Skip to content

Conversation

Gelbpunkt
Copy link
Member

@Gelbpunkt Gelbpunkt commented Sep 19, 2025

RISC-V has no PCI support at the moment, so we cannot test it there, but it will at least compile fine now.

@Gelbpunkt Gelbpunkt marked this pull request as draft September 19, 2025 14:40
Copy link
Contributor

@github-actions github-actions bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Benchmark Results

Benchmark Current: 0a19f43 Previous: b33fcf4 Performance Ratio
startup_benchmark Build Time 109.99 s 111.09 s 0.99
startup_benchmark File Size 0.89 MB 0.89 MB 1.00
Startup Time - 1 core 0.90 s (±0.05 s) 0.89 s (±0.04 s) 1.01
Startup Time - 2 cores 0.90 s (±0.03 s) 0.90 s (±0.04 s) 1.00
Startup Time - 4 cores 0.92 s (±0.04 s) 0.91 s (±0.03 s) 1.01
multithreaded_benchmark Build Time 111.00 s 110.24 s 1.01
multithreaded_benchmark File Size 1.00 MB 1.00 MB 1.00
Multithreaded Pi Efficiency - 2 Threads 89.38 % (±7.00 %) 84.90 % (±7.88 %) 1.05
Multithreaded Pi Efficiency - 4 Threads 44.95 % (±3.29 %) 43.03 % (±2.97 %) 1.04
Multithreaded Pi Efficiency - 8 Threads 25.93 % (±2.13 %) 24.82 % (±1.80 %) 1.04
micro_benchmarks Build Time 218.50 s 180.71 s 1.21
micro_benchmarks File Size 1.00 MB 1.00 MB 1.00
Scheduling time - 1 thread 102.78 ticks (±33.83 ticks) 82.82 ticks (±6.13 ticks) 1.24
Scheduling time - 2 threads 63.22 ticks (±19.08 ticks) 54.62 ticks (±15.23 ticks) 1.16
Micro - Time for syscall (getpid) 5.96 ticks (±2.91 ticks) 4.72 ticks (±0.52 ticks) 1.26
Memcpy speed - (built_in) block size 4096 57946.75 MByte/s (±41308.85 MByte/s) 58673.43 MByte/s (±41280.23 MByte/s) 0.99
Memcpy speed - (built_in) block size 1048576 26172.59 MByte/s (±22850.32 MByte/s) 35794.70 MByte/s (±29729.68 MByte/s) 0.73
Memcpy speed - (built_in) block size 16777216 17661.33 MByte/s (±14625.63 MByte/s) 18201.88 MByte/s (±15357.27 MByte/s) 0.97
Memset speed - (built_in) block size 4096 58388.15 MByte/s (±41629.56 MByte/s) 58256.35 MByte/s (±41091.22 MByte/s) 1.00
Memset speed - (built_in) block size 1048576 27166.47 MByte/s (±23639.72 MByte/s) 36668.98 MByte/s (±30237.05 MByte/s) 0.74
Memset speed - (built_in) block size 16777216 18173.82 MByte/s (±14962.30 MByte/s) 18892.83 MByte/s (±15915.58 MByte/s) 0.96
Memcpy speed - (rust) block size 4096 50663.74 MByte/s (±37454.92 MByte/s) 52108.99 MByte/s (±36973.17 MByte/s) 0.97
Memcpy speed - (rust) block size 1048576 28273.70 MByte/s (±25058.40 MByte/s) 36326.40 MByte/s (±30295.90 MByte/s) 0.78
Memcpy speed - (rust) block size 16777216 17339.71 MByte/s (±14318.18 MByte/s) 17909.66 MByte/s (±15119.83 MByte/s) 0.97
Memset speed - (rust) block size 4096 51516.06 MByte/s (±38025.78 MByte/s) 52636.95 MByte/s (±37323.35 MByte/s) 0.98
Memset speed - (rust) block size 1048576 28543.72 MByte/s (±25145.92 MByte/s) 37354.68 MByte/s (±30907.97 MByte/s) 0.76
Memset speed - (rust) block size 16777216 17932.53 MByte/s (±14754.19 MByte/s) 18580.08 MByte/s (±15656.00 MByte/s) 0.97
alloc_benchmarks Build Time 214.12 s 173.11 s 1.24
alloc_benchmarks File Size 0.95 MB 0.95 MB 1.00
Allocations - Allocation success 100.00 % 100.00 % 1
Allocations - Deallocation success 70.03 % (±0.27 %) 69.98 % (±0.29 %) 1.00
Allocations - Pre-fail Allocations 100.00 % 100.00 % 1
Allocations - Average Allocation time 12986.12 Ticks (±377.67 Ticks) 11757.27 Ticks (±180.78 Ticks) 1.10
Allocations - Average Allocation time (no fail) 12986.12 Ticks (±377.67 Ticks) 11757.27 Ticks (±180.78 Ticks) 1.10
Allocations - Average Deallocation time 1146.53 Ticks (±184.89 Ticks) 912.86 Ticks (±152.93 Ticks) 1.26
mutex_benchmark Build Time 216.86 s 181.24 s 1.20
mutex_benchmark File Size 1.00 MB 1.00 MB 1.00
Mutex Stress Test Average Time per Iteration - 1 Threads 21.20 ns (±3.03 ns) 17.30 ns (±0.81 ns) 1.23
Mutex Stress Test Average Time per Iteration - 2 Threads 22.68 ns (±2.96 ns) 19.02 ns (±1.97 ns) 1.19

This comment was automatically generated by workflow using github-action-benchmark.

This lets the driver be used on all architectures, not just x86_64.

While at it, drop the udelay requirement, it boils down to a spin loop
and we already have a spin loop here.

Signed-off-by: Jens Reidel <[email protected]>
Signed-off-by: Jens Reidel <[email protected]>
@Gelbpunkt Gelbpunkt marked this pull request as ready for review October 21, 2025 09:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant