diff --git a/WORKSPACE b/WORKSPACE
index 46bdeae55d..4b0501c47f 100644
--- a/WORKSPACE
+++ b/WORKSPACE
@@ -1,6 +1,6 @@
 workspace(name = "CompilerGym")
 
-load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
+load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive", "http_file")
 
 # === Google test ===
 
@@ -201,11 +201,81 @@ filegroup(
 )
 
 http_archive(
+    name = "ctuning-ai",
+    build_file = "//:compiler_gym/third_party/ctuning-ai.BUILD",
+    sha256 = "2e12233a3d898a96a2ce279123ee256bee357f649e3c579f974628b6c35ea53d",
+    strip_prefix = "ai-afaba3e2900523a0813693fd614a114b4a6f6729",
+    urls = ["https://github.com/ctuning/ai/archive/afaba3e2900523a0813693fd614a114b4a6f6729.tar.gz"],
+)
+
+# Datasets.
+
+http_file(
+    name = "cBench_consumer_tiff_data",
+    sha256 = "779abb7b7fee8733313e462e6066c16375e9209a9f7ff692fd06c7598946939a",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_consumer_tiff_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_office_data",
+    sha256 = "cfa09cd37cb93aba57415033905dc6308653c7b833feba5a25067bfb62999f32",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_office_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_telecom_data",
+    sha256 = "e5cb6663beefe32fd12f90c8f533f8e1bce2f05ee4e3836efb5556d5e1089df0",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_telecom_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_consumer_jpeg_data",
+    sha256 = "bec5ffc15cd2f952d9a786f3cd31d90955c318a5e4f69c5ba472f79d5a3e8f0b",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_consumer_jpeg_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_telecom_gsm_data",
+    sha256 = "52545d3a0ce15021131c62d96d3a3d7e6670e2d6c34226ac9a3d5191a1ee214a",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_telecom_gsm_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_consumer_data",
+    sha256 = "a4d40344af3022bfd7b4c6fcf6d59d598825b07d9e37769dbf1b3effa39aa445",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_consumer_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_bzip2_data",
+    sha256 = "46e5760eeef77e6b0c273af92de971bc45f33a59e0efc183073d9aa6b716c302",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_bzip2_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_network_patricia_data",
+    sha256 = "72dae0e670d93ef929e50aca7a138463e0915502281ccafe793e378cb2a85dfb",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_network_patricia_data.tar.gz"],
+)
+
+http_file(
     name = "cBench_network_dijkstra_data",
-    build_file_content = all_content,
+    sha256 = "41c13f59cdfbc772081cd941f499b030370bc570fc2ba60a5c4b7194bc36ca5f",
     urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_network_dijkstra_data.tar.gz"],
 )
 
+http_file(
+    name = "cBench_automotive_susan_data",
+    sha256 = "df56e1e44ccc560072381cdb001d770003ac74f92593dd5dbdfdd4ff9332a8e6",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_automotive_susan_data.tar.gz"],
+)
+
+http_file(
+    name = "cBench_automotive_qsort_data",
+    sha256 = "510b4225021408ac190f6f793e7d7171d3553c9916cfa8b2fb4ace005105e768",
+    urls = ["https://downloads.sourceforge.net/project/cbenchmark/cDatasets/V1.1/cDatasets_V1.1_automotive_qsort_data.tar.gz"],
+)
+
 # === C++ cpuinfo ===
 
 http_archive(
diff --git a/compiler_gym/envs/compiler_env.py b/compiler_gym/envs/compiler_env.py
index 0fbdbb9985..cbf822764d 100644
--- a/compiler_gym/envs/compiler_env.py
+++ b/compiler_gym/envs/compiler_env.py
@@ -795,7 +795,7 @@ def register_dataset(self, dataset: Dataset) -> bool:
             >>> env.benchmark = "my-dataset-v0/1"
 
         :param dataset: A :class:`Dataset` instance describing the new dataset.
-        :return: :code:`True` if the dataset was registered added, else :code:`False`.
+        :return: :code:`True` if the dataset was added, else :code:`False`.
         :raises ValueError: If a dataset with this name is already registered.
         """
         platform = {"darwin": "macos"}.get(sys.platform, sys.platform)
diff --git a/compiler_gym/envs/llvm/BUILD b/compiler_gym/envs/llvm/BUILD
index c65478ad7d..10daf39019 100644
--- a/compiler_gym/envs/llvm/BUILD
+++ b/compiler_gym/envs/llvm/BUILD
@@ -33,8 +33,12 @@ py_library(
 py_library(
     name = "datasets",
     srcs = ["datasets.py"],
+    data = [
+        "//compiler_gym/third_party/llvm:clang",
+    ],
     deps = [
         "//compiler_gym/datasets:dataset",
+        "//compiler_gym/util",
     ],
 )
 
diff --git a/compiler_gym/envs/llvm/datasets.py b/compiler_gym/envs/llvm/datasets.py
index a4cf787259..034c330313 100644
--- a/compiler_gym/envs/llvm/datasets.py
+++ b/compiler_gym/envs/llvm/datasets.py
@@ -3,9 +3,44 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 """This module defines the available LLVM datasets."""
-from typing import Callable, Dict, Optional
+import io
+import os
+import re
+import shutil
+import subprocess
+import sys
+import tarfile
+import tempfile
+from collections import defaultdict
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from contextlib import contextmanager
+from pathlib import Path
+from typing import Callable, Dict, List, NamedTuple, Optional
+
+import fasteners
+import gym
 
 from compiler_gym.datasets.dataset import Dataset
+from compiler_gym.util.download import download
+from compiler_gym.util.runfiles_path import cache_path, runfiles_path, site_data_path
+from compiler_gym.util.timer import Timer
+
+_CLANG = runfiles_path("CompilerGym/compiler_gym/third_party/llvm/clang")
+
+_CBENCH_DATA = site_data_path("llvm/cBench-v0-runtime-data/runtime_data")
+_CBENCH_DATA_URL = (
+    "https://dl.fbaipublicfiles.com/compiler_gym/cBench-v0-runtime-data.tar.bz2"
+)
+_CBENCH_DATA_SHA256 = "a1b5b5d6b115e5809ccaefc2134434494271d184da67e2ee43d7f84d07329055"
+
+
+if sys.platform == "darwin":
+    _COMPILE_ARGS = [
+        "-L",
+        "/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/lib",
+    ]
+else:
+    _COMPILE_ARGS = []
 
 LLVM_DATASETS = [
     Dataset(
@@ -20,13 +55,25 @@
     ),
     Dataset(
         name="cBench-v0",
-        url="https://dl.fbaipublicfiles.com/compiler_gym/llvm_bitcodes-10.0.0-cBench-v0.tar.bz2",
+        url="https://dl.fbaipublicfiles.com/compiler_gym/llvm_bitcodes-10.0.0-cBench-v0-macos.tar.bz2",
+        license="BSD 3-Clause",
+        description="https://github.com/ctuning/ctuning-programs",
+        compiler="llvm-10.0.0",
+        file_count=23,
+        size_bytes=7154448,
+        sha256="072a730c86144a07bba948c49afe543e4f06351f1cb17f7de77f91d5c1a1b120",
+        platforms=["macos"],
+    ),
+    Dataset(
+        name="cBench-v0",
+        url="https://dl.fbaipublicfiles.com/compiler_gym/llvm_bitcodes-10.0.0-cBench-v0-linux.tar.bz2",
         license="BSD 3-Clause",
         description="https://github.com/ctuning/ctuning-programs",
         compiler="llvm-10.0.0",
         file_count=23,
-        size_bytes=7150112,
-        sha256="498425efe8e335eb72623df50427eca87333d011f017ec3be5608883b4b7687a",
+        size_bytes=6940416,
+        sha256="9b5838a90895579aab3b9375e8eeb3ed2ae58e0ad354fec7eb4f8b31ecb4a360",
+        platforms=["linux"],
     ),
     Dataset(
         name="github-v0",
@@ -110,12 +157,554 @@
     ),
 ]
 
