Would it make sense to write wrappers of Data.Binary functions to emulate the networking functions that are usually expected? /jve On Sat, Dec 27, 2008 at 8:23 PM, Bryan O'Sullivan wrote:

2008/12/27 John Van Enk

...

Currently this only has:

- htons - htonl - ntohs - ntohl

This is all subsumed by the binary package (Data.Binary), where it makes a

lot more sense in any instance.

Didn't reply to all. Sorry for the double Johan. Johan asked:

will GHC be able to optimize away all of the Data.Binary stuff so the functions, if defined in terms of Data.Binary, are as efficient as a direct implementation?

Binary can be a bit slower than you might want. I'm guessing it would do fine for hton like operations, but in pureMD5 I noticed a 2x slow down when using Data.Binary instead of the low-level FFI implementation. In C one might want: inline uint32_t getWord32(unsigned char *bytestring) { uint32_t *p = &bytestring; return p[idx]; } And that is extremely fast code. In low level Haskell it looks more like: getNthWord n bs@(PS ptr off len) = inlinePerformIO $ withForeignPtr ptr $ \ptr' -> do let p = castPtr $ plusPtr ptr' off peekElemOff p n But even this low level code seems to be excessively slow with 33% of the MD5 run time attributed to this function. I ment to investigate this a couple months ago but have had zero time. Tom

On Sun, 2008-12-28 at 08:47 -0800, Bryan O'Sullivan wrote:

On Sun, Dec 28, 2008 at 1:38 AM, Thomas DuBuisson wrote: getNthWord n bs@(PS ptr off len) = inlinePerformIO $ withForeignPtr ptr $ \ptr' -> do let p = castPtr $ plusPtr ptr' off peekElemOff p n

The overhead here is very probably caused by withForeignPtr. In similar cases, I've seen much better performance from hoisting this to the outside of a loop.

Since ghc version 6.6, withForeignPtr has no overhead at all so there is no benefit to hoisting it out of loops. You can verify this by inspecting the core output. In ghc-6.6 the representation of ForeignPtr changed to put the Addr# directly in the ForeignPtr constructor (rather than an indirection further away). Duncan