Gpu util is 0 when run tensorflow training job, and context switch is very high - 


tensorflow: 1.2.0 gpu: titan x (pascal) driver: 370.28

i run distrubuted tensorflow train image classify model, see no gpu usage(actually,gpu util of mnist or other training job 0).

there's many poll system call when straced training process(poll fd /dev/nvidia0):

 poll([{fd=8, events=pollin}, {fd=12, events=pollin}, {fd=13, events=pollin}, {fd=14, events=pollin}, {fd=15, events=pollin}, {fd=17, events=pollin}, {fd=18, events=pollin}, {fd=19, events=pollin}, {fd=20, events=pollin}, {fd=21, events=pollin}], 10, 100 <unfinished ...>  futex(0x2d1eca4, futex_wait_bitset_private|futex_clock_realtime, 3340677, {1502763800, 428734182}, ffffffff) = -1 etimedout (connection timed out)

vmstat shows high context switch, millions cs per second.

had seen before?

i've had same problem before, it's because gpu not set run -- ran tensorflow on cpu, thought run on gpu. if right, won't this.

1) check use nvidia-smi check: despite gpu util 0%, gpu memory util 0% ? if there isn't process in processes @ all?

-- if so, gpu not used @ all, tensorflow must running on cpu (you may use top check cpu usage, if above 100% it's additional proof program paraleled on cpu)

in case, should check if you've installed gpu version of tensorflow. may find 2 different versions of installation introduction in www.tensorflow.org cpu/gpu respectly. cpu version tensorflow may never run on gpu.

plus above, machine enviroment requires specify gpu device want use explicitly. use command following check:

cuda_visible_devices=0 python rnn_mnist.py

(note =? must in right format, e.g. cuda_visible_devices=[0] invalid no warning promoted, program run on cpu instead)

2) if not case, python running on gpu , it's util 0%. there's possibility data fetching cost time, on cpu, , gpu waiting data, util avg 0%.

-- possible reason set batch_size small, try 128 or 1024.


Comments

Post a Comment

Popular posts from this blog

android - InAppBilling registering BroadcastReceiver in AndroidManifest -

i'm implementing inapp billing in android app. works well, however, i'm trying decouple broadcast receiver activity manifest. suggestion in android's trivialdrive sample: // important: dynamically register broadcast messages updated purchases. // register receiver here instead of <receiver> in manifest // because call getpurchases() @ startup, therefore can ignore // broadcasts sent while app isn't running. // note: registering listener in activity bad idea, done here // because sample. regardless, receiver must registered after // iabhelper setup, before first call getpurchases(). currently there's class extends broadcastreceiver : public class iabbroadcastreceiver extends broadcastreceiver { /** * intent action receiver should filter for. */ public static final string action = "com.android.vending.billing.purchases_updated"; private final iabbroadcastlistener mlistener; public iabbroadcastreceiver(iabbroadcastlistener listener) { m...
Read more

python Tkinter Capturing keyboard events save as one single string -

is possible store keyboard events 1 string? code below stores 1 char , prints it. card reader or bar code reader, contains collection of character/ string not 1 character @ time. goal save char pressed text variable. from tkinter import * root = tk() def key(event): text= event.char text+= event.char print ("pressed", text) def callback(event): frame.focus_set() print ("clicked at", event.x, event.y) frame = frame(root, width=100, height=100) frame.bind("<key>", key) frame.bind("<button-1>", callback) frame.pack() root.mainloop() currently, creating text variable , everytime key() function gets called, , text stores last character typed. you can define text module level variable , , use module level text inside key function - text = '' def key(event): global text text+= event.char print("pressed", text)
Read more

sql server - Why does Linq-to-SQL add unnecessary COUNT()? -

here tables: create table customer ( cid int primary key, name varchar(32) ) create table ord ( oid int primary key, cid int foreign key references customer, address varchar(20) ) here's linq-to-sql statement: var aaa = c in db.customer select new { c, o = c.ord.tolist() }; and here's inexplicable query generated linq-to-sql: select [t0].[cid] [cid], [t0].[name] [name], [t1].[oid] [oid], [t1].[cid] [cid2], [t1].[address] [address], (select count(*) [dbo].[ord] [t2] [t2].[cid] = [t0].[cid]) [value] [dbo].[customer] [t0] left outer join [dbo].[ord] [t1] on [t1].[cid] = [t0].[cid] order [t0].[cid], [t1].[oid] i want understand how rid of count(*) part. it's entirely uncalled for! linq needs count(*) in order determine number of entries returned o = c.ord.tolist() .
Read more