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

Thx! The bios is made a month ago by svet, should be unlocked... it fixed the 120hz screen problem tho

 

..cut..

Spoiler

from my old 6700K




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

 

Thank you! Meantime I have found the AI AC DC, it did the magic... the voltages are stable as hell, -10C comparing to the previous spikes. Thx also the BLCK, I have not found that I will try tomorrow.
4,5Ghz@1,18V (potato chip of my 7700k) but under 80C during stress, so I am fine with that.

 

The important thing is the TEMPS. Remember that VID is not current voltage. You don't know exactly what voltage is going into the CPU at all because of vdroop being applied.

With IA AC DC loadline of "1", which is the lowest resistance, the VID you see is the true starting VID of the CPU. But this is STILL not vcore! Because vcore has vdroop applied to it (voltage drop), which drops more at load than at idle. (aka the loadline slope). So on these laptops, without a "Core voltage" or "Vcore" sensor, you have no way of knowing what the live vcore actually is, only the base voltage if there were no vdroop.

The IA AC DC loadline "Auto" value has a reference value of 2.10 mOhms. What this means is, as the CPU current increases, the VID (due to resistance) increases, so the starting core voltage will also rise as well. Thus you have more heat and power draw. This does counter "vdroop" quite awkwardly (it's basically Intel's way of dealing with load with FIVRs), unlike the setting we do NOT have--Loadline Calibration (LLC), which works directly on vcore rather than default VID for adaptive voltage (or re-programmed VID if using static/override voltages). Since there will be substantial levels of vdroop with an IA AC DC loadline of 1 (since this disables the "VID Boost" at load), you may find yourself unstable at higher overclocks and voltages. If this happens, you need to raise the core voltage higher, or add a little bit to the IA AC DC Loadline value (values between 10-25 can work).

DO NOT EVER, EVER GO ANYWHERE NEAR THE UPPER LIMITS. DOING SO CAN DESTROY THE MAINBOARD VRM'S OR THE CPU.

 

thx for the explanation. Currently, 4,5Ghz, 75mV - offset, IA AC DC 5. Temps under overwatch 70-72C, under 3dmark physics test 80-82C. I tried to lower the IA AC DC to 1, but in 3dmark it failed. Also tried to increase the offset to 85-90, also failed. I will test with battlefront 2 (most demanding game - 100% gpu 100% cpu) and we'll see the temp. I also monitoring with hwinfo, the peak voltage was 1,179V, usually 1,15-1,16V.

 

Watch windows event viewer (eventvwr.msc -->System logs) for WHEA Correctable errors.
If you are getting those, it means you're not stable. Could sometimes be caused by high cache ratios, but not always. Usually gets triggered by heat when you are on borderline stability. To prevent, try lowering cache multiplier ratio, raising voltage or raising IA AC DC. WHEA correctable errors won't show up in prime95 or stress test (unless it just crashes, then you are VERY unstable). BSOD with "Whea uncorrectable error" is even more unstable.

 

Alert!

After latest Meltdown and Spectre patches for Windows7x64 (the KB4056897 and the cumulative one KB4056894) my overclocked 7700K needs more vcore than before (+0,10v from 1,23v before to 1,24v now, so from bios setting, adaptive +20mv to +30mv)
to not incur in L0/L1 cache error visible also from HWiNFO's WHEA Sensor Stats.

Thanks for attention, have a nice day!

 

Still need help ?

There are tons of settings shared in this thread for pretty much the entire F5.
But if you want something specific I do not mind helping.

I usually recommend adaptive, the blck should stay at 100 or wtvr it is stock, this doesn't entirely work like on a desktop where you can easily use the blck to clock the system. It should, but it doesn't work or is not as stable.

For your temps, I have a few things that come to mind, like are you sure the voltage reported is 100% correct ? Usually it's just under 1.3V at 4.5Ghz but never higher, so your okay there IMO.

Have you tried an undervolt ? Do you know where to start on this ? Most of us would use XTU to do it and try like -125mv - 175mv for 4.5Ghz.
This should affect temps by like 10C about. Also pasting with Liquid metal properly can help, but do not try it if your not comfortable with repasting, and know what your doing ahead of time.

