a[i] = b[i] = c[i] = i. The expression d = a + b * c computes d[i] = i + i * i for indices 0 through 63. Addition and multiplication are fused into a single vectorized loop, avoiding an intermediate array for the product. On Helium, the loop uses predication. Both tests print the resulting vector.test1 uses statically sized vectors and the allocator selected by TMP_ALLOC in Dsppp/Dsppp.cproject.yml. The default pool_allocator supplies four 256-byte buffers, one for each vector of 64 floats.test2 uses dynamically sized vectors with an explicit arena_allocator. A 1024-byte arena is allocated with 64-byte alignment. Each allocation rounds the arena offset up to a multiple of ALIGN (64 by default), assuming the arena base is already aligned. Individual deallocations do nothing; resetting the arena reclaims all its allocations. The test prints the consumed bytes before resetting and freeing the arena.TMP_ALLOC. When their allocator differs from the destination allocator, their data is copied instead of moved.TMP_ALLOC to stat_allocator and enable STAT_ALLOCATOR in Dsppp/Dsppp.cproject.yml. After test1, the example prints peak allocation counts and checks for outstanding allocations.a, b, c are the input vectors.d is the result vector.NB is the number of elements in each vector (64).ARENA_SIZE is the arena capacity in bytes (1024).See also Memory allocation for the custom allocator interface.