*** MSI 16L13 (Eurocom Tornado F5)/EVOC 16L-G-1080 15.6" Owner's Lounge ***

Here's what little I know, but I have to know your prime95 8 thread (Small FFT) AVX test. (large FFT will not cause enough load).

Offset "E3" in RW Everything contains the power ID. This powerID varies, and I do NOT know if EC firmwares contain power ID's of other systems of different EC's, but the 16L1's 330W powerID is "33", and the 230W ID is "32". Without a prema mod, a 32 "set by a GTX 1070 being installed' limits power draw to 230W, and 33 is for GTX 1080. People with stock 1070's (32 ID) will have CPU throttling if they exceed 230W.

From what I saw on the GT73VR, Skylake has two power ID's: 10=230W (1070), and 11=330W (1080).
Kaby lake has four power id's, probably due to lazy programming: the previous skylake power ID's and the kabylake ID's (90=230W->GTX 1070, 91=330W->GTX 1080).
Exceeding the power ID in total power draw will cap the CPU at 45W. If 45W cap STILL exceeds the power ID's max power draw, THEN it caps the CPU at 25W. It can't cap lower than 25W.

Now I did find something funny.
If you use a powerID that is not identified as belonging to a system (let's say, 89 or 09 or something), then the CPU gets hard capped at the TDP limit, regardless of what is set in the Bios OR current CPU speed. You can set the CPU to 3.5 ghz and still be capped at 45W by using an unidentified power ID.

 

Prime95 26.x won't draw enough power. Only uses Pentium 4/type 0 FFT. 29.x + Small FFT test will (cook an egg on your cpu).

 

And, there you go, bro. As I said... predictable.

Just over stock 91W TDP at only 4.5GHz with Prime95, held for about 8 seconds after it went over 91W before the MSI cancerware took over and screwed it up. Ain't it sweet? No thanks, MSI. I don't think so... losers. It could be sheer incompetence, but I'd put my nickle down on it being an intentionally engineered gimping exercise. Screw the customer... they're looking out for #1. It's a laptop, so it cannot be taken seriously, right? That's why BGA filth is OK, too.

 

@Mr. Fox
Can you do me a huge favor?

Sorry for asking you for all these favors but im trying to return all the help you gave me in the past.
Can you download RW Everything please, and post a screenshot of your EC RAM page (EC tab)?
Because that should not be happening...

 

Already had that one. Here you go.

 

It’s a reason Intel implemented AVX Offset Ratio feature. Old P95 is enough for stress testing. Aka more equal <normal> load than newest P95 version with avx.

 

@Mr. Fox
Change offset E3 to "33" then run prime95 again at the same mhz.

BTW this is me drawing 98W on a 7820HK at 4.3 ghz on prime AVX (45W TDP).
4.4 ghz will error out in a core. If I go too high on the voltage or loadline calibration (AC/DC loadline), the system doesn't throttle...it just shuts off and reboots. Note I can run prime at 4.5 ghz with AVX/FMA3 disabled (through local.txt from undoc.txt in prime folder) all day long but that draws less power.

 

Seems like that helps. It is still being throttled, but not as bad.

 

mine on win7

 

Just like I thought.
You may need to contact @Prema . Your powerID is incorrect.
The default powerID for a 330W 16L1 is 33, and for a 230W 16L1 is 32.
I knew this from other screenshots people posted about their 16L1 and someone showing me a 1070 ECRAM in a PM (32 ID).

"B3" is an unrecognized power ID. Remember what I told you about the GT73VR?
Any power ID that is unrecognized (e.g. not 10, 11, 90 or 91) hard caps the CPU at TDP (45W). <--these values only apply for the GT73VR. I tried using "33" on my GT73VR and got greeted by 45W TDP since it doesn't recognize 16L1's ID's.

I have no idea how your system got or why it has B3 as the power ID.
Prema must have changed something. But the EC has built in code where if it doesn't recognize the power ID, it will hard cap the CPU.

@aaronne are you using a premamod/Prema EC or stock EC/stock Bios?

 

@Falkentyne

I think I got the Prema's stuffs because my bios is full of God's gifts



Going to ask on Tornado F5's official thread

 

Well, I have stock EC files as well, and they throttle worse than with Prema's EC mod.