-# A map from benchmark name to a callback which takes as input an LlvmEnv
-# instance and returns None if the environment is valid, else a string error
-# message.
-#
-# TODO(cummins): Populate this map for cBench using the CK meta properties.
-# See: https://github.com/ctuning/ai/blob/main/program/cbench-bzip2/.cm/meta.json
-LLVM_BENCHMARK_VALIDATION_CALLBACKS: Dict[
-    str, Callable[["LlvmEnv"], Optional[str]]
-] = {}
+
+class BenchmarkExecutionResult(NamedTuple):
+    """The result of running a benchmark."""
+
+    walltime_seconds: float
+    """The execution time in seconds."""
+
+    error: Optional[str] = None
+    """An error message."""
+
+    output: Optional[str] = None
+    """The output generated by the benchmark."""
+
+
+def _compile_and_run_bitcode_file(
+    bitcode_file: Path,
+    cmd: str,
+    cwd: Path,
+    linkopts: List[str],
+    num_runs: int,
+    timeout_seconds: float = 60,
+) -> BenchmarkExecutionResult:
+    """Run the given cBench benchmark."""
+    binary = cwd / "a.out"
+
+    # cBench benchmarks expect that a file _finfo_dataset exists in the
+    # current working directory and contains the number of benchmark
+    # iterations in it.
+    with open(cwd / "_finfo_dataset", "w") as f:
+        print(num_runs, file=f)
+
+    # Generate the a.out binary file.
+    assert not binary.is_file()
+    subprocess.check_call(
+        [_CLANG, str(bitcode_file), "-o", str(binary)] + _COMPILE_ARGS + list(linkopts)
+    )
+    assert binary.is_file()
+
+    process = subprocess.Popen(
+        cmd,
+        shell=True,
+        stderr=subprocess.STDOUT,
+        stdout=subprocess.PIPE,
+        env=os.environ,
+        cwd=cwd,
+    )
+
+    try:
+        with Timer() as timer:
+            stdout, _ = process.communicate(timeout=timeout_seconds)
+    except subprocess.TimeoutExpired:
+        process.kill()
+        return BenchmarkExecutionResult(
+            walltime_seconds=timeout_seconds,
+            error=f"Benchmark failed to complete with {timeout_seconds} timeout.",
+        )
+    finally:
+        binary.unlink()
+
+    if process.returncode:
+        try:
+            output = stdout.decode("utf-8")
+            msg = f"Benchmark exited with returncode {process.returncode}. Output: {output}"
+        except UnicodeDecodeError:
+            msg = f"Benchmark exited with returncode {process.returncode}"
+        return BenchmarkExecutionResult(walltime_seconds=timer.time, error=msg)
+
+    return BenchmarkExecutionResult(walltime_seconds=timer.time, output=stdout)
+
+
+@fasteners.interprocess_locked(cache_path("cBench-v0-runtime-data.LOCK"))
+def download_cBench_runtime_data() -> bool:
+    """Download and unpack the cBench runtime dataset."""
+    if _CBENCH_DATA.is_dir():
+        return False
+    else:
+        tar_contents = io.BytesIO(
+            download(_CBENCH_DATA_URL, sha256=_CBENCH_DATA_SHA256)
+        )
+        with tarfile.open(fileobj=tar_contents, mode="r:bz2") as tar:
+            _CBENCH_DATA.parent.mkdir(parents=True)
+            tar.extractall(_CBENCH_DATA.parent)
+        assert _CBENCH_DATA.is_dir()
+        return True
+
+
+def _make_cBench_validator(
+    cmd: str,
+    linkopts: List[str],
+    os_env: Dict[str, str],
+    num_runs: int = 1,
+    compare_output: bool = True,
+    input_files: Optional[List[Path]] = None,
+    output_files: Optional[List[Path]] = None,
+    validate_result: Optional[
+        Callable[[BenchmarkExecutionResult], Optional[str]]
+    ] = None,
+    pre_execution_callback: Optional[Callable[[], None]] = None,
+):
+    """Construct a validation callback for a cBench benchmark. See validator() for usage."""
+    input_files = input_files or []
+    output_files = output_files or []
+
+    def validator_cb(env):
+        """The validation callback."""
+        for path in input_files:
+            if not path.is_file():
+                raise FileNotFoundError(f"Required benchmark input not found: {path}")
+
+        # Expand shell variable substitutions in the benchmark command.
+        expanded_command = expand_command_vars(cmd)
+
+        with tempfile.TemporaryDirectory(dir=env.service.connection.working_dir) as d:
+            # Execute the benchmark in a temporary working directory.
+            cwd = Path(d)
+            # Translate the output file names into paths inside the working
+            # directory.
+            output_paths = [cwd / o for o in output_files]
+
+            with benchmark_execution_environment(os_env):
+                if pre_execution_callback:
+                    pre_execution_callback()
+
+                # Produce a gold-standard output using a reference version of
+                # the benchmark.
+                if compare_output or output_files:
+                    gs_env = gym.make("llvm-v0")
+                    try:
+                        gs_env.reset(benchmark=env.benchmark)
+                        # Serialize the benchmark to a bitcode file that will then be
+                        # compiled to a binary.
+                        bitcode_file = Path(gs_env.observation["BitcodeFile"])
+                        try:
+                            gold_standard = _compile_and_run_bitcode_file(
+                                bitcode_file=bitcode_file,
+                                cmd=expanded_command,
+                                cwd=cwd,
+                                num_runs=1,
+                                linkopts=linkopts,
+                            )
+                            if gold_standard.error:
+                                raise OSError(
+                                    f"Failed to produce reference output for benchmark '{env.benchmark}' "
+                                    f"using '{cmd}': {gold_standard.error}"
+                                )
+                        finally:
+                            bitcode_file.unlink()
+                    finally:
+                        gs_env.close()
+
+                    # Check that the reference run produced the expected output
+                    # files.
+                    for path in output_paths:
+                        if not path.is_file():
+                            try:
+                                output = gold_standard.output.decode("utf-8")
+                            except UnicodeDecodeError:
+                                output = "<binary>"
+                            raise FileNotFoundError(
+                                f"Expected file '{path}' not generated\n"
+                                f"Benchmark: {env.benchmark}\n"
+                                f"Command: {cmd}\n"
+                                f"Output: {output}"
+                            )
+                        path.rename(f"{path}.gold_standard")
+
+                # Serialize the benchmark to a bitcode file that will then be
+                # compiled to a binary.
+                bitcode_file = Path(env.observation["BitcodeFile"])
+                try:
+                    outcome = _compile_and_run_bitcode_file(
+                        bitcode_file=bitcode_file,
+                        cmd=expanded_command,
+                        cwd=cwd,
+                        num_runs=num_runs,
+                        linkopts=linkopts,
+                    )
+                finally:
+                    bitcode_file.unlink()
+
+            if outcome.error:
+                return outcome.error
+
+            # Run a user-specified validation hook.
+            if validate_result:
+                validate_result(outcome)
+
+            # Difftest the console output.
+            if compare_output and gold_standard.output != outcome.output:
+                try:
+                    return (
+                        f"Benchmark output differs from expected.\n"
+                        f"Expected: {gold_standard.decode('utf-8')}\n"
+                        f"Actual: {outcome.output.decode('utf-8')}"
+                    )
+                except UnicodeDecodeError:
+                    return f"Benchmark output differs from expected (binary diff)"
+
+            # Difftest the output files.
+            for path in output_paths:
+                if not path.is_file():
+                    return f"Expected file not generated by benchmark {env.benchmark}: {path}.\nCommand: {cmd}"
+                diff = subprocess.Popen(["diff", str(path), f"{path}.gold_standard"])
+                stdout, _ = diff.communicate()
+                if diff.returncode:
+                    try:
+                        return f"Benchmark output file '{path}' differs from expected: {stdout}"
+                    except UnicodeDecodeError:
+                        return f"Benchmark output file '{path}' differs from expected (binary diff)"
+
+    return validator_cb
+
+
+@contextmanager
+def temporary_environment():
+    """Yield a temporary os.environ state."""
+    _environ = os.environ.copy()
+    try:
+        yield
+    finally:
+        os.environ.clear()
+        os.environ.update(_environ)
+
+
+@contextmanager
+def benchmark_execution_environment(env: Dict[str, str]):
+    """Setup the os.environ state for executing benchmarks."""
+    with temporary_environment():
+        for var in os.environ:
+            if not var.startswith("COMPILER_GYM_SITE_DATA") and var not in {
+                "PATH",
+                "RUNFILES_DIR",
+                "RUNFILES_MANIFEST_FILE",
+                "USER",
+                "SHELL",
+                "TERM",
+                "TMPDIR",
+                "HOME",
+            }:
+                del os.environ[var]
+        for key, val in env.items():
+            os.environ[key] = expand_command_vars(val)
+        yield
+
+
+def expand_command_vars(cmd: str) -> str:
+    """Expand shell variables in a command."""
+    with temporary_environment():
+        os.environ.clear()
+        os.environ["BIN"] = "./a.out"
+        os.environ["D"] = str(_CBENCH_DATA)
+        return os.path.expandvars(cmd)
+
+
+# A map from benchmark name to validation callbacks. Defined below.
+_VALIDATORS: Dict[str, List[Callable[["LlvmEnv"], Optional[str]]]] = defaultdict(list)
+
+
+def validator(
+    benchmark: str,
+    cmd: str,
+    data: Optional[List[str]] = None,
+    outs: Optional[List[str]] = None,
+    platforms: Optional[List[str]] = None,
+    compare_output: bool = True,
+    validate_result: Optional[
+        Callable[[BenchmarkExecutionResult], Optional[str]]
+    ] = None,
+    linkopts: List[str] = None,
+    env: Dict[str, str] = None,
+    pre_execution_callback: Optional[Callable[[], None]] = None,
+) -> bool:
+    """Declare a new benchmark validator.
+
+    TODO(cummins): Pull this out into a public API.
+
+    :param benchmark: The name of the benchmark that this validator supports.
+    :cmd: The shell command to run the validation. Variable substitution is
+        applied to this value as follows: :code:`$BIN` is replaced by the path
+        of the compiled binary and :code:`$D` is replaced with the path to the
+        benchmark's runtime data directory.
+    :data: A list of paths to input files.
+    :outs: A list of paths to output files.
+    :return: :code:`True` if the new validator was registered, else :code:`False`.
+    """
+    platforms = platforms or ["linux", "macos"]
+    if {"darwin": "macos"}.get(sys.platform, sys.platform) not in platforms:
+        return False
+    infiles = [_CBENCH_DATA / p for p in data or []]
+    outfiles = [Path(p) for p in outs or []]
+    linkopts = linkopts or []
+    env = env or {}
+
+    _VALIDATORS[benchmark].append(
+        _make_cBench_validator(
+            cmd=cmd,
+            input_files=infiles,
+            output_files=outfiles,
+            compare_output=compare_output,
+            validate_result=validate_result,
+            linkopts=linkopts,
+            os_env=env,
+            pre_execution_callback=pre_execution_callback,
+        )
+    )
+
+    return True
+
+
+def get_llvm_benchmark_validation_callback(
+    env: "LlvmEnv",
+) -> Optional[Callable[["LlvmEnv"], Optional[str]]]:
+    """Return a callback for validating a given environment state.
+
+    If there is no valid callback, returns :code:`None`.
+
+    :param env: An :class:`LlvmEnv` instance.
+    :return: An optional callback that takes an :class:`LlvmEnv` instance as
+        argument and returns an optional string containing a validation error
+        message.
+    """
+    validators = _VALIDATORS.get(env.benchmark)
+
+    # No match.
+    if not validators:
+        return None
+
+    def composed(env):
