To be able to create a proper set of transformations, the constraints on the
input and output of that transformation process should be properly specified.
This entails two parts: Specifying the MontiumC language, and specifying the
-Montium IR constraints, which is is the input to the backend.
+Montium IR constraints, which is the input to the backend.
Specifying Montium IR was relatively easy, since it is defined directly by the
-backend. The MontiumC specification is slightly more complex. There are two
+backend. The MontiumC specification was slightly more complex. There are two
different angles to it: What does the compiler support, and what do we want the
-compiler to support.
+compiler to support?
\subsubsection{What is supported?}
One angle for looking at MontiumC is seeing what the compiler can currently
of) the specification, because they will not work reliably in all cases.
The best way to detect these cases is making the compiler check its input using
-an the specification. This way, any code operating outside of the specification
-can be detected automatically. Writing such checks has not happened yet, mainly
+the specification. This way, any code operating outside of the specification
+can be detected automatically. Writing such checks has not happened so far, mainly
because the impact of the new hardware on MontiumC is not quite clear yet.
Existing transformations, on the other hand, might miss a few corner cases. When
detected automatically. A framework for this testing has been set up and
partially filled with small tests.
-Building this initial specification did pose a number of challenges. Since
-simply trying all possible C features to see if they are accepted by the
+Building this initial specification did pose a number of challenges.
+Simply trying all possible C features to see if they are accepted by the
MontiumC compiler and thus valid MontiumC is a lengthy process and only useful
in a limited way. A more constructive way would be to examine the compiler
components to see what transformations are applied and from that derive the
few C features.
\subsubsection{What is wanted?}
-A completely different angle of looking at this is from the requirements point
+A completely different angle of looking at this specification is from the requirements point
of view. What do we want MontiumC to support? This angle is even harder than the
previous one, since there are a lot of levels of requirements. Ideally, MontiumC
would not exist and our compiler would support the C language fully. However,
most efficient code. In the Montium case, a lot of things simply cannot be
mapped on the hardware at all.
-Considering that our ideal is not reachable (Though the new hardware might take
+Considering that our ideal is not reachable (though the new hardware might take
us a lot closer), every feature
considered for MontiumC was evaluated thoroughly for feasibility, both in hardware
and in the compiler. In practice, this meant that new language features would be
to make it run more efficiently.
In figure \ref{ExampleHigh} the same code is displayed, but this time
-using higer level C features (for loops, arra indexing). This is the
+using higer level C features (for loops, array indexing). This is the
level of code we are trying to achieve, but we're not there yet. It
should be noted that this is still not "normal" C, since we use the
"imul" function instead of the normal * operator. However, since the
It is not unlikely that the specification is still incorrect in a few places (or
rather, that the code does not implement the specification properly). Since
-so far there has been not any automated checking of programs against the
+so far there has not been any automated checking of programs against the
specification, these errors have not been uncovered. Once the new hardware is
more clearly defined and the MontiumC specification is updated for it, this
checking should be added so the specification and compiler can be better
LLVM has a pretty large amount of documentation, I spent most of my first
weeks with reading tutorials and documents. Since there was already a (very
-preliminary) version of the clang-based frontend, I also had some code to play
+preliminary) version of the Clang-based frontend, I also had some code to play
with.
During this period, it was not completely clear what the frontend should
prove very insightful, however, as to how the LLVM framework is built and what its
possibilities are.
-Additionally, during my working with the code in this internship I also produced
+Additionally, during my working with the code during this internship I also produced
a number of patches for LLVM, containing bugfixes, some cleanup and
documentation improvements. Since the best way to integrate with any open source
-project seems to be contributing code, I was giving commit access to the LLVM
+project seems to be contributing code, I was given commit access to the LLVM
tree not long thereafter. This access has proved very useful during the rest of
the internship, since it was now a a lot easier to make (simple) changes to the
LLVM framework to better suit the needs of Recore.
need are obvious. However, usually when making changes to the main LLVM
tree, just changing enough for Recore is not engough for LLVM. Since the LLVM
code must work on any program, not just MontiumC programs, extra changes are
-required (see also parapgrah \ref{StayingGeneric}). This is also an issue of
+required (see also parapgrah \ref{StayingGeneric}). Additionally, this is an issue of
building up credit within the LLVM community: The more you contribute to LLVM,
the more influence you have when things need changing.
with the hardware developers was not uncommon either. In practice, most
communication with the hardware developers went through the backend
developer, except for the design discussion concerning the new Montium
-hardware design (also see section \ref{Pipelining} below).
