diff --git a/stim/envi/bsq.h b/stim/envi/bsq.h
index 3df3701..c5b6d36 100644
--- a/stim/envi/bsq.h
+++ b/stim/envi/bsq.h
@@ -10,7 +10,7 @@
 #include <deque>
 #include <chrono>
 #include <future>
-
+#include <algorithm>
 
 
 namespace stim{
@@ -403,7 +403,7 @@ public:
 		dst[1] = (T*) malloc(max_batch_bytes);
 
 		size_t N[2];																		//number of slices stored in buffers 0 and 1
-		N[0] = N[1] = min(Y(), max_slices_per_batch);										//start with the maximum number of slices that can be stored (may be the entire data set)
+		N[0] = N[1] = std::min<size_t>(Y(), max_slices_per_batch);										//start with the maximum number of slices that can be stored (may be the entire data set)
 
 		std::ofstream target(outname.c_str(), std::ios::binary);					//open an output file for writing
 																		//initialize with buffer 0 (used for double buffering)
@@ -425,7 +425,7 @@ public:
 			if(y_load < Y()){													//if there are still slices to be loaded, load them
 				if(y_load + N[b] > Y()) N[b] = Y() - y_load;					//if the next batch would process more than the total slices, adjust the batch size
 				rthread = std::async(std::launch::async, &stim::bsq<T>::readlines, this, src[b], y_load, N[b]);
-				
+				//rthread.wait();	
 				y_load += N[b];													//increment the number of loaded slices
 			}
 
@@ -438,7 +438,7 @@ public:
 			t_end = std::chrono::high_resolution_clock::now();
 			t_batch = std::chrono::duration_cast<std::chrono::milliseconds>(t_end-t_start).count();
 			t_total += t_batch;
-			rthread.wait();
+			if(y_load < Y()) rthread.wait();					//if a new batch was set to load, make sure it loads after calculations
 		}
 
 		std::cout<<"Total time to execute: "<<t_total<<" ms"<<std::endl;
@@ -471,7 +471,7 @@ public:
 		dst[1] = (T*) malloc(max_batch_bytes);
 
 		size_t N[2];																		//number of slices stored in buffers 0 and 1
-		N[0] = N[1] = min(Y(), max_slices_per_batch);										//start with the maximum number of slices that can be stored (may be the entire data set)
+		N[0] = N[1] = std::min<size_t>(Y(), max_slices_per_batch);										//start with the maximum number of slices that can be stored (may be the entire data set)
 
 		std::ofstream target(outname.c_str(), std::ios::binary);					//open an output file for writing
 																		//initialize with buffer 0 (used for double buffering)
@@ -506,7 +506,7 @@ public:
 			t_end = std::chrono::high_resolution_clock::now();
 			t_batch = std::chrono::duration_cast<std::chrono::milliseconds>(t_end-t_start).count();
 			t_total += t_batch;
-			rthread.wait();
+			if(y_load < Y()) rthread.wait();					//if a batch was threaded to load, make sure it finishes
 		}
 
 		std::cout<<"Total time to execute: "<<t_total<<" ms"<<std::endl;
--
libgit2 0.21.4