+        download_cBench_runtime_data()
+
+        # Validation callbacks are read-only on the environment so it is
+        # safe to run validators simultaneously in parallel threads.
+        with ThreadPoolExecutor() as executor:
+            futures = [executor.submit(validator, env) for validator in validators]
+            result = None
+            for future in as_completed(futures):
+                result = future.result() or result
+            return result
+
+    return composed
+
+
+# ===============================
+# Definition of cBench validators
+# ===============================
+
+
+def validate_sha_output(result: BenchmarkExecutionResult):
+    """SHA benchmark prints 5 random hex strings. Normally these hex strings are
+    16 characters but occasionally they are less (presumably becuase of a
+    leading zero being omitted).
+    """
+    assert re.match(
+        r"[0-9a-f]{0,16} [0-9a-f]{0,16} [0-9a-f]{0,16} [0-9a-f]{0,16} [0-9a-f]{0,16}",
+        result.output.decode("utf-8").rstrip(),
+    )
+
+
+def setup_ghostscript_library_files():
+    """Pre-execution setup hook for ghostscript."""
+    # Copy input data into current directory since ghostscript doesn't like long
+    # input paths.
+    for path in (_CBENCH_DATA / "office_data").iterdir():
+        if path.name.endswith(".ps"):
+            shutil.copyfile(path, path.name)
+    # Ghostscript doesn't like the library files being symlinks so copy them
+    # into the working directory as regular files.
+    for path in (_CBENCH_DATA / "ghostscript").iterdir():
+        if path.name.endswith(".ps"):
+            shutil.copyfile(path, path.name)
+
+
+validator(
+    benchmark="benchmark://cBench-v0/bitcount",
+    cmd="$BIN 1125000",
+)
+
+validator(
+    benchmark="benchmark://cBench-v0/bitcount",
+    cmd="$BIN 512",
+)
+
+for i in range(1, 21):
+    validator(
+        benchmark="benchmark://cBench-v0/adpcm",
+        cmd=f"$BIN $D/telecom_data/{i}.adpcm",
+        data=[f"telecom_data/{i}.adpcm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/adpcm",
+        cmd=f"$BIN $D/telecom_data/{i}.pcm",
+        data=[f"telecom_data/{i}.pcm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/blowfish",
+        cmd=f"$BIN d $D/office_data/{i}.benc output.txt 1234567890abcdeffedcba0987654321",
+        data=[f"office_data/{i}.benc"],
+        outs=["output.txt"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/bzip2",
+        cmd=f"$BIN -d -k -f -c $D/bzip2_data/{i}.bz2",
+        data=[f"bzip2_data/{i}.bz2"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/crc32",
+        cmd=f"$BIN $D/telecom_data/{i}.pcm",
+        data=[f"telecom_data/{i}.pcm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/dijkstra",
+        cmd=f"$BIN $D/network_dijkstra_data/{i}.dat",
+        data=[f"network_dijkstra_data/{i}.dat"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/gsm",
+        cmd=f"$BIN -fps -c $D/telecom_gsm_data/{i}.au",
+        data=[f"telecom_gsm_data/{i}.au"],
+    )
+
+    # TODO(cummins): ispell executable appears broken. Investigation needed.
+    # validator(
+    #     benchmark="benchmark://cBench-v0/ispell",
+    #     cmd=f"$BIN -a -d americanmed+ $D/office_data/{i}.txt",
+    #     data = [f"office_data/{i}.txt"],
+    # )
+
+    validator(
+        benchmark="benchmark://cBench-v0/jpeg-c",
+        cmd=f"$BIN -dct int -progressive -outfile output.jpeg $D/consumer_jpeg_data/{i}.ppm",
+        data=[f"consumer_jpeg_data/{i}.ppm"],
+        outs=["output.jpeg"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/jpeg-d",
+        cmd=f"$BIN -dct int -outfile output.ppm $D/consumer_jpeg_data/{i}.jpg",
+        data=[f"consumer_jpeg_data/{i}.jpg"],
+        outs=["output.ppm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/patricia",
+        cmd=f"$BIN $D/network_patricia_data/{i}.udp",
+        data=[f"network_patricia_data/{i}.udp"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/qsort",
+        cmd=f"$BIN $D/automotive_qsort_data/{i}.dat",
+        data=[f"automotive_qsort_data/{i}.dat"],
+        outs=["sorted_output.dat"],
+        linkopts=["-lm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/rijndael",
+        cmd=f"$BIN $D/office_data/{i}.enc output.dec d 1234567890abcdeffedcba09876543211234567890abcdeffedcba0987654321",
+        data=[f"office_data/{i}.enc"],
+        outs=["output.dec"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/sha",
+        cmd=f"$BIN $D/office_data/{i}.txt",
+        data=[f"office_data/{i}.txt"],
+        compare_output=False,
+        validate_result=validate_sha_output,
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/stringsearch",
+        cmd=f"$BIN $D/office_data/{i}.txt $D/office_data/{i}.s.txt output.txt",
+        data=[f"office_data/{i}.txt"],
+        outs=["output.txt"],
+        linkopts=["-lm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/stringsearch2",
+        cmd=f"$BIN $D/office_data/{i}.txt $D/office_data/{i}.s.txt output.txt",
+        data=[f"office_data/{i}.txt"],
+        outs=["output.txt"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/susan",
+        cmd=f"$BIN $D/automotive_susan_data/{i}.pgm output_large.corners.pgm -c",
+        data=[f"automotive_susan_data/{i}.pgm"],
+        outs=["output_large.corners.pgm"],
+        linkopts=["-lm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/tiff2bw",
+        cmd=f"$BIN $D/consumer_tiff_data/{i}.tif output.tif",
+        data=[f"consumer_tiff_data/{i}.tif"],
+        outs=["output.tif"],
+        linkopts=["-lm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/tiff2rgba",
+        cmd=f"$BIN $D/consumer_tiff_data/{i}.tif output.tif",
+        data=[f"consumer_tiff_data/{i}.tif"],
+        outs=["output.tif"],
+        linkopts=["-lm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/tiffdither",
+        cmd=f"$BIN $D/consumer_tiff_data/{i}.bw.tif out.tif",
+        data=[f"consumer_tiff_data/{i}.bw.tif"],
+        outs=["out.tif"],
+        linkopts=["-lm"],
+    )
+
+    validator(
+        benchmark="benchmark://cBench-v0/tiffmedian",
+        cmd=f"$BIN $D/consumer_tiff_data/{i}.nocomp.tif output.tif",
+        data=[f"consumer_tiff_data/{i}.nocomp.tif"],
+        outs=["output.tif"],
+        linkopts=["-lm"],
+    )
+
+    # TODO(cummins): On macOS the following benchmarks abort with an illegal
+    # hardware instruction error.
+    if sys.platform != "darwin":
+        validator(
+            benchmark="benchmark://cBench-v0/lame",
+            cmd=f"$BIN $D/consumer_data/{i}.wav output.mp3",
+            data=[f"consumer_data/{i}.wav"],
+            outs=["output.mp3"],
+            compare_output=False,
+            linkopts=["-lm"],
+        )
+
+        validator(
+            benchmark="benchmark://cBench-v0/ghostscript",
+            cmd=f"$BIN -sDEVICE=ppm -dNOPAUSE -dQUIET -sOutputFile=output.ppm -- {i}.ps",
+            data=[f"office_data/{i}.ps"],
+            outs=["output.ppm"],
+            linkopts=["-lm", "-lz"],
+            pre_execution_callback=setup_ghostscript_library_files,
+        )
diff --git a/compiler_gym/third_party/cBench/BUILD b/compiler_gym/third_party/cBench/BUILD
index 59a3e40d7d..bb78136987 100644
--- a/compiler_gym/third_party/cBench/BUILD
+++ b/compiler_gym/third_party/cBench/BUILD
@@ -3,6 +3,76 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 
+genrule(
+    name = "cBench_tar",
+    srcs = [":cBench"],
+    outs = ["llvm_bitcodes-10.0.0-cBench.tar.bz2"],
+    cmd = "tar cjfh $@ -C $(@D) cBench-v0",
+    visibility = ["//visibility:public"],
+)
+
+filegroup(
+    name = "cBench",
+    srcs = [
+        ":adpcm",
+        ":bitcount",
+        ":blowfish",
+        ":bzip2",
+        ":crc32",
+        ":dijkstra",
+        ":ghostscript",
+        ":gsm",
+        ":ispell",
+        ":jpeg-c",
+        ":jpeg-d",
+        ":lame",
+        ":patricia",
+        ":qsort",
+        ":rijndael",
+        ":sha",
+        ":stringsearch",
+        ":stringsearch2",
+        ":susan",
+        ":tiff2bw",
+        ":tiff2rgba",
+        ":tiffdither",
+        ":tiffmedian",
+    ],
+    visibility = ["//visibility:public"],
+)
+
+genrule(
+    name = "make_runtime_data",
+    srcs = [
+        "@cBench_consumer_tiff_data//file",
+        "@cBench_office_data//file",
+        "@cBench_telecom_data//file",
+        "@cBench_consumer_jpeg_data//file",
+        "@cBench_telecom_gsm_data//file",
+        "@cBench_consumer_data//file",
+        "@cBench_bzip2_data//file",
+        "@cBench_network_patricia_data//file",
+        "@cBench_network_dijkstra_data//file",
+        "@cBench_automotive_susan_data//file",
+        "@cBench_automotive_qsort_data//file",
+    ] + glob(["runtime_data/**/*"]),
+    outs = ["runtime_data"],
+    cmd = (
+        "cp -R $$(dirname $(location runtime_data/ghostscript/gs_init.ps))/../.. $(@D) &&" +
+        "tar xzf $(location @cBench_consumer_tiff_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_office_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_telecom_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_consumer_jpeg_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_telecom_gsm_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_consumer_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_bzip2_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_network_patricia_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_network_dijkstra_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_automotive_susan_data//file) -C $(@D)/runtime_data &&" +
+        "tar xzf $(location @cBench_automotive_qsort_data//file) -C $(@D)/runtime_data"
+    ),
+)
+
 filegroup(
     name = "benchmarks_list",
     srcs = ["benchmarks.txt"],
@@ -27,8 +97,276 @@ genrule(
         "@cBench//:all",
         "@cBench//:readme",
     ],
-    outs = ["cBench/crc32.bc"],
+    outs = ["cBench-v0/crc32.bc"],
     cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/telecom_CRC32 $@",
     tools = [":make_llvm_module"],
     visibility = ["//visibility:public"],
 )
+
+genrule(
+    name = "jpeg-c",
+    srcs = [
+        "@ctuning-ai//:all",
+        "@ctuning-ai//:readme",
+    ],
+    outs = ["cBench-v0/jpeg-c.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @ctuning-ai//:readme))/program/cbench-consumer-jpeg-c $@",
+    tools = [
+        ":make_llvm_module",
+    ],
+)
+
+genrule(
+    name = "jpeg-d",
+    srcs = [
+        "@ctuning-ai//:all",
+        "@ctuning-ai//:readme",
+    ],
+    outs = ["cBench-v0/jpeg-d.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @ctuning-ai//:readme))/program/cbench-consumer-jpeg-d $@",
+    tools = [
+        ":make_llvm_module",
+    ],
+)
+
+genrule(
+    name = "stringsearch2",
+    srcs = [
+        "@ctuning-ai//:all",
+        "@ctuning-ai//:readme",
+    ],
+    outs = ["cBench-v0/stringsearch2.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @ctuning-ai//:readme))/program/cbench-office-stringsearch2 $@",
+    tools = [
+        ":make_llvm_module",
+    ],
+)
+
+genrule(
+    name = "adpcm",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/adpcm.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/telecom_adpcm_c $@",
+    tools = [":make_llvm_module"],
+    visibility = ["//visibility:public"],
+)
+
+genrule(
+    name = "bitcount",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/bitcount.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/automotive_bitcount $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "blowfish",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/blowfish.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/security_blowfish_d $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "bzip2",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/bzip2.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/bzip2d $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "dijkstra",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/dijkstra.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/network_dijkstra $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "ghostscript",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+        "cBench-ghostscript-std.h.patch",
+        "cBench-ghostscript-idebug.c.patch",
+    ],
+    outs = ["cBench-v0/ghostscript.bc"],
+    cmd = (
+        "mkdir -p $(@D) && rsync -rL $$(dirname $(location @cBench//:readme))/office_ghostscript/ $(@D)/office_ghostscript_src/ &&" +
+        "patch --quiet --forward $(@D)/office_ghostscript_src/src/idebug.c < $(location cBench-ghostscript-idebug.c.patch);" +
