edits

python/python-syntax/profiling/profiler_gui.py

@@ -6,59 +6,80 @@
 import matplotlib.pyplot as plt
 
 
-def updateLines(fig, ax, linesCpuUse, x, yarr, scaley=False):
-    for line, y in zip(linesCpuUse, yarr):
-        line.set_xdata(x)
-        line.set_ydata(y)
-    ax.set_xlim(x[0], x[-1])
-    if scaley:
-        ax.set_ylim(np.min(yarr)*0.9, np.max(yarr)*1.1)
+def memByProc(thr=1000, nameKey=None):
+    rez = {}
+    for proc in psutil.process_iter():
+        try:
+            #pinfo = proc.as_dict(attrs=['pid', 'name', 'username'])
+            key = (proc.pid, proc.name())
+            val = proc.memory_info().vms
 
+            goodName = (nameKey is None) or (nameKey in key[1])
+            if goodName and (val > thr):
+                rez[key] = val
+        except (psutil.NoSuchProcess, psutil.AccessDenied, psutil.ZombieProcess):
+            pass
+    return rez
 
-timeStep = 0.5          # Seconds, how frequently to update the plot
-displayWindow = 30      # How many timesteps to show on a plot
+stuff = memByProc(nameKey='python')
+print(stuff)
+print(len(stuff))
 
-nCpu = psutil.cpu_count()
-
-timeArr = np.array([-timeStep, 0])  # Time since start of measurement
-timeTrueArr = np.array([0,0])       # Absolute time
-cpuUseArr = np.zeros((nCpu, 2))     # Cpu usage in percent
-memUseArr = np.zeros(2)             # Mem usage in bytes
-
-# Initialize plots
-plt.ion()
-fig, ax = plt.subplots(ncols=2)
-ax[0].set_ylim(0, 100)
-ax[0].set_title("Cpu Load, %")
-ax[1].set_title("Memory Load, Bytes")
-linesCpuUse = [ax[0].plot(timeArr, cpuUseArr[i])[0] for i in range(nCpu)]
-linesMemUse = [ax[1].plot(timeArr, memUseArr)[0]]
-
-try:
-    while True:
-        print("time", timeArr[-1])
-        cpuPercent = psutil.cpu_percent(percpu=True)
-
-        timeArr = np.append(timeArr, timeArr[-1] + timeStep)
-        timeTrueArr = np.append(timeTrueArr, str(datetime.datetime.now().time()))
-        cpuUseArr = np.c_[cpuUseArr, np.array(cpuPercent)]
-        memUseArr = np.append(memUseArr, psutil.virtual_memory().used)
-
-        updateLines(fig, ax[0], linesCpuUse, timeArr[-displayWindow:], cpuUseArr[:, -displayWindow:])
-        updateLines(fig, ax[1], linesMemUse, timeArr[-displayWindow:], [memUseArr[-displayWindow:]], scaley=True)
-
-        fig.canvas.draw()
-        fig.canvas.flush_events()
-
-        time.sleep(timeStep)
-except:
-    print('User closed displayWindow')
-
-print('Writing results to HDF5')
-timeArrEncoded = [s.encode("ascii", "ignore") for s in timeTrueArr[2:]]
-with h5py.File('perfLog.h5', 'w') as logFile:
-    logFile.create_dataset('times', (len(timeArrEncoded), 1), 'S10', data=timeArrEncoded)
-    logFile.create_dataset('cpuUse', data=cpuUseArr[:, 2:])
-    logFile.create_dataset('memUse', data=memUseArr[2:])
-
-print('Done')
+#
+#
+# def updateLines(fig, ax, linesCpuUse, x, yarr, scaley=False):
+#     for line, y in zip(linesCpuUse, yarr):
+#         line.set_xdata(x)
+#         line.set_ydata(y)
+#     ax.set_xlim(x[0], x[-1])
+#     if scaley:
+#         ax.set_ylim(np.min(yarr)*0.9, np.max(yarr)*1.1)
+#
+#
+# timeStep = 0.5          # Seconds, how frequently to update the plot
+# displayWindow = 30      # How many timesteps to show on a plot
+#
+# nCpu = psutil.cpu_count()
+#
+# timeArr = np.array([-timeStep, 0])  # Time since start of measurement
+# timeTrueArr = np.array([0,0])       # Absolute time
+# cpuUseArr = np.zeros((nCpu, 2))     # Cpu usage in percent
+# memUseArr = np.zeros(2)             # Mem usage in bytes
+#
+# # Initialize plots
+# plt.ion()
+# fig, ax = plt.subplots(ncols=2)
+# ax[0].set_ylim(0, 100)
+# ax[0].set_title("Cpu Load, %")
+# ax[1].set_title("Memory Load, Bytes")
+# linesCpuUse = [ax[0].plot(timeArr, cpuUseArr[i])[0] for i in range(nCpu)]
+# linesMemUse = [ax[1].plot(timeArr, memUseArr)[0]]
+#
+# try:
+#     while True:
+#         print("time", timeArr[-1])
+#         cpuPercent = psutil.cpu_percent(percpu=True)
+#
+#         timeArr = np.append(timeArr, timeArr[-1] + timeStep)
+#         timeTrueArr = np.append(timeTrueArr, str(datetime.datetime.now().time()))
+#         cpuUseArr = np.c_[cpuUseArr, np.array(cpuPercent)]
+#         memUseArr = np.append(memUseArr, psutil.virtual_memory().used)
+#
+#         updateLines(fig, ax[0], linesCpuUse, timeArr[-displayWindow:], cpuUseArr[:, -displayWindow:])
+#         updateLines(fig, ax[1], linesMemUse, timeArr[-displayWindow:], [memUseArr[-displayWindow:]], scaley=True)
+#
+#         fig.canvas.draw()
+#         fig.canvas.flush_events()
+#
+#         time.sleep(timeStep)
+# except:
+#     print('User closed displayWindow')
+#
+# print('Writing results to HDF5')
+# timeArrEncoded = [s.encode("ascii", "ignore") for s in timeTrueArr[2:]]
+# with h5py.File('perfLog.h5', 'w') as logFile:
+#     logFile.create_dataset('times', (len(timeArrEncoded), 1), 'S10', data=timeArrEncoded)
+#     logFile.create_dataset('cpuUse', data=cpuUseArr[:, 2:])
+#     logFile.create_dataset('memUse', data=memUseArr[2:])
+#
+# print('Done')