Clevo Overclocker's Lounge

On the delid, the original article I read said the first was using an h110i, the second using the Kraken, so not all may be attributed to the delid, although a large portion would be...

Edit: After checking the original forum post, he did clarify it was the same cooler, meaning almost 30 degree difference solely from replacing the TIM.

Sent from my SM-G900P using Tapatalk

 

Did you check this post out?
http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-721#post-10403967
Your issue is not EC. It's temps or fps limited somewhere...
15335 Is an excellent score for you clock. 4.4 ghz
GT1 is low by 3 frames
GT2 is up by 1 frame.
112.66 seems to be frame limited at 120 fps. That's what use to happen to me from time to time. Actually it's going on right now on my main drive. That's why im running the sandisk SSD drive to verifiy it.
This is off that drive. I get a very low cpu score, but all i did was run as is.
Gpu score is above yours in test two. That's why I believe you are frame limited somewhere. You should be around where I am for that clock
i used 150/400
http://www.3dmark.com/fs/11122290

 

+15330 in FS physics with only 4.4GHz is difficult to explain. A very strange high score for 4.4GHz With a linear scaling would FS physics score been closer to +16750 for 4.8GHz

 

Well... The 15,335 physics score is actually from a little "trick" of using throttlestop & hotkeys to run 4.4GHz during GPU/CPU tests for minimal throttling. And during the physics test bump to 4.8GHz because the EC doesn't throttle me since the GPU isn't loaded, and then back to 4.4GHz during the combined test. I can run 4.8GHz stable, but not while the GPU is loaded. That was why it took so long to achieve that score, finding a good balance of max GPU clocks at minimal voltage I could run while maintaining some CPU overclock. Now the "secret" is out (if no one else had previously tried this) you can try it on yours and run the physics portion at 5+GHz for a higher overall score!

I "can" run the entire set of Fire Strike tests at 4.8GHz, but the throttling drops the clocks too much and reduces the overall score a lot.

@Johnksss@iBUYPOWER -- Yea I had seen your post and that was actually led me down the path of reducing voltage for the GPU, and therefore "power" (wattage) thinking that the EC was primarily keyed into the CPU/GPU wattage values to cause the throttling when a certain threshold was passed. And it ironically performed just as expected, with lowering the GPU voltage and even increasing clock over my previous runs I was able to achieve higher scores with less throttling.

Will be interesting when I get the new BIOS from HIDevolution to see if it truly is EC throttling as expected or something else. In the meantime I'll try shortly as you did in that post to see if the clocks hold running the GPU-Z render test

 

I know. I wanted be diplomatic The old trick @Mr. Fox had to do because Dellienware and Mr. Azor find it useful to crippling latest AW18 with a max cap on MB or firmware. Max 330w or there around.

 

So tested this on mine... Mine will stick the 2063 clock I ran in the Fire Strike bench without issue up to 55*C, using max fans and my cheapo laptop cooling pad. And with the max fans running the GPU-Z render test it's stable and won't downclock at all while constantly running at 51*C.

Spoiler

If I turn off my cheapo laptop cooling pad, and go to auto fans when it warms up to 55*C it'll downclock, but only to 2050... And the hottest it would get with the GPU-Z render test was 65*C where it stayed at 2050MHz. All that being said it would still drop lower than 2050 during the Fire Strike run when GPU/CPU was loaded at times....

Spoiler

 

Had not seen that, nice pictures and info

 

CCC is broken af.


I installed windows, fresh install, proceeded with all drivers install, downloaded latest CCC from Clevo site which was updated recently for P775.

Gave it a run, windows shut down (not crash, it actually did a shut down, What?!)

Tried to tinker with settings, gave it a run on Cinebench, shut down again.

Uninstalled CCC, restart, now it runs normal, 14 hours without a shut down, tried forcing the laptop at max limits using madVR and Cinebench, all works good.

My keyboard, on the other hand, is blue and nothing can turn the lights off without CCC, which is pretty bothersome. Temperatures and nosie levels are WAY better without CCC.

As @Mr. Fox pointed out, CCC is pure garbage, it is the worst try at controlling an EC and a power Bios ever made. From time to time, when there is a very high load, I get a notification that my brightness was changed even tho I didn't touch it o I guess something of CCC is not gone yet. I'll most probably reinstall windows clean, no CCC this time, everything should work fine.

Nvidia drivers - .03 thingy is still the most stable. Anything newer is broken and garbage, it crashes madVR when running in full screen if not exclusive, causes driver to disconnect, needs a windows restart to fix.


BTW, really interested in 7700K, can't wait to buy one! There's none available in Romania right now but as soon as they arrive, Imma get one as it seems my CPU unit in particular isn't the best out of it's batch when it comes to temps or OC abilities.

 

All I can say is I tried...
I do not want to argue your valued opinion on it so I'm going to stop there.
Let me know when your able to do this.
And this is with your 1.05.06 EC And Lower performaing 3dmark UI. As @Papusan puts it.
http://www.3dmark.com/fs/11123180

 

Same problems with the older .50 for DM1 also? Not that I want you to install

Edit. @Johnksss@iBUYPOWER UI 2.2.3509_64 is the fixed 3DM version

 

Further follow up on downclocking due to temps on the GTX 1080 with my custom freq/voltage curve... Run on auto fans without the laptop cooling pad on.