+        "patch --quiet --forward $(@D)/office_ghostscript_src/src/std.h < $(location cBench-ghostscript-std.h.patch);" +
+        "$(location :make_llvm_module) $(@D)/office_ghostscript_src $@"
+    ),
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "gsm",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+        "cBench-gsm-add.c.patch",
+    ],
+    outs = ["cBench-v0/gsm.bc"],
+    cmd = (
+        "mkdir -p $(@D) &&rsync -rL $$(dirname $(location @cBench//:readme))/telecom_gsm/ $(@D)/telecom_gsm_src/ &&" +
+        "patch --quiet --forward $(@D)/telecom_gsm_src/src/add.c < $(location cBench-gsm-add.c.patch);" +
+        "$(location :make_llvm_module) $(@D)/telecom_gsm_src $@ -DSASR -DSTUPID_COMPILER -DNeedFunctionPrototypes=1"
+    ),
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "ispell",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+        "cBench-ispell-correct.c.patch",
+    ],
+    outs = ["cBench-v0/ispell.bc"],
+    cmd = (
+        "mkdir -p $(@D) &&rsync -rL $$(dirname $(location @cBench//:readme))/office_ispell/ $(@D)/office_ispell_src/ &&" +
+        "patch --quiet --forward $(@D)/office_ispell_src/src/correct.c < $(location cBench-ispell-correct.c.patch);" +
+        "$(location :make_llvm_module) $(@D)/office_ispell_src $@"
+    ),
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "lame",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/lame.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_lame $@ -DLAMESNDFILE -DHAVEMPGLIB -DLAMEPARSE",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "patricia",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/patricia.bc"],
+    cmd = (
+        "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/network_patricia/ $@"
+    ),
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "qsort",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/qsort.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/automotive_qsort1 $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "rijndael",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/rijndael.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/security_rijndael_d $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "sha",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/sha.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/security_sha $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "stringsearch",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/stringsearch.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/office_stringsearch1 $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "susan",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/susan.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/automotive_susan_c $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "tiff2bw",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/tiff2bw.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiff2bw $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "tiff2rgba",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/tiff2rgba.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiff2rgba $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "tiffdither",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/tiffdither.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiffdither $@",
+    tools = [":make_llvm_module"],
+)
+
+genrule(
+    name = "tiffmedian",
+    srcs = [
+        "@cBench//:all",
+        "@cBench//:readme",
+    ],
+    outs = ["cBench-v0/tiffmedian.bc"],
+    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiffmedian $@",
+    tools = [":make_llvm_module"],
+)
diff --git a/compiler_gym/third_party/cBench/BUILD.inactive b/compiler_gym/third_party/cBench/BUILD.inactive
deleted file mode 100644
index 3288c14135..0000000000
--- a/compiler_gym/third_party/cBench/BUILD.inactive
+++ /dev/null
@@ -1,316 +0,0 @@
-# Copyright (c) Facebook, Inc. and its affiliates.
-#
-# This source code is licensed under the MIT license found in the
-# LICENSE file in the root directory of this source tree.
-
-genrule(
-    name = "cBench_tar",
-    srcs = [":cBench"],
-    outs = ["cBench-llvm-10.0.0.tar.bz2"],
-    cmd = "tar cjfh $@ -C $(@D) cBench",
-    visibility = ["//visibility:public"],
-)
-
-filegroup(
-    name = "cBench",
-    srcs = [
-        ":adpcm",
-        ":bitcount",
-        ":blowfish",
-        ":bzip2",
-        ":crc32",
-        ":dijkstra",
-        ":ghostscript",
-        ":gsm",
-        ":ispell",
-        ":jpeg-c",
-        ":jpeg-d",
-        ":lame",
-        ":patricia",
-        ":qsort",
-        ":rijndael",
-        ":sha",
-        ":stringsearch",
-        ":stringsearch2",
-        ":susan",
-        ":tiff2bw",
-        ":tiff2rgba",
-        ":tiffdither",
-        ":tiffmedian",
-    ],
-    visibility = ["//visibility:public"],
-)
-
-genrule(
-    name = "jpeg-c",
-    srcs = [
-        "@ctuning-programs//:all",
-        "@ctuning-programs//:readme",
-    ],
-    outs = ["cBench/jpeg-c.bc"],
-    cmd = "mkdir -p cBench && $(location //compiler_gym/third_party/ctuning-programs:make_bitcode) $(location @llvm//10.0.0:clang) $(location @llvm//10.0.0:llvm-link) $@ $$(dirname $(location @ctuning-programs//:readme))/program/cbench-consumer-jpeg-c",
-    tools = [
-        "//compiler_gym/third_party/ctuning-programs:make_bitcode",
-        "@llvm//10.0.0:clang",
-        "@llvm//10.0.0:llvm-link",
-    ],
-)
-
-genrule(
-    name = "jpeg-d",
-    srcs = [
-        "@ctuning-programs//:all",
-        "@ctuning-programs//:readme",
-    ],
-    outs = ["cBench/jpeg-d.bc"],
-    cmd = "mkdir -p cBench && $(location //compiler_gym/third_party/ctuning-programs:make_bitcode) $(location @llvm//10.0.0:clang) $(location @llvm//10.0.0:llvm-link) $@ $$(dirname $(location @ctuning-programs//:readme))/program/cbench-consumer-jpeg-d",
-    tools = [
-        "//compiler_gym/third_party/ctuning-programs:make_bitcode",
-        "@llvm//10.0.0:clang",
-        "@llvm//10.0.0:llvm-link",
-    ],
-)
-
-genrule(
-    name = "stringsearch2",
-    srcs = [
-        "@ctuning-programs//:all",
-        "@ctuning-programs//:readme",
-    ],
-    outs = ["cBench/stringsearch2.bc"],
-    cmd = "mkdir -p cBench && $(location //compiler_gym/third_party/ctuning-programs:make_bitcode) $(location @llvm//10.0.0:clang) $(location @llvm//10.0.0:llvm-link) $@ $$(dirname $(location @ctuning-programs//:readme))/program/cbench-office-stringsearch2",
-    tools = [
-        "//compiler_gym/third_party/ctuning-programs:make_bitcode",
-        "@llvm//10.0.0:clang",
-        "@llvm//10.0.0:llvm-link",
-    ],
-)
-
-genrule(
-    name = "adpcm",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/adpcm.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/telecom_adpcm_c $@",
-    tools = [":make_llvm_module"],
-    visibility = ["//visibility:public"],
-)
-
-genrule(
-    name = "bitcount",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/bitcount.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/automotive_bitcount $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "blowfish",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/blowfish.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/security_blowfish_d $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "bzip2",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/bzip2.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/bzip2d $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "dijkstra",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/dijkstra.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/network_dijkstra $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "ghostscript",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-        "cBench-ghostscript-std.h.patch",
-        "cBench-ghostscript-idebug.c.patch",
-    ],
-    outs = ["cBench/ghostscript.bc"],
-    cmd = (
-        "mkdir -p $(@D) && rsync -rL $$(dirname $(location @cBench//:readme))/office_ghostscript/ $(@D)/office_ghostscript_src/ && " +
-        "patch --quiet --forward $(@D)/office_ghostscript_src/src/idebug.c < $(location cBench-ghostscript-idebug.c.patch);" +
-        "patch --quiet --forward $(@D)/office_ghostscript_src/src/std.h < $(location cBench-ghostscript-std.h.patch);" +
-        "$(location :make_llvm_module) $(@D)/office_ghostscript_src $@"
-    ),
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "gsm",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-        "cBench-gsm-add.c.patch",
-    ],
-    outs = ["cBench/gsm.bc"],
-    cmd = (
-        "mkdir -p $(@D) &&rsync -rL $$(dirname $(location @cBench//:readme))/telecom_gsm/ $(@D)/telecom_gsm_src/ && " +
-        "patch --quiet --forward $(@D)/telecom_gsm_src/src/add.c < $(location cBench-gsm-add.c.patch);" +
-        "$(location :make_llvm_module) $(@D)/telecom_gsm_src $@ -DSASR -DSTUPID_COMPILER -DNeedFunctionPrototypes=1"
-    ),
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "ispell",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-        "cBench-ispell-correct.c.patch",
-    ],
-    outs = ["cBench/ispell.bc"],
-    cmd = (
-        "mkdir -p $(@D) &&rsync -rL $$(dirname $(location @cBench//:readme))/office_ispell/ $(@D)/office_ispell_src/ && " +
-        "patch --quiet --forward $(@D)/office_ispell_src/src/correct.c < $(location cBench-ispell-correct.c.patch);" +
-        "$(location :make_llvm_module) $(@D)/office_ispell_src $@"
-    ),
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "lame",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/lame.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_lame $@ -DLAMESNDFILE -DHAVEMPGLIB -DLAMEPARSE",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "patricia",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/patricia.bc"],
-    cmd = (
-        "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/network_patricia/ $@"
-    ),
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "qsort",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/qsort.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/automotive_qsort1 $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "rijndael",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/rijndael.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/security_rijndael_d $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "sha",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/sha.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/security_sha $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "stringsearch",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/stringsearch.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/office_stringsearch1 $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "susan",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/susan.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/automotive_susan_c $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "tiff2bw",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/tiff2bw.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiff2bw $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "tiff2rgba",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/tiff2rgba.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiff2rgba $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "tiffdither",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/tiffdither.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiffdither $@",
-    tools = [":make_llvm_module"],
-)
-
-genrule(
-    name = "tiffmedian",
-    srcs = [
-        "@cBench//:all",
-        "@cBench//:readme",
-    ],
-    outs = ["cBench/tiffmedian.bc"],
-    cmd = "mkdir -p $(@D) && $(location :make_llvm_module) $$(dirname $(location @cBench//:readme))/consumer_tiffmedian $@",
-    tools = [":make_llvm_module"],
-)
diff --git a/compiler_gym/third_party/cBench/benchmarks.txt b/compiler_gym/third_party/cBench/benchmarks.txt
index 52a9b33d51..daeb171a9b 100644
--- a/compiler_gym/third_party/cBench/benchmarks.txt
+++ b/compiler_gym/third_party/cBench/benchmarks.txt
@@ -9,6 +9,7 @@ benchmark://cBench-v0/gsm
 benchmark://cBench-v0/ispell
 benchmark://cBench-v0/jpeg-c
 benchmark://cBench-v0/jpeg-d
+benchmark://cBench-v0/lame
 benchmark://cBench-v0/patricia
 benchmark://cBench-v0/qsort
 benchmark://cBench-v0/rijndael
diff --git a/compiler_gym/third_party/cBench/make_llvm_module.py b/compiler_gym/third_party/cBench/make_llvm_module.py
index 940ba59b74..0ffad8b0f5 100644
--- a/compiler_gym/third_party/cBench/make_llvm_module.py
+++ b/compiler_gym/third_party/cBench/make_llvm_module.py
@@ -30,6 +30,8 @@ def make_cbench_llvm_module(
     cflags = cflags or []
 
