Sub segment read p2

.github/workflows/bench-pr.yml

@@ -73,9 +73,11 @@ jobs:
           extra_args: "--off-cpu-threshold=0.03"  # Personally tuned by @brancz
 
       - name: Run ${{ matrix.benchmark.name }} benchmark
+        continue-on-error: true
         shell: bash
         env:
           RUST_BACKTRACE: full
+          FLAT_LAYOUT_INLINE_ARRAY_NODE: true
         run: |
           bash scripts/bench-taskset.sh target/release_debug/${{ matrix.benchmark.id }} -d gh-json -o results.json
 

.github/workflows/bench.yml

@@ -86,9 +86,11 @@ jobs:
           extra_args: "--off-cpu-threshold=0.03"  # Personally tuned by @brancz
 
       - name: Run ${{ matrix.benchmark.name }} benchmark
+        continue-on-error: true
         shell: bash
         env:
           RUST_BACKTRACE: full
+          FLAT_LAYOUT_INLINE_ARRAY_NODE: true
         run: |
           bash scripts/bench-taskset.sh target/release_debug/${{ matrix.benchmark.id }} --formats ${{ matrix.benchmark.formats }} -d gh-json -o results.json
 

.github/workflows/sql-benchmarks.yml

@@ -198,8 +198,10 @@ jobs:
 
       - name: Run ${{ matrix.name }} benchmark
         if: matrix.remote_storage == null || github.event.pull_request.head.repo.fork == true
+        continue-on-error: true
         shell: bash
         env:
+          FLAT_LAYOUT_INLINE_ARRAY_NODE: true
           OTEL_SERVICE_NAME: "vortex-bench"
           OTEL_EXPORTER_OTLP_PROTOCOL: "http/protobuf"
           OTEL_EXPORTER_OTLP_ENDPOINT: "${{ (inputs.mode != 'pr' || github.event.pull_request.head.repo.fork == false) && secrets.OTEL_EXPORTER_OTLP_ENDPOINT || '' }}"
@@ -212,8 +214,10 @@ jobs:
 
       - name: Run ${{ matrix.name }} benchmark (remote)
         if: matrix.remote_storage != null && (inputs.mode != 'pr' || github.event.pull_request.head.repo.fork == false)
+        continue-on-error: true
         shell: bash
         env:
+          FLAT_LAYOUT_INLINE_ARRAY_NODE: true
           AWS_REGION: "us-east-1"
           OTEL_SERVICE_NAME: "vortex-bench"
           OTEL_EXPORTER_OTLP_PROTOCOL: "http/protobuf"

encodings/bytebool/public-api.lock

@@ -62,6 +62,10 @@ impl vortex_array::arrays::dict::take::TakeExecute for vortex_bytebool::ByteBool
 
 pub fn vortex_bytebool::ByteBool::take(array: vortex_array::array::view::ArrayView<'_, Self>, indices: &vortex_array::array::erased::ArrayRef, ctx: &mut vortex_array::executor::ExecutionCtx) -> vortex_error::VortexResult<core::option::Option<vortex_array::array::erased::ArrayRef>>
 
+impl vortex_array::arrays::filter::kernel::FilterReduce for vortex_bytebool::ByteBool
+
+pub fn vortex_bytebool::ByteBool::filter(array: vortex_array::array::view::ArrayView<'_, Self>, mask: &vortex_mask::Mask) -> vortex_error::VortexResult<core::option::Option<vortex_array::array::erased::ArrayRef>>
+
 impl vortex_array::arrays::slice::SliceReduce for vortex_bytebool::ByteBool
 
 pub fn vortex_bytebool::ByteBool::slice(array: vortex_array::array::view::ArrayView<'_, Self>, range: core::ops::range::Range<usize>) -> vortex_error::VortexResult<core::option::Option<vortex_array::array::erased::ArrayRef>>

encodings/bytebool/src/rules.rs

@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright the Vortex contributors
 
