check in

scripts/plot/load_miss_ratio.py

@@ -68,7 +68,7 @@ def _find_cache_size(datapath, metric):
  )
 
  if len(cache_sizes) < N_CACHE_SIZE:
- # the working sets of some traces are not large enough, when running with some cache sizes, 
+ # the working sets of some traces are not large enough, when running with some cache sizes,
  # e.g., 0.0001, there are too few objects to run, so we will have fewer cache sizes
  return []
 
@@ -113,7 +113,9 @@ def load_miss_ratio_reduction_from_dir(data_dir_path, algos, metric="miss_ratio"
  for f in sorted(glob.glob(data_dir_path + "/*")):
  # a list of miss ratio dict (algo -> miss ratio) at different cache sizes
  miss_ratio_dict_list = load_data(f, metric)
- assert len(miss_ratio_dict_list) == N_CACHE_SIZE, f"{f} {len(miss_ratio_dict_list)}"
+ assert (
+ len(miss_ratio_dict_list) == N_CACHE_SIZE
+ ), f"{f} {len(miss_ratio_dict_list)}"
  # print(f, sorted(miss_ratio_dict_list[0].keys()))
 
  if len(mr_reduction_dict_list) == 0:
@@ -157,6 +159,61 @@ def load_miss_ratio_reduction_from_dir(data_dir_path, algos, metric="miss_ratio"
  return mr_reduction_dict_list
 
 
+def load_miss_ratio_and_miss_ratio_reduction_from_dir(
+ data_dir_path, algos, metric="miss_ratio"
+):
+ data_dirname = os.path.basename(data_dir_path)
+
+ mr_reduction_dict_list = []
+ for f in sorted(glob.glob(data_dir_path + "/*")):
+ # a list of miss ratio dict (algo -> miss ratio) at different cache sizes
+ miss_ratio_dict_list = load_data(f, metric)
+ assert (
+ len(miss_ratio_dict_list) == N_CACHE_SIZE
+ ), f"{f} {len(miss_ratio_dict_list)}"
+ # print(f, sorted(miss_ratio_dict_list[0].keys()))
+
+ if len(mr_reduction_dict_list) == 0:
+ mr_reduction_dict_list = [
+ defaultdict(list) for _ in range(len(miss_ratio_dict_list))
+ ]
+ for size_idx, miss_ratio_dict in enumerate(miss_ratio_dict_list):
+ if len(miss_ratio_dict) == 0:
+ continue
+
+ mr_fifo = miss_ratio_dict["FIFO"]
+ if mr_fifo == 0 or 1 - mr_fifo == 0:
+ continue
+
+ miss_ratio_dict = {k.lower(): v for k, v in miss_ratio_dict.items()}
+
+ mr_reduction_dict = {}
+ for algo in algos:
+ if algo.lower() not in miss_ratio_dict:
+ logger.warning("no data for {} in {}".format(algo.lower(), f))
+ break
+ miss_ratio = miss_ratio_dict[algo.lower()]
+ mr_reduction = (mr_fifo - miss_ratio) / mr_fifo
+ if mr_reduction < 0:
+ mr_reduction = -(miss_ratio - mr_fifo) / miss_ratio
+
+ mr_reduction_dict[algo] = mr_reduction
+
+ if len(mr_reduction_dict) < len(algos):
+ # some algorithm does not have data
+ logger.info(
+ "skip {} because of missing algorithm result {}".format(
+ f, set(algos) - set(miss_ratio_dict.keys())
+ )
+ )
+ continue
+
+ for algo, mr_reduction in mr_reduction_dict.items():
+ mr_reduction_dict_list[size_idx][algo].append((mr_fifo, mr_reduction))
+
+ return mr_reduction_dict_list
+
+
 if __name__ == "__main__":
  r = load_data(
  "/disk/result/libCacheSim/result//all/meta_kvcache_traces_1.oracleGeneral.bin.zst"