 

Is Eurocomm a Prema Partner, though?
Remember, Eurocomm offers an "unlocked" Bios, that is not a prema bios.
I do NOT know if Eurocomm has a prema Bios AND their own unlocked Bios.
Prema's' Bios also requires the Prema EC (as far as I know).

HIDEvolution is a prema partner.

Mr Fox I wasn't saying to use the stock EC files. If you are not throttling (or throttling LESS) by changing the power ID to 33, then you made progress (perhaps MSI has a 110W power limit on the CPU rather than TDP, or you ran into some EDP Current limit issue). All I'm saying is, the power ID for your model should be 33, not B3.
Are you having any drawbacks from setting the ID to 33? Maybe B3 was set for some reason, but first you need to test 33 fully. You can always write a script that runs RWE through a batch file and a .RW file (I think the command in the .RW file is " >WEC 0xE3 0x33" without the quotes). I have one for my 330W ID, and another one for disabling the "artificial" power limit for CPU throttling when the battery is internally disconnected (Otherwise I'm limited to 150W, although the videocard can draw more than that, the CPU then gets capped to 25W!).

 

@Mr. Fox
Set offset E4 to "10". I dont know how it got set to 11, but I saw that happen before. 11 cause or is caused by PL2 or some other throttling (Bios (MSR) or PECI I don't know which), but I think something actually sets it to 11 as some sort of flag, yet I've only seen this happen ONE time to me. I actually had that happen ONCE on mine the day I found out about offset E3 (but before I knew E3 was the key) and I remember changing it to 10 and the PL2 throtting got turned off. I'm just a novice, but if some weird thing is setting it to 11 live, setting it back to 10 while at load might fix that. It seems setting it 'manually' from 10 to 11 doesn't do anything at all.

But something changed it to 11. You may be able to bypass the issue if you can keep it at 33 (E3) and 10 (E4) without E4 changing on you.

 

I will play around with it more and see if it is really better or not. I opened Prema EC with a hex editor and 0xE3 is not B3 or any other value mentioned above. So, I checked other stock EC versions and they are not either, so maybe something else is triggering it to flip to B3.

 

Mr Fox, are you talking about the "Raw" EC firmware?
You can't find these values with a hex editor. They are set in "EC RAM" on bootup. You would have to use Linux to disassemble the EC firmware into code (and that might require more checksum BS), and the EC code may be setting decimal values, but then it gets read as hexadecimal.

For example: Offset 42 is the battery charge level, but this is in hexadecimal. 64 is 100% (100 decimal. 16x6+4=100.

offsets 40 and 41 are charge on or off status. 8C 0A means the battery is charging. 00 00 means it's not charging. offset 31 with a value of "03" means charging, battery LED light is on. 09 means "Fully charged". I'm not sure what the "05" means in aaron's picture.

Could be the prema bios does something to prevent hybrid power draw if you have "8C 0A" status but the battery is not charging on your system.

I'm already seeing some strange stuff with your Prema EC vs the stock EC.
For example: You have 0C in offset F4. This is the normal value for GT73VR (all models) and GT73VR SLI. In fact your values look very similar to GT73VR.

Yet Aaron has 4C and another chinese user had 8C in his.
I don't know what that "85" is in @aaronne 's F5 offset. GT73VR has 03 in offset F6, the L1 seems to have 05 there.

Offset C6 seems to control some sort of "low power throttling" flag. I observed a change from C1 (or C0) to 4C (or 4 something) when battery was less than 30%, which causes a 150W AC power limit. Changing this manually back to C1 or C0.and it just gets reverted instantly back to 4X.

Offset 98 controls the turbo fans or some sort of fan curve that may need another offset in addition to it.

Offset 80 is GPU#1 temperature.
Offset 9F is GPU#2 temperature (For SLI systems only).

I have been unable to find any drawbacks from changing (offset F4) 0C to 8C on my GT73VR, however @sirgeorge changed it from 0C to 8C on his 1070 SLI, and changing it to 8C makes Orthos PSU test (GPU test) fail instantly, and makes his computer do a "windows shutdown" while gaming at load on his SLI system (it doesn't just power off like overloaded VRM's do, it shuts down gracefully). Once changed to 8C, changing it back to 0C didn't do anything; he had to power off the system to get it to not shutdown at GPU load.

