Further OC i7-4810mq 2.8ghZ using Intel Extreme Tuning help?

Anyone know how to further OC i7-4810mq 2.8ghZ using Intel Extreme Tuning?

Or where i can find info on it, to where i can maybe copy someone's specs?



thx

 

You need to flash an unlocked BIOS to your laptop. After flashing, you will be able to increase multipliers by 4 and change the BCLK. These BIOSes are made by Prema, you didn't mention the model of your Clevo machine, so we don't know if such BIOS is available for it. Anyway, Prema's BIOSes are available here: https://biosmods.wordpress.com/

Prema is a legend, donate to him if you can. He has done unimaginable amount of mods for the Clevo community.

 

I don't think you need to unlock it by the looks of that picture.

You up the multiplier to your target speed, see what voltage is the minimum for the speed and see if that is thermally stable. You will have to increase the current limit up to 112A.

 

correct, bios already unlocked!

some said max is 400mhz?
i dont wanna play around with it too much

 

Yes it's a max of 400mhz so if you want to target 4ghz (don't bother trying 4.1 and 4.2 for 2 and 1 core, keep them all the same) for 4 cores then see what voltage it's stable at and go from there.

 

Overclocked my CPU i7-4810mqCPU +400mhz 4.0ghz

** at first i had the first two cores at 42 & 41 but was unstable.
got 3 restarts and one with blue screen.

 

Yep which is why I said don't bother with it the extra voltage needed to make those states stable is not worth the quad core power hit.

 

still getting pretty hot, had a restart after 15min of bf4 gameplay.
was getting low 90s high 80s in high action and 100s in reg play, crazy ahah.

gonna need to get temps lower too, gonna still play around.
open to more tips

 

FINALLY GOT IT STABLE without any crash/restarts after like 10 restarts haha
passed stress test of 5-10ish min ( i know you should do at least 25min+) and also 10-15min of bf4!

now i just gotta fine tweak this shiet




moving it slowly, keeps on getting better

 

Try and keep the posts just to one thread

 

:-D what meaker said ^^

be gentle, hes just so excited

 

YEA! it's exciting and i wanna share my results. yea i will haha my bad. i'll keep it here lol

Here's my stable oc at 4.0GHz with a score of 757. Made the top 10 for our processors!
now to just play around.

if anyone got tips or anything keep me posted!
maybe move sliders to 42, 41?
is it really worth it?
processor cache ratio to 39 or 40?



template i made

 

You are at the max the CPU can do with 4 cores without touching the reference clock.

 

true.
maybe move sliders of core 1/2 to 42, 41?
is it really worth it?

or should i just keep on playing with values.
im sure i could get higher score.

 

Like I said that will increase your voltage requirements while making your 4 core speeds hotter (which is the most commonly used turbo state when you need the performance).

The only way to go further is adjusting the base clock which usually has a maximum adjustment of 7% (if you are lucky).

 

Just playing around still.

When it crashes, what's the first thing you try to adjust?

i like this, hope it's stable-ish. gonna test it out with xtu stress test 30ish min, 5to10 intel burn tests and prime


just some fun files

 

Crashing is usually a voltage issue.

 

bought this cooler to keep it cool. will definitely help with temps



also ran a quick test for my 4.0ghz clock, so far so good.
will run longer once i get my cooler

 

just beware that a laptop cooler also hastens dust build up inside ur machine, so u need to clean it more often than without cooler

 

At least it's not like asus where you need to take the whole machine apart to deep clean it, just remove the main hatch, take out the fans (which are secured separately from the heatsinks themselves.

 

How are your idle temps, OC'd and not? Just got mine the other day and want to make sure I'm in the correct range.


Sent from my iPhone using Tapatalk

 

Im running mine at 4Ghz like you as long as its cool in the room. Not sure you will push much further than that without special cooling changes, or else it starts tapping the throttle and reducing performance. Our processors are very similar, scores differ greatly tho.

 

40/50 i think

holy crap your benchmark score is amazing!!!!
what ram are you running and at what mhz?

EDIT: damn i just saw you running 4910MQ, wonder how i can use some of yur setup similar to mine

Everytime i tough dynamic cpu voltage offset, i always get bSOD!

Working on my benchmark, highest i got was 787.
so far with this setup voltage is way more constant. no little random dips etc. a lot smoother

 

Yes I have found that limit for this model too, it's as much as the heatsink can handle.

 

What size power brick do you have? Your power draw limits seem a little low for 4GHz. I've been able to draw 94W on my CPU package with 3.8GHz at stock voltage. If you have a 230W PSU (on the single GPU models) or a 330W PSU (on the SLI models) then you should relax the maximum power limits to about 100W for 4GHz just in case.

 

His power draw is definitely way too low. His XTU score @ 4GHz is lower than what my 4900MQ rakes in at a BONE STOCK 3.6GHz (~800)!!

I don't have my Clevo right now but if memory serves, for 4GHz on all 4 cores you need to push the power slider to at least 100W. I personally leave the power at 150W and be done with it, since I've never seen my 4900MQ go over 125W even when benching XTU.

Also, that current limit is also probably holding back the actual overclock. Even at stock my 4900MQ seems to do progressively better in XTU all the way up to 95A, and that's only at 3.6GHz. I'd go with at least 120A for 4GHz.

Now lsramair's 4910MQ @ 4GHz is right where I'd expect the XTU score to be at.

Ram speed matters in XTU for Haswell chips (but not for Ivy for some reason), going from 1600 to 2133 gave an 80 point increase everything else being equal IIRC.

