Stronger acquire scope for low-latency kernels

csrc/kernels/internode_ll.cu

@@ -260,7 +260,7 @@ dispatch(void* packed_recv_x, float* packed_recv_x_scales,
         int num_recv_tokens, recv_token_begin_idx;
         EP_STATIC_ASSERT(kNumWarpsPerGroup > 1, "Requires more than one warp per group");
         if (sub_warp_id == 1 and lane_id == 0) {
-            while ((num_recv_tokens = ld_acquire_global(rdma_recv_count + local_expert_idx * num_ranks + src_rank)) == 0);
+            while ((num_recv_tokens = ld_acquire_sys_global(rdma_recv_count + local_expert_idx * num_ranks + src_rank)) == 0);
             num_recv_tokens = -num_recv_tokens - 1;
             recv_token_begin_idx = atomicAdd(packed_recv_count + local_expert_idx, num_recv_tokens);
             shared_num_recv_tokens[warp_group_id] = num_recv_tokens;
@@ -450,7 +450,7 @@ combine(void* combined_x,
     if (responsible_expert_idx < num_experts) {
         EP_STATIC_ASSERT(kNumWarpsPerGroup > 1, "Invalid number of warps per group");
         if (sub_warp_id == 0 and lane_id == 0)
-            while (ld_acquire_global(rdma_recv_flag + responsible_expert_idx) == 0);
+            while (ld_acquire_sys_global(rdma_recv_flag + responsible_expert_idx) == 0);
     }
     cg::this_grid().sync();