M17X FPS issues

Hey guys,

I've been playing some of my games lately (as one should on an Alienware) and I've noticed that my FPS has plummeted compared to usual. Mostly I've been playing League of Legends, where I used to get 60 frames every time on max settings, now I'm lucky to get 60 on the lowest.

I called Dell Tech Support about this a few months back, where they advised I restore to factory defaults. I did as such and the issue went away, until a month ago. My issue is back and worse than ever, I play League of Legends and usually sit under 40 FPS, if more champions show up on the screen it gets lower. The lowest it has gotten to was 12 FPS. I called Dell Tech Support once again, where the guy had me run a bunch of tests via Alien Autopsy, and the results showed the PC running fine. He said my issue was "solved" (quotations because clearly it wasn't), closed my case and was then done with me.

Sure enough, the next time I boot up my PC the issue happens again.

Looking at my test results for the Gaming Test, is there a standard somewhere that the PC needs to hit? I ask this because on the DX11 and DX10 Shader Rendering Test I never actually hit 60 FPS. In fact, the DX10 test was at 40.

Has this happened to anyone else? I only got this computer in January, I'm a little disappointed with what I have here.

Edit: My bad guys, it's an M17 not M17X

 

Hi, sorry to hear your problems .

Could be a bunch of things but it's not too difficult to identify the most common ones.

One common one is poor pasting from the factory. I've seen this on one of my own AW's as well! To find out if it is getting too hot and throttling down the GPU you need to monitor the clock speed and temps while gaming. Find out the base clock speed for your dGPU and make sure you are achieving it.

I use msiafterburner, not for the overclocking but it has an on-screen component and can be set up to display stats while you game. Set it to show FPS, clock, temp, and voltage 'in OSD'.

It might also be the graphics switching (optimus) not working correctly. This means you are running on the built-in intel GPU which is far less powerful. If this is the problem then I believe you can test it by forcing the dGPU using a hot-key, Fn F7 - I/D GFX on mine - although it does nothing on my 120hz model so I'm not a lot of help if that is the problem.

As for the dell rep, sounds like you got a poor one. The last time I called on a similar issue he remotely connected and run the heaven4 benchmark in a window. He saw it getting hotter then it throttled crashed. New card and heatsink next day! If you refuse to allow them to close the call he HAS to escalate to a next level where hopefully you get a more experienced tech.

For checking out what you should be seeing I prefer to use in-game benchmarks and see what others are getting. The last two batman games age very good testers of performance. 3Dmark is commonly used for comparisons, I just prefer to stick to how it games .

What 'gaming test' and GPU are you using?

Good luck.

 

I'll definitely have to run msiafterburner, I forgot about it as I was using it to test prior to my first factory reset.

I've already checked the I/D GFX button and it is definitely using the proper GPU, which on mine is actually Fn F5.

I thought I had said something in my original post but going over it again I must have erased it, the gaming test I'm using is the one built into Alien Autopsy, and my GPU is a GTX 770M.

 

At least the switching is OK so we can forget about that .

Now it's a case of making sure the GPU is functioning at full speed. NvidiaInspector will also show you the 'P' (Power) state. P0 is max and this should be the setting while gaming. It will throttle down to the minimum (P8 on mine) as windows is not stressing it. P1 is the 2d/first throttle state - and enough to ruin a game session .

Have you seen this problem on any other games? I haven't played LOL but I would expect your GPU to eat it up.

When a throttle occurs you will also see a voltage drop which is a good indicator.

 

The minimum specs for League of Legends are as follows:
2 GHz processor
1 GB RAM (Windows Vista and 7 users will want 2 GB of RAM or more)
8 GB available hard disk space
Shader version 2.0 capable video card
Support for DirectX v9.0c or better
Windows XP (Service Pack 2 or 3 ONLY), Windows Vista, or Windows 7

You could certainly see how an Alienware would far surpass those and easily push to max graphics.

 

Yup, You should easily get 60+ fps on max settings. I didn't think it was a 'Watch Dogs', that game hits hard. Sometimes though a game just decides not to play nice. Updating/rolling back the video driver can help but lets look for a solid cause first .

Anyway, have you seen any numbers or temps?

 

I just played a match at max settings, floated around 30 fps the entire game, until fights including the entire team broke out. The file generated was a .hml and I don't have anything to open it, do you have a suggested application?

