Comments from The Vu

8 files changed, 41 insertions, 30 deletions

diff --git a/log.md b/log.md
index b6ac3c4..5f80202 100644
--- a/log.md
+++ b/log.md
@@ -620,3 +620,5 @@ need to look at register optimisations.
 - Capitalisation in the reference list
 - More optimisations: branch optimisation
 - More optimisations: tail recursion
+- Mention that the register counting method is simplistic (or has this been
+  mentioned already?)
diff --git a/thesis/further.tex b/thesis/further.tex
new file mode 100644
index 0000000..e499b7f
--- /dev/null
+++ b/thesis/further.tex
@@ -0,0 +1,11 @@
+\section{Further work}
+\label{sec:further}
+
+\begin{multicols}{2}
+
+\todo{
+	There are some things that are still to be done and some optimisation vectors that are not yet considered.
+	What will be written depends on how far I get in the coming weeks with this.
+}
+
+\end{multicols}
diff --git a/thesis/intro.tex b/thesis/intro.tex
index f92b21d..3d6347c 100644
--- a/thesis/intro.tex
+++ b/thesis/intro.tex
@@ -29,7 +29,7 @@ The main differences between ARM and Thumb-2 are the following:
 In ARM, every instruction has a 4-bit conditional field that allows for conditional execution.
 In the Thumb instruction set, all conditional instructions except branches have to be in an \enquote{IT block}.
 A first \ual{it} instruction gives the base condition and a then-else pattern.
-The following instructions are executed conditionally.
+The statements after the \ual{it} instruction are executed conditionally.
 For example:
 
 \begin{minted}{ual}
@@ -59,7 +59,7 @@ For these instructions there exist 32-bit variants that can address all sixteen
 		\begin{tikzpicture}[clean]
 			\node (start) {\clean{Start}};
 		\end{tikzpicture}
-		\caption{Initially}
+		\caption{Initially.}
 	\end{subfigure}%
 	\begin{subfigure}[b]{.2\linewidth}
 		\centering
@@ -68,7 +68,7 @@ For these instructions there exist 32-bit variants that can address all sixteen
 			\node[below=of fac.arg1.south] (4) {\clean{4}};
 			\draw (fac.arg1) -- (4.north);
 		\end{tikzpicture}
-		\caption{Applying \clean{Start}}
+		\caption{Applying \clean{Start}.}
 	\end{subfigure}%
 	\begin{subfigure}[b]{.3\linewidth}
 		\centering
@@ -84,7 +84,7 @@ For these instructions there exist 32-bit variants that can address all sixteen
 			\draw (minus.arg1) |- (4.east);
 			\draw (minus.arg2) -- (1.west);
 		\end{tikzpicture}
-		\caption{Applying \clean{fac n}}
+		\caption{Applying \clean{fac n}.}
 	\end{subfigure}%
 	\begin{subfigure}[b]{.2\linewidth}
 		\centering
@@ -97,16 +97,16 @@ For these instructions there exist 32-bit variants that can address all sixteen
 			\draw (times.arg2) -- (fac.west);
 			\draw (fac.arg1) -- (3.north);
 		\end{tikzpicture}
-		\caption{Applying \clean{-}}
+		\caption{Applying \clean{-}.}
 	\end{subfigure}
-	\caption{Rewriting a Clean node\label{fig:intro:rewriting}}
+	\caption{Rewriting a Clean node\label{fig:intro:rewriting}.}
 \end{figure*}
 
 \subsubsection{Interworking}
 \label{sec:intro:interworking}
-The ARM and Thumb instruction sets are designed to \emph{interwork}:
+The ARM and Thumb instruction sets are designed to interwork:
 	different parts of a program can be assembled for different instruction sets
-	and it is possible to switch instruction set when the program counter is written to~\parencite[A4.1]{armv7ar}.
+	and it is possible to switch instruction set when an instruction writes to the program counter~\parencite[A4.1]{armv7ar}.
 
 The Thumb-2 code generator proposed in this thesis does not produce ARM code,
 	though the existence of the interworking facility has effects on the techniques that can be used in it.
@@ -178,7 +178,7 @@ This function uses the B-stack, for basic values, for its arguments and return v
 More complex types would be passed on the A-stack.
 
