Clevo Overclocker's Lounge

Is there any way to see that graph without having bought 3DMark?

Also, is it normal for fans to spin up during screen saver without anything turned on? They blow full speed on me... I'm starting to strip windows down of espionage to see if I can fix this.

Anyone has one of those guides on windows optimizations for better scores? I'm a little curious on trying some of the things in there and see if I can fix random fans spinning up while idling.

 

Where is your desktop score at?

 

Stock Clevo VBIOS is capped at roughly 200W peak I believe. Desktop cards may be higher. You will also experience more Pascal thermal throttling due to worse cooling.

 

I think they're for 190W

 

I believe that's the intended limit, but I've also seen peaks of around 203W before. Probably just transient spikes, but I've never logged the readings.

 

Best way to check would be to add HWiNFO64's readings to MSI AB/RTSS displays and just play. See if you get sustained 200W+ loads, or if it just spikes.

 

A single Clevo 1080 will likely be able to pull between 250W and 300W (hopefully) once @Prema removes all of the cancer code. We need to be able to exhaust the capacity of the 330W AC adapter with a 1080 heavily overclocked.

I don't ever use the Clevo Hotkey GPU OC trashware. I don't care for it and its functional limits are too low. The benchmarks I have been running are +200 core and +600 memory.

I use NVIDIA Inspector, Precision X and Afterburner. Precision X is my favorite, but MSI Afterburner has some recent improvements that made it relevant to me again for the first time in years.

 

I almost always have OSD turned on and monitor this stuff like a hawk. Already confirmed it peaks out just under 200W and frequently sustained draw 160W per GPU.

 

https://www.guru3d.com/news-story/i...ore-i7-6700k-benchmarks-on-z270-platform.html

Sorry If it's posted already, the Pascal was an OCed Maxwell but to over 1800-2000MHz over air and stayed the same performance when down-clocked to the same GHz as that of Maxwell while the Kabyflake 7700K seems like a 6700K with a ~6-7% OC and when down to the 4.0 mark It looses the battle LMAO and doesn't scale even >1.8x% over the Skylake when positive that's all no more no less.

Welcome to the corporate fallacy version 2.0

@Papusan !!

 

Calling it cancer code would mean that even standard users should expect to see the card drawing that much

 

A summary of 7700K
http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-701#post-10402182

 

Well I do too, you just actually have one of these things LOL

Seems ok I suppose. What about OC'd and gaming? Have you tried something like Battlefield 1 or Witcher 3 with it?

 

I was a bit confused until I realised you linked the post after the one you meant

 

What I posted was referring to overclocked benching. I always game with the GPUs stock. I finished out my last 3 hours of Gears of War 4 with this beast. Power draw was similar. Nothing ever stood out as unusual. I will test witch Witcher 3. I don't remember if I still have it installled. May have to download it again if I do not.

 

Sounds good. Also, I want to PM you about something, are they open?

 

I always game overclocked personally, I want my gaming to benefit from the same gains the benchmarks get

 

By cancer code I am referring to the under-utilization, room temperature throttling and artificial capping of performance by not running in full boost mode. It should run at full boost until it reaches critical temperatures that induce thermal throttling to prevent damage. The GPU should behave more like a properly tuned CPU with a good BIOS by following orders explicitly and not having a mind of its own. NVIDIA has always had similar cancer code in the vBIOS that needs to be exorcised, but it is exaggerated with Pascal.

 

Imagine Volta. We'll need to pay them $$ to keep boost.

It'll be a monthly charge.

Edit: Relevant

 

OK, so on Witcher 3 set to Ultra and NVIDIA Crapworks enabled, everything totally maxed out, 2560*1440 resolution, I am seeing power draw in the range of 135W to 165W per GPU and averaging over 100 FPS running stock clocks.

 

Wow, nice. Those temps look lovely. How long was the game running?

 

Not very long. Maybe 10 or 15 minutes. But, the temps are pretty stable with the vapor chamber and that is fairly representative of what I see when gaming an hour or so. I think the highest I ever saw was like between 79°C to 82°C after a couple of hours playing GoW4, Doom or Metro LL. Also note, I do not have CCC/Hotkey installed so the fans are slower to react than they would be using the OC fan profile. The temps are cooler if I install that crapware and use the Overclock fan profile. If the final @Prema BIOS has the white keys by default and static lid lighting like the previous test version he provided for testing, then I will likely never reinstall CCC/Hotkey because I will have no use for it. The only reason I ever install it currently is to change the keys from blue to white because I don't like the blue default color.

 

Does FN + 1 not work for max fans without CCC installed? On my model it does; I can even use max fans in the BIOS.

I would like purple or red lights myself I think =D

 

Yes, it still works 100%. You don't even have to be in Windows for that to work. It will work during POST or in DOS.

The OC fan profile is CCC is not max fans. They still try to work automatically, it's just more aggressive at keeping the temps lower at the expense of a noisier system.

 

It helps keep things not too extreme for standard users and tweakers can always power mod it to remove clock variation based on power.

