The processor can't reach it's top core speed

Edit: http://forum.notebookreview.com/thr...its-top-core-speed.775350/page-6#post-9994273

Tuned



Hey guys, sorry if i posted in a wrong section

i have a laptop with core i7 4712HQ .. OCd to 3.5 Ghz

when i put the cpu under load with prime95. the maximum frequency it can achieve is 2.56 Ghz

what's the problem? it doesn't exceed 2.56. i feel no heat on laptop though

my other laptop with i7 3517U can reach it's maximum clock easily





no idea why is overclocking is available for this processor

 

You're hitting your power limits. Try upping the Turbo Boost Short and Boost Power Max Limits to something like 65 or 70W and see what happens.

You can also try and undervolt. I would start with something like -50mV on the dynamic CPU voltage offset. If thats stable, then continue to drop it down -10mV increments until you start getting BSODs. Once you get a BSOD, increase it back 5mV from the BSOD and check for stability. Repeat that until you find your lowest undervolt that maintains maximum clocks when under load.

 

you mean these options? wouldn't that harm the laptop? i paid a lot of money on it and it has no warranty



tried this but was so afraid that i disabled the test after 10 seconds .. it achieved 3.1 Ghz

 

According to notebookcheck, the i7-4712HQ reaches 3.0 GHz on all four cores, 3.2 GHz on two of the four cores, or 3.3 GHz on 1 of the four cores. So if the four cores are around 3.0 GHz at stock, you should be fine. I don't think a minor undervolt should have long term damage.
Since you overclocked it, though, I would think 3.1 to 3.2 GHz should be about the expected frequency.

 

Well there you have it. You asked why it was throttling down, and I told you why.

The CPU needs the power to hit those clocks you want. No it shouldn't harm the laptop as long as you keep your temps in check. Set a goal of keeping your CPU below 90C. It isn't going to allow you to actually set 70W, I just had you set it to that so it uses the maximum allowed by the chip/manufacturer. Its like getting free performance. You will achieve best results with an undervolt and using the higher power limits.

Undervolting your CPU will NOT cause damage to your CPU, if anything you will extend the life of your CPU. I don't know of anyone that has killed a CPU due to undervolting. Overvolting a CPU is a completely different case.

 

after increasing the Turbo Boost Short and Boost Power Max, the laptop limits it self to 37 W.

under volt -50 mV increased the TB to 3.2 Ghz then 2.9 Ghz!

EDIT: -70 now and the TB is between 2.9 and 3 Ghz

overclocking the stock clocks + undervolting was better for me .. -90 i get stable 3.1 Ghz with lower temps! i can undervolt more but i think -70 is the best for now

is that good for overclocked 3.5 Ghz CPU ? any side effects?

the CPU Utilization isn't a stable 100% will that cause any problem?

EDIT: Got BSOD on -90 .. back to -80 but the maximum i get now is 2.4 .. any idea what is happening?

 

Hey, are i7 Haswell CPUs at freq. 2.5 GHz adequate for gaming? No turbo boost, only stock speeds. Will it cause bottleneck in games?

 

When you BSOD, Intel XTU will reset default settings to prevent another BSOD. Make sure you set your overclocks and undervolts and apply them again.

 

Wow... 37W TDP... If you go look at NotebookCheck, the 4712HQ performs less than the older 4710HQ because of the 10W drop in TDP. You'll never get that chip to hit full sustained turbo speeds unfortunately. It is TDP locked to not pass 47W.

 

yeah noticed that. i made a table for the results i got

which one should i apply they are very close and anyway to get closer to 3.5 Ghz ? the clocks are stable on 3.0 3.1 Ghz

okay i want to cry.

 

It shouldn't throttle as much during gaming... But yeah, even my 1k 4940MX doesn't hit its full turbo out of the box. It has 20W higher default TDP than yours does yet when stressed to the max, it runs at 3.5GHz, a far cry from its turbo bins of 3.8GHz with 3 or 4 active cores, 3.9GHz with 2, and 4GHz with 1. My chip has an unlocked TDP but the heat it puts out still makes it difficult to actually maintain my turbo bins.

 

increasing the turbo boost power max results in a weird stable 3.2 - 3.1 Ghz .. TDP 37W

should that harm the laptop? considering the temps under 90 C ?

So the heat is a strong factor that causing the throttling.

 

Your temps are fine. Your problem is the power that it takes to sustain your turbo bins. I'm impressed that you have 3.1GHz at 37W

http://url.hwbot.org/1wnhwS4

Extract, right-click, run as administrator. Go to advanced settings, set threads to 8. Save. Run the 1024M test with XTU in the background and see what your clocks are. If they're 3.1GHz, you successfully fixed your throttle.

 

okay, that's scary?



Longer

 

Go to advanced settings and set the thread count to 8 and try again. I initially left that out of the instructions.

