Use Vec for primitive array unary operations

[kowanietz]

Which issue does this PR close?

• part of [Epic] Replace OffsetBufferBuilder / BufferBuilder usage with Vec, when possible #10245

Rationale for this change

Replacing BufferBuilder with Vec improves the performenace of primitive array unary operations.

What changes are included in this PR?

Replaces the BufferBuilder<...> usages in PrimitiveArray::try_unary and PrimitiveArray::unary_opt with Vecs.

Also adds benchmarks for both operations with and without input nulls

Are these changes tested?

All tests pass:

• cargo fmt --all -- --check
• cargo clippy -p arrow-array --all-targets --all-features --no-deps -- -D warnings
• cargo test -p arrow-array --all-features
• cargo bench -p arrow-array --bench primitive_array -- --test

Local benchmark results for 65,536 Int32 values:

Benchmark	Before	After	Change
try_unary, no input nulls	10.600 µs	8.865 µs	16.4% faster
try_unary, 20% input nulls	54.486 µs	54.623 µs	no clear change
unary_opt, no input nulls	48.641 µs	49.630 µs	no clear change
unary_opt, 20% input nulls	75.155 µs	75.903 µs	no clear change

I currently only have a MacBook available so benching this on Linux would probably make sense.

Are there any user-facing changes?

No.

[kowanietz]

run benchmark primitive_array

[adriangbot]

Hi @kowanietz, thanks for the request (#10783 (comment)). Only whitelisted users can trigger benchmarks. Allowed users: 2010YOUY01, Dandandan, Fokko, Jefffrey, Omega359, Rachelint, Rich-T-kid, adriangb, alamb, asubiotto, brunal, buraksenn, cetra3, codephage2020, coderfender, comphead, erenavsarogullari, etseidl, friendlymatthew, gabotechs, geoffreyclaude, grtlr, haohuaijin, jonathanc-n, kevinjqliu, klion26, kosiew, kumarUjjawal, kunalsinghdadhwal, liamzwbao, mbutrovich, mkleen, mzabaluev, neilconway, rluvaton, sdf-jkl, timsaucer, xudong963, zhuqi-lucas.

---

File an issue against this benchmark runner

[kowanietz]

Well, that's unfortunate.

Btw, in case the benchmark I added is out of scope, I'll remove it or move it to another PR :)

[Rich-T-kid]

run benchmark primitive_array

[adriangbot]

🤖 Arrow criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5374780982-1852-vcr25 6.12.85+ #1 SMP Sat Jun 27 09:31:30 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing perf/primitive-array-vec (52b4cdd) to c837ee8 (merge-base) diff

Run configuration

run benchmark primitive_array

BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench primitive_array
Results will be posted here when complete

---

File an issue against this benchmark runner

[Rich-T-kid]

Well, that's unfortunate.

Btw, in case the benchmark I added is out of scope, I'll remove it or move it to another PR :)

@kowanietz yea you'd need to add the benchmarks in a separate PR to use the benchmarks bot to compare the results. I dont think thats actually needed for this PR, see @/alamb comment here apache/datafusion#24071 (comment)

LGTM

[adriangbot]

🤖 Arrow criterion benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing perf/primitive-array-vec (52b4cdd) to c837ee8 (merge-base) diff

Run configuration

run benchmark primitive_array

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

New benchmark — branch-only results (no baseline comparison)

Details

group                                                perf_primitive-array-vec
-----                                                ------------------------
primitive_array_unary/try_unary/20pct_input_nulls    1.00     43.0±0.05µs        ? ?/sec
primitive_array_unary/try_unary/no_input_nulls       1.00      5.0±0.01µs        ? ?/sec
primitive_array_unary/unary_opt/20pct_input_nulls    1.00     48.6±0.28µs        ? ?/sec
primitive_array_unary/unary_opt/no_input_nulls       1.00     39.3±0.14µs        ? ?/sec

Resource Usage

branch

Metric	Value
Wall time	45.0s
Peak memory	10.0 MiB
Avg memory	6.1 MiB
CPU user	39.9s
CPU sys	0.0s
Peak spill	0 B

---

File an issue against this benchmark runner

[Jefffrey]

matches my bench results locally too

primitive_array_unary/try_unary/no_input_nulls
                        time:   [4.4402 µs 4.4502 µs 4.4598 µs]
+                       change: [−15.956% −15.699% −15.431%] (p = 0.00 < 0.05)
+                       Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) low severe
  1 (1.00%) low mild
  3 (3.00%) high mild
primitive_array_unary/unary_opt/no_input_nulls
                        time:   [25.532 µs 25.647 µs 25.761 µs]
-                       change: [+1.8110% +2.1490% +2.4997%] (p = 0.00 < 0.05)
-                       Performance has regressed.
Found 15 outliers among 100 measurements (15.00%)
  12 (12.00%) high mild
  3 (3.00%) high severe
primitive_array_unary/try_unary/20pct_input_nulls
                        time:   [27.961 µs 27.990 µs 28.021 µs]
+                       change: [−5.3402% −5.0562% −4.7822%] (p = 0.00 < 0.05)
+                       Performance has improved.
Found 11 outliers among 100 measurements (11.00%)
  3 (3.00%) low severe
  4 (4.00%) low mild
  1 (1.00%) high mild
  3 (3.00%) high severe
primitive_array_unary/unary_opt/20pct_input_nulls
                        time:   [38.286 µs 38.346 µs 38.414 µs]
                        change: [−1.6937% −1.1508% −0.6379%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 15 outliers among 100 measurements (15.00%)
  2 (2.00%) low mild
  13 (13.00%) high severe

• 1st one has improvement, other 3 probably just noise

[Jefffrey]

thanks @kowanietz & @Rich-T-kid