hsbencher-1.0: HSBencher/App.hs
{-# LANGUAGE BangPatterns, NamedFieldPuns, ScopedTypeVariables, RecordWildCards, FlexibleContexts #-}
{-# LANGUAGE CPP, OverloadedStrings, TupleSections #-}
--------------------------------------------------------------------------------
-- NOTE: This is best when compiled with "ghc -threaded"
-- However, ideally for real benchmarking runs we WANT the waitForProcess below block the whole process.
-- However^2, currently [2012.05.03] when running without threads I get errors like this:
-- benchmark.run: bench_hive.log: openFile: resource busy (file is locked)
--------------------------------------------------------------------------------
-- Disabling some stuff until we can bring it back up after the big transition [2013.05.28]:
#define DISABLED
{- |
This program runs a set of benchmarks contained in the current
directory. It produces two files as output:
results_HOSTNAME.dat
bench_HOSTNAME.log
ASSUMPTIONS -- about directory and file organization
----------------------------------------------------
This benchmark harness can run either cabalized benchmarks, or
straight .hs files buildable by "ghc --make".
---------------------------------------------------------------------------
<< TODO >>
---------------------------------------------------------------------------
* Replace environment variable argument passing with proper flags/getopt.
<Things that worked at one time but need to be cleaned up:>
* Further enable packing up a benchmark set to run on a machine
without GHC (as with Haskell Cnc)
* Clusterbench -- adding an additional layer of parameter variation.
-}
module HSBencher.App (defaultMainWithBechmarks, Flag(..), all_cli_options) where
----------------------------
-- Standard library imports
import Prelude hiding (log)
import Control.Applicative
import Control.Concurrent
import Control.Monad.Reader
import Control.Exception (evaluate, handle, SomeException, throwTo, fromException, AsyncException(ThreadKilled))
import Debug.Trace
import Data.Time.Clock (getCurrentTime, diffUTCTime)
import Data.Time.Clock.POSIX (utcTimeToPOSIXSeconds)
import Data.Maybe (isJust, fromJust, catMaybes)
import qualified Data.Map as M
import Data.Word (Word64)
import Data.IORef
import Data.List (intercalate, sortBy, intersperse, isPrefixOf, tails, isInfixOf, delete)
import qualified Data.Set as Set
import Data.Version (versionBranch, versionTags)
import GHC.Conc (getNumProcessors)
import Numeric (showFFloat)
import System.Console.GetOpt (getOpt, ArgOrder(Permute), OptDescr(Option), ArgDescr(..), usageInfo)
import System.Environment (getArgs, getEnv, getEnvironment)
import System.Directory
import System.Posix.Env (setEnv)
import System.Random (randomIO)
import System.Exit
import System.FilePath (splitFileName, (</>), takeDirectory)
import System.Process (system, waitForProcess, getProcessExitCode, runInteractiveCommand,
createProcess, CreateProcess(..), CmdSpec(..), StdStream(..), readProcess)
import System.IO (Handle, hPutStrLn, stderr, openFile, hClose, hGetContents, hIsEOF, hGetLine,
IOMode(..), BufferMode(..), hSetBuffering)
import System.IO.Unsafe (unsafePerformIO)
import qualified Data.ByteString.Char8 as B
import Text.Printf
import Text.PrettyPrint.GenericPretty (Out(doc))
-- import Text.PrettyPrint.HughesPJ (nest)
----------------------------
-- Additional libraries:
import qualified System.IO.Streams as Strm
import qualified System.IO.Streams.Concurrent as Strm
import qualified System.IO.Streams.Process as Strm
import qualified System.IO.Streams.Combinators as Strm
import UI.HydraPrint (hydraPrint, HydraConf(..), DeleteWinWhen(..), defaultHydraConf, hydraPrintStatic)
import Scripting.Parallel.ThreadPool (parForM)
#ifdef FUSION_TABLES
import Network.Google (retryIORequest)
import Network.Google.OAuth2 (getCachedTokens, refreshTokens, OAuth2Client(..), OAuth2Tokens(..))
import Network.Google.FusionTables (createTable, listTables, listColumns, insertRows,
TableId, CellType(..), TableMetadata(..))
#endif
----------------------------
-- Self imports:
import HSBencher.Utils
import HSBencher.Logging
import HSBencher.Types
import HSBencher.Methods
import HSBencher.MeasureProcess
import Paths_hsbencher (version) -- Thanks, cabal!
