(Claret a Ct ne oe EE STANFORD MEDICAL CENTER User Note J-32 MAR 30 197; Computing Newsletter Published Irregularly by the ACME Staff Vol.6 , No. 2 March 25, 1971 CONTENTS Item Page 1. Request for News _ 1 2. Recent Publications by ACME Users ~ 1 3. PL/ACME to IBM PL/1 Translator . J 1 4. Link Between ACME and Campus Facility Computers _ Ll 5. Overnight Service -L 6. ACME Grant Status 2 7. Communications Modems 2 8. ITEL ¢741 Type Terminal with Paper Tape Punch and Reader . . 2 9. Record Size Increase on ACME File System . - 2 10. New Plotting Program GRAPHH a) ll. Problem with Statistical Subroutines - 3 12. CLEAN Option on Subprogram Statement a) 13. Segment Large Programs 3 U4. ACME Library Function ERF(x) . . . 4 15. Some Statistical Subroutines placed in Overlay Segment 4 16. Improved Accuracy of Statistical Subroutines 4 17. Faster Program Compilation and Execution . a) 18. New and Updated ACME Notes Gince Last Newsletter . a) J-32 Page 1 lL. Request for News We are attempting to broaden the coverage of this newsletter to the entire Medical Center. Small machine users, Nospital ADP, medical users of the Sigma 5 at SEL, and other computer users are urged to send news items to Cindy Miller at ACME. 2. Recent Publications by ACME Users Several ACME users have had the results of their research published in re- cent months. The ACME staff appreciates notification of such publications and copies of reprints when available. Some of the most recent publications we have been made aware of include the following: A) Harman, Dr. Charles E., and Christopher S. Raymond, "Computer Prediction of Chronic Psychiatric Patients", The Journal of Nervous and Mental Disease, Vol. 150, No. 6, pp. 4¥90-530, Copyright 1970. B) Melges, Dr. Fredrick T., Dr. Jared R. Tinklenberg, Dr. Leo E. Hollister, and Hamp K. Gillespie, "Temporal Disintegration and Depersonalization During Marihuana Intoxication", Archives of General Psychiatry, Vol. 23, September 1970. _ C) Fries, Dr. James F., "Experience Counting in Sequential Computer Diagnosis", Archives of Internal Medicine, Vol. 126, pp. 647- 651, October 1970. 3. PL/ACME to IPM PL/1 Translator A new translator from PL/ACGs& to IBM's PL/1 will be canpleted and documented by mid April. The new translator will permit programs developed on ACME to be transferred to any large IBM 360 system in the country that supports IBM's PL/1 compiler. The translator should prove to be a great aid tc people transferring to other universities. It should also permit batch operation on the Campus Facility for production oriented users of programs developed under ACME. The User Note covering this topic will be BER-1. 4, Link Between ACME and Campus Facility Computers Hardware for the link between the ACME PDP-~1l and Campus Facility PDP-9 is nearly complete. It is hoped that by around the middle of April, ACME users will be able to transfer files to ORVYL on the 360/67 at Campus Facility. This will permit ACME users to run production programs in batch mode on the 360/67. In addition, Campus users will be able to transfer files from disk storage at Campus to ACME. The next ACME User Note will explain in detail how the link can be used. 5. Overnight Service For the past two months ACME users have been given a new overnight service. The individual user places his job in the overnight queue and an ACME operator runs the job during swing or owl shift. The advantage to the user is that page- minutes are provided at a time when very few users are on the system. Tn gen- eral this means that fewer pageminutes are used and therefore the result in cost is less. Response to the new service has been surprisingly high. Due to the J-3e2 Page 2 amount of systems effort that must be accomplished during the swing and owl shifts, overnight turnaround cannot be guaranteed. To take advantage of this service use the public program OVERNIGHT. 6. ACME Grant Status A nunber of users have inquired recently as to the status of the ACME Grant from NIH. The Biotechnology Resources Branch of NTH provided ACME a three year extension which will expire at the end of July, 1972. Tt may be possible to obtain an extension of approximately one year to the existing grant. Beyond that time a new method of supporting ACME must be found or the resource constituting ACME may have to be redefined. 