1

Topic: Re: "here my honor is touched"

2

Re: Re: "here my honor is touched"

Hello, , you wrote: > Nonsenses. There is a standard, for it there is a system of tests. And Russian Vicks writes the following: It contains such wide dial-up of syntactic constructions and intrinsic functions that, possibly, there is no compiler supporting all possibilities of language /1. But really, it is not simple so after all there was subset PL/1. I only remember that in its alphabet there were 48 characters instead of 60. I rarely refer on Vicks. But this information corresponds to that is in other sources. At me, for example, since ancient times Frolov's book and Oljunina was saved. > By the way, in language of 10 arithmetical and 22 mathematical intrinsic functions. Whether it is small for the English dictionary? In the book mentioned by me (2 issuing, M.Nauka, 1987) on page 22-24 is resulted the table in which the transcendental intrinsic functions are enumerated only. It is more than them 40. And after all in PL/1 more many that is. Itself told: One  for all occasions. And if syntactic words to add, so English also does not suffice. I can refer to the brief experience of usage PL/1 still. > And on what it is done by the remaining? Implementations on integer numbers somewhere came across to me.

3

Re: Re: "here my honor is touched"

Hello, , you wrote: > the transcendental functions only on page 22, remaining pages - conversions which it is full in any language And even lines, bit fields... In total also you will not remember. Here input-output has been made by operators, yes. > And it works faster built in commands DAS and DAA? Then what for developers of 8086 them entered? Just in case. How much I remember, 8086 any data treats as binary. But just in case entered some instructions and for BCD. DAS and DAA it is necessary to cause each time, how much I remember. On each 4 or 8 bits. Therefore and long. The number fixed can be much longer. If to dream up and implement fixed on a basis of pair the whole? In one to store an integer part, in other - fractional? Or, for example, to take 32-bit whole. 24 bits to lead out under the whole honor, remaining 8 - under fractional? Though all the same to consider  on PL/1 longer, than on the Fortran.

4

Re: Re: "here my honor is touched"

Hello, , you wrote: > And it works faster built in commands DAS and DAA? Certainly faster. > then what for developers of 8086 them entered? A heritage 8080... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

5

Re: Re: "here my honor is touched"

Hello, , you wrote: > Checked? Studied circuitry. Not 8086 or 8080, and simply circuitry. That will do. > and you the expert. The code of command DAS in 8080 do not prompt? I will not prompt, never used. Once for the general development looked at architecture 8080, support binary-decimal  even in 8080 looked ridiculously and ridiculously and except thoughts that on  it is possible to make the calculator, caused nothing. Or you what commands in 8080 and 8086 does not coincide, and in 8080 all are easier and is restricted to binary-decimal correction and transfer bit between  in a state word? > actually commands of operation with BCD (AAA, AAM, AAD, AAS and DAS) just in 8086 also appeared. 8086 in tens times it is more difficult 8080, there there was a possibility to implement all whims. > when it was clarified that the heritage of Kobola is difficult for supporting It was clarified that these commands are necessary to nobody also them did not use. And AMD them without  regrets threw out from 86-64, Intel joined this decision.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

6

Re: Re: "here my honor is touched"

Hello, , you wrote: > These commands not for the calculator, and for calculations with controllable accuracy in economic calculations. Bothered to explain. For calculations with controllable accuracy in economic calculations they it is better nothing . But many times more slowly and more difficult in circuitry. > the Intel was necessary to buy the license for architecture at AMD with all their jambs after them Itanium turned in Titanic. It was necessary, only it not a lack. > at AMD as appeared commands of control of overflow INTO and output control over boundaries BOUND are thrown out. Here I will tell nothing, I do not know their singularity. But likely it is necessary to look at frequency of usage of these commands in the real code. AMD likely , and  too. > What to do, fools everywhere suffice. At first threw out and SAHF/LANF, but then all the same returned. Happens and so.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

7

Re: Re: "here my honor is touched"

Hello, , you wrote: P>> Certainly faster. > checked? In this place I did not understand a question. We judge by result of the test. And it is that that it was necessary to replace structure member type: 3 offenseAgainstSineCondition float binary (49)//fixed decimal (16, 11) Then that float binary faster decimal fixed. However, if dec fixed to make on whole instead of BCD, the result, most likely, will be refined. > when it was clarified that the heritage of Kobola is difficult for supporting And by the way, in economic tasks in PL/1 declarations on a template were used. Somehow so: declare myvar picture ' 999,99 ' As the declaration in PL/1 precisely registered, I do not remember. Said that it not the same that decimal fixed. About the same was in Kobole. It was used PICT ' 9999V99 '.

8

Re: Re: "here my honor is touched"

Hello, pagid, you wrote: >> These commands not for the calculator, and for calculations with controllable accuracy in economic calculations. Bothered to explain. P> for calculations with controllable accuracy in economic calculations they it is better nothing . But many times more slowly and more difficult in circuitry. , and what for you repeat here discussion the Author:  Date: 26.02.17 in which both already participated?

9

Re: Re: "here my honor is touched"

Hello, , you wrote: P>> It was clarified that these commands are necessary to nobody also them did not use. And AMD them without  regrets threw out from 86-64, Intel joined this decision. > the Intel was necessary to buy the license for architecture at AMD with all their jambs after them Itanium turned in Titanic. About to "buy" did not hear. About obtaining it agree a cross connect-license to the agreement (on which in general AMD permanently receives more) - heard. But in general it is not too bad. > at AMD as appeared commands of control of overflow INTO and output control over boundaries BOUND are thrown out. Because are not necessary. At least in such type. A problem of this circuit of program exceptions - that they cannot be processed at the same level of privileges. In newer architecture it is accessible - for example, in RISC-V it is possible to adjust such delegation (and it is possible and not to admit). But the Intel did not begin to do. To make jo on a handling branch somehow normally is easier and more effective as a whole. I and exceptions of commands of the whole division threw out there. > What to do, fools everywhere suffice. Here they just cleverly arrived. > at first threw out and SAHF/LANF, but then all the same returned. And it already a curvature problem x86 as a whole. By the way, OF there does not enter, though in low byte of two not used bits.