Had to run AIDA64 GPU stress test to get the temps higher, tried 5x instances of GPU-Z simultaneously but it still wouldn't go higher than 65*C!

2063MHz for 54*C and less
2050MHz 55 - 66*C
2038MHz 67 - 76*C
2025MHz 77*C - ??*C (77 was hottest it was probably going to get as it was starting to level out)

 

Now look at my temps

First mobile chip ever to reach 2.2ghz


Just in case you were thinking that slider made a difference.



No way is it going to hold that with the power table we have now, but it is definitely possible should that time come.

 

Maybe it's just because I have the latest NVIDIA driver 376.33... But the best I could get (without "spoofing" the clocks up on the physics test) was 18,762 with 45x4 -50mV, 42x Cache, & +210/+500 GPU. You can see in the 3DMark results the CPU throttling



http://www.3dmark.com/fs/11123544

 

Actually no

The old .50 version worked just fine, I needed a fresh install of windows for something else for dev reasons, it was mostly super stable.

Don't install SVP if you plan to use madVR later on, it breaks video decoding and no amount of DDU or Nvidia reinstalls will fix it. It's nothing that big, but madVR crashes from hour to hour, it's still there and annoying tho.

 

That's awesome! And make me 2nd to join the GTX 1080 mobile 2.2+GHz club!

 

Maybe try with a higher undervoltage + lower cache to 4.0 and see how it goes. Even with 4.4GHz. Higher physics score doesn't do very much to increase overall total score in Firestrike.

 

Ironically was just doing that!!

Went from 45x -50mV to -70mV... And score up to 18,933, so highest one I've gotten yet. And without bumping to 48x during physics test still. I did bump the GPU core from +210 to +215 also (+500 RAM). I'll try dropping cache to 40x and see if that helps more

http://www.3dmark.com/3dm/16762862

 

I'll have to actually really test that then.

Edit: Just seen your new post

 

And go for max undervolt

I thought you already have tested the last UI
http://www.3dmark.com/fs/11123180

Edit. Or maybe I missed. You meant CCC .50

 

I'm trying to help afloyd. So i loaded up what he had on his machine.

See if you can get validated at 2.1 ghz @afloyd
http://www.3dmark.com/fs/11124128
Only run test 1 and dont worry about being above 4.6 ghz since that can run on stock voltage.

 

And people in the BGA camp will be very happy with the successor of the low budget Trash Skylake BGA 6700hq http://laptopmedia.com/news/exclusive-first-intel-core-i7-7700hq-benchmarks/ @Mr. Fox What a JOKE!!

 

First actual run...
http://www.3dmark.com/fs/11124276
I'll have to actually run through it to see if it can be improved upon later...
But right now it's losing.

Edit:
User error

 

https://www.techinferno.com/index.p...re-strike-produces-unreliable-results/&page=2

 

Been knew this for a long time...Why do you think @dspboys uses UI Version 2.0.2067 64

And in some test...Steam versions do better.

 

Since medium August. I have posted this several times in the forum And no one around couldn't even see the change in power draw with more and more crippled GUI versions. Although the difference wasn't so big
Edit. As you can see. Futuremark will do some changes to the custom settings

 

I can't seem to... I have to push the Core to +240 to get a 2100MHz reading in GPU-Z which works fine. But if I try to run either Graphics test 1 or 2 test in 3DMark the display goes corrupted and the benchmark exits. I tried clocking the CPU down but it's still glitching out...

 

Knew that as well.
Custom does not always jive together with full runs or some runs where the free version gave a better score.

That's to high. 200 to 215 should put you at 2100 if your are cold enough.

 

4.5 ended up being 982, 995, 997, 999, 999, 1005

 

Temperature wise though I had to set voltage around 1180mV with 4.5 x 4, 4.5 cache and the temperatures where in the low 80's when stress testing. I did find though that 4.3 x 4, 4.2 cache with voltage set to 1070mV running max 68C at just about anything I push at it. Played BF1 for a few hours and the CPU temps never went over 65C at 4.3/4.2 (that makes me happy).
Edit: Cinebench runs around 900 at 4.3/4.2 for me.

 

I'm just running on ambient 74*F (~23*C) air, haha, no A/C cooling for me yet.. During my bench runs the GPU peaks at 55*C though with FN-1 max fans & my laptop cooling pad fans on.

47*C GPU, +215 core === 2088MHz


49*C GPU, +228 core === 2100MHz


Guess +245 might have been a little too high....

Will try to run again with this

 

Haha, nice man. I'm not going to even attempt to try and reach the Firestrike scores you all have. Sounds like to much work and tweaking!! But congrats on your score and nice overclock.

Edit: I will have to download MSI afterburner again though now that I nuked the control center!

