*** Official Clevo P870DM/Sager NP9870-G Owner's Lounge - Phoenix has arisen! ***

So you confused me with another user and then linked to pages where you prove yourself wrong.

If we take out the two odd results that don't match the others (look like driver issues or the frame rates are already silly high).

(% improvement of FPS in 1070 to 980):
Crysis 3: +18%
Doom: +16%
GTA V: +25%
Middle Earth SOM: +27%
Metro LL : +18%
BF 4: +30%
Bioshock Infinite: +11%
The Division: +20%
Tomb Raider: +27%
Star Wars Battl.. + 20%
Fallout 4: +23%

Average:
21.4%

21.4% at stock (with a weighting that newer games see more of an improvement) and then with more headroom to overclock counts as a nice improvement to me.

 

What I dont understand is how the performances is different if they have the same hardware. I know they are downclocked but if u overclock it to the same speed. Shouldnt they have the same performances? Im talking about 1070 desktop-mobile.


Sent from my iPhone using Tapatalk

 

Rarely is it as simple as a direct die shrink

Because the 1070 is clocked much higher than a 980

Have fun clocking a 980 to 1600+mhz

@thegh0sts you better hope you get more than 10% or you've just done the single worst value price/performance upgrade ever...

I think you need to look for DX12 and 4K benches to see the 30% delta between the two. 1080p still very CPU limited and especially in notebooks there would be a skew towards the 980DT being paired with more powerful desktop and unlocked CPUs, compared with 1070 which is paired with locked BGA crap like the 6700HQ in a lot of thin and lights

 

Don't forget it's using the Mxm interface for power and I am getting decent scores that are reflected in benchmarks by notebookcheck.net. In terms of performance per watt it's quite efficient.

Sent from my SM-N910G using Tapatalk

 

A pic for those that don't believe i have the 1070 installed



my earlier diagram for the potential to have 1070 in SLI is still possible as that overlap is still very small.



It would also mean you would need to completely cut away the internal chassis/frame marked below.



and this is why you need to completely remove it.



So it's possible if you have the time, money, and patience.

 

Did you mod the heatsink? what are you temps in idle or gaming for short sessions?

 

no heatsink modding.

getting temps of mid 60s with fans at full blast or 70+ with the fans off.

 

they're not the same hardware... the 1070 notebook and a lower (but wider range) base and boost clocks but make up for it with an extra SMU cluster. the 1070 desktop has higher clocks but only has 1920 cores.

 

is this the laptop that DARCODER is trying to do 1070 SLI?



looks frickin' impossible.

 

xD will do my best with my lovely dremel. Allready fit a 4k inside.

 

let me know if it doesn't work out and I'll attempt it LOL

 

Can you tell my power splitter arrived?



Don't think I'll be running those settings daily

 

hmmm....GOW4 seems to run better with the 1070 even if i bump up the internal resolution to 1440p and i still get 60fps on ultra - that's something i couldn't do with the 980 for sure.

 

Base clocks and turbo clocks are two very different things especially with pascal.

 

I know. The base clock is slower in the mobile but the turbo is the same. But i mean if u overclock it. Should they have the same performance if working at same clock speed?


Sent from my iPhone using Tapatalk

 

Turbo boost 3.0 goes far beyond the listed turbo usually, especially on the desktop, if they were to be the same frequency the mobile 1070 would be faster as it has more shaders than the desktop 1070.

 

No, the boost clock is lower than the desktop variant by 38MHz. The boost range is wider on the notebook variant.

Sent from my SM-N910G using Tapatalk

 

I've seen my 1070 peak at 1885.5MHz. Nvidia inspector says my max theoretical boost is 1911MHz with the +38 overclock.

Sent from my SM-N910G using Tapatalk

 

Oh good, my extra ICD7 came through and just in time.

Sent from my SM-N910G using Tapatalk

 

This is where power mods and keeping the temperatures low will get you much closer to the desktop.

Nice

 

if @Prema gets the vbios i wonder if they can mod it.

 

i also hope he find some free time to play with this new vbios for our pleasure

 

i just compared my FS scores to the ones on notebookcheck.net and from looking at my scores to the averages on notebookcheck it appears that with my clocks the 1070 notebook just beats a 980TI by 4-5%. so I am sandwiched between the 980TI and the 1070 desktop.

most heavy games like BF1 and Dishonored 2 will be around 65-70 deg C for most of the time (with fans at full blast) and I don't think i should be all too concerned so long as it plateaus.

I might try overclocking based on the bases clocks so setting the clocks to 1506MHz instead of 1683MHz.

http://www.notebookcheck.net/Mobile-NVIDIA-GeForce-GTX-1070-Laptop.169549.0.html

 

The combined heatsink could be fun to play with too.

 

I wonder how difficult it is to separate the pipes from the plate? You want the plate of the single gpu but the pipes of the 980Dt.

Sent from my SM-N910G using Tapatalk

 

I've not checked but does any of it get in the way of a regular card or could you just do custom pads?

 

I will also try to mod the heatsink..will try to fit another fan to boost cooling. Just need to find slim measure but powerfull enough.