 

I'm on battery power at moment,what version of Prime95 to use to check it right?

 

29.x. And don't even think about it if you are not on AC power. You're going to throttle more than a drunk hobbit XD

However you are going to reach 100C within *SECONDS* if you try using Small FFT AVX if you are not using liquid metal ! A desktop is no problem with normal thermal paste because of MUCH higher static pressure and miles better cooling.

 

So, what I am hearing from all of these posts is MSI firmware is just an awful, disgusting abortion. But, I already knew that from the mess @Prema and I spent months working on, and before that the time I spent with Svet working on firmware for Eurocom. I have done some more playing with RW-E and setting it to 33 is actually worse than B3. Cinebench throttles all the way down at 4.7GHz (drops to 4.3GHz or less) and crashes before it completes unless I leave it set to B3. And, I am convinced that there are some hardware limitation or something else that MSI did to keep this machine from running full blast. If it has any shred of the hybrid power crap they are famous for, that may be screwing things up as well. Stupidity of this nature is why I am getting fed up with laptops. So sick of this kind of stupid crap.

I have the same EC version as @aaronne that I can flash, but I already tested it thoroughly and it throttles worse than the @Prema EC does. I could flash it and see if the RW-E values match his, but I already know from using it that it will only lower my performance. I tested about 8 different stock EC versions, 12 modded versions by Svet and more than that many tweaked by @Prema and the behavior is the same or worse.

 

Ah, so prema had to make a compromise.
basically, get silly crashes and absurd throttling with the default power ID (of 33), or get full power BUT a 91W unavoidable power limit with B3.

@Mr. Fox , see if Prema can re-write the mod to avoid this. I think he was not aware of this "issue" with the power ID's.
If you can tell him my findings, he may be able to make a new mod for you to remove ALL throttling. I'm in fact 90% sure he can do it now.

just give him my findings.
Tell him that MSI's cancer firmware has "built in machine detection" stored in offset E3.
GTX 1070=230W (= 32 on 16L1, 10 on GT73VR 6RE, 10 OR 90 on GT73VR 7RE, it defaults to 90 but the older 10 still works), GTX 1080=330W (=33 on 16L1, 11 on GT73VR 6RF, 11 OR 91 on GT73VR 7RF, it defaults to 91 on 7RF, but 11 still works).

On GT73VR, using any ID not identified as belonging to a preset enforces 45W TDP, at any CPU speed, unavoidable.

If the EC does not recognize the power ID, it force an absolute TDP rating on the CPU, ignoring anything in the Bios completely.

If you can give Prema this information, he may be able to rewrite his mod, as this seems to be a simple fix (since he might not have been aware of this TDP limit).

i'm willing to literally bet $1,000 that he can fix this problem for you, now that we know exactly WHY you have an unavoidable 91W TDP limit (because B3 is not identified by the firmware as a machine ID).

And yes you are 100% completely correct here.
Clevo clearly uses cancer firmware, but Clevo only gives you cancer, which can be cured by Chemotherapy (Aka Prema power). MSI actually gives you EBOLA. YEAH. EBOLA.

Clevo doesn't use some stupid ass power ID and then super hard caps the CPU at TDP if the ID doesn't match....

 

OK, I will ask him to look at these posts and see if he thinks anything can be done. Thanks for the help. Even if it is a true hardware limit and cannot be fixed I appreciate the effort.

 

From what little I know about coding when i studied it, this is simply a machine check.
Whatever prema did to make "B3" for your power ID would have to be recognized as a 330W power ID (or custom power ID, rather than "Unknown.". If Prema removed the 330W power restriction so you can draw as much power as you want until the machine shuts off, then the EC firmware still has to recognize that ID, which it doesn't, which is the reason that you have a 91W TDP at all cpu speeds.

I know this can be done because I can exceed 91W on a 45W BGA Trash chip, so you should not be limited. But yes, MSI is clearly recycling code from its BGAbooks @Papusan
It just raised 45W to 91W.