----------------------------------------------------------------------------------------------------
-- | USAGE
usageStr :: String
usageStr = unlines $
[
" ",
" Many of these options can redundantly be set either when the benchmark driver is run,",
" or in the benchmark descriptions themselves. E.g. --with-ghc is just for convenience.",
"\n ENV VARS:",
" These environment variables control the behavior of the benchmark script:",
"",
#ifndef DISABLED
" SHORTRUN=1 to get a shorter run for testing rather than benchmarking.",
"",
" THREADS=\"1 2 4\" to run with # threads = 1, 2, or 4.",
"",
" BENCHLIST=foo.txt to select the benchmarks and their arguments",
" (uses benchlist.txt by default)",
"",
" SCHEDS=\"Trace Direct Sparks\" -- Restricts to a subset of schedulers.",
"",
" GENERIC=1 to go through the generic (type class) monad par",
" interface instead of using each scheduler directly",
"",
" KEEPGOING=1 to keep going after the first error.",
"",
" TRIALS=N to control the number of times each benchmark is run.",
"",
#endif
#ifdef FUSION_TABLES
" HSBENCHER_GOOGLE_CLIENTID, HSBENCHER_GOOGLE_CLIENTSECRET: if FusionTable upload is enabled, the",
" client ID and secret can be provided by env vars OR command line options. ",
#endif
" ",
#ifndef DISABLED
" ENVS='[[(\"KEY1\", \"VALUE1\")], [(\"KEY1\", \"VALUE2\")]]' to set",
" different configurations of environment variables to be set *at",
" runtime*. Useful for NUMA_TOPOLOGY, for example. Note that this",
" can change multiple env variables in multiple distinct",
" configurations, with each configuration tested separately.",
"",
" Additionally, this script will propagate any flags placed in the",
" environment variables $GHC_FLAGS and $GHC_RTS. It will also use",
" $GHC or $CABAL, if available, to select the executable paths.",
" ",
#endif
" Command line arguments take precedence over environment variables, if both apply."
]
----------------------------------------------------------------------------------------------------
gc_stats_flag :: String
gc_stats_flag = " -s "
-- gc_stats_flag = " --machine-readable -t "
exedir :: String
exedir = "./bin"
--------------------------------------------------------------------------------
-- | Fill in "static" fields of a FusionTable row based on the `Config` data.
augmentTupleWithConfig :: Config -> [(String,String)] -> IO [(String,String)]
augmentTupleWithConfig Config{..} base = do
-- ghcVer <- runSL$ ghc ++ " -V"
-- let ghcVer' = collapsePrefix "The Glorious Glasgow Haskell Compilation System," "GHC" ghcVer
datetime <- getCurrentTime
uname <- runSL "uname -a"
lspci <- runLines "lspci"
whos <- runLines "who"
let runID = (hostname ++ "_" ++ show startTime)
let (branch,revision,depth) = gitInfo
return $
-- addit "COMPILER" ghcVer' $
-- addit "COMPILE_FLAGS" ghc_flags $
-- addit "RUNTIME_FLAGS" ghc_RTS $
addit "HOSTNAME" hostname $
addit "RUNID" runID $
addit "DATETIME" (show datetime) $
addit "TRIALS" (show trials) $
addit "ENV_VARS" (show envs) $
addit "BENCH_VERSION" (show$ snd benchversion) $
addit "BENCH_FILE" (fst benchversion) $
addit "UNAME" uname $
-- addit "LSPCI" (unlines lspci) $
addit "GIT_BRANCH" branch $
addit "GIT_HASH" revision $
addit "GIT_DEPTH" (show depth) $
addit "WHO" (unlines whos) $
base
where
addit :: String -> String -> [(String,String)] -> [(String,String)]
addit key val als =
case lookup key als of
Just b -> error$"augmentTupleWithConfig: cannot add field "++key++", already present!: "++b
Nothing -> (key,val) : als
-- Retrieve the (default) configuration from the environment, it may
-- subsequently be tinkered with. This procedure should be idempotent.
getConfig :: [Flag] -> [Benchmark DefaultParamMeaning] -> IO Config
getConfig cmd_line_options benches = do
hostname <- runSL$ "hostname -s"
t0 <- getCurrentTime
let startTime = round (utcTimeToPOSIXSeconds t0)
env <- getEnvironment
-- There has got to be a simpler way!
branch <- runSL "git name-rev --name-only HEAD"
revision <- runSL "git rev-parse HEAD"
-- Note that this will NOT be newline-terminated:
hashes <- runLines "git log --pretty=format:'%H'"
let
-- Read an ENV var with default:
get v x = case lookup v env of
Nothing -> x
Just s -> s
logFile = "bench_" ++ hostname ++ ".log"
resultsFile = "results_" ++ hostname ++ ".dat"
case get "GENERIC" "" of
"" -> return ()
s -> error$ "GENERIC env variable not handled yet. Set to: " ++ show s
maxthreads <- getNumProcessors
backupResults resultsFile logFile
rhnd <- openFile resultsFile WriteMode
lhnd <- openFile logFile WriteMode
hSetBuffering rhnd NoBuffering
hSetBuffering lhnd NoBuffering
resultsOut <- Strm.unlines =<< Strm.handleToOutputStream rhnd
logOut <- Strm.unlines =<< Strm.handleToOutputStream lhnd
stdOut <- Strm.unlines Strm.stdout
let -- Messy way to extract the benchlist version:
-- ver = case filter (isInfixOf "ersion") (lines benchstr) of
-- (h:_t) -> read $ (\ (h:_)->h) $ filter isNumber (words h)
-- [] -> 0
-- This is our starting point BEFORE processing command line flags:
base_conf = Config
{ hostname, startTime
, shortrun = False
, doClean = True
, benchsetName = Nothing
-- , trials = read$ get "TRIALS" "1"
, trials = 1
, pathRegistry = M.empty
-- , benchlist = parseBenchList benchstr
