EVGA SC17 1080 finally available

Its one of the downsides after applying new ucode. Earlier it wasn't like that.
Let it run in realtime mode and grab a coffee/tea/chocolates and benching will be complete with higher points.

 

I see.....just got to wait for it to complete, not really frozen. Thanks for the tip.

 

Lovely score. @Papusan are you happy with this score?

 

Don't do that....he is going to have me run Aida FPU+GPU at these clocks and creat a bonfire.

 

Why call it lovely score?

Dang, You react before me on @Vasudev's post. But here we go JK

 

Talked the heatsink today, and indeed the screws for the heatsink are on the other side of the motherboard, which is very odd and unfortunate.
Makes very hard to balance the heatsink correctly and then flip it over to screw it down.

 

At least it’s not a TRIPOD HS design

 

My CPU requires quite a bit of voltage, but at least now it is not overheating and can run cinebench at 4.5GHz. The extra voltage is only needed to overclock the Ring. Keeping the Ring at 3.6GHz the voltage is average I guess.

 

What if you run cashe 36x and instead go for 46x for the cores with same voltage?

 

Yeah at 4.5GHz and ring at 3.6GHz, it requires 40mV less.

 

I am getting very briefly some temperature spikes only core 1 when running AVX (FPU), but average temps are in line.....I think I put too little. I will repaste eventually and add a little more.

 

Stock paste Aida64 (cpu+cache+gpu) @4.1GHz


Conductonaut Aida64(cpu+cache+gpu) @4.1GHz

 

Design is similar to 15 r1/r2.

 

Since the power limits settings in Bios don't change anything, the only work around is Falkentyne's bypass, using IMON offset. But it has reached it limit, so when running at 4.6GHz it will run cinebench, but reaches the power limit. Can't run prime for example without it throttling due to power limit.
4.6GHz and ring at 4.2GHz.

 

Yeah, I know. But still a awful heatsink design

At least you crushed the new Alienware’s What Package Power was reached with use of negative IMON Offset? Can you shrink a bit on voltage if you run 36x Cashe?

 

Yeah its like stripping AW 15 clothes to repaste and repad.

 

But this laptop is even more crippled than alienware right?
Can't the alienware 17r4 BGA stuff run at a higher power limit than 75W?

what makes this a better laptop for example, than the alienware or even a MSIbook?
Alienware has tripod. MSI has battery boost but proper heatsink, but overcurrent protection (can't run Linpack or AVX small FFT prime95 at too high mhz or voltage or the laptop will shut off)

 

Did that Dell 330W supply ever work out for you? Sucks that they made this thing hard to repaste. Marketing terms such as "made for overclocking" really are awful when it comes to computers in general. Any PC made for overclocking would have non-soldered everything with easy access components.

 

I was able to hit 80 watts CPU draw easily.

I am just sad that the 8th gen CPUs are hexa core

Will i got the laptop for only 1100$ with 2 year warranty I think i am alright !

 

Hey Halo, 80W? certainly a lot better than 60W limit on mine. Can you get sustained 80W out of the CPU? Or does it throttle after a few seconds? Maybe the 1070 model has different power limits settings. If I show you how, could you send me the Bios dump for me to have a look? Thanks.

Enviado de meu Pixel 2 usando Tapatalk

 

I will test the charger when I get home, travelling at the moment.
Yes I agree their marketing was shameful and unfortunate. If they just sold this machine as a gaming laptop with an overclockable CPU and a GTX1080 without the wild claims, it would be fine. Very good build quality, one of the best I've seen. Decent all around gaming BGA book. The power limits are by design, to limit it to the cooling capacity of the system with stock paste and to facilitate their 2 year warranty. Pretty good out of the box performance and thermals, considering other disasters out there. However, as soon as they make wild marketing claims, true enthusiasts will and should have a field day, because the claims are ridiculous and a 60W power limit is really a joke. Very fitting of papusan's description of jokebook, turdbook etc.... overclockable yes, but limited to 4.0GHz on cores at the hardcoded 60W power limit.

Enviado de meu Pixel 2 usando Tapatalk

 

Could you post an image of hwinfo showing that power draw?

Enviado de meu Pixel 2 usando Tapatalk

 

Here the power draw around 75 to 79, I am using EVGA software not XTU ( I still having trouble overclocking there)

Muliplier is at 42
100mv Overvolt

The CPU DIDNT hit a power limit I think i simply don't have more cooling headroom since the CPU is reaching 92-95



Let me know How I I can dump the BIOS for you .

Side Question: This happens to me twice now, sometimes when i have longer play session my keyboard will lockup and wont work ! I tried to run HWinfo it will lock up it seems to me the southbridge is overheated and locked up even if I tried to restart it wont work I HAVE TO press the small reset whole above the power button. anyone had something like that ?

 

Man @Vistar Shook those Temps damn I really want to use Conductunat BUT I am still scared from galvanized corrosion.

 

This forum has many experienced enthusiasts, overclockers and such.... here in this thread, @Papusan, @Vasudez and @Falkentyne can attest that there is no issue with conductonaut and reactions with copper other than cosmetic, which can be removed, if desired, with lapping.

Enviado de meu Pixel 2 usando Tapatalk

 

Thank you, you know what I am going to do it today !

Let me know if you want the BIOS dump.
If you want I can try after a repaste to see if i can hit more power draw because it did not hit power limit in XTU.

Gallium here i come ! (ง'̀-'́)ง

 

NO issues with corrosion of copper with PROPERLY FITTING SANDED HEATSINK!
Bad fitting heatsink will cause absorption and hardening and less contact on surface so make sure you fix this. And apply enough (don't apply too little, you need to remember some will be absorbed by the copper).

On some surfaces, Galinstan will react with surface of copper (not inside the surface) and bond with it. Some parts of the LM left touching the surface will harden a little into a crystalline type substance. If your contact is good, this will not affect temps because the contact is not ruined. In fact for some users this may lower temps another 1C afterwards. If HS has excellent fit, there will still be LM left on the CPU or GPU. Problems are if HS is badly fitted, holes all over, badly sanded and pitted to begin with, then all the LM will be absorbed into the copper and there will be none left on the silicon except faint hardened residue. That will cause high temps, but no damage. Fix for that is to sand the heatsink better, make the fit better and apply a little more LM than before. Applying too little on a copper surface can actually be bad. That's fine to apply that little on nickle plated surface or CPU IHS.

Good trick is to apply a thin layer directly on the heatsink first, then "Rub" it in with a paintbrush or the Conductonaut swab, rub it in strongly and with strong strokes all over, so the "reaction" starts right away into little imperfections on the heatsink. This is the best thing. Then apply regular layer on the CPU or GPU.

for IHS delids, people got lower temps by also coating surface of the bottom IHS with LM also and rubbing it in, not just the CPU silicon only.

Read:
http://forum.notebookreview.com/threads/after-1-year-of-thermal-grizzly-conductonaut.799343/

 

Edit: coconut.

Enviado de meu Pixel 2 usando Tapatalk

 

even with the 36x cache it would run awefully close to the 60W PL limit under CB15, so obviously with prime it would throttle. The only setting I did to trick the power reading was the IMON Offset that falkentyne suggested, and it works great up to a point. So today I decided to change the IMON slope and voilá....tricks the power reading even further....so now it can run prime at 4.6 no issues.
Can't even try going over 4.6....hardcoded.....but good enough for the 7820HK and cooling capacity of the machine. Especially considering the hardcoded power limit would limit to 4.1 or 4.2 on full load. So thank you @Falkentyne for the IMON trick. Next upgrade, LGA.
Since I get temperature spikes on core 1 only with AVX/FMA I assume it was botched repaste, right? Will repaste again when conductonaut arrives and thermal pads.
Next project, mod the GTX1080 TDP from 180W to 200+ and see if anything blows...

 

now that the power limit can be bypassed, I can run the cache at 4.3GHz
CB at 46 and cache 43, silghtly higher than before.

 

Oh you can use an Imon slope of less than 1...and that makes it negative?
Wow nice trick.
I wonder if this would work on the MSI 16L13 to bypass 91 TDP?

While it "works" on the GT73VR, it does NOT bypass the CPU EC "current trip" limit.
The EC seems to record the actual AMPS drawn by the CPU and GPU in two EC registers (high value, low value, if low value=FF, high value increases by 01, then low value goes back to 00).
Seems like if the CPU is drawing "too much power", it will shut the entire laptop off at a certain temperature (94C or 92C?) and if the CPU requests more than "FF" amount of power, the laptop INSTANTLY shuts off as soon as the load signal is sent.

