ref: benchmark toy detector config

[niermann999]

Allow the toy detector grids to find all neighboring surfaces, this should increase the number of tracks that can be found and fitted. Furthermore, I restricted the simulation to an eta region, where the tracks should hit a decent number of sensitive surface.

I also tried removing the constrained step from the finder and fitter, since they are not used in performance relevant code and eat at least a bit of memory (it is an array or constraints that all need to be checked at every step).

I then tried to see if the performance hit comes from the grids, so I tried using the static grids of the toy detector instead of a grid with dynamic bin capacities. Apart from that, the detector is now copied to device instead of using managed memory.

[niermann999]

As soon as I let the grid actually search for neighbouring surfaces, the perfromance gets really bad...

Running ./bin/traccc_benchmark_cpu
Run on (48 X 2551.74 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x24)
  L1 Instruction 32 KiB (x24)
  L2 Unified 512 KiB (x24)
  L3 Unified 32768 KiB (x4)
Load Average: 0.92, 0.66, 1.04
WARNING: No entries in volume finder

Detector check: OK
-----------------------------------------------------------------------------------
Benchmark                         Time             CPU   Iterations UserCounters...
-----------------------------------------------------------------------------------
ToyDetectorBenchmark/CPU 1.8804e+10 ns   1.3231e+10 ns            1 event_throughput_Hz=7.55819/s

Running ./bin/traccc_benchmark_cuda
Run on (48 X 2270.35 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x24)
  L1 Instruction 32 KiB (x24)
  L2 Unified 512 KiB (x24)
  L3 Unified 32768 KiB (x4)
Load Average: 1.80, 2.18, 1.64
WARNING: No entries in volume finder

Detector check: OK
------------------------------------------------------------------------------------
Benchmark                          Time             CPU   Iterations UserCounters...
------------------------------------------------------------------------------------
ToyDetectorBenchmark/CUDA 1.3547e+10 ns   1.3471e+10 ns            1 event_throughput_Hz=7.42362/s

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud