1

Topic: Re: What for register SPL is necessary?

Hello, , you wrote: > article, is written one and a half years ago: > http://files.rsdn.org/122727/pl1ex23.doc > I Remind, what the author not I, and already printed here Loaves http://rsdn.org/users/99493.aspx What for in.doc to spread? To me can and it is interesting look, but swing.doc absolutely there is no desire. On a site not to add in any way unless?

2

Re: Re: What for register SPL is necessary?

Hello, , you wrote: Not "what for", and "why". I understand that the author too polite in relation to commands and Intel, and AMD. In a forum of such restriction is not present, and I can openly say that both firms frankly  all possibilities which to them were given at first by passage 16> 32, then 32> 64, in respect of manufacture of equal and convenient command system. Presence of all of these SIL - frank ghost effect of curve design. > to tell the truth, that there is such register, I only from this article and learned. > means, it precisely is not necessary Well I saw generation of its usage by compilers. In difference, for example, from ah/bh/ch/dh which do not lay down in the modern concepts of creation.

3

Re: Re: What for register SPL is necessary?

Hello, , you wrote: > Hello, netch80, you wrote: N>> Not "what for", and "why". I understand that the author too polite in relation to commands and Intel, and AMD. In a forum of such restriction is not present, and I can openly say that both firms frankly  all possibilities which to them were given at first by passage 16> 32, then 32> 64, in respect of manufacture of equal and convenient command system. Presence of all of these SIL - frank ghost effect of curve design. > well, I would not become so to accuse in curvature. Most likely, statistics of usage of commands was really processed by development of new architecture. Forces and means for it were selected the big. At first, said that AMD created amd64 in a terrible time trouble (before it there was an idea to make a sandwich from x86 in 32 and Alpha in 64, but failed because of sale Alpha in parts). So could be without "forces and means". Secondly, curvature unconditionally is. Generally to save a format of commands x86 it is not necessary, it madly expensive to parallel analysis. N>> well I saw generation of its usage by compilers. In difference, for example, from ah/bh/ch/dh which do not lay down in the modern concepts of creation. > it is curious, in what cases compilers used SPL? I.e. what it were for commands? Transfers? I will specify - not spl, certainly. SP to change in a normal context it is impossible, and compilers simply do not generate in such, where SP (esp, rsp...) - the normal register. Meant sil, dil. I do not remember about bpl, quite could be in a mode omit-frame-pointer in the big function.

4

Re: Re: What for register SPL is necessary?

Hello, , you wrote: > Hello, netch80, you wrote: N>> I Will specify - not spl, certainly. SP to change in a normal context it is impossible, and compilers simply do not generate in such, where SP (esp, rsp...) - the normal register. Meant sil, dil. I do not remember about bpl, quite could be in a mode omit-frame-pointer in the big function. > is not present, it is clear, even usage bpl in other article was, here: > http://rsdn.org/forum/design/6626256.flat the Author:  Date: 01.12 08:42 > I, probably, in vain in other section of a forum thrust it. Here already a problem of unambiguity of hierarchy without . > And here spl - it is similar valid any eructation, instead of the thought over singularity Well in any command system it is better to admit absurd, than to produce exceptions on everyone . Though it is possible to forbid specific combinations and to notify that after a while they acquire special sense of Intel so does regularly.

5

Re: Re: What for register SPL is necessary?

>>> And here spl - it is similar valid any eructation, instead of thought over singularity N>> Well in any command system it is better to admit absurd, than to produce exceptions on everyone . Though it is possible to forbid specific combinations and to notify that after a while they acquire special sense N>> Intel so does regularly. > in this case, I would prefer an exception, than the allowed command mov spl, al after which through one hundred-two following instructions too there will be an exception, only already figs knows what 86 the assembler for a long time is not intended for manual . And  it is not sealed up.