To make an executable program, the GHC system compiles your code and then links it with a non-trivial runtime system (RTS), which handles storage management, profiling, etc.
You have some control over the behaviour of the RTS, by giving special command-line arguments to your program.
% ./a.out -f +RTS -p -S -RTS -h foo bar
The RTS will snaffle
for itself, and the remaining arguments
-f -h foo bar
will be handed to your program if/when it calls
-RTS option is required if the
runtime-system options extend to the end of the command line, as in
% hls -ltr /usr/etc +RTS -A5m
As always, for RTS options that take
sizes: If the last character of
size is a K or k, multiply by 1000; if an
M or m, by 1,000,000; if a G or G, by 1,000,000,000. (And any
wraparound in the counters is your
NOTE: since GHC is itself compiled by GHC, you can change RTS
options in the compiler using the normal
+RTS ... -RTS
combination. eg. to increase the maximum heap
size for a compilation to 128M, you would add
+RTS -M128m -RTS
to the command line.
GHCRTS='-M128m' export GHCRTS
RTS options taken from the
variable can be overridden by options given on the command
Sets the interval that the RTS clock ticks at. The runtime uses a single timer signal to count ticks; this timer signal is used to control the context switch timer (Section 4.11, “Using Concurrent Haskell”) and the heap profiling timer Section 5.4.1, “RTS options for heap profiling”. Also, the time profiler uses the RTS timer signal directly to record time profiling samples.
Normally, setting the
directly is not necessary: the resolution of the RTS timer is
adjusted automatically if a short interval is requested with
-V is required in order to
increase the resolution of the time profiler.
There are several options to give you precise control over garbage collection. Hopefully, you won't need any of these in normal operation, but there are several things that can be tweaked for maximum performance.
[Default: 256k] Set the allocation area size
used by the garbage collector. The allocation area
(actually generation 0 step 0) is fixed and is never resized
(unless you use
Increasing the allocation area size may or may not give better performance (a bigger allocation area means worse cache behaviour but fewer garbage collections and less promotion).
With only 1 generation (
-A option specifies the minimum allocation
area, since the actual size of the allocation area will be
resized according to the amount of data in the heap (see
Use a compacting algorithm for collecting the oldest generation. By default, the oldest generation is collected using a copying algorithm; this option causes it to be compacted in-place instead. The compaction algorithm is slower than the copying algorithm, but the savings in memory use can be considerable.
For a given heap size (using the
option), compaction can in fact reduce the GC cost by
allowing fewer GCs to be performed. This is more likely
when the ratio of live data to heap size is high, say
NOTE: compaction doesn't currently work when a single
generation is requested using the
[Default: 30] Automatically enable
compacting collection when the live data exceeds
n% of the maximum heap size
-M option). Note that the maximum
heap size is unlimited by default, so this option has no
effect unless the maximum heap size is set with
[Default: 2] This option controls the amount of memory reserved for the older generations (and in the case of a two space collector the size of the allocation area) as a factor of the amount of live data. For example, if there was 2M of live data in the oldest generation when we last collected it, then by default we'll wait until it grows to 4M before collecting it again.
The default seems to work well here. If you have
plenty of memory, it is usually better to use
size than to
-F setting will be automatically
reduced by the garbage collector when the maximum heap size
setting) is approaching.
[Default: 2] Set the number of generations used by the garbage collector. The default of 2 seems to be good, but the garbage collector can support any number of generations. Anything larger than about 4 is probably not a good idea unless your program runs for a long time, because the oldest generation will hardly ever get collected.
Specifying 1 generation with
gives you a simple 2-space collector, as you would expect.
In a 2-space collector, the
-A option (see
above) specifies the minimum allocation
area size, since the allocation area will grow with the
amount of live data in the heap. In a multi-generational
collector the allocation area is a fixed size (unless you
-H option, see below).
[Default: 0] This option provides a “suggested heap size” for the garbage collector. The garbage collector will use about this much memory until the program residency grows and the heap size needs to be expanded to retain reasonable performance.
By default, the heap will start small, and grow and
shrink as necessary. This can be bad for performance, so if
you have plenty of memory it's worthwhile supplying a big
improving GC performance, using
usually a better bet than
(default: 0.3) In the threaded and SMP versions of the RTS (see
-threaded, Section 4.10.7, “Options affecting linking”), a
major GC is automatically performed if the runtime has been idle
(no Haskell computation has been running) for a period of time.
The amount of idle time which must pass before a GC is performed is
set by the
-I0 disables the idle GC.
For an interactive application, it is probably a good idea to use the idle GC, because this will allow finalizers to run and deadlocked threads to be detected in the idle time when no Haskell computation is happening. Also, it will mean that a GC is less likely to happen when the application is busy, and so responsiveness may be improved. However, if the amount of live data in the heap is particularly large, then the idle GC can cause a significant delay, and too small an interval could adversely affect interactive responsiveness.
