Aurora Kernel
Benchmarks.
Performance metrics of the Aurora Hybrid Kernel across emulated and bare-metal hardware environments.
Tested Benchmarks (QEMU)
Benchmarks of the Aurora Kernel on QEMU. Measurements demonstrate extreme low-latency responsiveness and determinism required for high-frequency hardware integration.
Average Boot-Up time : 0.70 seconds.
Application Load Time : 6.31ms - 15.26ms ( AuLoadExecToProcess ).
GPU Interrupt or Network Interrupt : 0.01ms - 13.90ms.
System Call & Memory Latencies
System Call Latency
- GetEnvironmentBlock : 0.00ms (Zero Latency).
- GetProcessHeapMem : < 0.07ms ( Near Zero Latency ).
- FileSetOffset call : < 42.31ms.
Memory Allocation Latencies
- ( kmalloc ) : 0.0ms - 4.41ms.
- ( kmalloc ) : < 10.94 ms.
- Shared Memory Request Latency: 1.05ms - 5.57ms.
Memory Allocation Profile
Faster Transitions
Core system calls are roughly 30-40% faster due to reduced abstraction layers.
Leaner Footprint
Minimal driver-linking on boot allows for a "Cold Boot" under
1 second to scheduler readiness.
Optimized Compositing
Direct shared-memory hooks for the user-space display server minimize the overhead.
Testing Environment & Methodology
Architecture: ARM64 (AArch64).
Platform: QEMU Emulation with UART debug output.
Timing: High Precision CPU cycle counter.
Tested Benchmarks (NXP i.MX 8M Plus)
Bare-metal hardware validation of the Aurora Kernel on the NXP i.MX 8M Plus SoC. Measurements demonstrate extreme low-latency responsiveness and determinism required for high-frequency hardware integration.
Bootup of Kernel up to schedular_start: 4678511 CPU cycles. Depends on initialization of subsystem and hardware initializations.
Application Load Time (AuLoadExecToProcess). Varies on usage of kernel file cache or reading the binary from disk.
Peripheral Interrupt handling latency: 0.01ms - 13.90ms. May vary with different peripherals.
System Call & Memory Latencies
System Call Latency
- FileStat : 0.00ms (Zero Latency).
- CreateMemMappings : < 3.12ms.
- OpenFile : < 3.22ms (2,644,525 CPU Cycles).
Memory Allocation Latencies
- Kernel Malloc ( kmalloc ) : 0.07ms - 0.08ms.
- Note: May vary with allocation sizes, assertion is made for aligned allocation.
Memory Allocation Profile
Faster Transitions
Core system calls are roughly 30-40% faster due to reduced abstraction layers.
Leaner Footprint
Minimal driver-linking on boot allows for a 2-second "Cold Boot" to scheduler readiness.
Optimized Compositing
Direct shared-memory hooks for the user-space display server minimize the overhead.
Testing Environment & Methodology
Architecture: ARM64 (AArch64).
Platform: Target SoC: NXP i.MX 8M Plus.
Timing: High Precision CPU timer with UART COM9 debug output.