7. Communications Modems The Bell Telephone Company 104A data sets in the ACME machine room have been replaced by Prentice Flectronics Company modems. This transition has oc- curred over the past 2 1/2 weeks. Some ACME users who enter the system via telephone lines have encountered difficulty with the new hardware. The shake down period should now be over; any further difficulty with the communications hardware should be reported promptly to the operator at extension 59043. The purpose of this change incidentally was to reduce annual operating costs. 8, ITEL 2741 Type Terminal with Paper Tape Punch and Reader The Infectious Disease Laboratory (Dr. Petralli and Ned Russell) has re- cently obtained through ACME an ITEL 1051 data terminal. This terminal is com- patible with the 2741 but has as an added feature the ability to punch and read paper tape. Paper tape can be punched off line from ACME and entered to the system at 15 characters per second, the normal operating speed of a 2741. Addi- tional such units can be provided at a monthly service fee of $280. This corres- ponds to the current service fee of $225 charged to holders of e74ls. This fee incidentally includes services in addition to hardware rental. 9. Record Size Increase on ACME File System On Wednesday, March 24, 1O71, the maximum record size on ACME will be in- creased from the current 1968 characters (492 numbers) to 65545 characters or 16385 numbers. This means that you will be able to write records which are larger than block size. However reading, writing, and re-writing large records will still take a fair amount of time -- any records larger than the old res- triction (block size: 1968 characters or 49° words) is a spanned record (uses multiple blocks) which will require multiple disk accesses. The ATTN key is disabled during the entire write. The following rules still apply: 1) A Single character string is not al- lowed to be larger than 1280 characters long (this is not true for character string arrays) and, 2) the end-of-string characver must still be included in calculations to determine the length of a string or string array. Example: Declare a(2) char(5) means that a is 12 characters Long. 7 NO AN PD Pare ° a The following are examples of what may be written: Declare a char(30); Declare b (16385) char(3); Declare c (16%84); Declare d (50); Declare ec doubi.; Declare f (500) char(10); Declare 1 r, 2a, 2d,2e; Declare 1 s, 2 f, 2 d3 The following are examples of variables too large to be written: Declare noa (70000); Declare nob (16500) char(3); Declare 1 nos, 2 noone (10000), 2 notwo (6000) char(3), 2 nodrei (383), 2 nono double; 10. New Plotting Program GRAPHH This program will plot an entire gravh complete with grid, grid labels, titles and plotted data on a display or plotter, given two vectors of coord- inates and the titles to be used. It is written in PL/ACME and is used as an EXTERNAL PROCEDURE. For further details, see ACME Note EG-1. ll. Problem with Statistical Subroutines During the last month, sore of the statistical subroutines would "my steri- ously" stop with error message 9184. This error has now been fixed, 12. CLEAN Option on Subprogram Statement The CLEAN option on the Subprogram statement has been modified to remove the Start and End of compilation messages, as well as the line numbers, from listings. The most obvious use of the feature is that by saying -- SUBPROGRAM name CLEAN NOLIST; a subprogram may be canpiled with absolutely no canpilation messages or Lines of text appearing on the terminal. 