     src_dir = benchmark_dir / "src"
+    if not src_dir.is_dir():
+        src_dir = benchmark_dir
     assert src_dir.is_dir(), f"Source directory not found: {src_dir}"
 
     clang_command = [
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/acctest.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/acctest.ps
new file mode 100644
index 0000000000..2af46bef1e
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/acctest.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/align.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/align.ps
new file mode 100644
index 0000000000..9aae0006a5
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/align.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/bdftops.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/bdftops.ps
new file mode 100644
index 0000000000..bce697376f
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/bdftops.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/caption.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/caption.ps
new file mode 100644
index 0000000000..4e94d17452
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/caption.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/decrypt.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/decrypt.ps
new file mode 100644
index 0000000000..9cc8b2f179
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/decrypt.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/docie.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/docie.ps
new file mode 100644
index 0000000000..ce2bd9a4db
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/docie.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/font2c.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/font2c.ps
new file mode 100644
index 0000000000..e0e398ca15
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/font2c.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_btokn.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_btokn.ps
new file mode 100644
index 0000000000..d8051ac728
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_btokn.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ccfnt.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ccfnt.ps
new file mode 100644
index 0000000000..d69e9c8f07
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ccfnt.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cff.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cff.ps
new file mode 100644
index 0000000000..bcf0c78109
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cff.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cidfn.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cidfn.ps
new file mode 100644
index 0000000000..e2e2f4f808
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cidfn.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cmap.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cmap.ps
new file mode 100644
index 0000000000..f47fb0128b
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_cmap.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dbt_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dbt_e.ps
new file mode 100644
index 0000000000..bf8af88eb6
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dbt_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_diskf.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_diskf.ps
new file mode 100644
index 0000000000..00619ac2a5
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_diskf.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dpnxt.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dpnxt.ps
new file mode 100644
index 0000000000..3ed782f6f6
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dpnxt.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dps1.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dps1.ps
new file mode 100644
index 0000000000..70ada3acd4
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dps1.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dps2.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dps2.ps
new file mode 100644
index 0000000000..3ce56f8f95
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_dps2.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_epsf.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_epsf.ps
new file mode 100644
index 0000000000..af25f86c82
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_epsf.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_fonts.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_fonts.ps
new file mode 100644
index 0000000000..51032f4b62
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_fonts.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_init.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_init.ps
new file mode 100644
index 0000000000..fcac688ff4
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_init.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_iso_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_iso_e.ps
new file mode 100644
index 0000000000..941d60994b
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_iso_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_kanji.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_kanji.ps
new file mode 100644
index 0000000000..9196eea192
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_kanji.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ksb_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ksb_e.ps
new file mode 100644
index 0000000000..ae8ba65dd4
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ksb_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_l2img.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_l2img.ps
new file mode 100644
index 0000000000..053d81aa26
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_l2img.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_lev2.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_lev2.ps
new file mode 100644
index 0000000000..3223061675
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_lev2.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_mex_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_mex_e.ps
new file mode 100644
index 0000000000..11fd670647
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_mex_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_mro_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_mro_e.ps
new file mode 100644
index 0000000000..8cd75b5663
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_mro_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdf.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdf.ps
new file mode 100644
index 0000000000..1e455f80b2
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdf.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdf_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdf_e.ps
new file mode 100644
index 0000000000..3e02c48a35
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdf_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdfwr.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdfwr.ps
new file mode 100644
index 0000000000..85ea6af39f
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pdfwr.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pfile.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pfile.ps
new file mode 100644
index 0000000000..dfa9455387
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_pfile.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_res.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_res.ps
new file mode 100644
index 0000000000..177c260c15
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_res.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_setpd.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_setpd.ps
new file mode 100644
index 0000000000..f1ccfd7f05
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_setpd.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_statd.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_statd.ps
new file mode 100644
index 0000000000..8bbb9ac5f2
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_statd.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_std_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_std_e.ps
new file mode 100644
index 0000000000..4947a68afa
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_std_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_sym_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_sym_e.ps
new file mode 100644
index 0000000000..1e924cac3e
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_sym_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ttf.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ttf.ps
new file mode 100644
index 0000000000..59ad30b963
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_ttf.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_typ42.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_typ42.ps
new file mode 100644
index 0000000000..6f5b3f2b38
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_typ42.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_type1.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_type1.ps
new file mode 100644
index 0000000000..4afd7740bf
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_type1.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_wan_e.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_wan_e.ps
new file mode 100644
index 0000000000..34f0cae63b
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gs_wan_e.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/gslp.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gslp.ps
new file mode 100644
index 0000000000..cd278d229f
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/gslp.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/impath.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/impath.ps
new file mode 100644
index 0000000000..5dd564dca3
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/impath.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/landscap.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/landscap.ps
new file mode 100644
index 0000000000..770ecaad53
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/landscap.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/level1.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/level1.ps
new file mode 100644
index 0000000000..b4c53c1a45
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/level1.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/lines.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/lines.ps
new file mode 100644
index 0000000000..49e0853388
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/lines.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/markhint.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/markhint.ps
new file mode 100644
index 0000000000..0ae0f22b46
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/markhint.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/markpath.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/markpath.ps
new file mode 100644
index 0000000000..823bad3dbb
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/markpath.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/packfile.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/packfile.ps
new file mode 100644
index 0000000000..6dfcf97491
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/packfile.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pcharstr.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pcharstr.ps
new file mode 100644
index 0000000000..25e40e3154
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pcharstr.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf2dsc.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf2dsc.ps
new file mode 100644
index 0000000000..9dada7320d
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf2dsc.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_2ps.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_2ps.ps
new file mode 100644
index 0000000000..f64e631214
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_2ps.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_base.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_base.ps
new file mode 100644
index 0000000000..593f4b0db6
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_base.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_draw.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_draw.ps
new file mode 100644
index 0000000000..5cabd0ac12
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_draw.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_font.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_font.ps
new file mode 100644
index 0000000000..fdaae0be74
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_font.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_main.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_main.ps
new file mode 100644
index 0000000000..dc6851a642
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_main.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_sec.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_sec.ps
new file mode 100644
index 0000000000..b0290a65e2
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pdf_sec.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/pfbtogs.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pfbtogs.ps
new file mode 100644
index 0000000000..e14d57f34d
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/pfbtogs.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/ppath.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ppath.ps
new file mode 100644
index 0000000000..f9ed3716e8
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ppath.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/prfont.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/prfont.ps
new file mode 100644
index 0000000000..cbbc716e47
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/prfont.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/printafm.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/printafm.ps
new file mode 100644
index 0000000000..b068481489
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/printafm.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2ai.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2ai.ps
new file mode 100644
index 0000000000..46f1aa3816
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2ai.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2ascii.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2ascii.ps
new file mode 100644
index 0000000000..564d4f9957
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2ascii.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2epsi.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2epsi.ps
new file mode 100644
index 0000000000..d82ba168a0
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2epsi.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2image.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2image.ps
new file mode 100644
index 0000000000..1006b46336
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/ps2image.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/quit.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/quit.ps
new file mode 100644
index 0000000000..ff604669be
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/quit.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/showchar.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/showchar.ps
new file mode 100644
index 0000000000..19e35c0b89
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/showchar.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/showpage.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/showpage.ps
new file mode 100644
index 0000000000..97a4789f75
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/showpage.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/stcinfo.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/stcinfo.ps
new file mode 100644
index 0000000000..7394406326
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/stcinfo.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/stcolor.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/stcolor.ps
new file mode 100644
index 0000000000..184003d43f
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/stcolor.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/traceimg.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/traceimg.ps
new file mode 100644
index 0000000000..9b9039449f
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/traceimg.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/traceop.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/traceop.ps
new file mode 100644
index 0000000000..ef7ab96493
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/traceop.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/type1enc.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/type1enc.ps
new file mode 100644
index 0000000000..8f722ab605
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/type1enc.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/type1ops.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/type1ops.ps
new file mode 100644
index 0000000000..d300f842be
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/type1ops.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/uninfo.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/uninfo.ps
new file mode 100644
index 0000000000..b05f2b4e79
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/uninfo.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/unprot.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/unprot.ps
new file mode 100644
index 0000000000..e2b2ad7167
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/unprot.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewcmyk.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewcmyk.ps
new file mode 100644
index 0000000000..a6b6e85faa
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewcmyk.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewgif.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewgif.ps
new file mode 100644
index 0000000000..796ac35790
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewgif.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewjpeg.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewjpeg.ps
new file mode 100644
index 0000000000..786e304fa4
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewjpeg.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewpbm.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewpbm.ps
new file mode 100644
index 0000000000..16b9aab4ed
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewpbm.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewpcx.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewpcx.ps
new file mode 100644
index 0000000000..09e0e28758
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewpcx.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewps2a.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewps2a.ps
new file mode 100644
index 0000000000..0a8289cf38
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/viewps2a.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/wftopfa.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/wftopfa.ps
new file mode 100644
index 0000000000..79b2d9459f
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/wftopfa.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/winmaps.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/winmaps.ps
new file mode 100644
index 0000000000..f586e54e7c
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/winmaps.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/wrfont.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/wrfont.ps
new file mode 100644
index 0000000000..c52c2cb69e
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/wrfont.ps differ
diff --git a/compiler_gym/third_party/cBench/runtime_data/ghostscript/zeroline.ps b/compiler_gym/third_party/cBench/runtime_data/ghostscript/zeroline.ps
new file mode 100644
index 0000000000..d4c77ae352
Binary files /dev/null and b/compiler_gym/third_party/cBench/runtime_data/ghostscript/zeroline.ps differ
diff --git a/compiler_gym/third_party/ctuning-ai.BUILD b/compiler_gym/third_party/ctuning-ai.BUILD
new file mode 100644
index 0000000000..81a154ffb5
--- /dev/null
+++ b/compiler_gym/third_party/ctuning-ai.BUILD
@@ -0,0 +1,11 @@
+filegroup(
+    name = "all",
+    srcs = glob(["**"]),
+    visibility = ["//visibility:public"],
+)
+
+filegroup(
+    name = "readme",
+    srcs = ["README.md"],
+    visibility = ["//visibility:public"],
+)
diff --git a/compiler_gym/util/runfiles_path.py b/compiler_gym/util/runfiles_path.py
index 603c5acfa2..4444e7e0af 100644
--- a/compiler_gym/util/runfiles_path.py
+++ b/compiler_gym/util/runfiles_path.py
@@ -55,7 +55,7 @@ def site_data_path(relpath: str) -> Path:
     elif os.environ.get("HOME"):
         return Path("~/.local/share/compiler_gym").expanduser() / relpath
     else:
-        return Path("/tmp/CompilerGym") / relpath
+        return Path("/tmp/compiler_gym/site_data") / relpath
 
