(Part 2) If U Can Solve This U R King

Improved wPrime score with no more throttling, clocks stable at 3.09GHz throughout the test:

 

@forum moderators, you can now change my username to happy user!

It kinda leaves a bitter taste in my mouth knowing that I had to undervolt to achieve stock CPU scores. Review units on notebookcheck.net, techradar, techporn, and a few other review sites got higher CPU benchmark scores without undervolting, right out of the box. Unless they are being dishonest with their readers but I doubt it.

Anyway, I will soon post a summary of all the working solutions to my problem in the original post, to help future readers with similar problems and save time. To be followed by an acknowledgement post thanking everyone who contributed, and crowning the winner King.

Thank you all.

 

Want your username changed? Request here

 

@Vasudev If you can provide an HWinfo screenshot like the one below while running CB R15 it would be great. It would help us compare power, clocks, multipliers, temps, etc...and understand why I cannot overcome the 673cb barrier even after undervolting by -100mV. Other MSI GT62VR users who undervolted using XTU are achieving 682cb-687cb.

These are my readings while running CB R15:

Spoiler: HWinfo during Test

 

Macmyc's score vs My score before undervolt: 687cb vs 660cb
Machmyc's score vs My score after undervolt: 684cb vs 678cb

Macmyc's OPENGL score vs My OPENGL score: 98fps (GTX1060) vs 81fps(GTX1070) !!

Screenshots in spoiler below.

Note: @macmyc has GTX1060 while I have GTX1070. And he is not using Throttlestop.

Extremely disappointing because i cannot think of any other reason why my performance is crippled like that. How come GTX1060 score much higher than my GTX1070 on OpenGL test?!

Spoiler: Macmyc Cinebench R15 Scores

Spoiler: My Cinebench R15 Scores -100mV Undervolt

 

Well my score dropped to 678 on normal mode with hwinfo with 68-75C on all cores.
What did I say? Try the undervolt profile I gave you. Uncheck speedstep,BD PROCHOT and check only speedshift and c1E.

 

My single core score in cinebench r15 dropped from 143cb to 126cb. Is this because of the undervolt or because of the recent intel ME firmware security patch which i applied? I ran the test 3 times after a cold restart.

 

Probably cores who handle other tasks in the background. Single core clock speed will drop if more than 1 core is in use. Read previous posts regarding clock speed for 6700Hq.

 

I monitored the core frequencies during single core test. 3 cores ran at 3.2GHz and one core ran at 3.3GHz. A core never hit 3.5GHz.

Anyway I always got a score of 143cb why now 126cb? I tested 3 times after cold boot, cpu was idle at beginning.

Could it be the Intel ME firmware patch or the undervolt? Anyone with i76700HQ and the latest intel ME firmware patch can run cinebench single core test please?

 

considering that single core performance is the most important aspect in gaming I'd want to know what single core performance you're getting @Vasudev

 

What is happening? Average score for this laptop is 143cb. i can't reach it anymore. This is @-100mV. Look at the maximum core clocks.

 

I restored the voltage back to default (0mv offset) and got perfect single core score again. Phew! What was the problem?? Power consumption increased by 3W when i removed undervolt, and so did the score.

 

So now I have to choose between good multicore performance (with undervolt) OR good single core performance (without undervolt) ?!

EDIT: No that is an inaccurate conclusion, see post below.

 

My above post turns out to be not accurate. I reran the MULTICORE test with NO undervolting and got the same high score I got for the first time when I undervolted (673cb). CPU Package Power did not exceed 41W this time without undervolting, which helped achieve a high score. How come CPU did not reach power limit without any undervolting? Why never before, why now?? Used to be 44.9W during this test at stock settings.

 

My highest score ever in CB R15 multicore test: 679cb and without any undervolting. I really do not understand what is happening now. Why did my CPU decide to not throttle anymore and consume less power at stock voltage? All i did was undervolt by -100mV using XTU then reverse it back to 0mV, restarting the laptop after each time I apply the changes. This is confusing.

The erratic behavior of my CPU power management concerns me.



680 after rerun without restart:

 

Told you my cpu is limited by FW.

 

Filty firmware.

And cpu (6700hq) will run on max boost on single core the whole time if you play games? What with other tasks in the background? Handled by the same single core? And are the rest of the 4 cores turned of or parked? Oh'well.