-- , benchversion = (benchF, ver)
, benchlist = benches
, benchversion = ("",0)
, maxthreads = maxthreads
, threadsettings = parseIntList$ get "THREADS" (show maxthreads)
, keepgoing = False
, resultsFile, logFile, logOut, resultsOut, stdOut
-- , outHandles = Nothing
, envs = read $ get "ENVS" "[[]]"
, gitInfo = (trim branch, trim revision, length hashes)
-- This is in priority order:
, buildMethods = [cabalMethod, makeMethod, ghcMethod]
, doFusionUpload = False
#ifdef FUSION_TABLES
, fusionTableID = Nothing
, fusionClientID = lookup "HSBENCHER_GOOGLE_CLIENTID" env
, fusionClientSecret = lookup "HSBENCHER_GOOGLE_CLIENTSECRET" env
#endif
}
-- Process command line arguments to add extra cofiguration information:
let
#ifdef FUSION_TABLES
doFlag (BenchsetName name) r = r { benchsetName= Just name }
doFlag (ClientID cid) r = r { fusionClientID = Just cid }
doFlag (ClientSecret s) r = r { fusionClientSecret = Just s }
doFlag (FusionTables m) r =
let r2 = r { doFusionUpload = True } in
case m of
Just tid -> r2 { fusionTableID = Just tid }
Nothing -> r2
#endif
doFlag (CabalPath p) r = r { pathRegistry= M.insert "cabal" p (pathRegistry r) }
doFlag (GHCPath p) r = r { pathRegistry= M.insert "ghc" p (pathRegistry r) }
doFlag ShortRun r = r { shortrun= True }
doFlag KeepGoing r = r { keepgoing= True }
doFlag (NumTrials s) r = r { trials=
case reads s of
(n,_):_ -> n
[] -> error$ "--trials given bad argument: "++s }
-- Ignored options:
doFlag ShowHelp r = r
doFlag ShowVersion r = r
doFlag NoRecomp r = r
doFlag NoCabal r = r
doFlag NoClean r = r { doClean = False }
doFlag ParBench r = r
--------------------
conf = foldr ($) base_conf (map doFlag cmd_line_options)
#ifdef FUSION_TABLES
finalconf <- if not (doFusionUpload conf) then return conf else
case (benchsetName conf, fusionTableID conf) of
(Nothing,Nothing) -> error "No way to find which fusion table to use! No name given and no explicit table ID."
(_, Just tid) -> return conf
(Just name,_) -> do
case (fusionClientID conf, fusionClientSecret conf) of
(Just cid, Just sec ) -> do
let auth = OAuth2Client { clientId=cid, clientSecret=sec }
tid <- runReaderT (getTableId auth name) conf
return conf{fusionTableID= Just tid}
(_,_) -> error "When --fusion-upload is activated --clientid and --clientsecret are required (or equiv ENV vars)"
#else
let finalconf = conf
#endif
-- runReaderT (log$ "Read list of benchmarks/parameters from: "++benchF) finalconf
return finalconf
-- | Remove RTS options that are specific to -threaded mode.
pruneThreadedOpts :: [String] -> [String]
pruneThreadedOpts = filter (`notElem` ["-qa", "-qb"])
--------------------------------------------------------------------------------
-- Error handling
--------------------------------------------------------------------------------
-- | Create a backup copy of existing results_HOST.dat files.
backupResults :: String -> String -> IO ()
backupResults resultsFile logFile = do
e <- doesFileExist resultsFile
date <- runSL "date +%Y%m%d_%s"
when e $ do
renameFile resultsFile (resultsFile ++"."++date++".bak")
e2 <- doesFileExist logFile
when e2 $ do
renameFile logFile (logFile ++"."++date++".bak")
path :: [FilePath] -> FilePath
path [] = ""
path ls = foldl1 (</>) ls
--------------------------------------------------------------------------------
-- Compiling Benchmarks
--------------------------------------------------------------------------------
-- | Build a single benchmark in a single configuration.
compileOne :: (Int,Int) -> Benchmark DefaultParamMeaning -> [(DefaultParamMeaning,ParamSetting)] -> BenchM BuildResult
compileOne (iterNum,totalIters) Benchmark{target=testPath,cmdargs} cconf = do
Config{shortrun, resultsOut, stdOut, buildMethods, pathRegistry, doClean} <- ask
let (diroffset,testRoot) = splitFileName testPath
flags = toCompileFlags cconf
paths = toCmdPaths cconf
bldid = makeBuildID flags
log "\n--------------------------------------------------------------------------------"
log$ " Compiling Config "++show iterNum++" of "++show totalIters++
": "++testRoot++" (args \""++unwords cmdargs++"\") confID "++ show bldid
log "--------------------------------------------------------------------------------\n"
matches <- lift$
filterM (fmap isJust . (`filePredCheck` testPath) . canBuild) buildMethods
when (null matches) $ do
logT$ "ERROR, no build method matches path: "++testPath
lift exitFailure
logT$ printf "Found %d methods that can handle %s: %s"
(length matches) testPath (show$ map methodName matches)
let BuildMethod{methodName,clean,compile,concurrentBuild} = head matches
when (length matches > 1) $
logT$ " WARNING: resolving ambiguity, picking method: "++methodName
let pathR = (M.union (M.fromList paths) pathRegistry)
when doClean $ clean pathR bldid testPath
-- Prefer the benchmark-local path definitions:
x <- compile pathR bldid flags testPath
logT$ "Compile finished, result: "++ show x
return x
--------------------------------------------------------------------------------
-- Running Benchmarks
--------------------------------------------------------------------------------
-- If the benchmark has already been compiled doCompile=False can be
-- used to skip straight to the execution.