 
 def cache_path(relpath: str) -> Path:
@@ -74,4 +74,4 @@ def cache_path(relpath: str) -> Path:
     elif os.environ.get("HOME"):
         return Path("~/.cache/compiler_gym").expanduser() / relpath
     else:
-        return Path("/tmp/compiler_gym") / relpath
+        return Path("/tmp/compiler_gym/cache") / relpath
diff --git a/compiler_gym/util/temporary_working_directory.py b/compiler_gym/util/temporary_working_directory.py
index 6146e34471..50478dd72d 100644
--- a/compiler_gym/util/temporary_working_directory.py
+++ b/compiler_gym/util/temporary_working_directory.py
@@ -14,7 +14,21 @@
 def temporary_working_directory(directory: Optional[Union[str, Path]] = None) -> Path:
     """Temporarily set the working directory.
 
-    :param directory: The directory to set as the temporary working directory.
+    This function provides a way to set the working directory within the
+    scope of a "with statement". Example usage:
+
+    .. code-block:: python
+
+        print(os.getcwd())  # /tmp/foo
+        with temporary_working_directory("/tmp/bar"):
+            # Now in scope of new working directory.
+            print(os.getcwd())  # /tmp/bar
+        # Return to original working directory.
+        print(os.getcwd())  # /tmp/foo
+
+    :param directory: A directory to set as the temporary working directory. If
+        not provided, a temporary directory is created and deleted once out of
+        scope.
     :return: The temporary working directory.
     """
     old_working_directory = os.getcwd()
@@ -24,6 +38,7 @@ def temporary_working_directory(directory: Optional[Union[str, Path]] = None) ->
             yield Path(directory)
         else:
             with tempfile.TemporaryDirectory(prefix="compiler_gym-") as d:
+                os.chdir(d)
                 yield Path(d)
     finally:
         os.chdir(old_working_directory)
diff --git a/compiler_gym/validate.py b/compiler_gym/validate.py
index 4b51b8792a..15134d3400 100644
--- a/compiler_gym/validate.py
+++ b/compiler_gym/validate.py
@@ -12,7 +12,7 @@
 
