Razer Blade 14 (7700HQ/1060) beatdown log

I probably haven't undervolted it enough, should be able to do more probably.

 

testing ram at a lower timing (16-16-16-35 1T vs 17-17-17-39 2T)

changing the dram voltage does not work on this bios

 

Maybe swap for a better BGA Easy task and very cheap @Mobius 1 @hmscott

 

-disassemblies to fan fins
-doesn't undervolt enough (most units are at -120mV with turbo on..)

You've been bragging about how bad notebooks are if they can't be overclocked, yet if tweaking something might make this unit run better: "oh yea, I could tweak some more".

Forget to be a hater and, please, make an impartial review with everything set up to its very best everytime or you'll look just like some random hater.

 

@Mobius 1 Is always honest

 

indeed. i once heard a story that he cut down his father's cherry tree (presumably with a Razor Blade). when confronted he said: "I cannot tell a lie, that was one god awful ugly tree"

 

ram lowest will go is 15-15-15-28 1T, some minor improvement to latency and speed



@Mr. Fox @bloodhawk @TBoneSan @Johnksss@iBUYPOWER @Papusan

continuing the trend of blade 14s that have a vrm failure due to an intentional placement (sabotage) of an uncooled vrm insulated by a plastic sheet on the back of the motherboard, the blade 14 2017 kabylake follows suit






@Makyura because not having enough time behind the wheels yet means being dishonest, right

go sit on a rusty nail while drinking liquid metal to cool off please

 

mkay sir, undervolting takes a few minutes if you stay safe (100mV is far from being problematic on what? 99% of the units?) and you've been speding hours doing stuff totally less necessary than this obvious tweak.

 

Maybe I have a life?


To verify undervolt stability I don't only use OCCT, many other programs to do so like wprime, tsbench, superpi (esp useful since it's reliant on RAM), aida, etc etc.

Less necessary, like what? Everything that you see here won't necessarily make it to the final review.

 

A 10h OCCT+FS 10 hours test to stress the hell out of a cooling system looks good and such, sure, but it's not very important in terms of real life performance like some undervolt; as I said going for a -100mV / -40mV is absolutely safe as the min I've read about in forums is -105 iirc.

This small (but major compared to usual performances) tweak could have cooled down the laptop by 3-4 C in average, which is indeed much more important for a normal user.

Disassembling the fans makes little sense for an users (even expert) oriented review compared to what the cooling system is capable of in idle-light prolonged sessions or even heavy gaming ones; I'm a space engineer and yet, although I use this notebook as my main drive as I'm curretly out of town, I have yet to stress the hell out of a notebook that much, I've probably heaten it up more while playing games, which brought it far from 100% CPU and GPU.

It would also make small sense for a hardware oriented review probably, still more than checking the fans model or fins design.

Note that I'm more than grateful about this work you are doing, wish there were more reviewvers doing this, (read: little sense =/= no sense) but I believe it makes little sense to do this forgetting basic stuff.

 

Every chip is different, I got some errors on tsbench using -115mv but not on OCCT.

If one benchmark passes and the other fails, then the system is not stable, period.

I'm still finding the sweet spot to pass all benchmark before calling it a stable undervolt.

If you want to be even fairer, I won't undervolt the system and just let it run at 55/45w stock Intel voltage profile, and we all know how well the CPU cooling works at that point.


You shouldn't look at undervolt only, take some time and look at the eventual voltage (vcore) under HWinfo. A good chip with a small undervolt will have close the eventual vcore against a bad chip with heavy unstable undervolt.

 

I'm not telling you to waste a week on that like someone would have to for an absolutely stable undervolt (which would require BIOS mod to begin with). I remember spending 15 days to find the sweet spot for my good ol' i7 860 before reaching stable 4.2 GHz, I'm far from believing that a reviewer would do something similar for a notebook he'll have to send back.

I'm just saying that 100mV is simple, fast, 99,99% stable and gives good results with close to no problems, while 60mV is as random as 100mV, no expression of the undervolt capabilities and gives little boost to the cooling system.

I'm not the only one who noticed it (I wasn't even the first one to do so) so, I didn't call anyone dishonest or anything alike, I just said it was strange and almost laughable.

This said, I'll still be waiting on this review as it looks in deep and I've been waiting for one for so long on a Razer device, but I'll take with a grain of salt the temperatures reached pre re-pasting and, would you not change said setting, the ones post re-pasting.

Thank you

 

As per your request, I'll try undervolting lower and doing one final pass of temp before taking it apart.

No guarantees.

 

Thank you very much!

 

How it could have gone:

Makyura - "Mobius, can you please try and undervolt it more? Most others are achieving 100mV."
Mobius 1 - "ok"

 

I am wondering, when you take it apart to repaste, how it will be done with liquid metal...since the RB has crappy heatsink that doesn´t have a copper shim, the copper tubes go directly on the cpu and gpu, so there are a lot of gaps....plus the aluminum around the heat sink, where I read that liquid metal may oxidize the aluminum....because of the lack copper shim for a flat contact, Razer ends up using a bunch of paste to fill in the gaps....so hopefully you can document your job for us, because I am interested in repasting mine soon. Thanks a again for the beatdown.