runOne :: (Int,Int) -> BuildID -> BuildResult -> Benchmark DefaultParamMeaning -> [(DefaultParamMeaning,ParamSetting)] -> BenchM ()
runOne (iterNum, totalIters) bldid bldres Benchmark{target=testPath, cmdargs=args_} runconfig = do
let numthreads = foldl (\ acc (x,_) ->
case x of
Threads n -> n
_ -> acc)
0 runconfig
sched = foldl (\ acc (x,_) ->
case x of
Variant s -> s
_ -> acc)
"none" runconfig
let runFlags = toRunFlags runconfig
envVars = toEnvVars runconfig
conf@Config{..} <- ask
----------------------------------------
-- (1) Gather contextual information
----------------------------------------
let args = if shortrun then shortArgs args_ else args_
fullargs = args ++ runFlags
testRoot = fetchBaseName testPath
log$ "\n--------------------------------------------------------------------------------"
log$ " Running Config "++show iterNum++" of "++show totalIters
-- ++": "++testRoot++" (args \""++unwords args++"\") scheduler "++show sched++
-- " threads "++show numthreads++" (Env="++show envVars++")"
log$ "--------------------------------------------------------------------------------\n"
pwd <- lift$ getCurrentDirectory
logT$ "(In directory "++ pwd ++")"
logT$ "Next run 'who', reporting users other than the current user. This may help with detectivework."
-- whos <- lift$ run "who | awk '{ print $1 }' | grep -v $USER"
whos <- lift$ runLines$ "who"
let whos' = map ((\ (h:_)->h) . words) whos
user <- lift$ getEnv "USER"
logT$ "Who_Output: "++ unwords (filter (/= user) whos')
-- If numthreads == 0, that indicates a serial run:
----------------------------------------
-- (2) Now execute N trials:
----------------------------------------
-- (One option woud be dynamic feedback where if the first one
-- takes a long time we don't bother doing more trials.)
nruns <- forM [1..trials] $ \ i -> do
log$ printf " Running trial %d of %d" i trials
log " ------------------------"
let doMeasure cmddescr = do
SubProcess {wait,process_out,process_err} <- lift$ measureProcess cmddescr
err2 <- lift$ Strm.map (B.append " [stderr] ") process_err
both <- lift$ Strm.concurrentMerge [process_out, err2]
mv <- echoStream (not shortrun) both
lift$ takeMVar mv
x <- lift wait
return x
case bldres of
StandAloneBinary binpath -> do
-- NOTE: For now allowing rts args to include things like "+RTS -RTS", i.e. multiple tokens:
let command = binpath++" "++unwords fullargs
logT$ " Executing command: " ++ command
doMeasure CommandDescr{ command=ShellCommand command, envVars, timeout=Just defaultTimeout, workingDir=Nothing }
RunInPlace fn -> do
-- logT$ " Executing in-place benchmark run."
let cmd = fn fullargs
logT$ " Generated in-place run command: "++show cmd
doMeasure cmd
------------------------------------------
-- (3) Produce output to the right places:
------------------------------------------
(t1,t2,t3,p1,p2,p3) <-
if not (all didComplete nruns) then do
log $ "\n >>> MIN/MEDIAN/MAX (TIME,PROD) -- got ERRORS: " ++show nruns
return ("","","","","","")
else do
-- Extract the min, median, and max:
let sorted = sortBy (\ a b -> compare (realtime a) (realtime b)) nruns
minR = head sorted
maxR = last sorted
medianR = sorted !! (length sorted `quot` 2)
let ts@[t1,t2,t3] = map (\x -> showFFloat Nothing x "")
[realtime minR, realtime medianR, realtime maxR]
prods@[p1,p2,p3] = map mshow [productivity minR, productivity medianR, productivity maxR]
mshow Nothing = ""
mshow (Just x) = showFFloat (Just 2) x ""
let
pads n s = take (max 1 (n - length s)) $ repeat ' '
padl n x = pads n x ++ x
padr n x = x ++ pads n x
-- These are really (time,prod) tuples, but a flat list of
-- scalars is simpler and readable by gnuplot:
formatted = (padl 15$ unwords $ ts)
++" "++ unwords prods -- prods may be empty!
log $ "\n >>> MIN/MEDIAN/MAX (TIME,PROD) " ++ formatted
logOn [ResultsFile]$
printf "%s %s %s %s %s" (padr 35 testRoot) (padr 20$ intercalate "_" args)
(padr 8$ sched) (padr 3$ show numthreads) formatted
return (t1,t2,t3,p1,p2,p3)
#ifdef FUSION_TABLES
when doFusionUpload $ do
let (Just cid, Just sec) = (fusionClientID, fusionClientSecret)
authclient = OAuth2Client { clientId = cid, clientSecret = sec }
-- FIXME: it's EXTREMELY inefficient to authenticate on every tuple upload:
toks <- liftIO$ getCachedTokens authclient
let
tuple =
[("PROGNAME",testRoot),("ARGS", unwords args),("THREADS",show numthreads),
("MINTIME",t1),("MEDIANTIME",t2),("MAXTIME",t3),
("MINTIME_PRODUCTIVITY",p1),("MEDIANTIME_PRODUCTIVITY",p2),("MAXTIME_PRODUCTIVITY",p3),
("VARIANT", show sched)]
tuple' <- liftIO$ augmentTupleWithConfig conf tuple
let (cols,vals) = unzip tuple'
log$ " [fusiontable] Uploading row with "++show (length cols)++
" columns containing "++show (sum$ map length vals)++" characters of data"
--
-- FIXME: It's easy to blow the URL size; we need the bulk import version.