Edit: Never mind I just opened it up with Notepad++.

The GPU temperature never exceeds 52 celsius, my memory clock jumps to 2000 a few times throughout the match, and the core clock jumps to almost 500 at those same times. The CPU temperatures aren't an issue either.

 

Afraid I don't. I only use the realtime numbers (write them down ) to see how it is doing. What temp? Any clock speed changes?

Do you have any more-recent games to test on it?

 

The CPU and GPU are about the same in temperature, around 50ish celsius. My FPS never exceeded 30, I'm about to do another match now. I'll test with a more recent game afterwards

 

I'd expect it to be a lot hotter in-game. Get the on-screen-display showing as you play. 50ish is idle temps

 

Are you on power saver mode power plan? Sounds like the CPU is not going up to 100% and turbo mode. Maybe clocked at <1GHz ?

My i7 at 0.8ghz will make a game give low FPS but temps stay low as a result.

 

My laptop is always on High Performance, never been in power saver mode (I always have it plugged in).

I also have the OSD going, and the temperature results say the 48-52 range for my temperature.

 

Do you have the clock speeds showing as well? What is your CPU at?

And GPU clock speeds?

 

Way too low for either CPU or GPU!

Something is not operating at full speed, that's for sure...

Same questions as DDDennniZZZ

 

Sorry there was a tornado warning in my area last night so I just hopped off ASAP!

The only clocks I have listed here are Core and Memory clock, and they are as follows
Core clock: 483.387 throughout the entire match and occasionally dropping below that (it never went higher)
Memory Clock: 810 through the whole match, with several spikes to 2004.750, when the Core clock would dip down below 483 it too would drop to 406

RAM usage was between 4-5000 for the entire match as well if that matters.

 

I don't know the default clock for that card but I would expect GPU numbers in the 600-800mhz range.

Get heaven4 and see if it runs at the same slow speeds: Heaven Benchmark | Unigine: real-time 3D engine (game, simulation, visualization and VR). Run it with afterburner OSD Just to rule out a game issue. Have you updated your video drivers?

I guess the temps are still low - under 65c?

 

Yeah they don't even reach 60.

I just played 10 minutes of Saint Row 4 on max graphics, this time the GPU temp hit 60 and the memory clock stayed at 2004.750 for a solid 5 minutes, otherwise it went between 405 and 810. The core clock however bounced between 135 and 483.387 during the loading screen and stayed at 483.387 during actual game play.

I'm going to run the Heaven Benchmark now.

 

Yeah, we might be homing in on it. If you get NvidiaInspector ( NVIDIA Inspector Download version 1.9.7.2) you can see what your core clock is supposed to be (default).

You can also try a workaround and in the driver control panel select 'Max Performance' for Power Management.

Forget the memory, that looks fine, it's the GPU clock that is at question right now .

 

On highest settings for the benchmark my PC got a very depressing score of 364, and my FPS never exceeded 30. Here's the screen cap from NvidiaInspector:
http://img.techpowerup.org/140728/nvidia_20140728_081726.png

Edit: I just ran the benchmark again on the lowest settings and got a score of 1021, where my average FPS was 40.5.

 

Thanks or the cap, that's what I needed to see. You should see a minimum of 706 clock when gaming and the boost feature will extend this to 797 (estimated) as long as temps are low and power draw is within limits. In windows the clock WILL drop to P8/135 but should climb to the max (p0) as soon as you game on it.

Setting max power preferred should also keep the card at P0.

I think we have removed the game from the equation now so it's 'things to try'.

Max power preferred in NV driver (also make sure any windows power saving is off). Test

Install new driver. Test

Remove and reseat video card. Test.

Stay with Heaven now that you can see the problem, you'll get more consistent testing

Edit: Run heaven in a window and see what P state nvidiainspector shows, with the benchmark the active window.

 

Still havent answered the clocks of your CPU. Get CPU-Z or HWinfo 64 to check CPU clock speeds, sounds like its the bottleneck to me.

Check your power profile (check the advanced settings), and make sure the maximum processor state is '100'.

 

I have done these two steps with no success, I can't remove and reseat my GPU without voiding my warranty (as it is a laptop)

I ran the test again and my GPU temperature went up between 60-71 celsius over the duration of the test. The P state was constantly at P0.

