Misc/abrash-black-book
Watch
1
0
Fork
You've already forked abrash-black-book
 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Use proper lists instead of DLs with bullet points

Browse source
This commit is contained in:
James Gregory 2013-12-30 13:34:37 +11:00
commit 7e957c0bb4
12 changed files with 83 additions and 83 deletions
Download patch file Download diff file Expand all files Collapse all files

22
11-03.html
View file

@ -40,19 +40,19 @@
</P>
<P>The picture is less clear in the 386 world since there are so many different memory architectures, but similar problems can occur in any computer built around a 286 or 386. The prefetch queue cycle-eater is even a factor&mdash;albeit a lesser one&mdash;on zero-wait-state machines, both because branching empties the queue and because some instructions can outrun even zero&mdash;5 cycles longer than the official execution time.)</P>
<P>To summarize:</P>
<DL>
<DD><B>&#149;</B>&nbsp;&nbsp;Memory-accessing instructions don&rsquo;t run at their official speeds on non-zero-wait-state 286/386 computers.
<DD><B>&#149;</B>&nbsp;&nbsp;The prefetch queue cycle-eater reduces performance on 286/386 computers, particularly when non-zero-wait-state memory is used.
<DD><B>&#149;</B>&nbsp;&nbsp;Branches often execute at less than their rated speeds on the 286 and 386 since the prefetch queue is emptied.
<DD><B>&#149;</B>&nbsp;&nbsp;The extent to which the prefetch queue and wait states affect performance varies from one 286/386 computer to another, making precise optimization impossible.
</DL>
<ul>
<li>Memory-accessing instructions don&rsquo;t run at their official speeds on non-zero-wait-state 286/386 computers.</li>
<li>The prefetch queue cycle-eater reduces performance on 286/386 computers, particularly when non-zero-wait-state memory is used.</li>
<li>Branches often execute at less than their rated speeds on the 286 and 386 since the prefetch queue is emptied.</li>
<li>The extent to which the prefetch queue and wait states affect performance varies from one 286/386 computer to another, making precise optimization impossible.</li>
</ul>
<P>What&rsquo;s to be learned from all this? Several things:
</P>
<DL>
<DD><B>&#149;</B>&nbsp;&nbsp;Keep your instructions short.
<DD><B>&#149;</B>&nbsp;&nbsp;Keep it in the registers; avoid memory, since memory generally can&rsquo;t keep up with the processor.
<DD><B>&#149;</B>&nbsp;&nbsp;Don&rsquo;t jump.
</DL>
<ul>
<li>Keep your instructions short.</li>
<li>Keep it in the registers; avoid memory, since memory generally can&rsquo;t keep up with the processor.</li>
<li>Don&rsquo;t jump.</li>
</ul>
<P>Of course, those are exactly the rules that apply to 8088 optimization as well. Isn&rsquo;t it convenient that the same general rules apply across the board?
</P>
<H4 ALIGN="CENTER"><A NAME="Heading7"></A><FONT COLOR="#000077">Data Alignment</FONT></H4>