When I tested the 17A1EMS1.105 EC firmware on my Gt73VR Kaby lake, I was wondering why I was super hard capped to 45W TDP Even at 3500 mhz. But that's because my power ID is 90 (230W). But this EC firmware does not recognize 90, because it was made for the Skylake (GT73VR 6rx), and only Kaby Lake added id's 90 and 91 in 17A1EMS1.108. If I had known at the time, which i didn't, I could have just used RWE to change the ID to 10 or 11, which that particular EC would recognize.

Oh man did we do the most epic thread crap, Mr Fox! i just realized that.

 

LOL for the drunken hobbit!
LCU for life but when my new 7700k arrive i'll go with condoctnaut that is already here!

So you right!, i'm at 4,3Ghz -165mv adaptive, after passing 91w the clocks throttle goes down to 4,1/4,2Ghz instead of 4,3Ghz!!

Argh!!

So there is no way to circumnvent this crippled thing?

 

I wont know until I get brave enough to run my GT73VR higher than 4.5 ghz to see if I can run it all the way up to the point where the VRM's can't even handle basic load anymore. Right now, instead of throttling, my system just shuts off if the VRM's get pushed too hard (at first I thought this was a CPU wattage protection from the EC, many of my older posts where I complained about "system shutting off" instead of throttling, I thought were due to using the wrong EC (17A1EMS1.112 vs .108) but thanks to sirgeorge's help, I found out this is VRM overload caused by "Automatic" loadline settings getting applied, which I then proved by setting STATIC voltage (instead of adaptive), and getting VERY high power draw (like, 90W) with 4.5 ghz prime 95 with AVX **DISABLED**, with the stock "Auto" AC/DC loadline (which defaults to 0) in Core IA Domain, but when I set loadline manually to 25/25 AC/DC, power draw drops from 90W to 76W!!!!!

"My" throttle issues were because I have a 230W Power ID because I have a 1070 installed. Going past 230W power draw enforces TDP limit. If TDP limit is still exceeded (like, by a modded video card), then 25W TDP limit gets triggered. Changing the power ID to 330W removes this, but I know of no way to remove the 330W limit, because the "SLI" power limit (base AC power x 2) seems to require activation of the second MXM slot, rather than a RW Everything register.

And then I can set 4.3 ghz, enable AVX prime small FFT and push 98W without the VRM's shutting down with 25/25 AC/DC. Using stock MSI settings (0/0)= instant shutdown INSTANTLY. Using 125/125= shutdown in a few seconds.

But i'm talking about shutdown issues, not throttling.

 

BTW - I flashed the same EC as @aaronne and look at the info. Same as @Prema BIOS... B3. So, I flashed EC 111. Same... B3.

@aaronne - did you change yours with RW-E, or was that the default value of 33?

 

Default, Mr.Fox this is inexplored path to me

Heres while running

 

I wonder if it's the BIOS that's setting the ID?
It could very well be the Bios itself, since the Bios does send information to the EC!
That would explain fully why you have B3 in your ID and @aaronne has 33. Since he has the unlocked "Stock" Bios while you have a special Prema Bios.

Ok that makes sense now.

Speaking of that, I need to speak with both of you.
From what i'm seeing from both you and @aaronne , NONE of you are able to bypass TDP, regardless of what you set in the Bios under power limits? Including Unlocked stock Bios and Prema Bios ?
So basically, the power limits in your bios are simply NOT working at all?

That's what I'm seeing here.

@aaronne
You're the key to this puzzle.
Please go into your Bios and set Power Limit 1 to 300W and Power Limit 2 to 300W. (also known as package power limit 1/2 or just PL1 and PL2).

(Do not change platform power limit 1 and platform power limit 2--they are ONLY to be used when using "Psys Pmax" (Power of System) and they're supposed to be used by system integrators to apply 'manual' cancer power limits, keep those disabled).

Then please check again prime 95 with AVX.

Also, change CPU CURRENT LIMIT to 200 amps, please. (or whatever divider gives you 200 amps...the GT73VR has a 1/4 divider, so 800=200A, not sure about yours). This is also called ICCMAX. Just don't set a direct value of 200--it's divided by something).

