diff --git a/systems/Perlmutter/dwf4.slurm b/systems/Perlmutter/dwf4.slurm
index f6d6a2a9..8a37a266 100644
--- a/systems/Perlmutter/dwf4.slurm
+++ b/systems/Perlmutter/dwf4.slurm
@@ -2,19 +2,26 @@
 #SBATCH -A m3886_g
 #SBATCH -C gpu
 #SBATCH -q debug
-#SBATCH -t 0:10:00
+#SBATCH -t 0:20:00
+#SBATCH -c 32
+#SBATCH -N 1
 #SBATCH -n 4
 #SBATCH --ntasks-per-node=4
-#SBATCH -c 32
-#SBATCH --exclusive
 #SBATCH --gpus-per-task=1
+#SBATCH --exclusive
 #SBATCH --gpu-bind=none
 
 export SLURM_CPU_BIND="cores"
 export MPICH_GPU_SUPPORT_ENABLED=1
-export MPICH_RDMA_ENABLED_CUDA=0
-export MPICH_GPU_IPC_ENABLED=0
+export MPICH_RDMA_ENABLED_CUDA=1
+export MPICH_GPU_IPC_ENABLED=1
 export MPICH_GPU_EAGER_REGISTER_HOST_MEM=0
-export MPICH_GPU_NO_ASYNC_MEMCPY=1
-OPT="--comms-overlap --comms-concurrent --shm-mpi 1"
-srun ./benchmarks/Benchmark_ITT --mpi 2.1.1.2 --grid 64.64.64.64 --accelerator-threads 8 --shm 2048 $OPT > ITT.log
+export MPICH_GPU_NO_ASYNC_MEMCPY=0
+#export MPICH_SMP_SINGLE_COPY_MODE=CMA
+
+OPT="--comms-overlap --shm-mpi 1"
+VOL=64.64.32.32
+srun ./benchmarks/Benchmark_dwf_fp32 --mpi 2.2.1.1 --grid $VOL --accelerator-threads 8 --shm 2048 $OPT
+#srun ./benchmarks/Benchmark_dwf_fp32 --mpi 2.1.1.4 --grid $VOL --accelerator-threads 8 --shm 2048 $OPT
+#srun ./benchmarks/Benchmark_dwf_fp32 --mpi 1.1.1.8 --grid $VOL --accelerator-threads 8 --shm 2048 $OPT
+