 The \clean{Start} rule compiles to%
-	\footnote{Again, the code has been shorted insignificantly for brevity.}:
+	\footnote{The code has been shorted insignificantly for brevity.}:
 
 \begin{minted}[tabsize=4]{abc}
 	__fac_Start
@@ -209,7 +209,7 @@ While the ABC-machine can be implemented in a straightforward manner using macro
 	it introduces several optimisations, some of which are target-dependent.
 
 The ARM code for the factorial example is as follows%
-	\footnote{Some irrelevant peculiarities have been removed for brevity and clarity.}:
+	\footnote{Some irrelevant peculiarities have been removed for brevity.}:
 
 \begin{minted}{ual}
 	sfac_P1:
diff --git a/thesis/reg-alloc.tex b/thesis/reg-alloc.tex
index 482cd8c..beb83c5 100644
--- a/thesis/reg-alloc.tex
+++ b/thesis/reg-alloc.tex
@@ -37,7 +37,7 @@
 				\draw ($({axis cs:85000,S0})-(0,2em)$) -- ($({axis cs:85000,A3})+(0,2em)$);
 			\end{axis}
 		\end{tikzpicture}
-		\subcaption{In the Clean compiler}
+		\subcaption{In the Clean compiler.}
 	\end{subfigure}%
 	\begin{subfigure}{.5\linewidth}
 		\begin{tikzpicture}
@@ -71,9 +71,9 @@
 				\draw ($({axis cs:155,S0})-(0,2em)$) -- ($({axis cs:155,A3})+(0,2em)$);
 			\end{axis}
 		\end{tikzpicture}
-		\subcaption{In the RTS\label{fig:reg-usage:rts}}
+		\subcaption{In the RTS.\label{fig:reg-usage:rts}}
 	\end{subfigure}%
-	\caption{Register usage with the Thumb-2 backend\label{fig:reg-usage}}
+	\caption{Register usage with the Thumb-2 backend.\label{fig:reg-usage}}
 \end{figure}
 
 % pi@rasppi:~/clean/src/compiler$ ~/count-regs.sh 
@@ -182,7 +182,7 @@ In Clean programs, the ARM registers are used as follows~\parencite[\texttt{arms
 			S0           & r12 & \highlight{r0} \\
 		\end{tabular}
 	\end{subfigure}%
-	\caption{Register allocation in the ARM and Thumb backends\label{tab:reg-alloc:arm-and-thumb}}
+	\caption{Register allocation in the ARM and Thumb backends.\label{tab:reg-alloc:arm-and-thumb}}
 \end{table*}
 
 In \cref{fig:reg-usage}, we count for each register the number of instructions it occurs in, in the code generated for the Clean compiler~\parencite{cocl}.
@@ -251,7 +251,7 @@ This is much more complicated, and for a rough estimate the simplistic method us
 			\addplot coordinates { (69896,0) [100\%] };
 		\end{axis}
 	\end{tikzpicture}
-	\caption{Code size for different backends\label{fig:reg-alloc:results}}
+	\caption{Code size for different backends.\label{fig:reg-alloc:results}}
 \end{figure*}
 
 \subsection{Optimisation}
diff --git a/thesis/results.tex b/thesis/results.tex
index 2f6de50..76afb27 100644
--- a/thesis/results.tex
+++ b/thesis/results.tex
@@ -18,7 +18,7 @@
 				& Diff. (\%)    & 11.9 &  3.6 &  5.1 &  5.9 & ---  &  3.7 & ---  & ---  & ---  &  5.4 &   0.9 &  2.6 &  5.2 &  0.0 & ---  &  8.0 & ---  & ---  \\
 		\end{tabular}
 	}
-	\caption{Code size and running time comparison for ARM and Thumb-2\label{tab:results}}
+	\caption{Code size and running time comparison for ARM and Thumb-2.\label{tab:results}}
 \end{table*}
 