stdRetry "insertRows" authclient toks $
insertRows (B.pack$ accessToken toks) (fromJust fusionTableID) cols [vals]
log$ " [fusiontable] Done uploading, run ID "++ (fromJust$ lookup "RUNID" tuple')
++ " date "++ (fromJust$ lookup "DATETIME" tuple')
-- [[testRoot, unwords args, show numthreads, t1,t2,t3, p1,p2,p3]]
return ()
#endif
return ()
-- defaultColumns =
-- ["Program","Args","Threads","Sched","Threads",
-- "MinTime","MedianTime","MaxTime", "MinTime_Prod","MedianTime_Prod","MaxTime_Prod"]
#ifdef FUSION_TABLES
resultsSchema :: [(String, CellType)]
resultsSchema =
[ ("PROGNAME",STRING)
, ("VARIANT",STRING)
, ("ARGS",STRING)
, ("HOSTNAME",STRING)
-- The run is identified by hostname_secondsSinceEpoch:
, ("RUNID",STRING)
, ("THREADS",NUMBER)
, ("DATETIME",DATETIME)
, ("MINTIME", NUMBER)
, ("MEDIANTIME", NUMBER)
, ("MAXTIME", NUMBER)
, ("MINTIME_PRODUCTIVITY", NUMBER)
, ("MEDIANTIME_PRODUCTIVITY", NUMBER)
, ("MAXTIME_PRODUCTIVITY", NUMBER)
, ("ALLTIMES", STRING)
, ("TRIALS", NUMBER)
, ("COMPILER",STRING)
, ("COMPILE_FLAGS",STRING)
, ("RUNTIME_FLAGS",STRING)
, ("ENV_VARS",STRING)
, ("BENCH_VERSION", STRING)
, ("BENCH_FILE", STRING)
-- , ("OS",STRING)
, ("UNAME",STRING)
, ("PROCESSOR",STRING)
, ("TOPOLOGY",STRING)
, ("GIT_BRANCH",STRING)
, ("GIT_HASH",STRING)
, ("GIT_DEPTH",NUMBER)
, ("WHO",STRING)
, ("ETC_ISSUE",STRING)
, ("LSPCI",STRING)
, ("FULL_LOG",STRING)
]
-- | The standard retry behavior when receiving HTTP network errors.
stdRetry :: String -> OAuth2Client -> OAuth2Tokens -> IO a ->
BenchM a
stdRetry msg client toks action = do
conf <- ask
let retryHook exn = runReaderT (do
log$ " [fusiontable] Retrying during <"++msg++"> due to HTTPException: " ++ show exn
log$ " [fusiontable] Retrying, but first, attempt token refresh..."
-- QUESTION: should we retry the refresh itself, it is NOT inside the exception handler.
-- liftIO$ refreshTokens client toks
-- liftIO$ retryIORequest (refreshTokens client toks) (\_ -> return ()) [1,1]
stdRetry "refresh tokens" client toks (refreshTokens client toks)
return ()
) conf
liftIO$ retryIORequest action retryHook [1,2,4,8,16,32,64]
-- | Get the table ID that has been cached on disk, or find the the table in the users
-- Google Drive, or create a new table if needed.
getTableId :: OAuth2Client -> String -> BenchM TableId
getTableId auth tablename = do
log$ " [fusiontable] Fetching access tokens, client ID/secret: "++show (clientId auth, clientSecret auth)
toks <- liftIO$ getCachedTokens auth
log$ " [fusiontable] Retrieved: "++show toks
let atok = B.pack $ accessToken toks
allTables <- stdRetry "listTables" auth toks $ listTables atok
log$ " [fusiontable] Retrieved metadata on "++show (length allTables)++" tables"
case filter (\ t -> tab_name t == tablename) allTables of
[] -> do log$ " [fusiontable] No table with name "++show tablename ++" found, creating..."
TableMetadata{tab_tableId} <- stdRetry "createTable" auth toks $
createTable atok tablename resultsSchema
log$ " [fusiontable] Table created with ID "++show tab_tableId
return tab_tableId
[t] -> do log$ " [fusiontable] Found one table with name "++show tablename ++", ID: "++show (tab_tableId t)
return (tab_tableId t)
ls -> error$ " More than one table with the name '"++show tablename++"' !\n "++show ls
#endif
--------------------------------------------------------------------------------
-- TODO: Remove this hack.
whichVariant :: String -> String
whichVariant "benchlist.txt" = "desktop"
whichVariant "benchlist_server.txt" = "server"
whichVariant "benchlist_laptop.txt" = "laptop"
whichVariant _ = "unknown"
-- | Write the results header out stdout and to disk.
printBenchrunHeader :: BenchM ()
printBenchrunHeader = do
Config{trials, maxthreads, pathRegistry,
logOut, resultsOut, stdOut, benchversion, shortrun, gitInfo=(branch,revision,depth) } <- ask
liftIO $ do
-- let (benchfile, ver) = benchversion
let ls :: [IO String]
ls = [ e$ "# TestName Variant NumThreads MinTime MedianTime MaxTime Productivity1 Productivity2 Productivity3"
, e$ "# "
, e$ "# `date`"
, e$ "# `uname -a`"
, e$ "# Ran by: `whoami` "
, e$ "# Determined machine to have "++show maxthreads++" hardware threads."