 

Don't say that-this

See 7700K@ 5775MHz http://hwbot.org/submission/3397126_aerotracks_cinebench___r11.5_core_i7_7700k_13.96_points
VS
6700K+-5775MHz http://hwbot.org/benchmark/cinebenc...d=processor_4012&cores=4#start=50#interval=50

Se also CB R15
http://hwbot.org/submission/3396439_aerotracks_cinebench___r15_core_i7_7700k_1279_cb

http://hwbot.org/benchmark/cinebenc...=processor_4012&cores=4#start=100#interval=50

 

http://www.3dmark.com/fs/11124726

 

Getting closer.... 2088 on the books with 3DMark: http://www.3dmark.com/3dm/16764483

EDIT: Had to use +215 core, then just tweak the freq/voltage curve to get that to work

Yes, it's definitely some work! But it also helps to "fine tune" and know your limits so you can get more FPS while gaming without going too crazy on the clocks.

One thing I really like about MSI Afterburner (beside the GPU freq/curve feature) is the Rivatuner it also comes with... Has some good OSD info you can see CPU/GPU usage %/temps and FPS while playing BF1.

 

Awesome, so close! And this is still with EC 1.05.06? Or have you went back to Prema/1.05.04??

 

That's a big 10/4 negative.
TIME MEASUREMENT INCONSISTENCIES DE... ( + 1 MORE )
And you ran the wrong test.


Edit:
Same EC. That wasn't your problem. You locking the voltage was. Once you stopped doing that you jumped from 112 to 117 fps.

Edit:
The osd is to get an idea of what is going on in real time instead of waiting on logs.
Once you see the error, you correct it and then run the bench with osd turned off. This will lower your scores about 98 percent of the time. And lower your fps in games as well.

 

Do not take from him the joy

 

Btw, @Johnksss@iBUYPOWER, thank you for your help to the community and tips for helping get our scores up! We all appreciate that!

Sent from my SM-G900P using Tapatalk

 

Thanks for that!

Now, as to 1.05.06 and SLI.....I WOULD BE TRYING TO KILL MY SELF RIGHT NOW!!!!!!!!!!!!!!!!!!!!!!!!!!!!

 

I assumes that you know something about this and not just the clever "Dell's engineers". Can you comment on this please Maybe also @Mr. Fox and @ajc9988 can chime in

 

Well, the cores can actually get to like 100C to 110C while the Tcase will show locked at 85C. Speculation of course.

 

Thank you for the hilarious post! What is even worse is the people still buying alienware that consider this acceptable. If I saw those temps regularly in the 3ghz range (for current existing x86 quad core processors), I would ask to be euthanized as life is not worth living. If I bought something like that, I'd ask to be declared mentally incompetent and use this as proof that I am incapable of making choices for myself and entering into contracts. This is becoming an intelligence test/marketing test as to whether brand loyalty can outweigh knowledge and understanding.It is this that makes a thin and light snapdragon laptop look like a logical choice because it can do as much as these throttling Intel CPUs while not running as hot... I know I've got more, but that is a good start!

Sent from my SM-G900P using Tapatalk

 

Trying to run bga cpus like they are socket ed seems to be their fatal mistake. You can't run the same voltages through them as you can with desktop cpus.

 

Indeed... that's some hara-kiri inducing throttling cah-cah with the borked EC and SLI.

Well... Just stop for a minute. Consider the source. Those clowns are lying to conceal their incompetence. There would be no legitimate excuse for one of those crappy turd CPUs to hit anywhere close to 100°C running stock had they actually cared enough to build it correctly. They can try to redefine what overheating means all they want to, but it won't change the fact that they are passing the buck and shirking their responsibility as the OEM.

Just because they can, doesn't mean they should under the circumstances in which their customers are complaining. Imagine how bad it would be if overclocked. They're just full of crap and taking those thermal thresholds out of context to make it seem like everything is OK so they don't have to feel obligated to fix their mess.

And, it's not clever. It's rotten and dishonest.

 

Considered Within the specification at 100C max temp with 1 or 2 cores, but outside with 3 or 4. I have never seen this info from Intel. Have you?

Would you if you were Dell support recommended re-paste if 1 or 2 core reach 100C, or only if multiple cores reach the same "low" temperature? Or would you say no... There is no need... Because this is within Intel's specifications

 

This is what you get when the new planned obsolescence of a device is to cause thermal degradation by design!

Sent from my SM-G900P using Tapatalk

 

But by the time that happens my 3 year $900 warranty should cover that. Geez all this BGA hate speech and fake* news

 

So, @LunaP popped back up for air in the P750ZM thread and I pointed her over here to join in the reindeer games!

Sent from my SM-G900P using Tapatalk

 

LETS BURN THIS CANDLE!!

 

@Johnksss@iBUYPOWER -- Lots of hours tweaking, but still can't quite catch your score @ 45x.... I did at least finally break 19,000 though!

http://www.3dmark.com/compare/fs/11125724/fs/11123180




EDIT: And thanks for taking the time to switch BIOS/etc and run the tests for a comparison for me!!