AfterDawn Forums

Periodic lag/freezing and no max performance?

This discussion thread has 39 messages.

#1
Hello everyone how is it going!
First of all, I apologize as I think this is in the wrong section, but I couldn't find a " help section " to post in, so please bear with me as this problem is driving me up the wall and this computer is like 3 days old only.
So I've built a new PC, specs :
CPU : Core i5 3450
Mobo : Gigabyte P61A-D3
GPU : GTX 560 Ti
RAM : 8 GB DDR3 RAM
HDD : 1 TB HDD
PSU : 700 Watts Cougar
So I've been experiencing a periodic lag, which I noticed both normally and in a game ( Crysis 2, which is the only game I've been playing since I got this ). The computer freezes for a sec, then goes on. Also it occured in Crysis 2 twice, once it froze for a sec or two then continuted, the other time the FPS dropped from constant 60 FPS to something like 45 FPS for a sec or two then returned to 60 FPS.
Also one time the computer totally froze and I had to do a reboot from the reset button.
I tried updating the GPU drivers to latest, same problem.
I think something is wrong and I'm not getting the max out of my PC, as it runs Crysis 2 on 50-60 FPS ( and sometimes drops to mid-late 40's ) only on 1280x1024 17'', V-sync on and hardcore settings ( and the benchmarks shows that it runs with better FPS on max settings and full HD, so it means that it must be something wrong ).
Thanks in advance and all help is appreciated.
▼▼ This topic has 38 answers - they are below this advertisement ▼▼
AfterDawn Advertisement
#2
Did you build it yourself/have you any experience self-building before?

Regardless, read this and run memtest86+ first and then Prime95. Also take note of GPU temp when in a game, e.g. Crysis. See how you get on and post back.

Originally posted by Ripper:
When I say "burn in", I'm referring to stress testing your CPU/RAM/GPU initially to make sure everything is stable, seated corrected etc. Following software is recommended:

CoreTemp - monitors CPU temp
GPUTemp - monitors GPU temp

Prime95 (32bit) - load up an instance for every core, and if something is wrong it won't run for very long (use 64bit version if you're building a 64bit machine, obviously)
FurMark - does the same thing, but for your GPU

If you get any weird freeze-ups or BSOD after you complete your build, the first thing you should do is run memtest86+ to check your RAM for faults.



#3
Thanks for the help.
I didn't build it myself, I had a guy in a shop build it for me.
The highest GPU temp I've recorded was 81C ( which isn't too much I think for A GPU on gaming mode on my highest possible settings ).
Also please note that I'm using a merely 1280x1024 17'' monitor, which is what makes the FPS Drops weird, cuz I thought that GTX 560 Ti shouldn't have any struggles in such a PC ( heck, it runs Crysis on Full HD according to benchmarks better than how I run it on only 1280x1024 )
Aren't Prime95 and FurMark a little bit dangerous for components since they stress them to the max? Also, how long should I use them?
#4
No, they aren't dangerous or I wouldn't have suggested them; run memtest and see if your RAM passes it.

Prime95 will generally highlight any heat/stress related errors very quickly, so 15/20 minutes is fine - whilst recording CPU temp. FurMark the same.


#5
Which options should I pick on " run a torture test " in prime95?
#6
Small FTTs and leave number of threads as whatever it is pre-filled with.

I'm suggesting you run memtest first, as it'd be best to eliminate that as a cause.


#7
Ok I couldn't run memtest first because I had no CD's available, so I ran prime95 for around 20 minutes, tests went smooth with no error detections. Was it really necessary that I ran memtest first?
P.S : Also, which options should I pick on FurMark? This is my first time using these tools, so please bear with me.
Thanks for all the help, appreciated.
P.S.2 : Oh btw, should I run the windows or the bootable version of memtest?
Thanks and sorry for many questions
This message has been edited since its posting. Latest edit was made on 30 Oct 2012 @ 10:45
#8
Originally posted by Skorpian:
I ran prime95 for around 20 minutes, tests went smooth with no error detections.
Okay, good.

Quote:
Was it really necessary that I ran memtest first?
No, but random freezing/lagging is more likely to be RAM which is what memtest will test.

Quote:
P.S : Also, which options should I pick on FurMark? This is my first time using these tools, so please bear with me.



Select the correct resolution and hit burn-in test, run for 15-20 mins. Monitor the temp yourself with GPU-Z, the Settings option is a backup if the GPU gets unusually hot. I personally think your 80 degree temp under load is a bit on the hot side but it's not dangerous til it hits nearer 95/100.

Quote:
P.S.2 : Oh btw, should I run the windows or the bootable version of memtest?
Thanks and sorry for many questions
You have to create a bootable disc, use the Windows version:
http://www.memtest86.com/memtest86-4.0a.iso.zip

Burn to CD with IMGBurn or similar and boot from it (assuming you know how to, but if not step 5 onwards here). You basically just let it run, there are a reasonable amount of tests to go through but if you're pushed for time then selecting test #5 should suffice.

If memtest fails or reports errors, power off your PC, remove the RAM and test each stick individually in the first RAM slot.

Edit: If you haven't already, make sure your graphics drivers are up-to-date.
This message has been edited since its posting. Latest edit was made on 30 Oct 2012 @ 11:26


#9
Graphics drivers are up to date.
How much time does memtest take? If it takes lots of hours, I'm cool with leaving it turned on throughout the night if it doesn't take like a day or so.
Again many thanks, I appreciate you took the time to help.
Also, which button should I click ( burn in test, benchmark, etc.. )?
Again many thanks for help.
This message has been edited since its posting. Latest edit was made on 30 Oct 2012 @ 12:42
#10
Burn in test on Furmark. You may also want to try removing your graphics card, and booting up without it to eliminate it as a cause - this is a relatively straight-forward task.