 % Program         Code size             Running time (+gc)   Nr. runs
diff --git a/thesis/storing-pc.tex b/thesis/storing-pc.tex
index bfb8442..4bae7fb 100644
--- a/thesis/storing-pc.tex
+++ b/thesis/storing-pc.tex
@@ -157,7 +157,7 @@ For this reason it is better to use S0 whenever possible.
 \subsection{Comparison}
 \label{sec:storing-pc:comparison}
 Assuming the worst case, that all instructions in the jump block are wide, we need four more bytes in Thumb than in ARM.
-As a benchmark, the Clean compiler has 41,006 jumps of this kind in 1,253,978 instructions, a rough 3.27\%.
+As a benchmark, the Clean compiler has 41,006 jumps of this kind in 1,253,978 instructions, or approximately 3.27\%.
 The four extra bytes in Thumb mean a size increase of $41006\cdot4\approx160$KiB on the 5.3MiB file, an increase of 3.00\%.
 
 As for the time complexity: every jump requires an extra instruction cycle.
diff --git a/thesis/thesis.bib b/thesis/thesis.bib
index f43bc26..4156ff0 100644
--- a/thesis/thesis.bib
+++ b/thesis/thesis.bib
@@ -1,41 +1,41 @@
 @manual{armv7ar,
-	label="ARM Ltd.",
+	author="{ARM Ltd.}",
 	key="ARM Ltd. 1996",
 	title="ARM Architecture Reference Manual. ARMv7-A and ARMv7-R edition",
 	year=1996
 }
 
 @manual{armv7m,
-	label="ARM Ltd.",
+	author="{ARM Ltd.}",
 	key="ARM Ltd. 2006",
 	title="ARMv7-M Architecture Reference Manual",
 	year=2006
 }
 
 @manual{armv8a,
-	label="ARM Ltd.",
+	author="{ARM Ltd.}",
 	key="ARM Ltd. 2013",
 	title="ARM Architecture Reference Manual. ARMv8, for ARMv8-A architecture profile (Beta)",
 	year=2013
 }
 
 @manual{armcallstd,
-	label="ARM Ltd.",
+	author="{ARM Ltd.}",
 	key="ARM Ltd. 2015",
 	title="Procedure Call Standard for the ARM Architecture",
 	year=2015
 }
 
 @manual{cleanlangrep,
-	title="Clean Language Report. Version 2.2",
+	title="Clean Language Report",
+	subtitle="Version 2.2",
 	year=2011,
 	author={Rinus Plasmeijer and Marco {\VAN{Eekelen}} and John {\VAN{Groningen}}},
 	organization="Department of Software Technology, University of Nijmegen"
 }
 
 @Online{clean,
-	label="Software Technology Research Group",
-	key="Software Technology Research Group",
+	author="{Software Technology Research Group}",
 	title="The Clean Home Page",
 	organization="Radboud University Nij\-me\-gen",
 	url="http://clean.cs.ru.nl/",
@@ -65,8 +65,7 @@
 }
 
 @Online{armcg,
-	key="ARM CG",
-	label="ARM CG",
+	author="{Software Technology Research Group}",
 	title="Clean's ARM code generator",
 	subtitle="Revision 295",
 	organization="Radboud University Nijmegen",
@@ -76,8 +75,7 @@
 }
 
 @Online{armrts,
-	key="ARM RTS",
-	label="ARM RTS",
+	author="{Software Technology Research Group}",
 	title="Clean's ARM run-time system",
 	subtitle="Revision 387",
 	organization="Radboud University Nijmegen",
@@ -87,8 +85,7 @@
 }
 
 @Online{cocl,
-	key="Clean compiler",
-	label="Clean compiler",
+	author="{Software Technology Research Group}",
 	title="The Clean compiler",
 	subtitle="Bootstrap from intermediate ABC files, 32-bit.",
 	organization="Radboud University Nijmegen",
diff --git a/thesis/thesis.tex b/thesis/thesis.tex
index 7ebb7ad..95abdee 100644
--- a/thesis/thesis.tex
+++ b/thesis/thesis.tex
@@ -54,6 +54,7 @@ It produces on average 20\% smaller code than the ARM code generator, which is o
 \input{load-offsets}
 \input{reg-alloc}
 \input{results}
+\input{further}
 
 \ifdraft\else
 \cleardoublepage