37bb2945cc38af48dfa5ad09392736c427008a80,perfkitbenchmarker/linux_benchmarks/redis_benchmark.py,,Run,#Any#,139

Before Change

  max_throughput_for_completion_latency_under_1ms = 0.0

  while latency < latency_threshold:
    iteration_results = {}
    threads += max(1, int(threads * .15))
    num_loaders = len(load_vms) * num_servers
    args = [((redis_vm, load_vms[i % len(load_vms)], threads / num_loaders +
              (0 if (i + 1) > threads % num_loaders else 1),
              FIRST_PORT + i % num_servers, i, iteration_results),
             {}) for i in range(num_loaders)]
    logging.error("BEFORE: %s", args)
    vm_util.RunThreaded(RunLoad, args)
    throughput = 0.0
    latency = 0.0
    logging.error("%s", iteration_results)
    for result in iteration_results.values():
      throughput += result[0]
    for result in iteration_results.values():
      latency += result[1] * result[0] / throughput

    if latency < 1.0:
        max_throughput_for_completion_latency_under_1ms = max(

After Change

          "Zero throughput for {} threads: {}".format(threads, client_results))

    // Average latency across clients
    latency = (sum(client_latency * client_throughput
                   for client_latency, client_throughput in client_results) /
               throughput)

    if latency < 1.0:
        max_throughput_for_completion_latency_under_1ms = max(
            max_throughput_for_completion_latency_under_1ms,

Instances