Overclocking in progress....Predator 17 (G9-791)

ChevyCam94
ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
edited March 2023 in 2016 Archives

So I decided to start messing with overclocking this guy.  I have the 970 with 3GB VRAM.

 

My initial Unigine score was 1102.  That was BEFORE I changed anything.

 

After the following tweaks (this is my system verification) I was able to bump that score up to 1391.

 

https://www.techpowerup.com/gpuz/details/4a6p9

 

Im using GPU Tweak as my app for setting speeds, and voltages.  And GPU-Z to monitor temps and loads.

 

 

With the fan at regular speed, the system went from a normal of 122 (F) to 129.2 (F).  If I turned the fan to MAX and benchmarked, it only reached 113 (F).

 

It seems this has potential, but I need to do more testing.  Im sure this isnt the furthest it can be overclocked, but Im playing it safe right now.

«1

Answers

  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon

    Screw it - Thought I'd run some BASIC tests myself for someone else to compare (Chevy - love to know how gtx 970m does tbh)

     

    If you'd like to compare - I ran Fire Strike 1.1 full default test via the 3DMark Demo on Steam.

    GTX 980M i7-6700HQ - ACER PREDATOR 17

     

    Normal no-overclocking results - gpu @1038MHz, memory @1253MHz and boost @1127MHz:

    http://www.3dmark.com/3dm/13620650

     

    I got a total score of 8366 ^(see above link for more details) - Funny because TechRadar got 8320 on their review.

    GPU temperature via MSI AfterBurner reached a (brief) maximum of 67°C on default automatic fans (3500+ rpm or something right?) - not "always on" mode.

     

    Fun added variables you could do the same test with would increase maximum fan speed through Acer Quick Access' "CoolBoost" - which lets the fans run to 4500 rpm I believe. And you could even add the extra frostcore fan that comes with all acer predators.

     

    Overclocking of +5% results - gpu @1090MHz, memory @1284MHz and boost @1179MHz:

    http://www.3dmark.com/fs/9509592

     

    Total score of 8443. GPU temperature via MSI AfterBurner reached a very brief maximum of 67°C again, on default automatic fans. It hit 67°C just before fading to black lol - technically better than before Very interesting how this had no real effect on the temperature.

     

    Overclocking of +10% results - gpu @1142MHz, memory @1315MHz and boost @1231MHz:

    http://www.3dmark.com/fs/9509829

     

    Total score of 8686. That's a bigger jump than from 0 to 5%. GPU temperature via MSI AfterBurner reached a lower maximum of 66°C on default automatic fans. So my tests show a decrease in temps the higher you clock...lol. Maybe ambient room temp was dropping ever so slightly. Or the laptops fans were working harder. A reason I bought this laptop was because it had a great cooling system, this helps prove that fact for me imo.

     

    Comparison between all results:

    http://www.3dmark.com/compare/fs/9509829/fs/9509592/fs/9509366#

     

    At 10%, it seems, according to Fire Strike 1.1, there's a minor gain of 2FPS. Not sure if that's worth the supposed extra strain? Do actual, real games running, have a better performance increase? Any examples? Be great to know! Considering CoolBoost wasn't even on and temps stayed stable; there's definitely more to gain should one choose to. Wonder if at full fans and maximum safe clock, if it can reach the desktop Gaming PC (HTC Vive/Oculus Rift minimum spec) GTX 970's score of 9271?

     

    This guy got close without any issues, so i'm sure the predator can perform on the same level if not better and get a tangible improvement of 10% extra horsepower in FPS when gaming https://forum-en.msi.com/index.php?topic=250531.0

     

    Note: The clock speeds that Fire Strike registered on its tests (links above) aren't the same as what GPU-Z reads (which is what I've written) when I increase them via AfterBurner. I don't know why. But the numerical % increases in clock speeds are correct nonetheless.

     

    I haven't come across any games that I'm unhappy with performance wise. So probably will leave it at stock for time being. Very interesting though and happy with how this laptop performs!

     

  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon

    You cannot alter the voltage on the 'm' series. That's where any potential damage in increased wear and tear would stem from anyway.

     

    I also did the Unigine test:

     

    Default

    FPS:
    46.1
    Score:
    1160
    Min FPS:
    8.2
    Max FPS:
    101.4

     

    + 10% overall clock increase

    FPS:
    48.7
    Score:
    1228
    Min FPS:
    11.7
    Max FPS:
    104.0

     

    Settings

    Render:
    Direct3D11
    Mode:
    1920x1080 8xAA fullscreen
    Preset
    Custom
    Quality
    Ultra
    Tessellation:
    Extreme

     

    What settings were you using Chevy? And what temperature reading are yours of? CPU, 'SYSTEM' OR GPU? Unless your overlcock is severe how would 970m best the 980m!? Mine was set to extreme except for obvious exclusion of multimonitor & 3D. Essentially mimmicks my findings above of 2-3 increase in fps with what appeared to be minimal if any temperature increase. I honestly forgot to pay attention to temps of GPU (64-65 range) but I think it didn't alter between different clock speeds. like before (maybe by 1 celsius)..