, e$ "# "
, e$ "# Running each test for "++show trials++" trial(s)."
-- , e$ "# Benchmarks_File: " ++ benchfile
-- , e$ "# Benchmarks_Variant: " ++ if shortrun then "SHORTRUN" else whichVariant benchfile
-- , e$ "# Benchmarks_Version: " ++ show ver
, e$ "# Git_Branch: " ++ branch
, e$ "# Git_Hash: " ++ revision
, e$ "# Git_Depth: " ++ show depth
-- , e$ "# Using the following settings from environment variables:"
-- , e$ "# ENV BENCHLIST=$BENCHLIST"
-- , e$ "# ENV THREADS= $THREADS"
-- , e$ "# ENV TRIALS= $TRIALS"
-- , e$ "# ENV SHORTRUN= $SHORTRUN"
-- , e$ "# ENV KEEPGOING= $KEEPGOING"
-- , e$ "# ENV GHC= $GHC"
-- , e$ "# ENV GHC_FLAGS= $GHC_FLAGS"
-- , e$ "# ENV GHC_RTS= $GHC_RTS"
-- , e$ "# ENV ENVS= $ENVS"
, e$ "# Path registry: "++show pathRegistry
]
ls' <- sequence ls
forM_ ls' $ \line -> do
Strm.write (Just$ B.pack line) resultsOut
Strm.write (Just$ B.pack line) logOut
Strm.write (Just$ B.pack line) stdOut
return ()
where
-- This is a hack for shell expanding inside a string:
e :: String -> IO String
e s =
runSL ("echo \""++s++"\"")
-- readCommand ("echo \""++s++"\"")
-- readProcess "echo" ["\""++s++"\""] ""
----------------------------------------------------------------------------------------------------
-- Main Script
----------------------------------------------------------------------------------------------------
-- | Command line flags.
data Flag = ParBench
| BinDir FilePath
| NoRecomp | NoCabal | NoClean
| ShortRun | KeepGoing | NumTrials String
| CabalPath String | GHCPath String
| ShowHelp | ShowVersion
#ifdef FUSION_TABLES
| FusionTables (Maybe TableId)
| BenchsetName (String)
| ClientID String
| ClientSecret String
#endif
deriving (Eq,Ord,Show,Read)
-- | Command line options.
core_cli_options :: (String, [OptDescr Flag])
core_cli_options =
("\n Command Line Options:",
[
#ifndef DISABLED
Option ['p'] ["par"] (NoArg ParBench)
"Build benchmarks in parallel (run in parallel too if SHORTRUN=1)."
#endif
Option [] ["no-recomp"] (NoArg NoRecomp)
"Don't perform any compilation of benchmark executables. Implies -no-clean."
, Option [] ["no-clean"] (NoArg NoClean)
"Do not clean pre-existing executables before beginning."
, Option [] ["shortrun"] (NoArg ShortRun)
"Elide command line args to benchmarks to perform a testing rather than benchmarking run."
, Option ['k'] ["keepgoing"] (NoArg KeepGoing)
"Keep executing even after a build or run fails."
#ifndef DISABLED
, Option [] ["no-cabal"] (NoArg NoCabal)
"A shortcut to remove Cabal from the BuildMethods"
#endif
, Option [] ["with-cabal-install"] (ReqArg CabalPath "PATH")
"Set the version of cabal-install to use for the cabal BuildMethod."
, Option [] ["with-ghc"] (ReqArg GHCPath "PATH")
"Set the path of the ghc compiler for the ghc BuildMethod."
, Option [] ["trials"] (ReqArg NumTrials "NUM")
"The number of times to run each benchmark."
, Option ['h'] ["help"] (NoArg ShowHelp)
"Show this help message and exit."
, Option ['V'] ["version"] (NoArg ShowVersion)
"Show the version and exit"
])
all_cli_options :: [(String, [OptDescr Flag])]
all_cli_options = [core_cli_options]
#ifdef FUSION_TABLES
++ [fusion_cli_options]
fusion_cli_options :: (String, [OptDescr Flag])
fusion_cli_options =
("\n Fusion Table Options:",
[ Option [] ["fusion-upload"] (OptArg FusionTables "TABLEID")
"enable fusion table upload. Optionally set TABLEID; otherwise create/discover it."
, Option [] ["name"] (ReqArg BenchsetName "NAME") "Name for created/discovered fusion table."
, Option [] ["clientid"] (ReqArg ClientID "ID") "Use (and cache) Google client ID"
, Option [] ["clientsecret"] (ReqArg ClientSecret "STR") "Use (and cache) Google client secret"
])
#endif
-- | TODO: Eventually this will be parameterized.
defaultMain :: IO ()
defaultMain = do
-- benchF = get "BENCHLIST" "benchlist.txt"
-- putStrLn$ hsbencher_tag ++ " Reading benchmark list from file: "
error "FINISHME: defaultMain requires reading benchmark list from a file. Implement it!"
