diff --git a/hpc/archer2/grid_search.slurm b/hpc/archer2/grid_search.slurm
index 9a71c5a9..91f7a93a 100644
--- a/hpc/archer2/grid_search.slurm
+++ b/hpc/archer2/grid_search.slurm
@@ -47,10 +47,10 @@ export SRUN_CPUS_PER_TASK=$SLURM_CPUS_PER_TASK
 n_points=1000000 # points per MPI process
 n_tasks=8
 global_depth=1 # Number of local roots matches the number of MPI processes, therefore the number of NUMA regions
-local_depth=(3 4 5 6)
-n_samples=(100 1000 5000)
-block_size=(16 32 64 128 256)
-n_threads=(2 4 8 16) # See if bandwidth saturates with different threading parameters for Rayon thread pool
+local_depth=(5 6)
+n_samples=(100 1000)
+block_size=(128 256)
+n_threads=(2 16) # See if bandwidth saturates with different threading parameters for Rayon thread pool
 export OMP_NUM_THREADS=1 # Need to set to 1 to avoid oversubsciption between Rayon and OpenMP
 
 # Create a CSV output file for analysis
@@ -60,7 +60,8 @@ echo "
 runtime, p2m, m2m, l2l, m2l, p2p, \
 source_tree, target_tree, source_domain, target_domain, layout, \
 ghost_exchange_v, ghost_exchange_u, gather_global_fmm, scatter_global_fmm, \
-source_to_target_data, source_data, target_data, global_fmm, ghost_fmm_v, ghost_fmm_u
+source_to_target_data, source_data, target_data, global_fmm, ghost_fmm_v, ghost_fmm_u \
+n_points, n_tasks, global_depth, local_depth, n_samples, block_size, n_threads
 " >> ${OUTPUT}
 
 # Perform grid search
@@ -68,7 +69,8 @@ for i in ${!local_depth[@]}; do
     for j in ${!n_samples[@]}; do
         for k in ${!block_size[@]}; do
             for l in ${!n_threads[@]}; do
-                srun --ntasks=$n_tasks --cpus-per-task=16 --distribution=block:block --hint=nomultithread \
+
+                runtime_output=$(srun --ntasks=$n_tasks --cpus-per-task=16 --distribution=block:block --hint=nomultithread \
                     ${WORK}/grid_search_mpi --n-points $n_points \
                     --expansion-order 3 \
                     --prune-empty \
@@ -76,7 +78,9 @@ for i in ${!local_depth[@]}; do
                     --local-depth ${local_depth[$i]} \
                     --n-samples ${n_samples[$j]} \
                     --block-size ${block_size[$k]} \
-                    --n-threads ${n_threads[$l]} >> ${OUTPUT}
+                    --n-threads ${n_threads[$l]})
+
+                echo "$runtime_output, $n_points, $n_tasks, $global_depth, ${local_depth[$i]}, ${n_samples[$j]}, ${block_size[$k]}, ${n_threads[$l]}" >> ${OUTPUT}
             done
         done
     done