Razer Blade Tinkering

[Pendragon]

I looked around on various forums but nobody has done it yet. Just wanted to know for those who have your Razer Blades, would you perform a few after market changes to see how far you can push it?

 

1) Undervolt (just want to see how low the temps will get from a stable undervolt)

2) Repaste with Gelid GC-Extreme or Thermal Grizzly Kryonaut (again, want to see how much temp improvement you could get)

3) Underclock (just want to see how low the temps would get on a underclock to like 2.5ghz with turbo boost disabled)

 

(As far as I can tell there's no off the shelf software for GPU undervolting. Would probably have to do custom bios edits or something, anyone know a solution for mobile GPU undervolting)

 

Purpose of this thread is just to know how far we can push it and whether we could have like a 70C extended full load Razer Blade.

[Swndlr]

I would probably only try undervolting on mine, not much of a tinkerer myself.

[Pendragon]

4 hours ago, Swndlr said:

I would probably only try undervolting on mine, not much of a tinkerer myself.

If you do, it would be amazing if you could post your results.

[Dackzy]

I work with fixing laptops and I wouldn't even bother trying to fix a razer blade. The undervolt might lower the temps a little and we talk very little, respasting will also help a little, but again we talk very little, underclock just not worth it, some CPUs will die from it. The biggests issues is that the fan they use are shit and should not be used in a laptop and the heatsink is a god damn joke I mean come on Razer do you guys really have such a little brain that you could not just stop and think that you should maybe make it bigger.

 

I have 3 suggestions for fixing the laptop (more like get rid of the problem)

1. Sell it to some idiot and buy a propper laptop

2. Take a sledgehammer and destroy it and buy a proper laptop

3. use it as a mouse pad or a plate and buy a proper laptop

[D2ultima]

10 hours ago, Pendragon said:

1) Undervolt (just want to see how low the temps will get from a stable undervolt)

2) Repaste with Gelid GC-Extreme or Thermal Grizzly Kryonaut (again, want to see how much temp improvement you could get)

3) Underclock (just want to see how low the temps would get on a underclock to like 2.5ghz with turbo boost disabled)

 

(As far as I can tell there's no off the shelf software for GPU undervolting. Would probably have to do custom bios edits or something, anyone know a solution for mobile GPU undervolting)

 

Purpose of this thread is just to know how far we can push it and whether we could have like a 70C extended full load Razer Blade.

1) this is dependent on CPU. Some undervolt like champs. Some don't. Temps usually don't drop more than 5c under load from undervolts. Cooler chip models (like skylake) gain less than hotter models (Like haswell) from undervolts in general, I believe.

2) IC Diamond would be the best case. Shin Etsu x23 7921, Arctic Ceramique 2 and Gelid GC Extreme are in this bracket as well, but ICD lasts longest. Note that this voids your warranty to repaste, because Razer warranty is fuckshit.

3) This does not count as "pushing" it. Especially since current blades only hit 3.1GHz under 4-core load, and adjusting of clock ratios is not possible on the 6700HQ. Either you have turbo on, or turbo off. 2.5GHz or 3.1GHz. Take your pick. Haswell would work, but then that's arguably a worse chip, especially for a crap cooling design like the Blade.

 

Maxwell cannot be undervolted. Desktop or laptop. Unless you use a custom video BIOS that forces voltage to a constant instead of the huge variance it has under normal operation parameters. Here, give this a read if you want to know: http://forum.notebookreview.com/threads/on-the-subject-of-maxwell-gpus-efficiency.773102/

 

For the record, Pascal has the same kind of voltage variance bullshit as Maxwell. It has a narrower band (~0.1v instead of ~0.21v) but it still has it. Pascal is basically a refined Maxwell for all intents and purposes. And again, flashing a custom vBIOS will void your Razer warranty.

 

70c full load Razer Blade getting the same or better performance it gets at stock is not possible unless you're basically living in a fidge. Or canada in winter. The machine is designed to operate near thermal limits without dropping below base clocks, under the claim that the boost clock is not a guarantee (which is bullshit; dropping under boost clocks for a CPU under load throws throttle flags in the CPU registers). The GPU is a slightly different story regarding boost and throttle flags I think, but still rings true. If they had a flat heatsink instead of using raw heatpipes touching the dies and the heatsinks were fully copper, you'd likely be able to pull a rabbit out of your hat with some liquid metal thermal paste, but that's basically about it for the machine. Propping it up for better airflow, maybe modding the case for greater fan intake capabilities, forcing the fans to max, best thermal paste you could use, maybe a shim soldered onto the heatpipes for use of liquid ultra on the CPU die (and maybe GPU die), and that's about the best cooling you could acquire for the blade.

