Non-throttling 5700HQ

We have a thread in the Gigabyte-section discussing the Aorus X5 which can be tuned to not throttle.
The reason I´m bringing it up to the gaming section is because we´ve successfully removed the thermal throttling with undervolt and repaste (note: BF4 still thermal throttles for me). But surprisingly it wont power throttle either. I can run 5 minutes XTU stress test and play GTA V and Crysis 3 at max clocks.

As far as I know it´s very rare, if not unheard of, that HQ Maxwells and Broadwells wont power throttle and I´m very curious if there are any more users out there with HQs that can hold their clocks.

Granted they can probably be pushed to power throttle with the correct settings and game levels but I´ve never even seen them pass 5 minutes of XTU.

Any thoughts?

Spoiler: Undervolted and repasted 5700HQ

 

Well your maximum power is less than TDP spec of 47W so shouldn't be surprising your not seeing power throttling. Any reason your not running the 2 extra bins for 3.7GHz?
Tjmax is 105C so seems there's plenty of headroom unless the notebook manufacturer is using a large TCC offset.

As far as Haswell or Broadwell are concerned I think it should be possible to bypass any power throttling, leaving just current and temperature throttling.

 

Well my point was I hadn´t seen an HQ pull it of at stock even. I´ve now tried 3.7 GHz and it does power throttle.

Spoiler: 3.7 GHz, undervolted & repasted

Notice the temp though, 84C is not too shabby for an overclocked CPU in a thin Aorus. But that´s another topic.

 

If you want to bypass the 47W limit let me know and I'll send you some software you can try.

 

Lots of the HQ chips as I told you are superbly overvolted out the box, but if you can undervolt it to be stable, depending on the workload, stock speeds can be had. But when you want to do the real heavy lifting so to speak, it's a different story. But so said so done, 47W be the limit of the HQ. Thankfully it seems skylake is not burdened by that limit. Now the only battle to fight is the soldered-only crap.

Do let me know if you're allowed to use Dufus' software and if it works.

 

What is Dufus's software? Dufus mind sharing?

 

Just trying to make use of an opportunity.

 

They might let me try something but I think they´d like a little more details regarding the software and/or what alterations would be involved.

 

I didn't realize it was a loaner, maybe best to leave it be then.

 

It's a review unit he has.

Might be interesting to see what Gigabyte says on the matter, though.

 

The 5700HQ is not as power hungry as the 47x0HQ were, so for instance in my P650SE-A, with a 40mV UV and the 4 core @ 3.7GHz, the power consumption is between 40 to 42W tops (wprime, prime, XTU bench).
In game (like BF4), it hovers well below 40W.
No thermal throttle either, that was implied.

 

Well it's a stand alone package so no installation, possibility of a BSOD. The main thing to look out for IMO would be running off the battery. Without software/firmware power limiting one would have to make sure to avoid too high a load on battery. Perhaps there are hardware controls in place for that if current limiting or multiplier limit isn't put into effect but for now battery operation is an unknown.

@pau1ow power limiting was shown during XTU stress and using 37x multi (post 3)

 

Haswell and broadwell draw about the same TDP; slightly more on the broadwell side (due to the L4 cache). The key is voltage and amps. A problem with many 47xxHQ chips were that their voltages were extremely high, with 1.2v+ being common for a relatively low 3.3-3.4GHz 4-core-turbo clockspeed. Most of the broadwell chips I've seen had more acceptable voltage (and thus TDP drain). If you were to overvolt a 5700HQ to say... 1.3v and compare it to a comparable overvolted-out-of-the-box 47xxHQ, they should draw about the same in stress tests.

Also, what is a P650SE-A? I've never heard of that machine before, and Clevo's website has never to my knowledge, contained a P6xxSx-A model of any kind. I also cannot find drivers or a manual for it in the product listing on their downloads page. I'm not arguing that you have a P650SE with a 5700HQ in it, but I would like if you can show me the P650SE-A model name somewhere.

 

XTU stress test:

Spoiler: 5700HQ @ 3.7GHz

As you can see, power & temps are quite low. The fan is quite inaudible as well.

 

That´s very low power draw, and your voltage is even what I have stock. When I´m at 1.110V (stock) I can´t even pass with stock speed, how can there be such a large difference?

 

Jeez your chip uses 1.11v? My 4940MX uses 1.09-1.095v at 3.7GHz and probably could go even lower since that's with my -80mv I use at 4GHz which uses 1.18-1.19v. My 4940MX is not a good specimen either... Average at best.

I'm also moving this out of gaming.

 

Could be any of a number of reasons.

1. Better chip
2. Inaccurate current scaling
3. Load is less. This is why I like running Linpack in its native state or through a wrapper such as Linx, the GFLOPS tells us if it is running optimally.

 

And @Convel s is at 1.144V (from the X5-thread).

 

Could this be down to Windows 8.1 & Windows 10 ? I see that you're using the former.

 

That's crazy. I don't even think my old 4770K which was a horrible chip even had voltage that high!

 

http://forum.notebookreview.com/threads/test-aorus-x5.787439/page-3#post-10190358

Don't know if I have a screenshot somewhere, but I had 1.147V stock on a Gigabyte P55W v4, also 5700HQ.