-- defaultMainWithBechmarks undefined
defaultMainWithBechmarks :: [Benchmark DefaultParamMeaning] -> IO ()
defaultMainWithBechmarks benches = do
id <- myThreadId
writeIORef main_threadid id
cli_args <- getArgs
let (options,args,errs) = getOpt Permute (concat$ map snd all_cli_options) cli_args
let recomp = NoRecomp `notElem` options
when (ShowVersion `elem` options) $ do
putStrLn$ "hsbencher version "++
(concat$ intersperse "." $ map show $ versionBranch version) ++
(unwords$ versionTags version)
exitSuccess
when (not (null errs && null args) || ShowHelp `elem` options) $ do
unless (ShowHelp `elem` options) $
putStrLn$ "Errors parsing command line options:"
mapM_ (putStr . (" "++)) errs
putStrLn$ "\nUSAGE: [set ENV VARS] "++my_name++" [CMDLN OPTIONS]"
mapM putStr (map (uncurry usageInfo) all_cli_options)
putStrLn$ usageStr
if (ShowHelp `elem` options) then exitSuccess else exitFailure
conf@Config{envs,benchlist,stdOut,threadsettings} <- getConfig options benches
hasMakefile <- doesFileExist "Makefile"
cabalFile <- runLines "ls *.cabal"
let hasCabalFile = (cabalFile /= []) &&
not (NoCabal `elem` options)
rootDir <- getCurrentDirectory
runReaderT
(do
logT$"Beginning benchmarking, root directory: "++rootDir
let globalBinDir = rootDir </> "bin"
when recomp $ do
logT$"Clearing any preexisting files in ./bin/"
lift$ do
-- runSimple "rm -f ./bin/*"
-- Yes... it's posix dependent. But right now I don't see a good way to
-- delete the contents a dir without (1) following symlinks or (2) assuming
-- either the unix package or unix shell support (rm).
--- Ok, what the heck, deleting recursively:
dde <- doesDirectoryExist globalBinDir
when dde $ removeDirectoryRecursive globalBinDir
lift$ createDirectoryIfMissing True globalBinDir
logT "Writing header for result data file:"
printBenchrunHeader
unless recomp $ log "[!!!] Skipping benchmark recompilation!"
let
benches' = map (\ b -> b { configs= compileOptsOnly (configs b) })
benchlist
cccfgs = map (enumerateBenchSpace . configs) benches' -- compile configs
cclengths = map length cccfgs
totalcomps = sum cclengths
log$ "\n--------------------------------------------------------------------------------"
logT$ "Running all benchmarks for all settings ..."
logT$ "Compiling: "++show totalcomps++" total configurations of "++ show (length benchlist)++" benchmarks"
let indent n str = unlines $ map (replicate n ' ' ++) $ lines str
printloop _ [] = return ()
printloop mp (Benchmark{target,cmdargs,configs} :tl) = do
log$ " * Benchmark/args: "++target++" "++show cmdargs
case M.lookup configs mp of
Nothing -> log$ indent 4$ show$ doc configs
Just trg0 -> log$ " ...same config space as "++show trg0
printloop (M.insertWith (\ _ x -> x) configs target mp) tl
-- log$ "Benchmarks/compile options: "++show (doc benches')
printloop M.empty benches'
log$ "--------------------------------------------------------------------------------"
if ParBench `elem` options then do
unless rtsSupportsBoundThreads $ error (my_name++" was NOT compiled with -threaded. Can't do --par.")
{-
--------------------------------------------------------------------------------
-- Parallel version:
numProcs <- liftIO getNumProcessors
lift$ putStrLn$ "[!!!] Compiling in Parallel, numProcessors="++show numProcs++" ... "
when recomp $ liftIO$ do
when hasCabalFile (error "Currently, cabalized build does not support parallelism!")
(strms,barrier) <- parForM numProcs (zip [1..] pruned) $ \ outStrm (confnum,bench) -> do
outStrm' <- Strm.unlines outStrm
let conf' = conf { stdOut = outStrm' }
runReaderT (compileOne bench (confnum,length pruned)) conf'
return ()
catParallelOutput strms stdOut
res <- barrier
return ()
Config{shortrun,doFusionUpload} <- ask
if shortrun && not doFusionUpload then liftIO$ do
putStrLn$ "[!!!] Running in Parallel..."
