This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] PR/27733, synth_mult cache missing too much


> My plan for 4.3 was to avoid the generation of
RTL in tree-ssa-loop-ivopts.c:multiply_by_cost.  We spend a lot of
time whilst deciding whether to hoist a multiplication out of a loop
choosing between low latency sequences that have the same "cost",
which is often easy to estimate based on bit-counts.  This will get
worse with "multiply-by-11".

You may still want to do that because the original testcase's compilation time, 30 seconds, is not ludicrous but still high. You can threshold and call choose_mult_variant directly, with a low cost limit (which may not even need to be target dependent), and compute the cost based on bitcounts if that fail.


Paolo


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]