This is an experimental feature, please let us know if it causes problems and/or could benefit from further tuning.
[Default: 1k] Set the initial stack size for new threads. Thread stacks (including the main thread's stack) live on the heap, and grow as required. The default value is good for concurrent applications with lots of small threads; if your program doesn't fit this model then increasing this option may help performance.
The main thread is normally started with a slightly larger heap to cut down on unnecessary stack growth while the program is starting up.
[Default: 8M] Set the maximum stack size for
an individual thread to
bytes. This option is there purely to stop the program
eating up all the available memory in the machine if it gets
into an infinite loop.
n of heap
which must be available for allocation. The default is
[Default: unlimited] Set the maximum heap size to
size bytes. The heap normally
grows and shrinks according to the memory requirements of
the program. The only reason for having this option is to
stop the heap growing without bound and filling up all the
available swap space, which at the least will result in the
program being summarily killed by the operating
The maximum heap size also affects other garbage
collection parameters: when the amount of live data in the
heap exceeds a certain fraction of the maximum heap size,
compacting collection will be automatically enabled for the
oldest generation, and the
will be reduced in order to avoid exceeding the maximum heap
Write modest (
-s) or verbose
-S) garbage-collector statistics into file
file. The default
is treated specially, with the output really being sent to
This option is useful for watching how the storage manager adjusts the heap size based on the current amount of live data.
Write a one-line GC stats summary after running the
program. This output is in the same format as that produced
-s, the default
is treated specially, with the output really being sent to
The RTS options related to profiling are described in Section 5.4.1, “RTS options for heap profiling”, those for concurrency in Section 4.11, “Using Concurrent Haskell”, and those for parallelism in Section 4.12.1, “Options to enable SMP parallelism”.
These RTS options might be used (a) to avoid a GHC bug, (b) to see “what's really happening”, or (c) because you feel like it. Not recommended for everyday use!
Sound the bell at the start of each (major) garbage collection.
Oddly enough, people really do use this option! Our pal in Durham (England), Paul Callaghan, writes: “Some people here use it for a variety of purposes—honestly!—e.g., confirmation that the code/machine is doing something, infinite loop detection, gauging cost of recently added code. Certain people can even tell what stage [the program] is in by the beep pattern. But the major use is for annoying others in the same office…”
An RTS debugging flag; varying quantities of output
depending on which bits are set in
num. Only works if the RTS was
compiled with the
Produce “ticky-ticky” statistics at the
end of the program run. The
business works just like on the
“Ticky-ticky” statistics are counts of
various program actions (updates, enters, etc.) The program
must have been compiled using
(a.k.a. “ticky-ticky profiling”), and, for it to
be really useful, linked with suitable system libraries.
Not a trivial undertaking: consult the installation guide on
how to set things up for easy “ticky-ticky”
profiling. For more information, see Section 5.7, “Using “ticky-ticky” profiling (for implementors)”.
(Only available when the program is compiled for
profiling.) When an exception is raised in the program,
this option causes the current cost-centre-stack to be
This can be particularly useful for debugging: if your
program is complaining about a
error and you haven't got a clue which bit of code is
causing it, compiling with
-auto-all and running with
-RTS will tell you exactly the call stack at the
point the error was raised.
The output contains one line for each exception raised in the program (the program might raise and catch several exceptions during its execution), where each line is of the form:
< cc1, ..., ccn >
a cost centre in the program (see Section 5.1, “Cost centres and cost-centre stacks”), and the sequence represents the
“call stack” at the point the exception was
raised. The leftmost item is the innermost function in the
call stack, and the rightmost item is the outermost
Turn off “update-frame squeezing” at garbage-collection time. (There's no particularly good reason to turn it off, except to ensure the accuracy of certain data collected regarding thunk entry counts.)
GHC lets you exercise rudimentary control over the RTS settings for any given program, by compiling in a “hook” that is called by the run-time system. The RTS contains stub definitions for all these hooks, but by writing your own version and linking it on the GHC command line, you can override the defaults.
Owing to the vagaries of DLL linking, these hooks don't work under Windows when the program is built dynamically.
ghc_rts_optslets you set RTS
options permanently for a given program. A common use for this is
to give your program a default heap and/or stack size that is
greater than the default. For example, to set
-K1m, place the following definition in a C source
char *ghc_rts_opts = "-H128m -K1m";
Compile the C file, and include the object file on the command line when you link your Haskell program.
These flags are interpreted first, before any RTS flags from
GHCRTS environment variable and any flags
on the command line.
You can also change the messages printed when the runtime system “blows up,” e.g., on stack overflow. The hooks for these are as follows:
For examples of the use of these hooks, see GHC's own
versions in the file
ghc/compiler/parser/hschooks.c in a GHC