Window benchmarks

parallel-orch/node.py

@@ -14,10 +14,14 @@
 import analysis
 
 STATE_LOG = '[STATE_LOG] '
+OVERHEAD_LOG = '[OVERHEAD_LOG] '
 
 def state_log(s):
     logging.info(STATE_LOG + s)
 
+def overhead_log(s):
+    logging.info(OVERHEAD_LOG + s)
+
 class NodeState(Enum):
     INIT = auto()
     READY = auto()
@@ -526,13 +530,15 @@ def parse_env(content):
         return conflict_exists
 
     def kill_children(self):
+        overhead_log(f"KILL|{self.cnid}")
         assert self.state in [NodeState.EXECUTING, NodeState.SPEC_EXECUTING]
         process = self.exec_ctxt.process
         try:
             os.killpg(process.pid, signal.SIGKILL)
         except ProcessLookupError:
             pass
         process.wait()
+        overhead_log(f"KILL_END|{self.cnid}")
 
     ##                                      ##
     ##          Transition Functions        ##
@@ -568,8 +574,9 @@ def reset_to_ready(self, spec_pre_env: str = None):
         # TODO: make this more sophisticated
         if self.state in [NodeState.EXECUTING, NodeState.SPEC_EXECUTING]:
             self.kill_children()
-
+        overhead_log(f"DELETE_SANDBOX|{self.cnid}")
         util.delete_sandbox(self.exec_ctxt.sandbox_dir)
+        overhead_log(f"DELETE_SANDBOX_END|{self.cnid}")
         self.exec_ctxt = None
         self.exec_result = None
         if spec_pre_env is not None:
@@ -615,7 +622,9 @@ def commit_frontier_execution(self):
             self.trace_fd = None
             self.trace_ctx = None
         self.update_loop_list_context()
+        overhead_log(f"COMMIT|{self.cnid}")
         executor.commit_workspace(self.exec_ctxt.sandbox_dir)
+        overhead_log(f"COMMIT_END|{self.cnid}")
         # util.delete_sandbox(self.exec_ctxt.sandbox_dir)
         self.state = NodeState.COMMITTED
         self.trace_state()
@@ -635,7 +644,9 @@ def finish_spec_execution(self):
 
     def commit_speculated(self):
         assert self.state == NodeState.SPECULATED
+        overhead_log(f"COMMIT|{self.cnid}")
         executor.commit_workspace(self.exec_ctxt.sandbox_dir)
+        overhead_log(f"COMMIT_END|{self.cnid}")
         # util.delete_sandbox(self.exec_ctxt.sandbox_dir)
         self.state = NodeState.COMMITTED
         self.trace_state()

parallel-orch/partial_program_order.py

@@ -7,13 +7,17 @@
 
 PROG_LOG =  '[PROG_LOG] '
 EVENT_LOG = '[EVENT_LOG] '
+OVERHEAD_LOG = '[OVERHEAD_LOG] '
 
 def event_log(s):
     logging.info(EVENT_LOG + s)
 
 def progress_log(s):
     logging.info(PROG_LOG + s)
 
+def overhead_log(s):
+    logging.info(OVERHEAD_LOG + s)
+
 def simulate_loop_iter_env(env, var, loop_list_context, loop_iters):
     loop_list = loop_list_context.get_top()
     try:
@@ -401,7 +405,9 @@ def valid(self):
 
     def fetch_fs_actions(self):
         for node in self.get_executing_normal_and_spec_nodes():
+            overhead_log(f"TRACE_FETCHING|{node.cnid}")
             node.gather_fs_actions()
+            overhead_log(f"TRACE_FETCHING_END|{node.cnid}")
 
     def _has_fs_deps(self, concrete_node_id: ConcreteNodeId):
         node_of_interest : ConcreteNode = self.get_concrete_node(concrete_node_id)

report/benchmarks/bio4/large/setup

@@ -0,0 +1,6 @@
+#!/bin/bash
+
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+export BIODIR="${PASH_SPEC_TOP}/report/benchmarks/bio4"
+$BIODIR/setup.sh large/list
+