 

That's fine and I know that's what they like to say, but I still think that stock NVIDIA GPU behavior sucks. I've always thought that and Pascal is no exception. A standard user can use the power saving or optimal performance profile in NVCP. If I set NVCP to maximum performance that's what I should get. I should see my clock speeds locked on full boost until thermal intervention at critical temps, not the spastic "dynamic" clocking nonsense. But, it will soon hopefully be a non-issue with a @Prema vBIOS.

 

It's a lot closer to the max at stock than it has ever been percentage wise.

 

Yes, that part does seem to be true compared to the past. It's just not quite where it should be. But, improvement is always good. Considering they essentially have a monopoly and little incentive to do better, we should all be thankful for anything that is an improvement coming from NVIDIA.

 

Okay, I did some testing.
Here is a comparison of the laptop GTX 1080 to my desktop GTX 1080. SPOILER: The results are shocking.

DESKTOP



Noticeable increase in performance also in games compared to factory (FE) clocks.

Highest GPU temp: 74C

LAPTOP




Highest GPU temp: 78C

Both systems running a 6700K clocked at 4.2 Ghz.

 

There are setups where the vBIOS throttles the cards when they reach 70C... That is 70C!...

They would also throttle the limit of cells per hour to a rate that they advertised higher, but it never works because it would heat up, and everything would move with a stutter. The good side is that there would be no viruses tho

 

Forgot to add Rise of the Tomb Raide result...super overclocked laptop VGA gives 3-4 FPS less than the desktop VGA with lower OC. In matter of fact, I can get higher FPS if I set 0.975 V for 1898 Mhz in MSI Afterburner curve.
tl;dr: the higher the OC the less is the increase in FireStrike points, also in games I get less FPS with OC.

 

Aside from FS/TS runs, how do you verify that memory is overclocking with positive scaling?

 

I was thinking, is there a chance that my results are low because 330W PSU is not enough ?

 

I dont know but I can say that with 140/300 OC I got lower fps in RoTTR compared to factory clocks on the laptop.

 

18,583 is my highest Fire Strike run so far (stock BIOS/EC)... But I had lots of CPU throttling, and likely GPU as well.
47x4 CPU, 47x Cache @ 1.265v Static
+100/+175 GPU core/memory

As far as the PSU, I think the current CPU/GPU throttling is keeping the single 330W viable. I hooked up a wattage meter to mine and during benching I have seen it hit 331W once, but generally it hovers around 280W or so. Once the Prema BIOS comes into play I'm sure the single 330W will be a limitation. I already have a second 330W (got an eBay steal for $100), just trying to find a good price on the converter box

 

Yea, I was afraid that a single 330W might be a reason why it is performing worse than a desktop variant (besides thermal throttling)

Hopefully the new single 700W psu will be available for the public next year.

 

Just testing to see if the image posted doing imgur.

 

I went ahead and ordered some 2x16 3000mhz gskill ripjaws. Figured I would try my luck and hopefully xmp profiles will work.

 

nVIDIA GeForce Drivers 376.19 Findings & Fixes

 

Hi mate. Sorry for the late reply. Busy as hell at the moment. I only use throttlestop for monitoring these days because I have everything enabled in the bios.

Sent from my SM-N910F using Tapatalk

 

Very interesting!

Thanks for keeping us up to date!

Looking forward to impressions and if this will become the new .06 most stable Nvidia driver

 

works great, installs fine on laptops, GSYNC is there, no bugs so far

 

Any change in performance / bench results?

 

Optimal should look something like this for a laptop with your clocks.
And yes, I can do it with my stock Sager bios That came shipped with my laptop.
Don't mind the cpu dips. I left that there to show it didn't make a difference.
http://www.3dmark.com/spy/831110


http://www.3dmark.com/fs/10989146

Those runs were done on a single PSU. So i'm guessing we can rule that out.

Also, they are rated up to 440W Max before cut off. Unless something has changed.

And here is @dspboys Maxing out over clocking on a sinlge PSU
http://www.3dmark.com/fs/10683952

 

That's awesome to know they can "burst/peak" to 440W! I imagine it couldn't sustain that for a lengthy amount of time though...

 

It's not burst peak. I will sit anywhere under that, till you cross that. Then it just shuts off.
We all ready tested it. Well I did anyway.

Edit: I see what you mean about burst/peak. Understood

 

What was your temp during the test? My temps are in the 70s with a P775DM3-G. I see you have a newer vBIOS (mine is 86.04.2A.00.2B) , maybe that has a different power table which throttles the core clock at higher temps? Would it break the GTX 1080 in my P775DM3-G if I would flash it with a vBIOS from a GTX 1080 which was in a P870DM3-G?
Also, what program did you use to OC ?

 

You know, somehow, I think your "stock" is different from everyone else's "stock".

In everything you own.

 

Maybe 40C at best. Give or take. That's why I said optimal. I'm pretty sure there is nothing wrong with your cards.
As to flashing. I don't know. I haven't tried flashing another bios on this mobile card.

@D2 Ultima

Those are HIS clocks, not mine.

 

Well they will deliver 330w to the system this does not include the losses in the brick so all can draw more than their rating if they are not lying/faulty.

 

330W RMS / 440W Max