Turbo Frequency 3,500 MHz (1 core),
3,300 MHz (2 cores),
3,200 MHz (3 cores),
3,100 MHz (4 cores)

 

Any comments Captain Papusan for my bench results?

 

Lower than your best

 

@Vasudev, your single core result is low, I achieved 147cb by removing the undervolt. Maybe you would like to try the same? 694cb in multicore is awesome. I wonder if it is that speed shift setting in TS + undervolt that pushes it that far.

When you say firmware what do you mean exactly? The Intel Management Engine Firmware? Or the System Firmware provided by MSI? or BIOS? or...?

And any clue why my scores are high all of a sudden after I removed undervolting?

 

I forgot to turn off WD.
Did you see a icon to the left of TS? Its a UWP version of Powerplan Switcher.

 

Both.
Put some intel services to manual or stop them.[/USER]

 

Instead of completely removing undervolt - try dropping it down +10mV at a time and benchmarking, until you reach a happy balance of lower temps and best performance.

When you could unlock the cores from the cache undervolt, the best performance came leaving the cache undervolt less than the cores undervolt. You would tune for the best undervolt for the cache vs performance, then tune the undervolt for the cores based on stability.

Have fun

 

Here you go.

 

@Vasudev thank you very much. This confirms it: Undervolting reduces the single core performance significantly. I will avoid undervolting for now, and I am getting high multicore scores @41W without any undervolting. Confusing but whatever.

Now why is my OpenGL Score so low? 81fps.

 

Your driver.
Maybe my ambients being high is the cause. Now its 30-32C room temps and cpu temps are 35-38C.

 

Read!!

Regarding Undervolt. All to massive undervolt and thing will be screwed up. Use the middle way as @hmscott pointed out.

No need for it.

 

Why don't you test BF3 again instead of benchmarking over and over again...

 

-50mv in my case it improved it slightly.

Spoiler

EDIT:
Reran with no undervolt:

Spoiler

I'll assume margin of error, so basicially the same performance for me.

Edit:

Spoiler

Ran again without undervolt, it seems I do indeed lose a little bit on undervolt in single core. Which is weird considering how my multiscore isn't changed by it at all. With undervolt 146 - 147, without undervolt 148-149, done with 4 tests each, I only once achieved 148 with undervolt on my very first test.

 

Yeah, it takes time before people learn it the hard way. This is Old knowledge @Mr. Fox

 

Guys do i need to restart everytime I apply new volt settings in XTU?

@Danishblunt, yeah
with undervolting -100mV:
Single core score range: 126cb-134cb
Without undervolting:
Single core score range: 147cb-148cb (only today, historically it was 143cb)

Undervolting indeed hurts single core performance. Too much of it or not, i have to find out by calibrating for the sweet spot.

As of now, 2 things are very puzzling:

1. How come my multicore score today is 680cb (no undervolting) while it has always been 660cb. I did not change anything today except undervolting to -100mV and then reverting it back to 0mV. And no new windows updates. if Bad Firmware is causing this behavior how do i fix it?

2. The OpenGL score 81fps for my GTX1070 while the GTX1060 is scoring 88fps-98fps. I agree it is irrelevant since 3DMark scores are very good, but why.

 

There you go. I am getting my highest recorded framerates ever. Between 110fps and 160fps in metro map multiplayer, average is 138fps running around the map. Better than when I first activated dual-channel mode but I don't know why.This is in line with my better Cinebench scores today too.

Today: Dual-Channel Mode + New Cooling Components + New GTX1070 (+20fps increase)


November 2017: Single-Channel Mode, Old & Replaced GTX1070, Faulty Cooling Components:


When I first activated my dual-channel mode this month, my max fps was 150fps in BF3 multiplayer metro map. Now it is 160fps and my CB R15 scores are higher than ever @680cb and 148cb. i have no idea why and I am afraid to shut down or restart my laptop and lose this momentum.

EDIT: i installed the latest xtu from intel website today. That's the only change i made.

 

Your GPU is still running at only 60-65%.
I think if it worked at 100% you would have about 200 fps

 

Are you sure people can run bf3 with their i76700HQ and gtx1070 at 99% gpu load? Do u have proof? If yes how can i fix this? My gpu runs at 100% in mass effect andromeda and the witcher 3 but not bf3.