report/benchmarks/bio4/medium/setup

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+export BIODIR="${PASH_SPEC_TOP}/report/benchmarks/bio4"
+$BIODIR/setup.sh medium/list

report/benchmarks/bio4/setup.sh

@@ -1,7 +1,5 @@
 #!/bin/bash
 
-set -e
-
 export PATH=$PATH:$HOME/.local/bin
 export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
 

report/benchmarks/bio4/small/setup

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+export BIODIR="${PASH_SPEC_TOP}/report/benchmarks/bio4"
+$BIODIR/setup.sh small/list

report/benchmarks/dgsh/run_all.sh

@@ -0,0 +1,22 @@
+#!/bin/bash
+
+export PATH=$PATH:$HOME/.local/bin
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+export PASH_TOP=${PASH_TOP:-$PASH_SPEC_TOP/deps/pash}
+
+output_dir="$PASH_SPEC_TOP/report/output/dgsh"
+download_dir="$PASH_SPEC_TOP/report/resources/dgsh"
+result_dir="$PASH_SPEC_TOP/results/dgsh"
+
+./run --target sh-only
+rm -rf /tmp/*
+mkdir -p $result_dir/base
+mv $output_dir/* $result_dir/base
+
+for window in 0 5 10 20 40
+do
+    ./run --target hs-only --window $window
+    rm -rf /tmp/*
+    mkdir -p $output_dir/$result_dir
+    mv $output_dir/* $result_dir/$window
+done

report/benchmarks/max_temp/max_temp_optimal.sh

@@ -0,0 +1,17 @@
+#!/bin/bash
+
+## This is used to test against the optimal parallel ordering (theoretical best)
+
+for year in $(seq $FROM $TO); do
+    (
+        echo "Processing year: $year"
+        find "$RESOURCE_DIR/$year" -type f -name '*.gz' -print0 | xargs -0 -I{} sh -c 'gunzip -c "{}" >> "$RESOURCE_DIR/$year.txt"'
+
+        # Process in parallel
+        cut -c 89-92 "$RESOURCE_DIR/$year.txt" | grep -v 999 | sort -rn | head -n1 &
+        cut -c 89-92 "$RESOURCE_DIR/$year.txt" | grep -v 999 | sort -n | head -n1 &
+        cut -c 89-92 "$RESOURCE_DIR/$year.txt" | grep -v 999 | awk '{ total += $1; count++ } END { print total/count }' &
+        wait
+    ) &
+done
+wait

report/benchmarks/max_temp/run_all.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+
+output_dir="$PASH_SPEC_TOP/report/output/max_temp" # Adjust the path as necessary
+download_dir="$PASH_SPEC_TOP/report/resources/max_temp"
+result_dir="$PASH_SPEC_TOP/results/max_temp"
+echo $output_dir
+echo $download_dir
+echo $result_dir
+
+rm -f $download_dir/*.txt
+
+./run --target sh-only
+rm $download_dir/*.txt
+mkdir -p $result_dir/base/
+mv $output_dir/* $result_dir/base/
+
+for window in 40 30 20 10 0
+do
+    ./run --target hs-only --window $window
+    rm $download_dir/*.txt
+    rm -rf /tmp/*
+    mkdir -p $result_dir/$window
+    mv $output_dir/* $result_dir/$window
+done
+

report/benchmarks/max_temp/setup.1.sh

@@ -0,0 +1,24 @@
+#!/bin/bash
+
+export PATH=$PATH:$HOME/.local/bin
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+export PASH_TOP=${PASH_TOP:-$PASH_SPEC_TOP/deps/pash}
+
+download_dir="$PASH_SPEC_TOP/report/resources/max_temp" # Adjust the path as necessary
+
+
+# Loop through the years from 2000 to 2018
+for year in {1901..1901}; do
+    # Create the URL for the specific year
+    url="ftp://ftp.ncdc.noaa.gov/pub/data/noaa/$year"
+
+    # Create a temporary directory for each year
+    mkdir -p "$download_dir/$year"
+    mkdir -p $year
+
+    # Download files from the current year's directory
+    lftp -c "open '$url' && mirror --parallel=5 --verbose . '$year'"
+
+    mv "$year" "$download_dir/$year"
+
+done