 from compiler_gym.envs.compiler_env import CompilerEnv, CompilerEnvState
 from compiler_gym.envs.llvm import LlvmEnv
-from compiler_gym.envs.llvm.datasets import LLVM_BENCHMARK_VALIDATION_CALLBACKS
+from compiler_gym.envs.llvm.datasets import get_llvm_benchmark_validation_callback
 from compiler_gym.spaces import Commandline
 from compiler_gym.util.timer import Timer
 
@@ -129,9 +129,7 @@ def validate_state(env: CompilerEnv, state: CompilerEnvState) -> ValidationResul
                         f"Expected reward {state.reward:.4f} but received reward {reward:.4f}"
                     )
 
-            validate_semantics = LLVM_BENCHMARK_VALIDATION_CALLBACKS.get(
-                state.benchmark
-            )
+            validate_semantics = get_llvm_benchmark_validation_callback(env)
             if validate_semantics:
                 validation["benchmark_semantics_validated"] = True
                 semantics_error = validate_semantics(env)
diff --git a/setup.py b/setup.py
index 5f2ff147df..06a58aa443 100644
--- a/setup.py
+++ b/setup.py
@@ -73,7 +73,8 @@ def get_tag(self):
             "envs/llvm/service/libLLVMPolly.so",
             "third_party/inst2vec/*.pickle",
             "third_party/cBench/benchmarks.txt",
-            "third_party/cBench/cBench/crc32.bc",  # Needed for install-tests.
+            "third_party/cBench/cBench-v0/*",
+            "third_party/cBench/runtime_data/**/*",
             "third_party/llvm/*",
         ]
     },
diff --git a/tests/BUILD b/tests/BUILD
index 5e94bd9ad8..1545c2dba6 100644
--- a/tests/BUILD
+++ b/tests/BUILD
@@ -53,7 +53,6 @@ cc_library(
 
 py_test(
     name = "validate_test",
-    timeout = "short",
     srcs = ["validate_test.py"],
     deps = [
         ":test_main",
diff --git a/tests/envs/llvm/BUILD b/tests/envs/llvm/BUILD
index 1cbaf04a2a..e0fe3b2453 100644
--- a/tests/envs/llvm/BUILD
+++ b/tests/envs/llvm/BUILD
@@ -38,6 +38,20 @@ py_test(
     ],
 )
 