scripts/plot/plot_miss_ratio2.py

@@ -0,0 +1 @@
+python3.12

scripts/plot/plot_mr_red_box.py

@@ -21,7 +21,7 @@ def update_algo_name(algo_name):
  name_dict = {
  "WTinyLFU-w0.01-SLRU": "TinyLFU",
  "S3FIFO-0.1000-2": "S3FIFO",
- "Cloud2QPlus-0.1000-2-0.50": "Cloud2QPlus",
+ "Cloud2QPlus-0.1000-2-0.50": "Cloud2Q+",
  }
 
  return name_dict.get(algo_name, algo_name)
@@ -74,7 +74,6 @@ def plot_box_algo(datapath, size_idx=0, metric="miss_ratio"):
  algo_list = [
  "S3FIFO-0.1000-2",
  "Cloud2QPlus-0.1000-2-0.50",
- # "Cloud2QPlus-0.1000-2-1.00",
  "LIRS",
  "TwoQ",
  "ARC",
@@ -108,7 +107,7 @@ def plot_box_algo(datapath, size_idx=0, metric="miss_ratio"):
  plt.clf()
 
 
-def plot_box_ignore_frac(datapath, size_idx=0, metric="miss_ratio"):
+def plot_box_corr(datapath, size_idx=0, metric="miss_ratio"):
  """
  plot the miss ratio reduction box
 
@@ -151,12 +150,13 @@ def plot_box_ignore_frac(datapath, size_idx=0, metric="miss_ratio"):
  medianprops=dict(color="black", linewidth=1.6),
  )
 
+ plt.xticks(range(1, len(algo_list) + 1), name_list, rotation=0)
  plt.xlabel("Correlation window size (fraction of small FIFO size)")
  plt.ylabel("Miss ratio reduction from FIFO")
- plt.xticks(range(1, len(algo_list) + 1), name_list, rotation=0)
  plt.grid(linestyle="--")
- plt.savefig("{}_corr_box_{}.png".format(metric, size_idx))
- plt.savefig("{}_corr_box_{}.pdf".format(metric, size_idx))
+ plt.subplots_adjust(bottom=-0.2)
+ plt.savefig("{}_corr_box_{}.png".format(metric, size_idx), bbox_inches="tight")
+ plt.savefig("{}_corr_box_{}.pdf".format(metric, size_idx), bbox_inches="tight")
  plt.clf()
 
 
@@ -167,17 +167,16 @@ def plot_box_ignore_frac(datapath, size_idx=0, metric="miss_ratio"):
  parser.add_argument("--datapath", type=str, help="path to the cachesim result")
  ap = parser.parse_args()
 
- # plot_percentiles("{}/all".format(ap.datapath), size_idx=0, metric="miss_ratio")
- # plot_percentiles("{}/all".format(ap.datapath), size_idx=2, metric="miss_ratio")
-
- plot_box_algo("/disk/libCacheSim/_build/result2/", size_idx=0, metric="miss_ratio")
- # plot_box_algo("/disk/libCacheSim/_build/result2/", size_idx=2, metric="miss_ratio")
- plot_box_algo("/disk/libCacheSim/_build/result2/", size_idx=4, metric="miss_ratio")
+ plot_box_algo(
+  "/disk/libCacheSim/_build/result_data/", size_idx=0, metric="miss_ratio"
+ )
+ plot_box_algo(
+  "/disk/libCacheSim/_build/result_data/", size_idx=4, metric="miss_ratio"
+ )
 
- plot_box_ignore_frac(
- "/disk/libCacheSim/_build/result2/", size_idx=0, metric="miss_ratio"
+ plot_box_corr(
+ "/disk/libCacheSim/_build/result_data/", size_idx=0, metric="miss_ratio"
  )
- # plot_box_ignore_frac("/disk/libCacheSim/_build/result2/", size_idx=2, metric="miss_ratio")
- plot_box_ignore_frac(
- "/disk/libCacheSim/_build/result2/", size_idx=4, metric="miss_ratio"
+ plot_box_corr(
+ "/disk/libCacheSim/_build/result_data/", size_idx=4, metric="miss_ratio"
  )