 

^single player not multiplayer and desktop cpu @4.6ghz. It keeps the gpu busy. My i76700HQ is a mobile chip at 3.1ghz. I believe it is a bottleneck in bf3 that does not push enough frames for the gtx1070 to keep it fully busy hence the low gpu load. When i activated dual channel gpu load increased.

Still if u come across anyone with a GTX1070 and i76700HQ with full gpu load in bf3 plz let me know.

EDIT: I ran singleplayer and gpu load is 85%-99%. This is because the cpu is under more pressure in multiplayer due to the presence of human players so it becomes a bottleneck and does not feed the gpu fast enough. My fps in singleplayer is now 180-200fps. Multiplayer always has lower framerates in my case.

@Semmy
Single player:

 

Even the GTX 1080 uses up to 90%. I think with good optimization it should use up to 100%

 

See his description i7-5820k @4.5ghz so no cpu bottleneck. Usually the GPU runs at a lower load when there is a CPU bottleneck. U will never find 100% GTX1070 load with i7-6700HQ running BF3 multiplayer.

@Vasudev do u have this game multiplayer?

 

No. I have Witcher 3, Hitman 2016, Witcher 2, Sleeping Dogs Def. edition, GTA 5,3dmark, Thief 4 and some others as well.

 

I have posted same answer... I don't know how many times in this and other theads Firmware!! + drivers.

You could read my answers and be done with it... Or simply use a Google search. As you can see... Big variation depending on what OEM - Model - Firmware and drivers. A 1060 can beat 1080. Hope this will clear up!!
Cinebench R15 openGL bench for 1080


Cinebench R15 openGL bench for 1070


Cinebench R15 openGL bench for 1060

 

Your i7 6700HQ is only used on 50-60% of your videos. So I do not think this is a problem for the mobile processor. Rather, it is a problem of poor Windows optimization and game optimization, which can not use the processor and graphics card as it should

 

Solution?

@Papusan, Thank you I read you well. I will ignore OpenGL scores. Here are my firmware and driver versions:

I have the latest EC Firmware from MSI Support Page version 16L2EMS1.107

I have latest BIOS version E16L2IMS.116 from MSI Support Page

I have latest Intel ME Firmware 11.8 from MSI Support Page

Device Manager says I have Intel Management Engine Driver v11.6.0.1032

Device Manager says I have Intel Chipset C230 LPC Controller HM170 driver v10.1.1.38

What exactly is faulty among these and needs to be updated?

 

This is HP OMEN 17 with a GTX of 1070 and a mobile I7 of 6700HQ by 3.1 GHz. in the multiplayer is under great pressure due to the presence of players in this heavy game BF1. But the GPU holds 99%

 

If you look at his video description on youtube it's because he has Thanks for watching in his specs. I dont know whats that i dont have it.

 

I don't have MSI (I despise BGA). But do not have too big hopes that the OEM's aka the brand will make something decent. Crippled hardware, Firmware and software are all they create. Now and in the future!! Read also my thread... How Dell cripple performance

 

This MSI GT62VR 6RE holds up to 99% of the GPU in BF1

 

I have to disagree on this one. Some of the Other users who have better performance are msi owners. I should ask macmyc and danishblunt or maybe phoenix for their version of firmwares and drivers used. If you are an MSI owner, please report your:

EC Firmware version

BIOS version

Intel ME Firmware version

Intel Management Engine Driver version

Intel Chipset C230 LPC Controller HM170 driver version

Thanks.

@Semmy, thanks for the insightful videos but so what can i do? I mean look it is 40 pages now and I already replaced my GPU with a new one, replaced the whole cooling module, and installed a second RAM stick, and reinstalled windows 10, and still GPU does not reach 99% in BF3 multiplayer. What can be done dear? What else can be done?! Maybe bf3 multiplayer is not optimized.

When i installed win10 I blocked automatic driver updates of all devices using the policy editor. Could that be a "crippling" problem? Shall i enable automatic driver updates through windows updates?

 

try a few other games (3-4) with multiplayer .. To check and compare how the GPU will work there

 

dno some weird old firmware from 2016 is on my notebook, playing bf3 now on 200FPS multiplayer.

 

You can also enable the adaptive power management mode in the Nvidia settings and disable CpuStates in bios.
And disable the CPU core parking