA Subject: Model 1 Midas Release To: Dorado Users Group: New Model 1 Midas is released. Fixes Map read-write problem and several other bugs. Prettyprint of errors is enhanced. ------- *start* 00626 00024 US Date: 26 DEC 1978 1236-PST From: MCDANIEL Subject: new mema for model 1 To: Dorado Core Group: [ivy]<dorado>mema.dm contains a new version of the memory diagnostics. There are two main changes: 1) The cache patterns now include a left cycled zero bit. 2) The xBoard tests include a map addressing test. To simplify remembering how to acquire versions of mema, I've left two new command files on the model1 debugging disk. "NewDoradoMema.cm" will automatically load mema from [ivy]<dorado>mema.dm and "NewMcDMemA.cm" will automatically load mema from [ivy]<mcdaniel>dorado>mema.dm. Gene ------- *start* 00550 00024 US Date: 29 DEC 1978 1234-PST From: DEUTSCH Subject: Release of MicroD 7.9 To: Micro-Micro(D) Users: Another change in the placement algorithm. If you have a Dorado microprogram that fills more than 7/8 of IM, you should run it through the new MicroD to make sure it will still place. There is a new global and local switch /A, which lists only the locations of instructions which are absolutely placed (i.e. with an AT[...] clause). /A, /C, /L (local), and /N are mutually exclusive; local switches override global. ------