 

It"s been changed with RB 2016 v1 iirc, it's nolonger a pipes direct contact heatsink.

 

gpu is still direct contact without slug

 

You forgot to add in the failure rate. Which is astronomical and the integrity is never the same.

Although this was answered, I wanted to add that there is a lot more to it than just a simple under volt and calling it a day.

We do what we do to understand the system, find the flaws and to hopefully be able to fix the flaws so that even the simple tweaks are taken full advantage of.

Simply under-volting is beginner stuff and calling it a day is beginner stuff.

Disassembling the laptop and understanding it is not for beginners so most won't understand this. That's okay.

If you're a space engineer, then you out of most people would understand that there is a lot more to it than just the obvious.

The things that are being done and explored are not duct tape fixes... we don't send things off into space with a duct tape over a hole on the frame of the spaceship.

Again, this stuff is not for beginners and they won't understand it.

The goal for most is to just get their laptops to run....our goal is to get these things to run at their absolute peak potential, without taking any short cuts and being content with it.

That is the reason for all these tedious tear downs, tests etc... ie... there is a purpose and it's not just for giggles.

It's all about finding the sweet spot.

Most people don't understand that not all chips are created equal. That's okay...

Your answer pretty much sums it up...

 

It's been done on an early 2016 model. The latest model should be the same.
Instructions here:
http://forum.notebookreview.com/thr...ste-of-razer-blade-2016-gpu-overclock.803592/

 

-110mv done


hwinfo reset after 2 hours of stress under 24c ambient

repaste tomorrow

 

expired

use mega.nz maybe

 

I had set it to last 3 days only; anw:

https://uniroma1it-my.sharepoint.co...rid=2_063f2984ed1ec40beb681ed9fbd8a33ef&rev=1

 

nvm I got it
what area you want me to test around novigrad?

 

Witcher 3 loop sitting in novigrad

 

+74C with wimpy 15W Cpu Package power, Oh well
Is it possible to put maximum fan speed in RazerBook's? And when will you apply a new paste?

 

paste today

fan control not possible

 

Oh'well. Almost the same as Dellienware. Why bother with fan control. Have you download and use latest firmware?

 

Latest firmware is installed.



Fan curve is "worse" imo, the machine is passively cooled up to 65c which I think is too high.



I also noticed that the heatpipe is warped here, I'll see if I can get a replacement from our supplier.

Warped heatpipe means that heat transfer slower between the source and radiator fins.

 

Filler set

 

Can you fit a .5mm copper shim in there? I think it would make a world of difference if the die had a flat, even, smooth surface. Ridiculous to have spaced heat-pipes on there.

 

I do have some, will try

You mean sandwich pipe - shim - die right? all with LM

 

Exactly

 

I'll do a run with and without shim.

Either way I have to do it twice since this unit is getting replaced due to a kinked heatpipe.

 

So far the liquid metal is good, now I have to burn in the device before doing a temperature test.

 

@Mobius 1 , what is that gooey stuff you are using instead of thermal pads?

 

K5-Pro(very thick thermal paste for replacing pads).

 

It looks absolutely insane using LM directly on the pipe, makes my eyes hurt.

 

BIOS mod still on to-do list? Not trying to rush you, just wondering.

The reason I ask is because I'm interested if it's possible to disable Optimus and attempt to OC the display. I'm unsure if that option is unlocked in the stock BIOS.

 

Blade skylake / kabylake version is fully muxless, it's permanent optimus save for the HDMI port.

The physical hardware itself is not present. And in turn the bios only shows the option to use optimus.

 

Thanks for clearing that up.

 

Lol i am surprised that one heatsink still has the direct heatpipe contact. Really one of the worst choices they ever made.

@RazerCS Maybe you guys should get with the program for once and deliver a real well engineered machine.

 

yea this is weird also

I'll try to see the temps using a shim, but for now I will run the blade through a liquid metal temperature test since it has been burned in for 24 hours

 

I am going to repaste mine soon, but I am not confident enough to use the LM, do you think kryonaut would work? Even though the gpu heatsink is the hdt type?

 

What I find strange as well is that I was watching a video of a repaste on the original RB from 2013, and the cpu and gpu don´t have the direct copper tube contact...so it seems like a step backwards....they did redo the cpu but not the gpu side yet.

 

cpu heatsink is not hdt, only gpu

as for the gpu, use ic diamond because the intended purpose is to fill in gaps on uneven heatsink, this is the one laptop that you can justify using icd on


what model do you have? 970m or 1060?

 

Thanks mobius for the tip...I have the skylake 1060.

Sent from my SM-G930F using Tapatalk

 

hahah I remember when it was posted by Razer that their thermal design was "industry leading" and they were "years ahead" with their thermal designs. And yet they went from spreader to HDT to spreader again lolol

 

I remember that

But i just dont get why they do this. Iys not like they save money with this.

 

well... a small amount on the spreader?