<li>Graphs - It would be great to have gnuplot graphs to keep track of how the
tree is changing over time. We already gather a several statistics, it
- just neccesary to add the script-fu to gnuplotize it.
+ just necessary to add the script-fu to gnuplotize it.
<li>Regression tests - We should run the regression tests in addition to the
program tests...
</ol><p>
+
<!-- ======================================================================= -->
</ul><table width="50%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
<tr><td> </td><td width=100%>
href="http://mail.cs.uiuc.edu/pipermail/llvmbugs/">llvm-bugs</a> list. If you
get the program to compile, it would be extremely useful to convert the build
system to be compatible with the LLVM Programs testsuite so that we can check it
-into CVS and the automated tester can use it to track progress of the compiler.
+into CVS and the automated tester can use it to track progress of the
+compiler.<p>
+
+When testing a code, try running it with a variety of optimizations, and with
+all the back-ends: CBE, llc, and lli.<p>
<!-- ======================================================================= -->
</b></font></td></tr></table><ul>
<ol>
-<li>Transform <tt>setjmp</tt> and <tt>longjmp</tt> calls to use the LLVM
- <tt>invoke</tt> mechanism.
-<li>Add support for a volatile attribute on loads and stores
-<li>Support for variable argument functions
<li>Add a new conditional move instruction: <tt>X = select bool Cond, Y, Z</tt>
-<li>Add support for platform independ
ant prefetch support. The GCC <a
+<li>Add support for platform independ
ent prefetch support. The GCC <a
href="http://gcc.gnu.org/projects/prefetch.html">prefetch project</a> page
has a good survey of the prefetching capabilities of a variety of modern
processors.
</b></font></td></tr></table><ul>
<ol>
-<li>Improve the efficiency of the bytecode loader/writer, allow streaming lazy
- loading of functions from the bytecode (for use by the JIT, for example)<br>
-<li>Rework the PassManager
-<li>Do not encode zero intializers for large arrays into the bytecode
-
+<li>Someone needs to look into getting the <tt>ranlib</tt> tool to index LLVM
+ bytecode files, so that linking in .a files is not hideously slow. They
+ would also then have to implement the reader for this index in
+ <tt>gccld</tt>.<br>
+
+<li>Improve the efficiency of the bytecode loader/writer<br>
+<li>Extend the FunctionPassManager to use a ModuleProvider to stream functions
+ in on demand. This would improve the efficiency of the JIT.
+<li>Rework the PassManager to be more flexible
<li>Some transformations and analyses only work on reducible flow graphs. It
would be nice to have a transformation which could be "required" by these passes
which makes irreducible graphs reducible. This can easily be accomplished
<a name="pointeranalysis">Pointer and Alias Analysis
</b></font></td></tr></table><ul>
-We have a strong base for development of both pointer analysis based
-optimizations as well as pointer analyses themselves. It seems natural to want
-to take advantage of this...<p>
+We have a <a href="AliasAnalysis.html">strong base for development</a> of both
+pointer analysis based optimizations as well as pointer analyses themselves. It
+
seems natural to want to take advantage of this...<p>
<ol>
<li>Implement a flow-sensitive context-sensitive alias analysis algorithm<br>
- Pick one of the somewhat efficient algorithms, but strive for maximum
precision
-<li>Implement a flow-sensitive context-insensitive alias anlaysis algorithm<br>
+<li>Implement a flow-sensitive context-insensitive alias analysis algorithm<br>
- Just an efficient local algorithm perhaps?
<li>Implement an interface to update analyses in response to common code motion
<li>Implement alias analysis based optimizations:
<ul>
<li>Dead store elimination
- <li>Location invariant Code Motion (LcICM)
- <li>Store+Reload or "store forwarding" elimination:<p>
- Change:
- <pre>
- store int X, int* P
- Y = load int* P</pre>
- into:
- <pre>
- store int X, int *P
- Y = X</pre>
- <li>Register promotion (move loads and stores out of loop bodies, for example)
</ul>
</ol>
- A linear time or nearly so algorithm
<li>Implement a strength reduction pass
<li>Value range propagation pass
-<li>Implement a tail recursion elimination pass
<li>Implement an unswitching pass
+<li>Write a loop unroller, with a simple heuristic for when to unroll
</ol>
<!-- ======================================================================= -->
<ol>
<li>Implement a global register allocator
<li>Implement a better instruction selector
-<li>Implement a static compiler in addition to the JIT (easy project)
+<li>Implement support for the "switch" instruction without requiring the
+ lower-switches pass.
</ol>
<!-- ======================================================================= -->
<address><a href="mailto:sabre@nondot.org">Chris Lattner</a></address>
<!-- Created: Tue Aug 6 15:00:33 CDT 2002 -->
<!-- hhmts start -->
-Last modified: Thu Feb 6 13:38:41 CST 2003
+Last modified: Wed Oct 1 16:48:54 CDT 2003
<!-- hhmts end -->
</font></body></html>
projects / oota-llvm.git / blobdiff