Hope this helps, catch you around, cheers.

 

I can probably help you, what seems to be the issue exactly ?

 

seems these fans have a shelf life, wish we all knew this ahead of time. Some seem still okay after a year going forward.

however I have to wonder if it's because the fans were running full speed all the time, from overclocking the systems. The CPU and GPU do share some parts of the heatsink in most models nowadays so that doesn;t help having a desktop CPU, can not wait for the mobile lineup to come back. (PGA)

Anyways - do any of you guys have the modified bottom panels ? Generally better air flow will extend the life of the fans too, because they will not have to work so hard to push the air non-stop, and being 4-5C cooler at a minimum usually also means less RPM's, and the fans themselves can stay cooler as well, and wear out slower.

HIDevolution sells modified panels, I've purchased some for my clients plenty of times, though sometimes it increases wait times for delivery.

I started personally modifying the bottom panels myself instead when wait times are too long now. Very simple process too.

 

well if you have the eDP cable that comes with the laptop, the standard 60hz panel I think uses the same cable, it's when you want the use those 4K panels I think you needed the 40/30 pin cable instead.

 

It's more a matter of how will your heatsink work, because the standard 1070 heatsink will not work with the 1080, also if you do not change the motherboard and simply get the heatsink, then you will not be able to bolt down the heatsink and GPU the same as it is stock with the motherboard that comes with the 1080 for instance.
There also might be power issues, for the MXM 3.0 connector, using the regular board, but unsure, haven't tested and might just be speculation.

 

reseat the cables for the LCD, all of them. Could have been tugged maybe or poorly installed, if not, I would replace the cable or the panel, but if you can show us a picture we can tell you if its panel related or cable related.

You can inspect it too, if it is more the crystals lining up vertically or horizontally from pressure and such, or damage, or if its simply the cable it would not affect performance or be detected as an issue in windows at all, but it would still show those artifacts everywhere.

if its the GPU there would be signs of performance degradation when using it, and temps and who knows what else, depending. But sounds like Cable connection somewhere to me.

 

Why ? What sort of help do you need exactly or whats the concern/issue ?

 

Not only 10% for students but many sales throughout the year and end of the month sales for various models, not always directly through EUROCOM, sometimes through their resellers too, as they offer us specials as incentives to pass onto the customers, for instance free 1080 upgrade from the 1070 was 3 months long and just just ended, was like 500$ in savings as an example, which over all was more then 10% and this lasted for quite a while, and was for the F5, they had that for October to end of December, I myself had a few people who took up the offer, but I know some other resellers had plenty of people taking advantage of it. Also keep in mind this laptop is now going on 1 year + so picking them up for a slight bargain nowadays is potentially possible, I see many even in this thread suggesting they are going to sell thier model and I wouldn't be surprised to find out half of them are in near mint condition or something too, just look around, ask around.

 

Well, there were a few different models they were testing playing with the idea of almost 7-8 months ago at this point, and none fell through and some were even AMD related.

 

@woodzstack The VID is being reported, not the voltage. And the VID can vary wildly by as much as 100mv even at full complete idle.
Without a true vcore sensor, it's hard to know what actual voltage is being used by the CPU. In addition, VID does not show vdroop at all. Setting IA AC DC loadline in Core I/A from 1-10 will greatly help stabilize the VID, so the VID shown is much closer to the target voltage (before vdroop) but you may need to tweak the voltage, because the Auto setting uses a reference resistance of 2.10 mOhms (value=210 in that Bios setting).

 

That's quite some specific knowledge there, you became the go to expert on this laptop hahah !
I'm almost sick of working on these... but I can work on this laptop with my eyes closed now, after having done so about 200 times.

 

It was other issues, I think, or from what I was told in the summer when they were all in Taiwan and doing stuff.