report/benchmarks/unix_50/full.sh

@@ -1,25 +1,25 @@
 #!/bin/bash
 
-IN1=$IN/100M-1.txt
-IN2=$IN/100M-2.txt
-IN3=$IN/100M-3.txt
-IN4=$IN/100M-4.txt
-IN5=$IN/100M-5.txt
-IN6=$IN/100M-6.txt
-IN7=$IN/100M-7.txt
-IN8=$IN/100M-8.txt
-IN91=$IN/100M-9.1.txt
-IN92=$IN/100M-9.2.txt
-IN93=$IN/100M-9.3.txt
-IN94=$IN/100M-9.4.txt
-IN95=$IN/100M-9.5.txt
-IN96=$IN/100M-9.6.txt
-IN97=$IN/100M-9.7.txt
-IN98=$IN/100M-9.8.txt
-IN99=$IN/100M-9.9.txt
-IN10=$IN/100M-10.txt
-IN11=$IN/100M-11.txt
-IN12=$IN/100M-12.txt
+IN1=$IN/1G-1.txt
+IN2=$IN/1G-2.txt
+IN3=$IN/1G-3.txt
+IN4=$IN/1G-4.txt
+IN5=$IN/1G-5.txt
+IN6=$IN/1G-6.txt
+IN7=$IN/1G-7.txt
+IN8=$IN/1G-8.txt
+IN91=$IN/1G-9.1.txt
+IN92=$IN/1G-9.2.txt
+IN93=$IN/1G-9.3.txt
+IN94=$IN/1G-9.4.txt
+IN95=$IN/1G-9.5.txt
+IN96=$IN/1G-9.6.txt
+IN97=$IN/1G-9.7.txt
+IN98=$IN/1G-9.8.txt
+IN99=$IN/1G-9.9.txt
+IN10=$IN/1G-10.txt
+IN11=$IN/1G-11.txt
+IN12=$IN/1G-12.txt
 
 #1.sh 1.0: extract the last name
 cat $IN1 | cut -d ' ' -f 2
@@ -128,8 +128,3 @@ cat $IN11 | grep 'UNIX' | cut -f 1
 
 #36.sh 11.2: most repeated first name in the list?
 cat $IN11 | cut -f 2 | cut -d ' ' -f 1 | sort | uniq -c | sort -nr | head -n 1 | fmt -w1 | sed 1d
-
-
-# # 12.1: transform this list of instructions such that if the snake follows the
-# #       new instructions top to bottom, it ends on the location of the apple.
-# cat $IN12 > /dev/null

report/benchmarks/unix_50/run_all.sh

@@ -0,0 +1,20 @@
+export PATH=$PATH:$HOME/.local/bin
+export PASH_SPEC_TOP=${PASH_SPEC_TOP:-$(git rev-parse --show-toplevel --show-superproject-working-tree)}
+export PASH_TOP=${PASH_TOP:-$PASH_SPEC_TOP/deps/pash}
+
+output_dir="$PASH_SPEC_TOP/report/output/unix_50"
+download_dir="$PASH_SPEC_TOP/report/resources/unix_50"
+result_dir="$PASH_SPEC_TOP/results/unix_50"
+
+./run --target sh-only
+rm -rf /tmp/*
+mkdir -p $result_dir/base
+mv $output_dir/* $result_dir/base
+
+for window in 0 10 20 30 40
+do
+    ./run --target hs-only --window $window
+    rm -rf /tmp/*
+    mkdir -p $result_dir/$window
+    mv $output_dir/* $result_dir/$window
+done

report/benchmarks/unix_50/setup.sh → report/benchmarks/unix_50/setup

@@ -33,9 +33,6 @@ for i in ${inputs[@]}; do
     if [[ ! -f "$inflate_size-$i.txt" ]]; then
         $inflate "$i.txt" 10M
         $inflate "10M-$i.txt" $inflate_size &
-        # $inflate "100M-10M-$i.txt" 1G
-
-        # mv "1G-100M-10M-$i.txt" "1G-$i.txt"
     fi
 done
 wait