(strms,barrier) <- parForM numProcs (zip [1..] pruned) $ \ outStrm (confnum,bench) -> do
outStrm' <- Strm.unlines outStrm
let conf' = conf { stdOut = outStrm' }
runReaderT (runOne bench (confnum,totalcomps)) conf'
catParallelOutput strms stdOut
_ <- barrier
return ()
else do
-- Non-shortrun's NEVER run multiple benchmarks at once:
forM_ (zip [1..] allruns) $ \ (confnum,bench) ->
runOne bench (confnum,totalcomps)
return ()
-}
else do
--------------------------------------------------------------------------------
-- Serial version:
-- TODO: make this a foldlM:
let allruns = map (enumerateBenchSpace . configs) benches
allrunsLens = map length allruns
totalruns = sum allrunsLens
let
-- Here we lazily compile benchmarks as they become required by run configurations.
runloop :: Int
-> M.Map BuildID (Int, Maybe BuildResult)
-> M.Map FilePath BuildID -- (S.Set ParamSetting)
-> [(Benchmark DefaultParamMeaning, [(DefaultParamMeaning,ParamSetting)])]
-> BenchM ()
runloop _ _ _ [] = return ()
runloop !iter !board !lastConfigured (nextrun:rest) = do
-- lastConfigured keeps track of what configuration was last built in
-- a directory that is used for `RunInPlace` builds.
let (bench,params) = nextrun
ccflags = toCompileFlags params
bid = makeBuildID ccflags
case M.lookup bid board of
Nothing -> error$ "HSBencher: Internal error: Cannot find entry in map for build ID: "++show bid
Just (ccnum, Nothing) -> do
res <- compileOne (ccnum,totalcomps) bench params
let board' = M.insert bid (ccnum, Just res) board
lastC' = M.insert (target bench) bid lastConfigured
-- runloop iter board' (nextrun:rest)
runOne (iter,totalruns) bid res bench params
runloop (iter+1) board' lastC' rest
Just (ccnum, Just bldres) ->
let proceed = do runOne (iter,totalruns) bid bldres bench params
runloop (iter+1) board lastConfigured rest
in
case bldres of
StandAloneBinary _ -> proceed
RunInPlace _ ->
-- Here we know that some previous compile with the same BuildID inserted this here.
-- But the relevant question is whether some other config has stomped on it in the meantime.
case M.lookup (target bench) lastConfigured of
Nothing -> error$"HSBencher: Internal error, RunInPlace in the board but not lastConfigured!: "++(target bench)
Just bid2 ->
if bid == bid2
then do logT$ "Skipping rebuild of in-place benchmark: "++bid
proceed
else runloop iter (M.insert bid (ccnum,Nothing) board) lastConfigured (nextrun:rest)
initBoard _ [] acc = acc
initBoard !iter ((bench,params):rest) acc =
let bid = makeBuildID $ toCompileFlags params
base = fetchBaseName (target bench)
dfltdest = globalBinDir </> base ++"_"++bid in
case M.lookup bid acc of
Just _ -> initBoard iter rest acc
Nothing ->
let elm = if recomp
then (iter, Nothing)
else (iter, Just (StandAloneBinary dfltdest))
in
initBoard (iter+1) rest (M.insert bid elm acc)
zippedruns = (concat$ zipWith (\ b cfs -> map (b,) cfs) benches allruns)
unless recomp $ logT$ "Recompilation disabled, assuming standalone binaries are in the expected places!"
runloop 1 (initBoard 1 zippedruns M.empty) M.empty zippedruns
{-
do Config{logOut, resultsOut, stdOut} <- ask
liftIO$ Strm.write Nothing logOut
liftIO$ Strm.write Nothing resultsOut
-}
log$ "\n--------------------------------------------------------------------------------"
log " Finished with all test configurations."
log$ "--------------------------------------------------------------------------------"
liftIO$ exitSuccess
)
conf
-- Several different options for how to display output in parallel:
catParallelOutput :: [Strm.InputStream B.ByteString] -> Strm.OutputStream B.ByteString -> IO ()
catParallelOutput strms stdOut = do
case 4 of
-- First option is to create N window panes immediately.
1 -> do
hydraPrintStatic defaultHydraConf (zip (map show [1..]) strms)
2 -> do
srcs <- Strm.fromList (zip (map show [1..]) strms)
hydraPrint defaultHydraConf{deleteWhen=Never} srcs
-- This version interleaves their output lines (ugly):
3 -> do
strms2 <- mapM Strm.lines strms
interleaved <- Strm.concurrentMerge strms2
Strm.connect interleaved stdOut
-- This version serializes the output one worker at a time:
4 -> do
strms2 <- mapM Strm.lines strms
merged <- Strm.concatInputStreams strms2
-- Strm.connect (head strms) stdOut
Strm.connect merged stdOut
----------------------------------------------------------------------------------------------------
-- * GENERIC HELPER ROUTINES
----------------------------------------------------------------------------------------------------
-- These should go in another module.......
collapsePrefix :: String -> String -> String -> String
collapsePrefix old new str =
if isPrefixOf old str
then new ++ drop (length old) str
else str
didComplete RunCompleted{} = True
didComplete _ = False
-- Shorthand for tagged version:
logT str = log$hsbencher_tag++str
hsbencher_tag = " [hsbencher] "
-- Compute a cut-down version of a benchmark's args list that will do
-- a short (quick) run. The way this works is that benchmarks are
-- expected to run and do something quick if they are invoked with no
-- arguments. (A proper benchmarking run, therefore, requires larger
-- numeric arguments be supplied.)
--
shortArgs :: [String] -> [String]
shortArgs _ls = []
-- shortArgs [] = []
-- DISABLING:
-- HOWEVER: there's a further hack here which is that leading
-- non-numeric arguments are considered qualitative (e.g. "monad" vs
-- "sparks") rather than quantitative and are not pruned by this
-- function.
-- shortArgs (h:tl) | isNumber h = []
-- | otherwise = h : shortArgs tl
----------------------------------------------------------------------------------------------------