 

Not necessarily just this laptop. Just MSI uses similar base code on all their laptops. And you see in the above posts people lowered their temps and stabilized the VID by tweaking the IA AC DC loadline setting.
Vdroop is part of the specification to this day (even on the 8700K desktops). The IA AC DC loadline setting was supposed to 'help' counter vdroop by raising the VID based on CPU current (then vdroop is applied to that higher VID target) at load. Raja@Asus was the one who told me that when I asked him in a private message. But unless the IA AC DC is set to a manual low value (like between 1-10), it varies highly from the actual VID shown at load. (you can tell something is bizarre, because at full idle, for example, with a VID of 1.20v, you will see the VID fluctuate wildly from 1.20v-1.32v even with no load on the CPU. That means that the 1.32V VID is the VID used at high load and 1.20 at idle. But then when you put a big load on, like 8 thread prime 95, you see 1.22V (AVX disabled) for the VID, or 1.24V if AVX/FMA3 is enabled. Yet the power draw and temps show that the actual VID (or voltage) is MUCH higher than 1.24v. You can tell if u set static vcore of 1.24v and set IA AC DC Loadline to 1 .......instead of auto....

 

Could be more viable with setting it to 1 and keeping voltage static, it would be slightly more responsive and your temps would be what you manage and expect depending on the rest of the settings.

However, it still runs too hot, and usually on apative you get better performance to temp from what I've seen (with an undervolt).

it's like they gave us half the settings we need for a desktop and it just was not everything we needed to be 100% perfectly viable, but close enough that we would waste our time on it trying anyways.

 

@woodzstack ... That was a LONG nap.

 

From my own testing, basically, adaptive and static voltages act EXACTLY the same *IF* and only *IF* you can find the default VID for your chip, at your "target" mhz speed, and then find what 'stable' static voltage is actually needed, and then test that combined with adaptive+offset to match it .. But your true TARGET DEFAULT VID can ONLY be accurately found if IA AC DC is set to 1 (lowest that is not zero). For example:

Let's say your TARGET DEFAULT VID at 4.5 ghz is 1.08v. Now if IA AC DC loadline were set to auto (default MSI value), this VID AT FULL IDLE would be jumping around randomly for no reason, between 1.08 to 1.22v. Besides this being annoying as hell, it makes it hard to find the default VID if its bouncing all over. If it were set to 1, however, then your VID would show up as exactly 1.08v at idle, and 1.0794v during NON AVX prime95 and 1.11v during AVX prime 95. (from what was mentioned on overclock.net, AVX loads cause the CPU VID (or vcore) to rise by about 30mv). But again this is NOT your live vcore because is STILL Vdroop unaccounted for, so it's very possible you would instantly BSOD as soon as you started that prime stress test (depends on your CPU). You know vdroop is not accounted for, because the VID would barely even move (maybe by 0.4mv) from full idle to full NON AVX prime95.

Now back to the "Auto" setting:
With a 1.08v default VID, in this case (USING ADAPTIVE VOLTAGE to find it), the reference mOhms resistance would cause the VID to fluctuate between the minimum and maximum limits (no idea why) at full idle. This would cause misreporting at load: full idle: VID would jump around between 1.08 to 1.23v. Prime95 non AVX would report 1.12v full load, and prime AVX (FMA3) would report 1.14v. Yet the temps would be MUCH higher with higher power draw, clearly higher, than if you set a static voltage of 1.12v (1112mv) with AC DC loadline=1. In this case, the non AVX and FMA3 loads would still report (very close) to the same VID as the auto setting I mentioned, but the temps and power draw would be much lower, which clearly means something is being reported wrong. (this is assuming you were actually stable at this setting to begin with---this is purely hypothetical. In MY case, adaptive vcore on my 7820Hk @ 4.5 ghz is 1.08v default VID. That's how I know this example).

Even if you set a static vcore of 1175mv (in this still same case scenario) with AC DC loadline=5, the power draw and temps (sometimes power draw can be reported a bit wrong--pay attention to the temps instead!), will STILL be lower temps than 1.08v (adaptive voltage) with AC DC loadline=Auto (0). That's because the actual VID in this case (with the auto setting) would be boosted to around 1.22v. If you compared this with a static voltage of 1.22v override and set IA AC DC loadline=1, you would notice the same temps (but this time, the VID would be shown accurately). That's your proof right there.

