diff --git a/.gitignore b/.gitignore
index b5686f7..6ce11c8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -3,4 +3,5 @@
 /workspace.xml
 .idea
 primes.txt
-timings.csv
\ No newline at end of file
+*.csv
+*.csv#
\ No newline at end of file
diff --git a/benchmarks/block_size.py b/benchmarks/block_size.py
index c30da4c..9b62659 100644
--- a/benchmarks/block_size.py
+++ b/benchmarks/block_size.py
@@ -1,4 +1,7 @@
+import datetime
 import math
+import time
+
 import pycuda.autoinit
 import pycuda.driver as drv
 import numpy as np
@@ -34,10 +37,11 @@ def bench_block_size(grid_size: int, block_size: int, calc_count: int):
     endEvent.synchronize()
 
     kernel_execution_time = startEvent.time_till(endEvent)
-    return (block_size * grid_size)/kernel_execution_time
+    return (block_size * grid_size) / kernel_execution_time
 
 
 if __name__ == "__main__":
+    start_time = datetime.datetime.now()
     parser = OptionParser()
     parser.add_option("--bench-output", dest="bench_output",
                       help="The output file for timings", default="timings.csv", type="string")
@@ -45,7 +49,10 @@ if __name__ == "__main__":
                       help=" How many calculations steps should be done per GPU thread", default="1000000", type="int")
 
     parser.add_option("--grid_size", dest="grid_size",
-                      help="number of blocks", default="1024", type="int")
+                      help="number of blocks", default="8192", type="int")
+    parser.add_option("--number_of_threads", dest="number_of_threads",
+                      help="If set the grid size is ignored and is adjusted so that the number of threads is the same "
+                           "in all cases", type="int", default="1048576")
     parser.add_option("--block_size_start", dest="block_size_start",
                       help="initial number of threads per block", default="4",
                       type="int")
@@ -61,18 +68,33 @@ if __name__ == "__main__":
     (options, args) = parser.parse_args()
     print(
         "Benchmarking block size. Grid Size: {}, Start: {}, Step: {} ,Stop: {}, Calculations: {}, Repetitions: {}".format(
-            options.grid_size, options.block_size_start, options.block_size_step, options.block_size_stop, options.calculation_steps, options.repetitions))
+            options.grid_size, options.block_size_start, options.block_size_step, options.block_size_stop,
+            options.calculation_steps, options.repetitions))
 
-    with open(options.bench_output, 'w') as file:
+    filename = str(time.strftime("%Y%m%d-%H%M%S")) + "_block_size_" + str(options.number_of_threads) + "threads_" + str(
+        options.calculation_steps) + "claculations_" + str(options.repetitions) + "repetitions_" + options.bench_output
+    with open(filename, 'w') as file:
         file.write("block size,calls per second\n")
 
     current_block_size = options.block_size_start
 
     while current_block_size <= options.block_size_stop:
         total_duration = 0
+        if options.number_of_threads is not None:
+            if options.number_of_threads % current_block_size != 0:
+                print(str(options.number_of_threads) + " is not dividable by block size of " + str(
+                    current_block_size) + " thus will be skipped")
+                current_block_size += options.block_size_step
+                continue
+            else:
+                current_grid_size = int(options.number_of_threads / current_block_size)
+        else:
+            current_grid_size = options.grid_size
         for i in range(0, options.repetitions):
-            duration = bench_block_size(options.grid_size, current_block_size, options.calculation_steps)
+            duration = bench_block_size(current_grid_size, current_block_size, options.calculation_steps)
             total_duration += duration
-        with open(options.bench_output, 'a') as file:
-            file.write(str(current_block_size) + "," + str(total_duration/options.repetitions) + "\n")
+        with open(filename, 'a') as file:
+            file.write(str(current_block_size) + "," + str(total_duration / options.repetitions) + "\n")
         current_block_size += options.block_size_step
+
+    print("finished in " + str((datetime.datetime.now()-start_time)))