 

Please go into your Bios and set Power Limit 1 to 300W and Power Limit 2 to 300W. (also known as package power limit 1/2 or just PL1 and PL2).

Ok (now are set both 160W), go to do and re post here

(Do not change platform power limit 1 and platform power limit 2--they are ONLY to be used when using "Psys Pmax" (Power of System) and they're supposed to be used by system integrators to apply 'manual' cancer power limits, keep those disabled).

^_^ Oops, maybe I changed too many stuff in my bios, because I set all PL (1,2,3,4) at same 160W, going to disable em all soon

Then please check again prime 95 with AVX.

Yes!

Also, change CPU CURRENT LIMIT to 200 amps, please. (or whatever divider gives you 200 amps...the GT73VR has a 1/4 divider, so 800=200A, not sure about yours). This is also called ICCMAX. Just don't set a direct value of 200--it's divided by something).

Hope to see what are you talking..

 

Leave Power Limit 3 and 4 to disabled. They do weird stuff.
And do NOT change PLATFORM POWER LIMIT 1 and PLATFORM POWER LIMIT 2. Leave those "Disabled".

Only change package power limit (CPU power limit 1/2) or PL1 and PL2, whatever they are called.

ICCMax is also called "CPU Current Limit." The Bios should call it ICCMax in the help thingy. This must be at least 200 amps to do AVX. 100 amps is not enough.
If your Bios has an 1/8 divider, then you would set it to 1600. If your Bios has a 1/4 divider, then you set it to 800.
You can also set it to 600 if your Bios has a 1/4 divider. That's fine too, even though that's 150 amps. you're not going to be exceeding 150 amps without the system overloading the VRM's anyway.

 

This is a waste of time. Not doing it any more. If me and Svet and then me and Master @Prema, could not have already figured this out with months of testing and hundreds of mods it is futile. I can tell you that Prema removed all traces of throttle code and it still throttles. It's either secret crap that changes on the fly or it is a hardware limitation. All three of our MSI machines are a firmware mess, and I suspect the rest are as well and people that think they are OK haven't tried pushing the envelope hard enough to watch it fail (or their BGA CPUs cannot push it far enough to fail). One one hand I cannot complain because the 16L13 beats anything else but the DM3. On the other hand, it is junk for serious overclocking and clearly messed up because MSI messed it up on purpose or because they are stupid. I cannot identify which is the correct response. MSI and Clevo are two peas in a pod with their retarded laptop gimping nonsense. (So where does that leave the rest that are inferior, LOL?) At least Clevo can be fixed.

Here's what you asked for...

 

mine is not downclocking at moment

 

Your CPU clock speeds are too low. Set it to 4.8GHz and see if it holds if the CPU goes past 100W.

 

Mr Fox, calm down a moment. We're trying to troubleshoot a bug.

@aaronne are you running prime AVX?
Try 4.5 ghz or higher. You should be past 100W.
Please confirm your results. Wait at least 2 minutes after starting the test.
Then do it 100 mhz higher and keep going.

You may not be stable doing AVX load at 4.8 ghz. I don't think a laptop can cool a 4.8 ghz 8 thread AVX load CPU. Even desktops have problems doing this. But if you blue screen or shut down instead of throttle, well...we're getting somewhere.

If you can NOT do prime avx at 4.8 or 4.9 ghz, run WPRIME like mr fox did earlier. But the FIRST STEP is to make SURE you can EXCEED 91 watts without throttling!. 1 step at a time, guys, please. 1 step at a time.

Remember aaronne does not have good english. please be patient with him.

 

I set something wrong because hit only 97W

 

Thank you all men. ^_^

Seems th temps is not an issue here

 

Please run Throttlestop 8.48 and check Limit Reasons.
Close HWinfo first.
I want to see that window.
Keep prime running.

I need to see if you are getting power limit (PL2) limit, PL1 limit, or EDP OTHER only (cpu and ring). Usually, with power limit 2, you get PL2 under "CPU" and "EDP OTHER" in RING, at the same time.

I noticed you are *NOT* getting "CPU power limit exceeded" in HWinfo, but this usually takes TIME to appear. It's easier if I see this in Throttlestop Limit Reasons (less information=easier to see).