15. Segment Large Programs Users with large programs which stay in core for long periods of tire (and consequently cost large anounts of money) might be advised to segment their programs in two or more stsaller programs. ‘These smaller programs could trans~ mit data to each other through data or text files and each program in the se- quence could automatically call the next program when it is needed. The CLEAN option mentioned in the previous paragraph, can be used to suppress compilation messages for the succeeding programs. An ACME consultant may be able to help you segnent your programs if you need help. J-32 Page 4 Vs. ACME Library Function ERF (x) One of the mathematical functions available in the ACME Function Library is x 2 BRP(x) = = eat In the update to the PL/ ACME manual, issued in November 1970, users were asked to change the constant in this formula. This instruction was incorrect. Users are advised to restore the formula on page 2.3.5.2 of the manual to the above form. 15. Some Statistical Subroutines placed in Overlay Segment The following statistical subroutines have been placed in overlay segment: ANOVAL, ANOVA2, AUTO, CLRATTO, CROSS, CROSTAB, EIGEN, CDATA, MULTR, ONETAB, ORDER, and TRANSPOS. This means that not all of the above subroutines are kept in core at any one time. Consequently, when a user wishes to use one of then, he might have to wait an extra time around the commutator for the subroutine he wants to be brought in from disk. Our studies have shown that the above subroutines have very low use counts. Putting them into an overlay segment was a favorable al- ternative to taking them out of the system since they are still available should they be desired, and the time waiting for them to be brought in off of disk should be negligible for the average user. 16. Improved Accuracy of Statistical Subroutines ACME Subroutines ATOVAL, ANOVAS, BASTAT, ONETAB, SIMREG, and AUTO have been modified to improve the nwnerical accuracy of their results with regard to com- puting means, standard deviations, sums of squares, and suas of cross products. Previously, these subroutines gave seriously inaccurate results in certain cases. The error was worst when the mean of the data was very large compared with the standard deviation. If the ratio of the mean to the standard deviation was of the order 16k, then 2k significant digits might be lost. (Sixteen because ACME's 460 is a hexidecimal machine.) If the data was at least roughly normally distributed, then about 95% of the observations should fall within two standard deviations of the mean, and a visual check of the data would indicate whether a serious loss of accuracy had cecurred. User Note J-30 dated December 1, 1970, explained similar changes for the ACME Subroutine BASIC, but there was an error. The article in J-30 said, "If the ratio of the mean to the standard deviation was of the order 16k," but it should have said, "Tf the ratio of the mean to the standard deviation was of the order 16K," as in the paragraph immediately preceeding. The accuracy of ANOVAL, ANOVA®, BASTAT, ONTTAB, SIMREG, and AUTO has now been significantly improved. For a detailed description of the changes made in these subroutines see their respective ACME Notes: EAT-6 for ANOVAL, EAU-6 for ANOVA2 , RAG for BASTAT, EAC-5 for ONETAB, FEAL-6 for SIMREG, and FRAR-& for AUTO. Tf you have any questions please contact Jane Whitner, extension 6120 or Bob Hale at extension 6126, Page 5 17. Paster Program Compilation and lxecution After you huve tested and debugeed your progrem use "PROGRAM CRUNCH PUPLIC" to crunch it into a newfile. To easily achieve the maximun line length of 200 characters, respond to line length prompt with the word STANDARD. See ACME ree EDR. If there are any questions, please contact Bob Bassett at extension 120. 18. New and Undated ACME Notes Since Last Newsletter New PSEC-1 File Security (G. Wiederk 101d), February 25, 1971 HPDP11-1 PDP-11 to 360 Connection (Van der j.ans/Osborne), February 243, 1971 BG-1 ACME Subprogram GRAPHH: Creates Display and Plotter Output (ale), March 22, 1971 Updated CT-13 ACME Terminal Listing (Class), March 10, L971 EAR-4 ACME Subroutine AUTO (Shih/whitner), February 10, 1971 EAT-6 ACME Subroutine ANOVAL (Liere/Whitner), March 17, 1971 BAU-6 ACME Subroutine ANOVA2 (Liere/Whitner), March 12%, 1971 EBG-3 ACME Program Library JACXNIFS: Confidence Limits for a Ratio of Two Means (Mocre/Whitner}, February 11, 1971 EFB-2 ACME Program Library LISTAKER: Listing/P Punching Service (Bassett), February 1, 1972 KL-3 Tables in ACME Dependent on System Perameters (Wiederhold/Girardi), February 23, 1971 OL-7 Loading ACME, ACME29, and ACMES2 Systems (Class, Granieri, Sutter), February 19, 1971 WPDA-2 2e7Ol PDA: External Devices Connection (Girardi/Class), February 17, 19/1