 

The problem is that there's a whole bunch of surface mount components that are higher than the die. This along with the contours of the contact plate, it leaves a good 2-3mm gap enough where I can see between the layers.

In other words there's no direct contact between the heatsink and gpu die.

Sent from my SM-N910G using Tapatalk

 

You can use a shim to bridge the gap. If you solder it onto the heatsink, there's basically zero added thermal resistance, or you can use CLU if you're comfortable with it.

 

Just be careful if you will use CLU to fulfill the gap. I recommend the other option. CLU is good for the normal cases with no gaps.

Sent from my SM-N920C using Tapatalk

 

The point of the shim is to ensure that there is no gap.

 

The problem is that you have surface mount components that are directly touching the heatsink plate and I am not comfortable with.

Sent from my SM-N910G using Tapatalk

 

I got some grey 0.5mm 5w/mk phobya pads from pccg they would do the trick for insulating, they're pretty tough and very thin.

I also got a bunch of big squares of yumcha ones everything from 1mm through 4mm which arrived yesterday. Good timing I think some of the ones on there are not doing their job as I have occasionally had some artifacting during benches at way lower clocks than it should happen. The big 4mm ones up the back on the VRMs are looking dry and shaggy they've been reused that many times now

The Conductonaut is absolutely awesome, a +250mhz core overclock pulling close to 200W according to HWInfo and it doesn't break 70C.

I smeared more on the CPU and it made no difference really, think I need to delid. I'm a bit of a noob at the Skylake OC game, 4.4 at -60mV and 4.5 at -25mV stable is decent fwict but 4.4 or 4.5 on aida64 FPU stress or 4.6 on anything else thermal throttles. Add it to the fiddling to-do list...

 

I find it interesting that my 1070 ranks between a 980TI and the 1070 desktop - a gap of a about 10% (and i think i am squarely in the middle of that at 5% faster than the 980TI) when notebookcheck.net ranks it lower than the 980TI.

I've posted my FS scores and i think what's pulling the 1070 laptop score in NBC.N's review is because many of the test models come with mobile CPUs and not a desktop CPU that i have.

You can compare it to their median FS scores and you can see i am at least 2000 points ahead so it's safe to believe that i am besting a 980TI using less power than 2 980Ms.

 

You looking at just graphics score or total FS including CPU? As I said a couple days ago a lot of 1070s are paired with 6700HQs in thin n lights which brings the average total FS score down. Look at the size of the red bar that's the range of the FS graphics results, the highest are higher than the 980ti graphics FS score and eyeballing it roughly halfway between 980ti and 1070desktop.

 

FS graphic score of thegh0sts 1070 is around 17.5k, 980 ti have almost 17k max and desktop 1070 around 18.5k max so...

980m sli have around 19k. Still that 1070 is great performance of single card. Possibility to put that on sli will be winner in case that upgrade to 1080 is impossible.

thegh0sts what was the cost of upgrade to 1070?

 

so I floated onto this website and noticed they are selling the 1080 with heatsink for this system http://rjtech.com/shop/index.php?dispatch=categories.view&category_id=239 will I need to modify anything to slap that in there? I have the 2 980m variant now.

 

Yes, it won't even come close to fitting without dremeling out a screw hole. Then you would need to get ahold of one of the 4 pin power cables and mod it to the 8 pin variant in the DM2/3. Even then there may be additional soft/hard modding necessary to bypass any firmware-level roadblocks.

 

The answer for now is: 1080 won't work in this system.

For now nobody selling 1080 for P870DM (1) and no one confirm that 1080 works in P870DM (1)

Look for laptop models with 1080 there: P775DM3 P870DM2/3. Look for description also "1. This is for Clevo selected model laptops only"

 

got it, so what's the best route to upgrade my 980ms?

 

Driver and system variations can cause score changes. The first reviews were on early drivers.

 

Did you read my previous posts? I have already stated this.

Sent from my SM-T560NU using Tapatalk

 

A whole new laptop!

Sent from my SM-N910G using Tapatalk

 

Two words: A LOT!

Sent from my SM-N910G using Tapatalk

 

@thegh0sts If you find some free time,what is your Firestrike extreme results? this is a result with 1070N in p375sma:
http://www.3dmark.com/3dm/16809043? temp reachs 80-85c with Clu and some little heatsink mod. I will add an extra fan to boost cooling then..

 

What about normal firestrike?

Sent from my SM-N910G using Tapatalk

 

My FS Extreme scores compared with @DARCODER score.

http://www.3dmark.com/compare/fs/11162544/fs/11146775

only a 5-6% lead (in favor of me) really, probably due to CPU, different GPU drivers, and i've overclocked both the CPU and GPU.

 

Thanks! well its not my score, it is from a chinese forum but i m happy score are very close. my package is ready to be picked up...but i am not in town.

 

you're never in town when a GPU arrives. LOL

well, to be fair do both FS standard and FS ultra with the same +38MHz OC. if you got the custom bios for the laptop then oc if you can.

 

lol... yea maybe on the 26 gonna be able to pick it up. After modding, will try each benchmarkts. Yes my laptop has prema Bios.