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    Honestly I don't remember offhand. We should all agree on a specific benchmark app, and monitoring software, that everyone can use, so our control is the same.
  • beststevie
    beststevie ACE Posts: 506 Pioneer

    with my last notebook with 970M, too, I also did such crazy things like loading an unofficial Vbios. It worked great wiht adjusting the v-core and performance was very similar to 980M. But I think the bigger difference in most games I play is the bigger Ram of the 980M.

     

    cheers,

     

    steve

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    I have yet to fill more than about 2.3GB of VRAM. But I'll set up a baseline benchmark setup for us tonight.
  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon

    I'll do the Unigine test with the same setting you have squishymuffin.

     

    Also, how are you getting more than 60fps?  I cant seem to get it to get any higher then maybe 61.3fps.  Its like there is a framerate lock going on.  I even put it on the lowest settings, and still, locked at 60fps max.

     

    Edit:  vsync is killing my fps.  BUT, it looks like ***** with vsync off.

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon

    Unigine 4.0:

    Render:  DirectX11

    Mode:  1920x1080 8xAA fullscreen

    Preset:  Custom

    Quality:  Ultra

    Tessellation:  Extreme

     

    Vsync:  OFF

     

     

    Stock settings:

    Memory Clock:  2505 MHz

    FPS:  35.1

    Score:  885

    Min:  7.6

    Max:  74.7

    GPU Temp:  116 (F)

     

     

    +12% increase (memory clock only):

    Memory Clock:  2800 MHz

    FPS:  37.2

    Score:  936

    Min:  11.9

    Max:  79.3

    GPU Temp:  122 (F)

     

     

    +16% increase (memory clock only):

    Memory Clock:  2905 MHz

    FPS:  38.3

    Score:  945

    Min:  13.6

    Max:  83.5

    GPU Temp:  125 (F)

     

     

    None of those were done while adjusting the core clock.  This was ONLY adjusting the Memory Clock.  Any higher, and I began to feel uneasy about it.  So I called it at +16% max.  Next round, I'll adjust the core clock also.

  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon

    Those are interesting comparative results. Thanks. Not sure if I should be concerned on why your temps are so much cooler than mine though?? Mine reach up to 65C (150F) when gaming/doing these tests. It's definitely GPU? Are you running full power fans? As in on all the time and at highest rpm using that coolboost feature?

     

    Edit: Did research (interesting like below) as soon i posted this and I think my temps are completely fine http://laptopmedia.com/highlights/acer-predator-17-vs-alienware-17-r3-detailed-stability-and-temperature-battle-report/

     

    "NVIDIA GeForce GTX 980M runs at 67°C while the same GPU in the Alienware enclosure reached 73°C with the same default clock speeds." It's intesting to note the Predator doesn't do as good a job with the CPU though; "the Predator CPU temperature had already reached ~90°C while Alienware 17 remained around 70°C"

     

    Guess it's just the difference in GPU between 970m and 980m? (EDIT: it's not) Be good if someone else on here can give us some field research of their own Smiley Happy Mine is actually on a cooling pad; as in it's elevated above the table somewhat. No extra fans used though. No aircon, 7f building, probably ambient normal temps.

     

    I just noticed on my Unigine tests above I mentioned 'overall clock increase'. In that I mean memory is also included. Can't edit it for some reason.

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon

    Strangely, I tried to increase the core clock also, and it made my scores LOWER.  VERY weird.

     

    Not sure why mine is cooler than yours.  Could be ambient air temp.  My desk is directly under a ceiling mounted vent.  And the AC is on.  Plus my desk has a glass top, so that adds to the cooling effect.

     

    Im using GPU-z to monitor my temps.  I set them to display the MAX stats for each section, so I dont have to keep looking at it, I can just check it when the benchmark is finished.

     

    Also, just a suggestion.  Go into the XBOX program, and under Settings -> Game DVR, DISABLE IT!  That should give you a little boost in performance, now that you dont have the extra overhead.

  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon

    I'll check those xbox settings. Yes I think your circumstances definitely aid cooling lol. Are you running full power fans? - As in on all the time and at highest rpm using that coolboost feature? 

