This doesn't have the effect I expected: loop xs = case blah of One thing -> ... loop The other -> ... realloc ... where {-# NOINLINE realloc #-} realloc = do something loop ... My intention here was that the loop would not contain the code for realloc and that it'd be done as a call at the cmm level. My intention is to take the slow and rarely taken realloc path out of the code for the fast path. It seems the {-# NOINLINE realloc #-} pagma did not have the effect I intended. Looking at the -ddump-simpl and -ddump-cmm, the code for the realloc gets expanded in place in a branch of a case statement. In the cmm code we end up with just what I didn't want: loop_info: if (offset != 4096) goto later; ... ... lots of realloc code taking up space ... in the instruction / trace cache ... later: .. do the fast bits, read a byte, write a byte jump loop_info; Not only does the slow path take up space but it's in the location favoured by the hardware's static branch prediction. Reversing the test doesn't help because either way ghc turns it into: case thing of _DEFAULT -> 4096 -> and from that generates CMM: if (thing != 4096) goto much_later; ... much_later: ... The reason I was looking at this is because I've been trying to figure out why our lazy byte string fusion primitives are much slower than the strict versions. It's improving though, it's now only half the speed rather than a tenth of the speed. :-) The ByteString.Lazy code is an interesting mixture of strict and lazy. We must strictly read/write the chunks but lazily generate/consume the list of chunks. I just discovered that I should have been reading STG all along rather than core from the simplifier or CMM. STG takes out all the type annotations which tend to make things quite verbose. Mind you, seeing the types can be handy too to see if/how things are unboxed. Even so, I kind of wish there were a stage between STG and CMM that showed the imperative model of STG with linear layout, control flow and notes to indicate thunk/closure allocations. I expect most of my problem is that I do not understand the STG evaluation model sufficiently well to see how it maps to basic blocks, jumps/calls etc. Duncan
[Narrowing to ghc users only] That's odd. I tried it (the HEAD) and it worked fine. Input and output below You'll notice that the binding for 'realloc' got floated into the branch of the case (that's FloatIn), but it is never inlined. | Even so, I kind of wish there were a stage between STG and CMM that | showed the imperative model of STG with linear layout, control flow and | notes to indicate thunk/closure allocations. I expect most of my problem | is that I do not understand the STG evaluation model sufficiently well | to see how it maps to basic blocks, jumps/calls etc. Try -ddump-prep. It's essentially STG with a bit less clutter. Simon ======= Input ========= {-# OPTIONS -fglasgow-exts #-} module Foo7 where data Thing = One Thing | The Bool loop xs ys = case xs of One thing -> loop thing ys The other -> case realloc of True -> False False -> True where {-# NOINLINE realloc #-} realloc = case ys of One thing -> True The other -> False ======= Output ========= Rec { Foo7.loop :: Foo7.Thing -> Foo7.Thing -> GHC.Base.Bool [GlobalId] [Arity 2 NoCafRefs Str: DmdType SS] Foo7.loop = \ (xs_add :: Foo7.Thing) (ys_ade :: Foo7.Thing) -> case xs_add of wild_B1 { Foo7.One thing_adv -> Foo7.loop thing_adv ys_ade; Foo7.The other_adz -> let { realloc_seH :: GHC.Base.Bool [Str: DmdType] realloc_seH = case ys_ade of wild1_Xc { Foo7.One thing_adn -> GHC.Base.True; Foo7.The other1_adr -> GHC.Base.False } } in case realloc_seH of wild1_Xe { GHC.Base.False -> GHC.Base.True; GHC.Base.True -> GHC.Base.False } } end Rec } | -----Original Message----- | From: glasgow-haskell-bugs-bounces@haskell.org [mailto:glasgow-haskell-bugs-bounces@haskell.org] | On Behalf Of Duncan Coutts | Sent: 13 August 2006 23:34 | To: GHC-users list; GHC-bugs list | Subject: behaviour of {-# NOINLINE #-} in where clauses | | This doesn't have the effect I expected: | | loop xs = | case blah of | One thing -> ... loop | The other -> ... realloc ... | | | where | {-# NOINLINE realloc #-} | realloc = do | something | loop ... | | My intention here was that the loop would not contain the code for | realloc and that it'd be done as a call at the cmm level. My intention | is to take the slow and rarely taken realloc path out of the code for | the fast path. | | It seems the {-# NOINLINE realloc #-} pagma did not have the effect I | intended. Looking at the -ddump-simpl and -ddump-cmm, the code for the | realloc gets expanded in place in a branch of a case statement. In the | cmm code we end up with just what I didn't want: | | loop_info: | if (offset != 4096) goto later; | ... | ... lots of realloc code taking up space | ... in the instruction / trace cache | ... | later: | .. do the fast bits, read a byte, write a byte | jump loop_info; | | | Not only does the slow path take up space but it's in the location | favoured by the hardware's static branch prediction. | | Reversing the test doesn't help because either way ghc turns it into: | | case thing of | _DEFAULT -> | 4096 -> | | and from that generates CMM: | | if (thing != 4096) goto much_later; | ... | much_later: | ... | | | The reason I was looking at this is because I've been trying to figure | out why our lazy byte string fusion primitives are much slower than the | strict versions. It's improving though, it's now only half the speed | rather than a tenth of the speed. :-) | | The ByteString.Lazy code is an interesting mixture of strict and lazy. | We must strictly read/write the chunks but lazily generate/consume the | list of chunks. | | I just discovered that I should have been reading STG all along rather | than core from the simplifier or CMM. STG takes out all the type | annotations which tend to make things quite verbose. Mind you, seeing | the types can be handy too to see if/how things are unboxed. | | Even so, I kind of wish there were a stage between STG and CMM that | showed the imperative model of STG with linear layout, control flow and | notes to indicate thunk/closure allocations. I expect most of my problem | is that I do not understand the STG evaluation model sufficiently well | to see how it maps to basic blocks, jumps/calls etc. | | | Duncan | | _______________________________________________ | Glasgow-haskell-bugs mailing list | Glasgow-haskell-bugs@haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
On Mon, 2006-08-14 at 10:12 +0100, Simon Peyton-Jones wrote:
[Narrowing to ghc users only]
That's odd. I tried it (the HEAD) and it worked fine. Input and output below
You'll notice that the binding for 'realloc' got floated into the branch of the case (that's FloatIn), but it is never inlined.
Ok, I'll try with the latest head. In my code the realloc name has disappeared in the -ddump-simpl and the only remaining 'let' expressions are for other things (for allocating prim byte arrays and ForeignPtrContents).
| Even so, I kind of wish there were a stage between STG and CMM that | showed the imperative model of STG with linear layout, control flow and | notes to indicate thunk/closure allocations. I expect most of my problem | is that I do not understand the STG evaluation model sufficiently well | to see how it maps to basic blocks, jumps/calls etc.
Try -ddump-prep. It's essentially STG with a bit less clutter.
Right'o. Thanks.
======= Input =========
{-# OPTIONS -fglasgow-exts #-}
module Foo7 where
data Thing = One Thing | The Bool
loop xs ys = case xs of One thing -> loop thing ys The other -> case realloc of True -> False False -> True
where {-# NOINLINE realloc #-} realloc = case ys of One thing -> True The other -> False
======= Output =========
Rec { Foo7.loop :: Foo7.Thing -> Foo7.Thing -> GHC.Base.Bool [GlobalId] [Arity 2 NoCafRefs Str: DmdType SS] Foo7.loop = \ (xs_add :: Foo7.Thing) (ys_ade :: Foo7.Thing) -> case xs_add of wild_B1 { Foo7.One thing_adv -> Foo7.loop thing_adv ys_ade; Foo7.The other_adz -> let { realloc_seH :: GHC.Base.Bool [Str: DmdType] realloc_seH = case ys_ade of wild1_Xc { Foo7.One thing_adn -> GHC.Base.True; Foo7.The other1_adr -> GHC.Base.False } } in case realloc_seH of wild1_Xe { GHC.Base.False -> GHC.Base.True; GHC.Base.True -> GHC.Base.False } } end Rec }
Yes, that looks more like what I want, I think. Duncan
Hello Duncan, Monday, August 14, 2006, 2:33:31 AM, you wrote:
Reversing the test doesn't help because either way ghc turns it into:
case thing of _DEFAULT -> 4096 ->
to be exact, ghc passes code for default case separately from code for other cases. look at emitSwitch procedure to know why it's required
and from that generates CMM:
if (thing != 4096) goto much_later; ... much_later: ...
can't you use 'if' expressions?
Even so, I kind of wish there were a stage between STG and CMM that showed the imperative model of STG with linear layout, control flow and notes to indicate thunk/closure allocations. I expect most of my problem is that I do not understand the STG evaluation model sufficiently well to see how it maps to basic blocks, jumps/calls etc.
i once tried to understood STG->CMM code generation. it's all in codeGen directory of ghc sources. and now i think that STG is pretty low-level imperative language, not harder than C for example. i even had the idea to write STG->C translator that generates efficient code. on the other side, jhc already contains pretty the same thing (John claims that jhc's internal language is close to STG) -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com
participants (3)
-
Bulat Ziganshin -
Duncan Coutts -
Simon Peyton-Jones