Fix aarch64 timer interrupt configuration - #2585
Open
Gelbpunkt wants to merge 2 commits into
Open
Conversation
Dividing first and then multiplying could hypothetically result in losing accuracy. If, for example, wt is 999, then wt / 1000 is 0 and we end up with a deadline that is 0. If we perform the multiplication first, we preserve precision.
All of our absolute times are boot-relative, so we need to add the BOOT_COUNTER offset to the absolute value before programming CNTP_CVAL_EL0. Otherwise, if BOOT_COUNTER has a nonzero value, our timer interrupts would always be in the past and fire constantly.
There was a problem hiding this comment.
Benchmark Results
Details
| Benchmark | Current: 14a9820 | Previous: 2e23902 | Performance Ratio |
|---|---|---|---|
| startup_benchmark Build Time | 84.23 s |
80.34 s |
1.05 ❗ |
| startup_benchmark File Size | 0.79 MB |
0.80 MB |
1.00 ❗ |
| Startup Time - 1 core | 0.76 s (±0.02 s) |
0.75 s (±0.02 s) |
1.01 |
| Startup Time - 2 cores | 0.76 s (±0.02 s) |
0.74 s (±0.02 s) |
1.03 |
| Startup Time - 4 cores | 0.77 s (±0.04 s) |
0.74 s (±0.02 s) |
1.03 |
| multithreaded_benchmark Build Time | 86.27 s |
82.11 s |
1.05 ❗ |
| multithreaded_benchmark File Size | 0.90 MB |
0.86 MB |
1.05 ❗ |
| Multithreaded Pi Efficiency - 2 Threads | 91.72 % (±5.11 %) |
85.89 % (±6.61 %) |
1.07 |
| Multithreaded Pi Efficiency - 4 Threads | 45.71 % (±2.79 %) |
43.43 % (±2.56 %) |
1.05 |
| Multithreaded Pi Efficiency - 8 Threads | 25.86 % (±1.58 %) |
25.76 % (±1.53 %) |
1.00 |
| micro_benchmarks Build Time | 86.17 s |
80.40 s |
1.07 ❗ |
| micro_benchmarks File Size | 0.90 MB |
0.86 MB |
1.05 ❗ |
| Scheduling time - 1 thread | 65.98 ticks (±3.66 ticks) |
62.65 ticks (±4.06 ticks) |
1.05 |
| Scheduling time - 2 threads | 36.95 ticks (±5.49 ticks) |
34.08 ticks (±4.10 ticks) |
1.08 |
| Micro - Time for syscall (getpid) | 3.85 ticks (±0.66 ticks) |
3.45 ticks (±0.58 ticks) |
1.11 |
| Memcpy speed - (built_in) block size 4096 | 83180.02 MByte/s (±57453.60 MByte/s) |
82448.38 MByte/s (±56997.13 MByte/s) |
1.01 |
| Memcpy speed - (built_in) block size 1048576 | 30335.58 MByte/s (±24510.37 MByte/s) |
30585.98 MByte/s (±24707.84 MByte/s) |
0.99 |
| Memcpy speed - (built_in) block size 16777216 | 26194.12 MByte/s (±21662.98 MByte/s) |
26340.06 MByte/s (±21720.96 MByte/s) |
0.99 |
| Memset speed - (built_in) block size 4096 | 83238.16 MByte/s (±57496.92 MByte/s) |
82292.76 MByte/s (±56891.50 MByte/s) |
1.01 |
| Memset speed - (built_in) block size 1048576 | 31076.82 MByte/s (±24945.88 MByte/s) |
31323.85 MByte/s (±25145.86 MByte/s) |
0.99 |
| Memset speed - (built_in) block size 16777216 | 26959.59 MByte/s (±22154.85 MByte/s) |
27104.68 MByte/s (±22209.94 MByte/s) |
0.99 |
| Memcpy speed - (rust) block size 4096 | 73455.06 MByte/s (±51221.51 MByte/s) |
74097.96 MByte/s (±51811.44 MByte/s) |
0.99 |
| Memcpy speed - (rust) block size 1048576 | 30192.66 MByte/s (±24461.51 MByte/s) |
30361.60 MByte/s (±24602.37 MByte/s) |
0.99 |
| Memcpy speed - (rust) block size 16777216 | 27013.20 MByte/s (±22375.87 MByte/s) |
27625.34 MByte/s (±22806.88 MByte/s) |
0.98 |
| Memset speed - (rust) block size 4096 | 73905.87 MByte/s (±51551.12 MByte/s) |
74373.47 MByte/s (±51976.48 MByte/s) |
0.99 |
| Memset speed - (rust) block size 1048576 | 30918.67 MByte/s (±24887.59 MByte/s) |
31110.89 MByte/s (±25033.24 MByte/s) |
0.99 |
| Memset speed - (rust) block size 16777216 | 27773.31 MByte/s (±22846.33 MByte/s) |
28386.93 MByte/s (±23265.03 MByte/s) |
0.98 |
| alloc_benchmarks Build Time | 78.39 s |
74.76 s |
1.05 ❗ |
| alloc_benchmarks File Size | 0.87 MB |
0.87 MB |
1.00 ❗ |
| Allocations - Allocation success | 91.31 % |
91.31 % |
1 |
| Allocations - Deallocation success | 100.00 % |
100.00 % |
1 |
| Allocations - Pre-fail Allocations | 61.44 % |
61.44 % |
1 |
| Allocations - Average Allocation time | 4236.81 Ticks (±103.92 Ticks) |
5860.58 Ticks (±98.43 Ticks) |
0.72 ❗ |
| Allocations - Average Allocation time (no fail) | 5042.30 Ticks (±104.07 Ticks) |
6554.81 Ticks (±92.86 Ticks) |
0.77 ❗ |
| Allocations - Average Deallocation time | 1174.89 Ticks (±186.24 Ticks) |
1805.01 Ticks (±250.35 Ticks) |
0.65 ❗ |
| mutex_benchmark Build Time | 78.86 s |
79.82 s |
0.99 ❗ |
| mutex_benchmark File Size | 0.90 MB |
0.86 MB |
1.05 ❗ |
| Mutex Stress Test Average Time per Iteration - 1 Threads | 12.10 ns (±0.41 ns) |
12.10 ns (±0.41 ns) |
1.00 |
| Mutex Stress Test Average Time per Iteration - 2 Threads | 40.04 ns (±2.30 ns) |
40.26 ns (±1.68 ns) |
0.99 |
This comment was automatically generated by workflow using github-action-benchmark.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
We currently misconfigure the timer interrupt on aarch64, which results in it constantly firing if we have a
BOOT_COUNTERthat is not zero. We were also losing precision in some hypothetical edge cases. After fixing both, I see around 7 Gbit/s of throughput in the networking benchmarks rather than 5 Mbit/s and no more constant timer interrupts firing.