+use vortex_array::arrays::filter::FilterReduceAdaptor;
 use vortex_array::arrays::slice::SliceReduceAdaptor;
 use vortex_array::optimizer::rules::ParentRuleSet;
 use vortex_array::scalar_fn::fns::cast::CastReduceAdaptor;
@@ -12,4 +13,5 @@ pub(crate) static RULES: ParentRuleSet<ByteBool> = ParentRuleSet::new(&[
     ParentRuleSet::lift(&CastReduceAdaptor(ByteBool)),
     ParentRuleSet::lift(&MaskReduceAdaptor(ByteBool)),
     ParentRuleSet::lift(&SliceReduceAdaptor(ByteBool)),
+    ParentRuleSet::lift(&FilterReduceAdaptor(ByteBool)),
 ]);

encodings/bytebool/src/slice.rs

@@ -6,8 +6,10 @@ use std::ops::Range;
 use vortex_array::ArrayRef;
 use vortex_array::ArrayView;
 use vortex_array::IntoArray;
+use vortex_array::arrays::filter::FilterReduce;
 use vortex_array::arrays::slice::SliceReduce;
 use vortex_error::VortexResult;
+use vortex_mask::Mask;
 
 use crate::ByteBool;
 
@@ -22,3 +24,18 @@ impl SliceReduce for ByteBool {
         ))
     }
 }
+
+impl FilterReduce for ByteBool {
+    fn filter(array: ArrayView<'_, Self>, mask: &Mask) -> VortexResult<Option<ArrayRef>> {
+        if array.buffer().is_on_host() && mask.true_count() * 2 > mask.len() {
+            return Ok(None);
+        }
+        Ok(Some(
+            ByteBool::new(
+                array.buffer().filter(mask, size_of::<u8>())?,
+                array.validity()?.filter(mask)?,
+            )
+            .into_array(),
+        ))
+    }
+}

encodings/fastlanes/src/bitpacking/array/unpack_iter.rs

@@ -95,7 +95,7 @@ impl<T: BitPacked> BitUnpackedChunks<T> {
     pub fn try_new(array: &BitPackedData, len: usize) -> VortexResult<Self> {
         Self::try_new_with_strategy(
             BitPackingStrategy,
-            array.packed().clone().unwrap_host(),
+            array.packed().clone().try_into_host_sync()?,
             array.bit_width() as usize,
             array.offset() as usize,
             len,

encodings/fsst/src/canonical.rs

@@ -60,28 +60,23 @@ pub(crate) fn fsst_decode_views(
         .execute::<PrimitiveArray>(ctx)?;
 
     #[expect(clippy::cast_possible_truncation)]
-    let total_size: usize = match_each_integer_ptype!(uncompressed_lens_array.ptype(), |P| {
-        uncompressed_lens_array
-            .as_slice::<P>()
-            .iter()
-            .map(|x| *x as usize)
-            .sum()
-    });
-
-    // Bulk-decompress the entire array.
-    let decompressor = fsst_array.decompressor();
-    let mut uncompressed_bytes = ByteBufferMut::with_capacity(total_size + 7);
-    let len =
-        decompressor.decompress_into(bytes.as_slice(), uncompressed_bytes.spare_capacity_mut());
-    unsafe { uncompressed_bytes.set_len(len) };
-
-    // Directly create the binary views.
     match_each_integer_ptype!(uncompressed_lens_array.ptype(), |P| {
+        let uncompressed_lengths = uncompressed_lens_array.as_slice::<P>();
+        let total_size: usize = uncompressed_lengths.iter().map(|x| *x as usize).sum();
+
+        // Bulk-decompress the entire array.
+        let decompressor = fsst_array.decompressor();
+        let mut uncompressed_bytes = ByteBufferMut::with_capacity(total_size + 7);
+        let len =
+            decompressor.decompress_into(bytes.as_slice(), uncompressed_bytes.spare_capacity_mut());
+        unsafe { uncompressed_bytes.set_len(len) };
+
+        // Directly create the binary views.
         Ok(build_views(
             start_buf_index,
             MAX_BUFFER_LEN,
             uncompressed_bytes,
-            uncompressed_lens_array.as_slice::<P>(),
+            uncompressed_lengths,
         ))
     })
 }