Liquid Metal Repaste of Razer Blade 2016 + GPU overclock

After undervolting and repasting with Thermal Grizzly Conductonaut, my CPU temps dropped about 23ºC

Like most Blade owners, the GPU in my Razer Blade 14" (970m early 2016) would thermally throttle soon after starting a gaming session, when the GPU temp would reach 87 ºC. The throttling would effectively turn off boost speeds and often reduce the clock speed to around 900MHz or less. Sure, this laptop gets really hot and the fans get super loud, so I should just accept the notion that it can't be cooled down very well because it's a thin laptop, right?

After spending some time searching for info about repasting I quickly came across a common theme regarding the Razer Blade 14":

"The Blade is too thin, repasting won't help!"
"OMG It will make temps worse!"
"Razer is all-knowing and uses a special paste!"
"It will void your warranty!"

I found only two reported success stories online and even with those the temp reduction they achieved was not amazing. The problem is the contact surface area between the heat sink and the GPU & CPU is very poor in these laptops. In fact, the gap is so large that the CPU of my laptop literally has NO PHYSICAL CONTACT with the heat sink! Because of this, heavy amounts of thermal paste must be used to fill in the large gaps.



Original factory paste visible on GPU (left) and CPU (right).


Original factory paste visible on GPU heat sink (left) and CPU heat sink (right). Also, for some reason the plastic backing was not removed from a thermal pad.


Unfortunately this isn't very efficient, as the more paste you use, the less effective it becomes at transferring the heat away. Luckily this isn't really a problem for liquid metal (debated by some people) which is why I decided it would be the most effective solution for this laptop.
"Wait but liquid metal won't work with aluminum! Also it won't work with HDT heat sinks! Also it's dangerous if you use too much because it will squirt out and get everywhere!!!"

Liquid metal quickly oxidizes aluminum and damages it when they come in contact with each other. To prevent this I protected the exposed aluminum by coating it with epoxy. I used JB Weld epoxy in this case. I mixed it with some rubbing alcohol in order to make it easier to apply as a thin layer.


CPU heat sink after sealing exposed aluminum with epoxy.


GPU heat sink after sealing exposed aluminum with epoxy.


Liquid metal is also electrically conductive and will short out any metal connections it comes in contact with. I covered the exposed circuitry that could possibly come in contact with the liquid metal by using the blue tacky stuff used to hang paper/posters to the wall. This stuff also very easily handles heat over 100 degrees Celsius. (Some people have used electrical tape instead)


Exposed contacts of CPU have been insulated.


Exposed contacts of GPU have been insulated.


Just like with the regular thermal paste, I had to use extra amounts of liquid metal in order to fully fill in the large gaps between the heat sink and GPU/CPU. Otherwise little or no contact would be made between them after reinstalling the heat sink. Before I applied the liquid metal I first cleaned the copper with a strong brazing flux paste, then wiped it off with rubbing alcohol. The flux contains acids that remove the surface oxidation on the copper, allowing the liquid metal the best possible contact to the copper after application. (I don't know how well this improved thermal efficiency, but it was an easy and quick step so it was worth a shot) I then used more blue tacky stuff to surround the liquid metal in order to prevent the chance of it spilling out onto the motherboard.


Copper cleaned and liquid metal applied. Blue tack will "seal" in the liquid metal as it molds around the CPU after reinstalling the heat sink.


Liquid metal applied. Blue tack will "seal" in the liquid metal as it molds around the GPU after reinstalling the heat sink.




After putting everything back together, I fired up the laptop and started stress testing the temps.
Below are my observations with ambient temperatures at 21 ºC (70 ºF):

• CPU temps dropped by about 10ºC (fans set to quiet mode)
• CPU idle temps didn't change much, but now it cools down to idle temps much quicker
• GPU no longer thermally throttles
• GPU temps average 77ºC and max out at about 81-82ºC
• Fans never reach max speed anymore
• Fans spin down to idle much quicker after a gaming session
• Laptop surface temps don't get nearly as hot as they used to

CPU stress test temp before repaste: 82ºC (fans set to quiet mode)



CPU stress test temp after repaste: 72ºC (fans set to quiet mode)



CPU stress test temp after repaste + undervolt (160mV): 59ºC



While the liquid metal made a big difference in temps, undervolting made an even bigger difference. After hours of stress testing with prime95 and Intel XTU I was able to safely undervolt the CPU by 160mV.

