The dataset viewer is not available for this split.
Error code: FeaturesError
Exception: ArrowInvalid
Message: Schema at index 1 was different:
workload segment weight: string
vs
workload duckdb_tpch_q1_sf10: string
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 243, in compute_first_rows_from_streaming_response
iterable_dataset = iterable_dataset._resolve_features()
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 4379, in _resolve_features
features = _infer_features_from_batch(self.with_format(None)._head())
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2661, in _head
return next(iter(self.iter(batch_size=n)))
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2839, in iter
for key, pa_table in ex_iterable.iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2377, in _iter_arrow
yield from self.ex_iterable._iter_arrow()
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 580, in _iter_arrow
yield new_key, pa.Table.from_batches(chunks_buffer)
~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
File "pyarrow/table.pxi", line 5039, in pyarrow.lib.Table.from_batches
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
return check_status(status)
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: Schema at index 1 was different:
workload segment weight: string
vs
workload duckdb_tpch_q1_sf10: stringNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
HPCA 2027 validated database traces
Validated DynamoRIO instruction traces for database application-plus-input configurations used in HPCA 2027 backend-bound characterization.
Included workloads
| Directory | Application and input | Captured fetched instructions | Scarab backend bound |
|---|---|---|---|
mongodb |
MongoDB 7.0.37, 10M-record YCSB database, sort/aggregate, 1 GiB WiredTiger cache, 3 GiB container | 100,098,204 global across server threads | 41.76% on the 89,501,578-instruction dominant worker |
postgresql |
PostgreSQL 16.10, TPC-H SF10 Q7 | 120,961,104 | 76.97% |
Both reported backend-bound values were recomputed from Scarab raw top-down
slot counters under the Golden Cove configuration with --dcache_assoc 8.
MongoDB and PostgreSQL exceed the 30% acceptance threshold. MySQL is not
included because its final Scarab result was 23.57%, despite passing native
screening.
Layout
The trace layout follows the established ASPLOS BFS bundle convention:
<workload>/
traces/whole/<drmemtrace-dir>/
raw/
trace/
bin/
validation/
<workload>/
Each bin/ directory contains the exact traced server executable,
libdrmemtrace.so, libdynamorio.so, and modules.log. The validation/
directories add alignment timestamps, basic_counts, ZIP integrity logs,
trace summaries, Scarab commands/configuration, raw counters, and top-down
breakdowns. SHA256SUMS covers every uploaded file except itself.
Important MongoDB counting note
MongoDB's 100,098,204 fetched instructions are the global DynamoRIO server capture across 73 thread ZIPs. The dominant worker selected for one-core Scarab simulation contains 89,501,578 instructions. The repository preserves all 73 raw and converted thread traces and does not misrepresent the global capture count as a per-thread count.
Validation policy
- Every converted trace ZIP passes
unzip -t. - DynamoRIO
basic_countsrecords the captured fetched-instruction count. - Scarab simulations reached the end of the selected trace and emitted raw top-down counters.
- Backend-bound percentages are computed as the residual after frontend bound, bad speculation, and retiring from raw slot counts.
- Repository-wide hashes can be checked with
sha256sum -c SHA256SUMS.
SimPoint bundle formalization
Each workload follows the complete historical SimPoint bundle contract:
fingerprint/
simpoints/
traces/whole/
traces_simp/{trace,raw,bin}/
trace_clustering_info.json
Fingerprints were generated in Scarab trace_bbv_distributed mode with 10M instruction segments. SimPoint 3.2 used maxK = round(sqrt(segments)) and 99% coverage. Each minimized ZIP contains chunk zero, one preceding warmup chunk when available, and the selected target chunk, matching the established ASPLOS BFS convention.
| Workload | Selected segments | SimPoint weights |
|---|---|---|
| MongoDB | 1, 4, 8 | 0.782109, 0.111730, 0.106161 |
| PostgreSQL | 12, 1, 2 | 0.751986, 0.165342, 0.0826712 |
PostgreSQL segment 12 is the genuine final partial interval of the 120.96M instruction trace. Its target contains 961,103 instructions. It is retained and disclosed rather than represented as a full 10M interval. Every minimized SimPoint ZIP was replayed successfully in Scarab.
- Downloads last month
- 61