perf: configurable float storage dtype (default float32, ~½ coeffs memory)#839
Draft
FBumann wants to merge 3 commits into
Draft
perf: configurable float storage dtype (default float32, ~½ coeffs memory)#839FBumann wants to merge 3 commits into
FBumann wants to merge 3 commits into
CodSpeed HQ / CodSpeed Performance Analysis
succeeded
Jul 16, 2026 in 0s
Performance Gate Passed
⚡ 99 improved benchmarks
✅ 74 untouched benchmarks
⏩ 173 skipped benchmarks1
Performance Changes
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ⚡ | Memory | test_op[expr_add_scalar] |
187.5 KB | 93.8 KB | ×2 |
| ⚡ | Memory | test_op[var_div_scalar] |
562.5 KB | 281.3 KB | +100% |
| ⚡ | Memory | test_op[expr_mul_array_match] |
468.8 KB | 234.4 KB | +100% |
| ⚡ | Memory | test_op[expr_mul_scalar] |
468.8 KB | 234.4 KB | +100% |
| ⚡ | Memory | test_to_lp[expression_arithmetic-n=250] |
2.9 MB | 1.4 MB | +99.3% |
| ⚡ | Memory | test_to_lp[knapsack-n=10000] |
82.2 KB | 43.9 KB | +87.34% |
| ⚡ | Memory | test_to_lp[nodal_balance-severity=0] |
129.1 KB | 71.3 KB | +81.19% |
| ⚡ | Memory | test_op[expr_sum_all] |
212.3 KB | 118.6 KB | +79.05% |
| ⚡ | Memory | test_op[expr_sum_dim] |
212.3 KB | 118.6 KB | +79.05% |
| ⚡ | Memory | test_op[con_eq_expr] |
1,079.7 KB | 610.9 KB | +76.73% |
| ⚡ | Memory | test_op[expr_sub_expr_match] |
1,079.7 KB | 610.9 KB | +76.73% |
| ⚡ | Memory | test_op[var_add_var_match] |
1,079.9 KB | 611.1 KB | +76.71% |
| ⚡ | Memory | test_op[var_sub_var_match] |
1,079.9 KB | 611.1 KB | +76.71% |
| ⚡ | Memory | test_op[expr_add_var] |
986 KB | 564.1 KB | +74.79% |
| ⚡ | Memory | test_op[var_add_var_bcast] |
4.4 MB | 2.6 MB | +73.17% |
| ⚡ | Memory | test_op[expr_add_expr_bcast] |
4.4 MB | 2.5 MB | +72.68% |
| ⚡ | Memory | test_op[var_mul_var] |
891.8 KB | 516.8 KB | +72.57% |
| ⚡ | Memory | test_op[expr_add_expr_match] |
892.1 KB | 517.1 KB | +72.52% |
| ⚡ | Memory | test_op[expr_add_masked] |
892.1 KB | 517.1 KB | +72.52% |
| ⚡ | Memory | test_build[cumsum-severity=0] |
16.4 KB | 9.6 KB | +70.24% |
| ... | ... | ... | ... | ... | ... |
ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.
Tip
Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.
Comparing fluxopt:perf/values-float32 (d23e0d7) with master (c2607b6)
Footnotes
-
173 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩
Loading