 

stable 3.2 then the drop. 37W TDP is an evil

 

I decided to try my chip with your chip's constraints

That's with a full locked 37W TDP the whole time.
I guess again, its the Haswell silicon lottery :/
My chip has a 3.1GHz base clock - without the 80mv undervolt, it runs 3.1GHz @ 38W

 

about luck, when i bought my laptop from HP online store i paid 1500$ and got a different weaker model with a very bad 720p screen, HDD instead of SSHD and a non working msata slot

lost 300$, got a weaker model and i can't replace it.

i guess my laptop results are good for an HQ 37W processor and i got a good chip .. that's made me feel better

thank you very much

 

You should be able to get higher speeds under a lighter load as well. What happens when you change the 8 threads to 4? That's probably going to be more realistic in every day usage scenarios.

Actually... can you raise the 53A to 60A and try it? I forgot to change the 53A back to 70A for my chip and it remained at 37W. If you can raise that, you may be able to get the chip to run up to 47W.

 

Looks much better.

Now the processor current limit... will it let you bump that from 53 to 60?

 

wouldn't that harm the laptop or battery? i can change the limit to 60

the laptop is HP Envy 15 not a high gaming laptop

 

You can set it much higher w/o worrying. It only draws as much as it needs.

 

Exactly.

From testing my chip just now, it needs 70A for 47W.

 

Increased the processor current limit to 60 .. the difference is the utilization is stable 100%

8 threads test

 

Try 70A

 

70 A

what should i keep? 60 or 70 or stock? the difference i notice is CPU Utilization and actually i don't know what it means

 

Unfortunately its not making a difference. Do you have a setting that allows you to change the turbo boost short power time from 28s? If you do, max that out.

 

this option is disabled. should i reset the current limit or keep it 60, 70?

i guess we squeezed the most juice out of the CPU

 

It doesn't matter what the current limit is at, 53A is all that it needs. You can try -85mv for your undervolt to see if that gets you a solid 3GHz but it looks like we've hit the wall.

 

actually on most of the tests it starts with 3 Ghz but then a drop to 2.98 "with prime"

-85 undervolt and 47.250 turbo boost power max gets me better results .. stable 3.09- 3.1 Ghz with XTU Stress test, i guess that's the maximum power.

 

Well its certainly better than where you started.

 

Is there any point in getting 4712MQ over 4702MQ, with XTU tuning in mind, as a replacement for i3-4000M? Is it safe to try using full-voltage CPU and undervolting it heavily if the notebook was offered with 4702MQ max?

 

There's no point in getting either unless they're MQ instead of HQ

 

Absolutely not. Clearly Intel isn't binning these chips right. I wouldn't get a 37W chip.

The only difference between them is the extra 100MHz base clock and the 100MHz higher max turbo bin on the 4712HQ, which we can see obviously isn't getting hit.

http://ark.intel.com/compare/78932,75118

 

octiceps, Ethrem, thank you for your answers, however I've corrected the question a bit, could you please take another glance?

 

Depends on if the mobo can supply more than 37W to the CPU

 

octiceps, what happens if not? Something burns out, or it simply doesn't work? How do I discover if it's able or not, other than trying? Schematics are not available.

 

If it works with 4702MQ, it may or may not POST with 4712MQ. It depends on if the BIOS recognizes it.

 

octiceps, Ethrem, I meant 4700MQ, just in case.

 

If it works, it just won't be able to use more than 37W so there will be TDP throttling like what OP has. Speaking of which, OP's laptop seems to have this mobo power limitation as his 4712HQ won't even boost up to its short duration power limit which is odd.

 

octiceps, so, it shouldn't cause motherboard failure, and 4700MQ will either throttle to 4702MQ levels, or simply refuse to work, am I right?

 

If it works with the 4700MQ, that's a 47W chip.

 

Right

 

*sheds a tear for a user with a HQ chip who cannot acquire the performance*

 

*hugs his P370SM3 tighter*

 

I Would like to share my settings for tuning I7-4712hq to get better clocks than stock, if any one cares.
i think they are good comparing to the stock results.

frequency recorded with XTU

Test With wPrime

1 thread > 3.4 Ghz - 3.3X Ghz at least
4 threads > Solid 3.2 Ghz
8 threads > 3.0X Ghz

Test with XTU Stress test

3.1 Ghz

Appreciated ;_;

 

I have a question. is there any benefit from undervolting processor cache voltage offset and is it safe?

and is there any way prevent XTU from resetting the profile i made. some times it resets it self to stock

--EDIT: for the one who asked about processor performance in gaming. i tried GTA V for several hours max settings

the processor didn't throttle from 3.2 Ghz as i have a decent cooling system and the temperature doesn't exceed 65 C

weird thing from HP

 

Always an advantage with lowest possible voltage... With all too low voltage at processor cache voltage offset, you risk instability... No.

 

I would even recommend testing with a reduced Processor cache ratio. I have little to zero performance loss going from 36 to 30. And every little you manage to undervolt stably, you get a bit of headroom in power, that you can use to maintain higher clocks under stress without reaching TDP limits.