As for what you're requesting DDDenniZZZ, I downloaded CPU-Z and see no drastic fluctuations, I'm not entirely certain what it is I'm looking for out of the program.

 

Opening up the machine will not void the warranty. One big plus with Dell, their reps sign off with a clear statement that you can open it up and even change parts yourself - but of course return it to original for support.

Was the clock speed fluctuating during the test (afterburner OSD) ? That temp sounds more 'normal'.

We are now at the point where dell may be useful. You call them up with an initial problem of games running slowly'. When asked, you have already factory defaulted and checked temps which are low. Invite them to connect and show the heaven benchmark (in a window or they will loose visibility) and its miserable scores and low temps(let's just confirm them, past a screenie of the Heaven4 settings and I'll see what number I get for evidence).

Let them figure it out at this point?

Good luck and if the P0 state and full clock are maintained during the benchmark then it could be the CPU throttling or something else altogether (game not playing nice with your setup?) .

Edit: Do you use v-sync? If so turn that off .

 

Here's the picture of the benchmark results, it doesn't show the temperature with it, which I find strange but oh well.

http://i.imgur.com/idJcoPm.png

 

You'll only get the temp in combo with afterburner's OSD - wrong, I mised it, just note the max. I'll give your settings a quick run and post back..



Numbers for full screen are 811 @ 32fps. Min 14.7 max 75.9.

 

Hmm does seem a bit weird, try HWinfo 64, once you run it, press the sensors button, and scroll down the list of sensors until you see the CPU clock speed. Or look for CPU clock speed in CPU-Z. I would expect around 3500MHz for your CPU to get up to at least.

I have a feeling your CPU is being a bottleneck as its not running at full speed.

Intel extreme tuning utility includes a CPU benchmark test you can use (it only takes a minute or two I think). And you can compare it to others with the 4700MQ.

GPU looks like its fine for me, especially if the CPU is not running at full speed.

 

The weirdest thing that's happening is that it'll be fine one minute and bad the next. I shut down to head to class (doesn't start for another 30 minutes), tried another game of League of Legends and it ran flawlessly on max settings. Mind you that's after a reboot, but this is something that happens from time to time. It happens several times a week, and isn't always solved by a reboot, but it happens often enough to be a giant pain.

 

Updated post above

 

Mine wasn't even full screen, when my computer starts acting up again I'll run it in full screen with the same settings again.

 

I downloaded the Intel Extreme Tuning Utility and ran the benchmark and got 693 marks, then ran both the CPU and Memory stress tests and passed. Is there anything you would like to see from the program?

 

Does the stress test indicate any thermal throttling? It will show it on the bottom right hand side if I remember correctly. You can still pass the stress test but have thermal throttling kicking in if your paste isn't working as well as it should. I have seen the test pass but still be dealing with thermal throttling.....you would think they would classify that as a failure.

 

Right now even when not doing a stress test (I just finished one) it says 100% thermal throttling. I assume that's a bad thing, when I highlight it it says "Current 100%"

 

Overheating CPU, check the heatsink fins (behind the fan) is not clogged up, otherwise it might be time for a repaste. My 4900QM with 'Dell' paste throttled as well. You either need to clean, or if it appears clean then a repaste (not as scary as it sounds either).

 

So when Intel Extreme Tuning Utility says it's average temp is 47 Celsius (116 Fahrenheit) and thermal throttling is at 100%, that means that the CPU is waaaaaaaaaaay too hot? I know GPUs can safely go to high temperatures, I wasn't sure if this was the same with CPUs or not.

 

It shouldn't be throttling at such a low temperature. Maybe artificial throttling. What did it say for the maximum clock speed?

 

I'm not sure where to find that, if you could point me in the right direction that would be lovely.

 

Well thank you guys for all of your help, but I got off the phone with tech support last night, and with the help of all the troubleshooting you guys helped me through, they've finally decided to send me an on-site technician (part of my warranty).

He's fairly certain that there's something wrong with my GPU, and currently I'm having a power adapter issue as well where it's not charging, so he's sending parts for that as well.

Once again, thank you guys very much!

 

Good news.

If the power adaptor is reporting not charging then this could well be the whole problem! The GPU card will throttle if the power adaptor is not charging (or reporting not charging).

 

Check your power profile. See if maximum processor state is set too 100. Then by using HWinfo 64, see what your maximum clock speed is. Run a virus scan or something to stress the CPU.