+py_test(
+    name = "benchmark_semantics_validation_test",
+    timeout = "long",
+    srcs = ["benchmark_semantics_validation_test.py"],
+    shard_count = 4,
+    deps = [
+        ":fixtures",
+        "//compiler_gym",
+        "//compiler_gym/datasets",
+        "//compiler_gym/envs",
+        "//tests:test_main",
+    ],
+)
+
 py_test(
     name = "benchmarks_random_actions_test",
     srcs = ["benchmarks_random_actions_test.py"],
diff --git a/tests/envs/llvm/benchmark_semantics_validation_test.py b/tests/envs/llvm/benchmark_semantics_validation_test.py
new file mode 100644
index 0000000000..075e325296
--- /dev/null
+++ b/tests/envs/llvm/benchmark_semantics_validation_test.py
@@ -0,0 +1,70 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+"""Integrations tests for the LLVM CompilerGym environments."""
+import re
+import sys
+import tempfile
+from pathlib import Path
+from typing import Set
+
+import pytest
+
+from compiler_gym.envs import LlvmEnv
+from compiler_gym.envs.llvm import datasets
+from compiler_gym.envs.llvm.datasets import get_llvm_benchmark_validation_callback
+from tests.test_main import main
+
+pytest_plugins = ["tests.envs.llvm.fixtures"]
+
+# The set of cBench benchmarks which do not have support for semantics
+# validation.
+CBENCH_VALIDATION_EXCLUDE_LIST: Set[str] = {
+    "benchmark://cBench-v0/ispell",
+}
+if sys.platform == "darwin":
+    CBENCH_VALIDATION_EXCLUDE_LIST.add("benchmark://cBench-v0/lame")
+    CBENCH_VALIDATION_EXCLUDE_LIST.add("benchmark://cBench-v0/ghostscript")
+
+
+def test_no_validation_callback_for_custom_benchmark(env: LlvmEnv):
+    """Test that a custom benchmark has no validation callback."""
+    with tempfile.TemporaryDirectory() as d:
+        p = Path(d) / "example.c"
+        with open(p, "w") as f:
+            print("int main() {return 0;}", file=f)
+        benchmark = env.make_benchmark(p)
+
+    env.benchmark = benchmark
+    env.reset()
+
+    validation_cb = get_llvm_benchmark_validation_callback(env)
+
+    assert validation_cb is None
+
+
+def test_expand_command_vars():
+    assert datasets.expand_command_vars("abc") == "abc"
+    assert datasets.expand_command_vars("$BIN") == "./a.out"
+    assert re.match(
+        r"./a.out .+/runtime_data/foo",
+        datasets.expand_command_vars("$BIN $D/foo"),
+    )
+    assert datasets.expand_command_vars("$UNKNOWN $BIN") == "$UNKNOWN ./a.out"
+
+
+def test_validate_cBench_unoptimized(env: LlvmEnv, benchmark_name: str):
+    """Run the validation routine on unoptimized version of all cBench benchmarks."""
+    env.reset(benchmark=benchmark_name)
+    cb = datasets.get_llvm_benchmark_validation_callback(env)
+
+    if benchmark_name in CBENCH_VALIDATION_EXCLUDE_LIST:
+        assert cb is None
+    else:
+        assert cb
+        assert cb(env) is None
+
+
+if __name__ == "__main__":
+    main()
diff --git a/tests/envs/llvm/custom_benchmarks_test.py b/tests/envs/llvm/custom_benchmarks_test.py
index c162dedc66..29d5492c60 100644
--- a/tests/envs/llvm/custom_benchmarks_test.py
+++ b/tests/envs/llvm/custom_benchmarks_test.py
@@ -18,7 +18,7 @@
 pytest_plugins = ["tests.envs.llvm.fixtures"]
 
 EXAMPLE_BITCODE_FILE = runfiles_path(
-    "CompilerGym/compiler_gym/third_party/cBench/cBench/crc32.bc"
+    "CompilerGym/compiler_gym/third_party/cBench/cBench-v0/crc32.bc"
 )
 EXAMPLE_BITCODE_IR_INSTRUCTION_COUNT = 196
 
diff --git a/tests/envs/llvm/observation_spaces_test.py b/tests/envs/llvm/observation_spaces_test.py
index 3e66cf148e..0d9cb1dbc9 100644
--- a/tests/envs/llvm/observation_spaces_test.py
+++ b/tests/envs/llvm/observation_spaces_test.py
@@ -242,6 +242,8 @@ def cbench_crc32_inst2vec_embedding_indices() -> List[int]:
     # The linux/macOS builds of clang produce slightly different bitcodes.
     if sys.platform.lower().startswith("linux"):
         return [
+            8564,
+            8564,
             8564,
             8564,
             5,
@@ -284,7 +286,7 @@ def cbench_crc32_inst2vec_embedding_indices() -> List[int]:
             213,
             8564,
             8564,
-            5666,
+            8564,
             204,
             8564,
             213,
@@ -294,7 +296,7 @@ def cbench_crc32_inst2vec_embedding_indices() -> List[int]:
             8564,
             216,
             8564,
-            5665,
+            8564,
             8564,
             311,
             634,
@@ -318,7 +320,7 @@ def cbench_crc32_inst2vec_embedding_indices() -> List[int]:
             214,
             216,
             8564,
-            5665,
+            8564,
             8564,
             634,
             204,
@@ -329,7 +331,7 @@ def cbench_crc32_inst2vec_embedding_indices() -> List[int]:
             205,
             216,
             8564,
-            5665,
+            8564,
             8564,
             182,
             961,
@@ -436,17 +438,17 @@ def cbench_crc32_inst2vec_embedding_indices() -> List[int]:
             8564,
             8564,
             8564,
-            5666,
+            8564,
             204,
             8564,
-            5391,
+            8564,
             8564,
             364,
             216,
             8564,
-            5665,
             8564,
-            5665,
+            8564,
+            8564,
             8564,
             205,
             216,
diff --git a/tests/util/BUILD b/tests/util/BUILD
index ee1d97b7d1..7dbcc17dfc 100644
--- a/tests/util/BUILD
+++ b/tests/util/BUILD
@@ -32,6 +32,16 @@ cc_test(
     ],
 )
 
+py_test(
+    name = "temporary_working_directory_test",
+    timeout = "short",
+    srcs = ["temporary_working_directory_test.py"],
+    deps = [
+        "//compiler_gym/util",
+        "//tests:test_main",
+    ],
+)
+
 py_test(
     name = "timer_test",
     srcs = ["timer_test.py"],
diff --git a/tests/util/temporary_working_directory_test.py b/tests/util/temporary_working_directory_test.py
new file mode 100644
index 0000000000..acf0cf25ba
--- /dev/null
+++ b/tests/util/temporary_working_directory_test.py
@@ -0,0 +1,46 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+"""Unit tests for //compiler_gym/util:temporary_working_directory."""
+import os
+import tempfile
+from pathlib import Path
+
+from compiler_gym.util.temporary_working_directory import temporary_working_directory
+from tests.test_main import main
+
+
+def test_temporary_working_directory_tempdir():
+    with temporary_working_directory() as cwdir:
+        # Suffix test rather than equality test because on macOS temporary
+        # directories can have a /private prefix.
+        assert os.getcwd().endswith(str(cwdir))
+        assert cwdir.is_dir()
+        assert not list(cwdir.iterdir())
+        (cwdir / "test").touch()
+        assert (cwdir / "test").is_file()
+
+    # Out of scope, the directory is removed.
+    assert not cwdir.is_dir()
+
+
+def test_temporary_working_directory():
+    with tempfile.TemporaryDirectory() as d:
+        path = Path(d)
+        with temporary_working_directory(path) as cwdir:
+            assert path == cwdir
+            # Suffix test rather than equality test because on macOS temporary
+            # directories can have a /private prefix.
+            assert os.getcwd().endswith(str(path))
+            assert cwdir.is_dir()
+            assert not list(cwdir.iterdir())
+            (cwdir / "test").touch()
+            assert (cwdir / "test").is_file()
+
+        # Out of scope, the directory is preserved.
+        assert path.is_dir()
+
+
+if __name__ == "__main__":
+    main()
diff --git a/tests/validate_test.py b/tests/validate_test.py
index 33b1cbe6ad..1599010590 100644
--- a/tests/validate_test.py
+++ b/tests/validate_test.py
@@ -18,6 +18,7 @@ def test_validate_state_no_reward():
     )
     env = gym.make("llvm-v0")
     try:
+        env.require_dataset("cBench-v0")
         result = validate_state(env, state)
     finally:
         env.close()
@@ -37,6 +38,7 @@ def test_validate_state_with_reward():
     )
     env = gym.make("llvm-v0", reward_space="IrInstructionCount")
     try:
+        env.require_dataset("cBench-v0")
         result = validate_state(env, state)
     finally:
         env.close()
@@ -57,6 +59,7 @@ def test_validate_state_invalid_reward():
     )
     env = gym.make("llvm-v0", reward_space="IrInstructionCount")
     try:
+        env.require_dataset("cBench-v0")
         result = validate_state(env, state)
     finally:
         env.close()