 

Is it ok?

 

Yes, now try 4.7 ghz please. Thank you
BTW NICE TEMPS!

 

in a bios click and a reboot!
Ambient temperature at 19c°/66F Fan at auto (Silent Option on half) on notebook cooler thermaltake massive23 overvolted (12v instead of 5v) ^_^

@Falkentyne thanks for time spent
Edit-



PL1 on Core before I start prime95, what that mean? too low vcore set?

 

Just clear it. This is often a false alarm. I usually get "PL2" and "Thermal" and "EDP Other" randomly every time I reboot my PC!

 

red and yellow pl1, I saw max 97W btw 144secs



and on hwinfo32 max 100W and clock down to 4,3Ghz

 

Can you exceed 95-97W at lower cpu speeds?
This is disturbing. This means that this model can not exceed TDP at all, so the "power limit" settings simply do not work at all?
So what is the point of a power limit setting if it can't exceed TDP?

I sent you a PM, by the way.

Sorry for wasting your time, Mr Fox. I was trying my best to help, because I know that usually companies are lazy and they just re-use assets (that's why the EC RAM has the same structure on so many different MSI models), but this has actually upset me. The BGA soldered filth can exceed TDP to more than double stock TDP (until the VRM's just give out) while the barebones is limited to TDP because of PECI (EC) power limits?

Dufus would be able to break this but no one has seen him in years...

 

no downclocked by itself, maybe too low vcore because I'm using offset?

 

No, because you exceeded package TDP, PL1 made you downclock because it set you back to 95W. The Bios power limit override is not working. At the lower overclock you only reached 93W.

 

Nooo buyed yesterday a fck new 7700k..nooo

nooo..noo. only 95W of fun? now is a real household appliance

 

Thank you.

I have a question about that black stuff on the heatsink, see the cross pipe leading from GPU to CPU? You don't have it on your heatsink but mine does, will the black stuff stop heat transfer from it to another copper piece if I placed it on it?

I'm thinking of using a new unused water AIO kit laying around and sticking it to that copper piece to help absorb the heat during benchmarking process, I'm just curious as on how it will affect the cooling system overall.

 

We will get drivers updated soon. Please let our support team to know so they can provide you and others with newer drivers. Also for these of you upgrading from Skylake to Kaby Lake CPU please contact our support team and get KBL BIOS.

 

@Eurocom Support
There seems to be a very large flaw in the barebones MSI. We have been discussing it in the clevo forum oddly enough.
It seems the "Power Limit" Bios option is being completely ignored. The CPU's seem capped to 91W TDP at all times, whether on stock, unlocked or Prema Bios.
This 91W limit seems to be unavoidable. Do you know anything about this?

I can pull 100W from the CPU on a BGA 7820HK 45W with GT73, but 7700K cannot exceed base 91W TDP on 16L1.

http://forum.notebookreview.com/thr...870tm1-owners-lounge-phoenix-4.809589/page-11

Can you investigate this?
I believe I know exactly WHY this is happening because I can make my GT73 do the same thing by using an "unknown" power ID. I think MSI did not allow this CPU to have its PECI power limits removed at all.

 

Would you please fill online support request with our Support Team and provide them all details. Are you using Eurocom unlocked BIOS?

 

I have the GT73VR. @aaronne and @Mr. Fox were the ones who found this out.
Aaronne needs to do this.

This is a serious error by MSI.

@aaronne is using the unlocked Eurocomm Bios.
@Mr. Fox is using the Prema Bios on the Evoc (HIDevolution) model but he tested stock EC also--same problem.

I'm almost 100% sure that this is an EC coding error by MSI because the EC firmware does not seem to have overrides for the 7700K CPU. The Bios supports it but the EC does not "identify" it (VERY complicated but I know why).

I don't know if this affects other CPU's.
The GT73VR does the exact same thing with 7820HK if you use an EC firmware that does not have the kaby lake ID in it (Example 17A1EMS1.107 EC that was written for skylake 6820HK instead of 17A1EMS1.108).

This issue cannot be fixed by Eurocomm directly. MSI must fix this.

If you PM me I can tell you EXACTLY why this problem is happening because its...uh...complicated and rather ridiculous...