Depends on the amount of RAM you have, a while for 8GB - I can't recall exactly off the top of my head sorry. It is fine (and common practise for some) to run it overnight.

Have you run any virus scans? I know it is a new PC but it wouldn't be unheard of.
This message has been edited since its posting. Latest edit was made on 30 Oct 2012 @ 13:01


#11
I'm completely newbie when it comes to deal with hardware from inside of the PC, so removing the graphics card, straight-forward as it may seem, is a kind of hard task for me. I don't want to damage it either, because all parts are still under warranty and I may be able to replace it if it's faulty.
Does the prime95 test indicate that the CPU is not the issue?
Yes I've made a complete virus scan.
Is it normal for the FPS to dramatically drop down all of a sudden in a game? Like I've been playing Battlefield bad company 2, it drops to around 30-40 in intensful scenes ( explosions etc, that's normal isn't it?). But suddenly, in an area of no intense at all, like walking or a few bullets, the FPS dropped to like 15 FPS for a sec or two then went up. That's not normal, is it?
Thanks again for all the help, I'll run the furmark test later on today and will report back ( because I can't run it now as I have to go ). Also will try to run the memtest over the night.
#12
FPS fluctuating like that sounds normal enough; I don't game on PC at all though.

Quote:
Does the prime95 test indicate that the CPU is not the issue?
Yes, if the CPU was at fault it would likely overheat and shut the PC down.

To be honest, if your machine is indeed still under warranty then your time is probably better spent taking it back to him and asking to perform said tests on your components himself, or ask for his opinion.

See how you get on with memtest anyway.


#13
Sorry I'm late but was busy was the past couple of days and couldn't do the furmark test.
I did the test but the temperature reached 85C after only 3 minutes ( which made it produce the beeping alarm so I closed furmark) and the fan speed never exceeded 53%. What does this mean?
This message has been edited since its posting. Latest edit was made on 02 Nov 2012 @ 9:47
#14
Bump
#15
Had a busy weekend - have you run memtest yet?


#16
Yes, memtest ran smooth without errors.
So I got a couple questions to sum everything up :
1-The GPU reached 85C after only 3 minutes of furmark testing, and the fan usage never exceeded 53%, does that mean that something is wrong with the GPU/GPU fan? ( I turned off the test when it reached 85C )
2-I have got 8 GB 1600 MHz ram but my mobo supports only 1333 MHz, could that be causing issues?
#17
Good that memtest ran smoothly.

Quote:
1-The GPU reached 85C after only 3 minutes of furmark testing, and the fan usage never exceeded 53%, does that mean that something is wrong with the GPU/GPU fan? ( I turned off the test when it reached 85C )
Temperature wise, under full load (in an environment running Furmark provides), that's probably alright to be honest. They should run safely up to 100C or so before cutting out themselves due to safety - see this. I assume the GPU fan is running correctly? An easy way to check would be to have the side of the case off and just look.

Quote:
2-I have got 8 GB 1600 MHz ram but my mobo supports only 1333 MHz, could that be causing issues?

Not ideal but it shouldn't be a problem as your motherboard will have clocked it down to the fastest speed that it supports - 1333MHz. What RAM is it specifically?

Are your graphics drivers up-to-date etc?
This message has been edited since its posting. Latest edit was made on 05 Nov 2012 @ 10:59


#18
I think I might return this RAM and get a 1333 MHz ram instead since this is a waste of money. The Ram is Kingston KMX1600C9D3/8G.
Yes, I updated both the graphics card and motherboard drivers
About the fan, what I am asking is if it's normal for the GPU fan not to work more than 53%.
#19
Yes you may as well be using 1333Mhz RAM unless you plan on upgrading your mobo any time soon.

The fan should spin at more than 53% speed, yes - perhaps something to mention when you return your RAM as it obviously sounds like a graphics related issue if RAM is not the problem.

Let me know how you get on; unfortunate we couldn't get to the bottom of it here, maybe someone else will weigh in.


#20
Hmm, I discovered that windows gadgets increases its ram usage as the time passes. Once, I discovered after the computer has been running for many hours that it was taking up 4.5 GB of ram! I closed it from windows task manager but still same problem, however does this mean that the windows install could be corrupt ( it could be a software not a hardware issue)?
This message has been edited since its posting. Latest edit was made on 07 Nov 2012 @ 6:08
#21
Yes it could be a software issue; I have no idea about Windows gadgets but obviously they shouldn't be using 4.5GB of RAM.

I assume you are running 64bit Windows, as you have 8GB of RAM installed?


#22
Yes I am using 64bit windows.
#23
Then perhaps you could try a fresh install.

Before doing that, go in to BIOS and you should be able to check the SMART status - and therefore health, generally speaking - of your HDD, or if not use an app such as http://panterasoft.com/hdd-health/


#24
What does this smart thingy do?
#25
checks your hard drive out somewhat if going bad.
This discussion thread has been automatically closed, as it hasn't received any new posts during the last 180 days. This means that you can't post replies or new questions to this discussion thread.

If you have something to add to this topic, use this page to post your question or comments to a new discussion thread.