Even if I change Imon offset to -31999 and Imon slope to 25 or something extremely low, the CPU says its pulling like 20W of power, but the EC Value says "D7" (prime95 AVX small FFT @ 4.4 ghz, with IA AC DC Loadline=1 and 1.178v static voltage). same if I put the values back at default (CPU wattage=100W).

the video card uses the same register. They seem to be combined...
This is starting to make sense now.

Why Mr Fox's laptop would shut off at 5.1 or 5.2 ghz when a *GPU* LOAD was put on it, when Prema removed both the CPU and GPU power restrictions and had the GPU at 250 watts.TDP. That could only happen if the EC recognized the total -amps- (rather than watts) draw.

This is the "Devil"register (EC registers C8 and C9).

 

Yeah, I set my slope to 50 and offset to 0, because it cuts the power draw in half, so easier to see how much power is actually being drawn. Haven't reached any amp limit yet to see if it behaves the same as the MSI, because the BIOS limit is 46x.

 

I did manage to get it running at 47x with XTU, for some reason with TS it won't stick....need to mess around a little more. Either way not viable, this badly binned cpu needs loads of voltage to run at 47x4 44x cache. At static 1.390 it shut off in CB, I think it needs over 1.4V, will try later to see if I see smoke.

 

shuts off? what do you mean? the laptop turned off or you got a BSOD?

 

Core is KING!! Lower Cashe ratio and try again.

 

That's a REALLY good x46 score.

 

my cache won't overclock to 4.4 without lot's and lot's of voltage, so kept it at 4.3 for this 47x4 CB run.

 

@Vistar Shook

Holy hell.
How are you getting such low temps?
4700 mhz on all cores with a VID that high? What in the living hell?

What is your IA AC DC Loadline set to ?

 

Temps are wrong, I ran CB in realtime. Temps are in the 80s.
IA AC DC at 0 since I haven't noticed any difference in VID like on your MSI.

 

@Vistar Shook
Try setting it to 1
you have the unlocked Bios
Experimentation is free

Throw the old man a bone please

 

Sorry not mid 80s high 70s.
correct temps.

 

I already messed with that before and didn't see any change in reported VID on my machine.
Here is a run with AC DC set to 1 in the BIOS. The same results, 1C hotter, though that is probably not related.

 

Unstable at 4.8 even at 1.45V....
but this is trying to increase the cache multiplier, which mine just won't go above 43x.

 

So your EVGA actually has a NON CANCER BIOS which actually only applies the IA AC DC loadline setting for adaptive voltage, like desktop boards do.
Then it locks it down so you can't overclock....(without eh...magic)......impressive......

While BOTH the MSI *AND* Alienware (afaik, one person tried using static voltages and from his test, it was also boosting the real voltage extremely high based his temps reaching 100C...) laptops apply the IA AC DC loadline resistance IMPROPERLY to STATIC voltages!! Which makes static voltages completely UNUSABLE on MSI hardware, because if you set 1.17v static voltage with MSI hardware, with default IA AC DC loadline (2.10 mOhms is the Intel reference value), the voltage boost from the resistance causes the voltage to rise to 1.3v+ during heavy load (the less load, the less boost)......., so your temps get completely out of control or your laptop just shuts off....

 

I managed to get it stable at 48x4 but cache at 43x at a whopping 1.430V.

 

As a comparison, Papusan's 7700K runs 48x4 at just 1.218V and scores 11pts higher on CB15.

 

@Vistar Shook please check windows event viewer for WHEA correctable errors really quick (might have to scroll down). you may not be stable at that 4800 mhz.

Also: important:

Can you run x47 and x48 again for me, and check the "maximum" power draw button in Throttlestop?
I want to see whats going on.

And how are you beating my scores?
@Papusan what is with these BGA chips?

1) prime95 4700 mhz, x44 cache, 1.275v, 25 minutes (AVX and FMA3 disabled, IA AC DC loadline=1): if i had IA AC DC at default values laptop would power off immediately.




2) cinebench after 3 consecutive runs, max temps, 4700 mhz, x44 cache, 1.275v, IA AC DC loadline=1. If AC DC were default, laptop would shut off during the test and temps reach 95C before that!

 

yeah probably getting loads of WHEA at 48, haven't checked yet.

 

It's fine. Just run the x47 and x48 cinebench tests for me again and press "maximum power" in throttlestop, so we can compare what's going on.
I want to see if MSI is up to their normal BS.

 

yeah but I have a 50 in IMON slope, so I assume doubling should be the reported cpu package power.