     

    But to further add to the calming of the nerves, notebookcheck and their 970m model on a 30 min load benchmark shows my temps are good (*phew):

     

    "Neither the processor nor the graphics card can utilize their full performance in the stress test (Prime95 and FurMark for at least one hour), even though the temperatures are not critical at 84 °C for the CPU and 69 °C for the GPU. The processor is probably limited by the TDP and runs at 2.8 GHz, while the GPU cannot quite maintain its nominal clock of 924 MHz at just 885 MHz. A 3DMark run immediately after the stress test did not determine a performance reduction." - Room Temperature 22 °C

     

    Max load: (surface areas of laptop):

     
     37.5 °C41.7 °C41 °C 
     30.9 °C41.7 °C38.8 °C 
     23.8 °C29.2 °C30.2 °C 
     
    Maximum: 41.7 °C
    Average: 35 °C
     
     41.6 °C41.3 °C41.1 °C 
     38.5 °C37.3 °C25.6 °C 
     28.4 °C26.4 °C24.2 °C 
     
    Maximum: 41.6 °C
    Average: 33.8 °C

     

    See more: http://www.notebookcheck.net/Acer-Predator-17-G9-791-Notebook-Review.158933.0.html

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    Hmmmm. Might have to rethink my temps. I thought they were getting a bit high
  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon

    Maybe used to desktop temps? Everything I've found points to you, even me, being below the norm. Even then, there's a good buffer before critical temps arrive - which is what? 85C plus? (I don't know) my previous machine would shutdown if and when it reached 90C. And 100C is maximum allowed operating temperature I believe.

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    Yeah, that could be. This is by far the most powerful laptop I've ever owned. I should expect more from it. Lol
  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    Is there a good, FREE app that allows easy setting of the speeds? GPU Tweak always wants to crash when applying settings. If I'm LUCKY, 1 in about 30 attempts actually works.
  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon

    You referring to increases clock and memory speeds? I've only used MSI Afterburner (has good in app/game monitoring of fps,cpu,mhz,gpu and temps of all.)

     

    From laptopmags tests "The GPU ran at around 67-68 °C, which is considered to be extremely low since most notebooks go as high as 78-79 °C." http://laptopmedia.com/highlights/how-acer-predator-g9-791-notebook-deals-with-the-extra-heat/

     

    So we now have a good healthy range for anyone who wanted to know...lol.

     

    EDIT: what has greater effect on fps? Memory or Clock speed? Or both equally?

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    I'm seeing a bigger benefit from memory speed increases. Need to do more testing though
  • SquishyMuffin
    SquishyMuffin Member Posts: 30 Die Hard WiFi Icon
    FPS:
    51.4
    Score:
    1296
    Min FPS:
    17.5
    Max FPS:
    112.7

     

    Same benchmark settings on Unigine. The clock speed was set to the max of +135mhz on afterburner (gpu-z 1173mhz) and I stupidly didn't note down the memory. I think it was 1492mhz gpu-z total. I think I noticed artifacts, so I wont actually ever use these settings. Be interesting if it was the max clock speed or increased memory that caused it.

     

    Fire Strike with same speeds as above reached a score of 8875. Again, I wouldn't use the settings for stability reasons. So somewhere just below is optimal I think.

    http://www.3dmark.com/fs/9535876

     

    Temps, again, didn't really change.

  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon
    Nice. I'm going to try these settings with vsync ON, because let's face it, crazy screen tearing sucks the fun out of everything.

    Interestingly, if you go to the NVIDIA control panel, and select the list of things you can change for 3D settings, there are a few missing.

    One is under vsync. Don't click on it, just hover the mouse over it, and a description shows up under the list of what the different options do.

    One of them is "fast". Basically, it's supposed to buffer at MORE than 60fps, but drop excess frames and display ONLY those matching the 60hz refresh rate, and supposedly not tear.

    I only see the options for " On" and "Off".
  • ChevyCam94
    ChevyCam94 Member Posts: 178 Skilled Fixer WiFi Icon

    OK, so I did a little more tweaking, and I think I found the MAX that my system feels comfortable at.  Any higher memory speeds, and I began to see artifacts.  On my second round of 3000+ memory speeds, the system locked up on test 17 of 26.

     

    Remember, this is with vsync ON!

     

    Here are my results from what I would consider to be my max system boost.

     

    https://www.techpowerup.com/gpuz/details/xupw

     

    Unigine results:

    FPS:  35.5 (stock was 26.5)     +9 fps

    Score 895 (stock was 668)     +227

    Min FPS:  18.5 (stock was 14.0)     +4.5 fps

    Max FPS:  60.9 (stock was 56.9)     +4 fps (capped at "60fps")

    GPU temp:  128 (F) (stock was 122 (F))     +6 (F)

     

    Also, I AM using CoolBoost, but I am NOT using Max Fan.  My power profile is custom and set to CPU min @ 1%, and CPU max @ 100%.

  • beststevie
    beststevie ACE Posts: 506 Pioneer

    Doing benchmarks for such things have shown some facts to me like:

     

    the older the benchmark is, the less is the difference between the old and the current system

     

    This might depend to the not updated benchmark software which does not support the current features.

     

    So I would ask you to do an alternatve, a current benchmark as well.

     

    Such as Firestrike and Timespy would be appreciated.

     

    But also congrats for your hard work Smiley Wink