I'm very happy with the results. My only concern is the possibility of the liquid metal flowing away from the mating surfaces if the laptop is bumped or handled aggressively. However, if Conductonaut solidifies after use like some other brands of liquid metals do then maybe this won't be an issue. So far I've thrown the laptop into my backpack and walked around campus a few times without any problems.



With the extra thermal headroom I decided to try overclocking the 970m. After extensive testing I finally settled on raising the boost clock speed by 200 MHz and RAM clock speed by 400MHz. (I modified the bios using Maxwell II BIOS Tweaker.)
I even tried boost speeds at 1380MHz but within minutes of gameplay the GPU would begin to throttle down to the 1200's due to temps or power limitations. At 1240MHz thermal throttling is uncommon as long as I have the laptop slightly propped up for greater airflow.

Stock GPU boost clock: 1038MHz
New boost clock: 1240MHz

Stock GPU memory clock: 2505MHz (1253MHz)
New memory clock: 2905MHz (1453MHz)





Before overclock



After overclock





UPDATE:

Almost two months later and the temps are still the same as when I first applied the Conductonaut. I've also opened up the laptop several times to make sure no LM has squeezed past the blue tack/putty.

Playing Mass Effect Andromeda for about 30 min (high graphics, rendered resolution 1920x1080) the max CPU temp is 59 ºC, and the overclocked GPU maxes out at 86 ºC without throttling. I have the laptop propped up about 2 inches for better airflow.

 

Hot damn dude. This is awesome. Wonder if this is possible with the Stealth as well...


Sent from my iPhone using Tapatalk

 

Glad you liked it!
I don't see why this wouldn't work with the Stealth either. Perhaps the temps wouldn't drop as drastically but only one way to find out.

 

Picture doesn't show for me @PRSnow

 

Pics work fine for me. My only concern is if the fit is airtight. If not, you're going to have temps skyrocketing very soon due to metal paste drying out if the seal is not airtight.

 

I'll be sure to post status updates if anything changes but so far the temps have been solid after two weeks. I'm confident the CPU is sealed airtight, but as for the GPU I can't say for sure. They are definitely both sealed well enough to prevent the liquid metal from getting onto the motherboard though. Just for peace of mind I recently opened up the laptop to have a look at the motherboard and the seals look good as far as I can tell.

 

Pic doesn't show for me though, only says [ img ]

 

the images show for me.

 

Yeah I can't see the images either. OP, would you mind linking them as well? I'd love to see your work.

 

Ditto.

 

I added links to each photo. Let me know if that doesn't work...

 

disable https everywhere and the photos will show


@PRSnow imgur is a much better choice, photobucket is cancer

 

So true.

 

this is pretty legit. let us know the long term implications of this. wonder if a method could be standardized to ensure HDT + LM. Wonder if the LHZ crew can start LM repastes on Razer Blades (plz dont, though purely business wise I think it's great because tons of people will want to fix their broken machines but I really don't want the world encouraging Razer to keep making broken ****).

 

Cancer free now! I updated the links to go through imgur instead.
I also didn't realize how much photobucket was reducing the image sizes making them look like crap.

 

thanks

 

Sweet! I like it a lot, good job bro

 

Thank you good sir!

 

Update:

Over a month later and temps are still rock solid. I also recently overclocked the RAM to 2667MHz with no need to increase voltage. The RAM temps also didn't increase from stock either (used IR temp sensor gun directly on ram to check).
According to the system analyzer software I used (SiSoftware Sandra) the RAM chips in my laptop are rated at a max speed of 2667MHz from the manufacturer SanMax. Maybe Razer underclocks them from the factory?

 

what software did you use to oc ram?

 

Can you do some real life tests like checking temps after a long gaming session?

I really want to see how it performs with both CPU and GPU under full load after 30+ minutes of gaming.

Thank you!

 

You can follow my thread on the Blade 14 Kabylake, will be posting temperatures and whatnot.

 

Appreciate it, but what i wanted to see are the temps on a Razer Blade 14 using a liquid metal cooling solution, since your temps look like a RB using normal TIM.

 

That's using the stock TIM

 

Are those at stock frequencies? Cause mine has never even hit 90s with -125mV underVolt (which is there from day 1) CPU wise.

Also, the fan profile makes ALOT of difference, I'm on the "cool" one all the time cause of this, and unistalling Synapse makes it bad (looks like the FW profile is different from any SW one and it's also different from the BIOS one, as I tried using the BIOS editing tool with no success).

 

stock frequency with -55mv undervolt

will try if I can get it any lower, kaby is more optimized I think

 

Yup, I'm talking about Kaby Lake one too.

 

Fan profile is set to "high" on synapse, doesn't seem to make different from auto

Even then, seems like max fanspeed is only maintained for short bursts of time even though temps are critical

 

What it does, at least on my notebook (might be a FW update crap?) is to ramp up at 70ish% at 65 C and to helicopter at 80 C.

I just took this pic with my phone (looks clear enough tho, ahah) after having had OW in background (for the zSilver rewards) and while playing the Witcher 3 at max setting which allow me to have 60 steady FPS (mix of ultra and high).

 

Use OCCT and firestrike ultra graphics test 1 loop to replicate my temps

 

Done, still far from your values tho

 

run it in a loop for 10 minutes for the temp to stabilize, then reset your monitoring software (hwinfo etc)

I am using hwinfo in a 500ms polling period

 

Still absolutely nothing similar.

 

90+c

it's close to what I get, maybe you have lower ambient

have hwinfo ss on min/max/avg after temps stabilize?

 

Well, 5C less is not really a small amount and no, the ambient is pretty hot tbh, I live in Rome and I've had to keep my windows closed whole day cause of a fire a few km away from here which lasted 20 hours+, I'd say I'm at about 25+ C, got no thermometer to measure it tho.

I'm attaching OCCT temp graphs for the CPU:

Average was around 82C I'd say, nothing really bad; that's cause of their weird curve which makes the fans go helicopter like just beyond 90C (which was happening at peaks only, very rearely). Would they kick in full blast quicker, it'd be colder probably.



Turbo boost kept goin on-off, which isn't really good, but better this than higher temps I guess.

Not to mention the fact that this scenario is unreal; I'm pretty sure my PC hasn't ever gone over 85C peak, CPU wise, and I'm checking it very often and I've been playing a few games lately (OW, Doom, Kerbal, Quake (beta), Battlefield 1, The Witcher 3).

 

That's the thing, my turboboost is on the entire time, it reached critical temp but cpu is able to hold 3.4ghz

My ambient is 24-25c on average when tested (3+ hour period for that screenshot)

 

Hi, Ive one question about the conductonaut syringe itself. Do you think 1g is enough for one cpu and gpu ? (with the 2 heatsink)

 

enough for 2 blade

 

Yes, there is plenty. I was initially worried about the same thing but like Mobius said there is enough for two Blades in fact.

 

I used the bios after editing it to unlock the overclocking menus.

 

Sure, I'll post the exact numbers in a couple days as soon as I have time.

 

link me

 

This guide explains how to enable the hidden menus in the bios. The rest I had to figure out on my own, but I'm open to answering questions if you run into a problem.

https://insider.razerzone.com/index.php?threads/razer-blade-2016-bios-mod.14368/

 

will do today

 

Thx guys

 

Any update on the status of your machine?

 

Still running great!
After playing Mass Effect Andromeda for about 30 min the max CPU temp was 59 ºC, and the GPU maxed out at 86 ºC without throttling. I have the laptop propped up about 2 inches for better airflow.

 

Im cleaning everything right now, just waiting for epoxy to dry out on the aluminum then im good to apply conductonaut.









I hope i can see awesome results like yours!

 

Alright!, it's done!

Idle temps, 5 minutes after boot:



20 minutes of Intel XTU Stress Test: (fans barely spining, cant hear them)



So far so good, gonna update this post after a long gaming session.

EDIT 1:

45 minutes of Battlefield 1 on a flat surface with no cooling pad:



Temperatures before conductonaut were 99c max spikes, average of 93c on CPU and 87c on GPU (flat surface)

With conductonaut average temps are 75c CPU and 80-83c GPU.

 

Is this a 2015?

Anyway, gj