 

He's on 130A for current limit though, in the last screenshot I saw him use. But definitely, power draw is too low. Linpack grabs 80W from me at stock with an undervolt, and at 3.8GHz at stock voltage it can pull 94W or more, and I STILL think I'm voltage limited there somehow.

Also the boost power time should go as high as possible; I think that's 128 seconds?

 

Well if he enables the TDP and throttle % graphs in XTU it should display what is going on fairly nicely.

 

I think for the SLI Clevo laptops (P3xxSM), there's no turbo power lock (at least with Prema's modded BIOS), so the boost time is a pointless slider. As long as the CPU doesn't pop its turbo power cap or thermal throttle it will hold the turbo for as long as it can. I've tried setting it to the shortest and longest time and it made no difference as long as the voltage and power and current limits were set correctly.

 

That's true. I was wondering why my clocks were able to sit 24/7 at their max multis in the past since that was set to 28 seconds.

 

Just got my NP9377 last week. Here are my scores in XTU:



My cpu is a 4910MQ. Test is a 4GHz run and Turbo is the 3.9GHz run. Strangely my 3.9G oc is faster than my 4G oc. I did another run at 3.8G oc and it scored the same as my 3.9G run so I may end up keeping the cpu at 3.8G for 24/7 use with no throttling and some overhead to allow for higher room temps in summer.

 

Reran at 3.8G and got this score. Srsly, with a lower multiplier for 4 cores I score much higher even with power throttling on part of the run... Go figure!

 

Try the following changes:
40x on 1, 2, 3 and 4 active cores
turbo boost power max 120W
turbo boost short power max 120W
Processor current limit 130A
voltage offset -30mV

And see if that works. Remember to use max fans and prop the machine up so it has better airflow so it passes the test. If that doesn't work, increase the voltage by 5mV until it's stable (I.E. from -30mV to -25mV then to -20mV until stable).

If it's stable at -30mV try -35mV and go less voltage until it performs incorrectly (lower scores).

 

Or you get a BSOD of course

 

I don't know why people are so afraid of bumping the current and power sliders. Those are limits not "levels" meaning it's basically a cap. The CPU will only take as much as it needs.

Voltage on the other hand is a different story.

 

Well it will make it hotter by unleasing it for longer, care wants to be taken with all the settings when starting out. But yes the voltage is the one to be very careful with.

 

That's true, my response was more of a "why is my XTU score than that guy on HWBot? Hmm seems like his power and current limits are set much higher let me raise mine by 5% and see what happens" type of thing.

 

A lot of people are unsure about what each setting physically does, they are just raising sliders. In that case caution is at least safe.

 

doing small tweaks here and there.
my oc failed after a couple of days while browsing :\
back to the drawing board

 

It may be the idle voltages that are out rather than the load ones if you are browsing.

 

He's been messaging me, saying he left it at stock voltage, so I told him to raise it slightly as he's using adaptive. Unless there's a way to bump idle voltages without bumping load voltages? If so, I'm new to that and would like to learn myself XD

 

No it's all linked as far as I am aware so you have to balance it out for the whole clock range.

 

Current OC, hasn't failed since 12/27
Played aprox 6 hours of Bf4 on it, watched videos, quick prime95 test, stress test from intel, hd video stream, etc
I'm gonna do a stronger test once i upgrade my laptop cooler with small high rpm fans that ordered at 5800RPM.
BF4 temps are in the 80s.
Idling temps are high 40s, or at 50.
Only played with 4 settings.( red boxes)

*also is it worth it to get little heat sinks for msata drive? i have 2X SAMSUNG 840 EVO msata drives (250gb & 120gb)
link: heat sink for memory chip | eBay

 

I suggest setting your cache ratio to 40x, but that's the only thing I can think of that's off about this. You could set current limit to 120A if you want. What machine do you have? If you have the P17xSM-A model, you could get a 330W power adapter and really release those power draw limits XD.

 

I dunno, something seems off, that XTU score of 760 is way way way too low for 4GHz, even assuming 8GB of 1600 ram. Like I said with my stock 4900MQ and 16GB of 1600 ram I was getting around 805 in XTU.

 

If he's got single channel RAM the low score could be explained, but if he isn't using throttlestop or making sure it keeps a high multi, that might explain it. No idea.

 

My 3940XM does about 793 @4.2GHz which is bloody low... I have 1 4GB stick and 1 8GB stick... is that the problem?

 

Could be part of it, but XTU is fairly ram agnostic for Ivy. XTU gives a huge boost if it detects a Haswell chip, and with Haswell both speed and quantity of ram seem to matter a lot.

My 4930K (hex core) @ 4.5GHz with quad channel 1866 ram gives around 1350. My 4900MQ @ 4.2GHz with dual channel 2400 ram gives almost 1100 in XTU. That a hex core is only ~20% faster than a quad core clearly points to something being off about XTU.

If I drop the ram speed to 1866 my 4900MQ @ 4.2 gives around 1020. With the 4930K, dropping ram speed to 1600, or boosting to 2133 had negligible impact (<5 points) on XTU.

 

how about just using a different cpu benchmark? seems like xtu just gives non-reproducible crappy results

 

No the results are very reproducible, but trying to compare CPU performance across different generations using XTU is not a good idea because Haswell has such a ridiculous advantage.

Oh and I forgot to mention, if you have a Haswell chip, HT gives you a nice boost in XTU but if you have an Ivy (and possibly earlier chip), HT is actually detrimental to the XTU results. With HT disabled I get 1400 in XTU with my 4930K, but with HT on it drops to around 1350. Crazy huh?