+hardware design (also see section \ref{Pipelining}).
In addition, discussions regarding design issues at various levels often happen
out in the open, which invites people with an opinion about something to
In some cases, these extra constraints can be formulated in a generic way, such
that the LLVM code can handle architectures with or without the constraint.
+Examples include providing Montium specific parameters through LLVM's
+``\verb TargetData '' class and modifications to the loop unroller to support
+custom policies.
+
In a lot of cases, however, the constraint is so Montium specific that changing
-LLVM to support it is not feasible. In a few cases, this meant that the backend
-was changed to support more features of the LLVM IR. In other cases, a
-Recore-specific transformations was added to solve these problems. In a few more
-cases, the problems are still unresolved, effectively resulting in additional
-constraints on the MontiumC language.
+LLVM to support it is not feasible.
+
+In a few of these cases, this meant that the backend was changed to support more
+features of the LLVM IR. An example of this is finding datapath constants (which
+are the result of a function in MontiumC, so hard to track by LLVM).
+
+In other cases, Recore-specific transformations were added to solve these
+problems. Examples of these are a transformation that removes all global
+variables and passes them as arguments and return values instead, a
+transformation that forcibly inlines all functions marked as ``\verb inline ''
+and a transformation that limits variable lifetimes by introducing extra phi
+nodes.
+
+In a few more cases, the problems are still unresolved, effectively resulting in
+additional constraints on the MontiumC language. Examples of these are
+preventing instructions from being moved out of if/else blocks (which is
+perfectly fine from an LLVM IR point of view, but does not take into account the
+extra meaning that an if statement has in MontiumIR) and removal of unused bits
+from a constant (which could introduce more different constants than the Montium
+has registers for them).
+
\subsection{New hardware design}
\label{Pipelining}
required for the new (pipelined) hardware design and the hardware design itself.
Even though this is completely outside of the area of my assignment, the initial
prototype of that scheduler was created by someone else using LLVM. Because of
-my experience with LLVM, I have been assisting him with that. Initially mostly
-helping out with hints on LLVM coding, but later also with thinking about the
+my experience with LLVM, I have been assisting him with it. Initially I
+helped him by giving hints on LLVM coding, but later also with thinking about the
scheduler and hardware design.
I will not go into much detail about the new hardware and its scheduler here,
different from the loop "kernel", the number of instructions needed for
a pipelined loop can easily increase a lot.
-However, all pipelined loops share a very distinct structure (first
+However, all pipelined loops share a very distinct structure: first
stage 1, then stage 1+2, then stage 1+2+3, etc, then all stages at the
-same time, similar for the epilogue). Also, every instruction in the
+same time, similar for the epilogue. Also, every instruction in the
prologue and epilogue are a strict subset of the instructions in the
kernel. By adding some hardware support for exactly this structure, the
code size increase for the prologue and epilogue can be effectively
reduced to a fixed number of instructions (which take the number of stages as a
-parameter and uses preloaded instructions with explicit stage annotation).
+parameter and use preloaded instructions with explicit stage annotation).
The tradeoff here is that this hardware is only usable specifically for these
inner loops, any other code will leave this extra hardware unused. However,
On the new hardware, however, function calls are more powerful, which should
lead to a lot less code duplication. For this reason, putting every instruction
in configuration registers might actually take more space instead of less. It
-should be noted that, the configuration registers of the old Montium are
+should be noted that the configuration registers of the old Montium are
effectively a compiler controlled cache that is mandatory and static
(instructions must be in the cache and the cache cannot be modified at runtime).
By lifting these limitations, we get a cache that is a lot more flexible.
projects / matthijs / projects / internship.git / blobdiff