I have included a the Benchmarks of the 11 ways of achieving this task where the version above is 'SIMD_no_heap_allocation'. Please see the thread for the code.

I have included a the Benchmarks of the 11 ways of achieving this task where the version above is 'SIMD_no_heap_allocation'. Please see the thread for the code of the other versions and to see how far we have come.

I'm trying to find the fastest way to multiply every element in an array by a given number, in this case 2. So far I have written a SIMD instruction that performs at 65 ns multiplying every number in an array of size 1000 by 2.

The techniques used so far are: SIMD instructions, MemoryMarshal method for the loop, Spans, Vector256 Unsafe loading, BitHacks for the multiplication and calculating upper limit of SIMD instructions.

I have included a the Benchmarks of five ways of achieving this task where the version above is 'SIMD_no_heap_allocation'

| Method                              |  Mean   | Error |StdDev |  Gen0  | Allocated |
|-------------------------------------|--------:|------:|------:|-------:|----------:|
| LINQ                                | 1,747.4 | 16.52 | 12.90 | 0.6485 |    4072 B |
| for_loop                            |   613.0 |  4.35 |  4.07 | 0.6409 |    4024 B |
| SIMD                                |   221.1 |  4.39 |  4.31 | 0.6413 |    4024 B |
| SIMD_no_heap_allocation             |   217.5 |  0.78 |  0.69 |      - |         - |
| SIMD_no_heap_allocation_in_parallel | 5,969.2 | 68.29 | 63.88 | 0.2899 |    1832 B |

I'm trying to find the fastest way to multiply every element in an array by a given number, in this case 2. So far I have written a SIMD instruction that performs at 33 ns multiplying every number in an array of size 1000 by 2.

The techniques used so far are: SIMD instructions, MemoryMarshal method for the loop, Spans, Vector256 Unsafe loading and storing, BitHacks for the multiplication and calculating upper limit of SIMD instructions.

I have included a the Benchmarks of the 11 ways of achieving this task where the version above is 'SIMD_no_heap_allocation'. Please see the thread for the code.

| Method                                                | Mean         | Error      | StdDev     | Gen0   | Gen1   | Allocated |
|------------------------------------------------------ |-------------:|-----------:|-----------:|-------:|-------:|----------:|
| PLINQ                                                 | 22,086.67 ns | 301.232 ns | 235.182 ns | 2.9907 | 0.0305 |   17968 B |
| LINQ                                                  |  1,674.13 ns |   3.968 ns |   3.518 ns | 0.6485 |      - |    4072 B |
| for_loop                                              |    597.42 ns |   2.075 ns |   1.839 ns | 0.6409 |      - |    4024 B |
| memory_marshall_method_bit_hacked                     |    373.40 ns |   0.442 ns |   0.392 ns |      - |      - |         - |
| SIMD                                                  |    208.13 ns |   0.870 ns |   0.727 ns | 0.6413 |      - |    4024 B |
| SIMD_no_heap_allocation                               |    209.12 ns |   0.448 ns |   0.374 ns |      - |      - |         - |
| SIMD_no_heap_allocation_bit_hacked                    |    129.27 ns |   0.766 ns |   0.679 ns |      - |      - |         - |
| SIMD_memory_marshall_method                           |     59.86 ns |   0.942 ns |   0.925 ns |      - |      - |         - |
| SIMD_Vector256_Memory_marshal                         |     58.63 ns |   0.171 ns |   0.160 ns |      - |      - |         - |
| SIMD_Vector256_Memory_Marshal_unrolled                |     37.67 ns |   0.114 ns |   0.101 ns |      - |      - |         - |
| SIMD_Vector256_Memory_Marshal_unrolled_quicker_bounds |     33.85 ns |   0.110 ns |   0.103 ns |      - |      - |         - |