[robin5693]

1. I'm in the process of undervolting my CPU, it's sitting at -75mV running a stress test atm. Last I checked it was peaking at 86C down from 92 at -60mV.

2. I purchased the Xigmatek Xi-3 HDT thermal paste from ebay, going to repaste once that comes in. Reasoning for paste here: http://forum.notebookreview.com/threads/repaste-those-razer-blade-2015s-at-your-own-risk.771707/page-17#post-10067269

3. Haven't underclocked/disabled turbo on my cpu yet. Just bought Watch_Dogs and will be testing gameplay/temps once I'm done all my tweaks.

[Pendragon]

19 hours ago, robin5693 said:

2. I purchased the Xigmatek Xi-3 HDT thermal paste from ebay, going to repaste once that comes in. Reasoning for paste here: http://forum.notebookreview.com/threads/repaste-those-razer-blade-2015s-at-your-own-risk.771707/page-17#post-10067269

 

Reason why that didn't work is because Razer heatsinks are direct heatpipe but done poorly so there's plenty of gaps inbetween. From the pictures you can see that there would be significant gaps in the contact of the dies and the heatpipe. There's another article on the same forum where the guy just smeared thermal paste all over the heatpipes and that worked better. Btw. the blade's heatpipe's have changed since the 2014 version.

 

On 6/9/2016 at 11:22 AM, D2ultima said:

Or canada in winter.

That is exactly where I live LOL, almost never have heat problems on any electronics. Btw, i've used IC diamond in the paste and the micro chunks of "diamond" always scratch my heat spreader, which is why I never considered it. Always thought that would be a poorly idea to use it directly on the die. 

 

Oh well. Hopefully Apple does better with its Polaris 11 Macbook Pros.

 

 

 

 

 

 

[D2ultima]

3 minutes ago, Pendragon said:

Btw, i've used IC diamond in the paste and the micro chunks of "diamond" always scratch my heat spreader, which is why I never considered it.

There is no problem with ICD... people just remove it badly. Soak the thing with alcohol and then gently wipe off. If all doesn't come off, soak again and then gently wipe off more. Alcohol soaked Q-tips are great for the wiping off part.

 

Here's a pic of my GPU die after removing ICD like that. Mirror finish; you can see my roof clearly. Before. After.

[robin5693]

14 hours ago, Pendragon said:

Reason why that didn't work is because Razer heatsinks are direct heatpipe but done poorly so there's plenty of gaps inbetween. From the pictures you can see that there would be significant gaps in the contact of the dies and the heatpipe. There's another article on the same forum where the guy just smeared thermal paste all over the heatpipes and that worked better. Btw. the blade's heatpipe's have changed since the 2014 version.

 

Xi 3 - HDT is designed for direct heat pipe contact, and I'll see what they look like when I take it apart

 

FYI, Still working on undervolting, It's working surprisingly well.... Currently at -165mV and counting....

[D2ultima]

52 minutes ago, robin5693 said:

FYI, Still working on undervolting, It's working surprisingly well.... Currently at -165mV and counting....

Make sure your undervolting isn't introducing current limit throttle. Less volts = more amps needed for working. It's possible to undervolt so far that you need to raise current limit to stop throttling. Be wary of that.

[robin5693]

13 minutes ago, D2ultima said:

Make sure your undervolting isn't introducing current limit throttle. Less volts = more amps needed for working. It's possible to undervolt so far that you need to raise current limit to stop throttling. Be wary of that.

Where would I find out if that is happening in ThrottleStop?

 

Edit: my CPU speed is not dropping at all, if that's what you're talking about...

Edited June 12, 2016 by robin5693
Clarification

[D2ultima]

10 minutes ago, robin5693 said:

Where would I find out if that is happening in ThrottleStop?

 

Edit: my CPU speed is not dropping at all, if that's what you're talking about...

You'd need to run limit reasons to tell.

 

I'm not saying your CPU speed *WILL* drop. I'm simply saying to make sure that you hold your proper turbo boost speeds under stress scenarios. If you undervolt and it's stable but you start getting current limit throttle, relax the undervolt.

[robin5693]

Okay yeah that makes sense. Is there an application called "limit reasons" that I can run to make sure I'm not screwing myself over? I'm running the Prime95 torture test for 2 hours each time I change the voltage to check if it crashes. Turning it off every once in a while to make sure it still works under idle as well.

[D2ultima]

6 hours ago, robin5693 said:

Okay yeah that makes sense. Is there an application called "limit reasons" that I can run to make sure I'm not screwing myself over? I'm running the Prime95 torture test for 2 hours each time I change the voltage to check if it crashes. Turning it off every once in a while to make sure it still works under idle as well.

[robin5693]

Mine's grayed out. Architecture-specific driver not available yet?