Anyway, for the 1175mv static override voltage, compared with the 1.08v 'default' VID at 4.5 ghz, in this case, if you wanted the temps and power draw to be identical with adaptive vcore at 4.5 ghz (since AC DC loadline=1 would be unstable with 1.08v), and static vcore with 1175mv (with AC DC loadline=5)--stable, in this case, you would use adaptive voltage, +100mv offset, and AC DC loadline=5 (or 1).

Maybe this went over some people's heads, but for anyone who is willing to actually test this, and watch the temps and power draw, it will be a "oh this makes sense!" moment.

Again for reference; the IA AC DC =0 (auto) setting is the same as using a value of 210 (2.10 mOhms).

Sorry for the wall of text. I did find that with static voltages, for some reason, the voltage will not 'drop' at idle, and speedstep won't work, on GT73VR (speed shift will work, although AGAIN voltages will not drop but that's another story), but with adaptive voltage + an offset (to match the static voltage that you found was stable), the voltage WILL drop at idle and speedstep will work.

*Edit*
it seems from reading OCN, that on SOME desktop motherboards, when using STATIC voltages, while the VID still will "boost" up with the 2.10 reference Auto setting, that the actual VID gets completely ignored in favor of the override static voltage. e.g. at 4.8 ghz, VID = 1.4v, vcore=1.32v (temps=75C load), while setting IA AC DC=1 (or 0.01), 4.8 ghz, VID=1.325v, vcore=1.32v, temps=75C (same temps as before). This is assuming a medium value of "Loadline Calibration (LLC), which is a completely different setting, which flattens VDROOP on the vcore, instead of changing the VID itself. But some mainboards will boost the vcore also from the static value, based on VID. It seems that the MSI laptops are also boosting the vcore based on VID. Since we don't have loadline calibration :/

 

The fans MSI have used is trash. Not only for the Barbones. And What a wall of posts. Only 9 or 10 in a row

Edit. Use this one

Spoiler

It's called the edit button

 

MSI has the best cooling in a taptop that I've seen brah

 

Haven't seen j00 push +130w from your BJCi7... bruh
Edit. But My point on this... All too much failed fans (not only for the MSI barebones) Seen the same on MSI's main notebook portfolio... The Jokebooks.
I have run my fans on max a lot of times over prolonged times... I can't remember I have seen failed fans due this. Both for my AW17 and two of my Clevo's.

 

It seems on Amazon lots of MSI laptops have under average reviews. I mostly read the ones with 1060 and above graphics cards.
I'm selling this laptop and going with a 17incher with a 7700hq cpu and 1070 that I got for free. I ordered a laptop with a 1050 for my sister and got one with a 1070. The seller sent the wrong model laptop, I messaged the seller about it, waited a week, no response, so I am keeping it. A Gigabyte P57Xv7-KL3 with 2 year warranty when my F5's warranty expired last month.

 

A lot of inexperienced people buy the entry level laptops and complain about stuff easy to fix, or they are hypercritical, you have to read them with understanding of the common complaints / solutions, and you'll find most are overboard in criticism - all vendors. But, there are trends to look for, so don't discount bad reviews altogether.

 

I bought two extra fans to keep on hand for when mine eventually go bad. At least they're easy to swap.

 

The warranty is the laptops, that should stay no matter who owns it.

 

haha I love you guys !

Yeah I was in this extremely tight spot waiting on restock for like 10 weeks for every product. Totally out of my control, since this was from my supplier. But I'm BACK ! muahaha

 

Mine is modified. So far, I've had no issue with either fan, but now I probably will. Thanks a lot! Hahaha

 

Makes me wonder how much sooner your laptop fans would have died if you did not have open vents for the fans.

 

Seems like they last about a year... who knows? They are easy to replace and I'd rather burn up several sets of fans than my cpu or gpu...

 

Hi guys, thx for the support, I managed to fix the cpu voltage, eliminate the spikes I have purchased a PCIe Plextor ssd to upgrade, and move the OS from sata to pcie 4x. In bios info page or raid menu I see the ssd, however I cannot select for boot, and under windows 10 I cannot see it either.... what should I do to use it? thx

 

I've t this day never seena burnt out CPU. GPU's don't get me started, haha

 

you will need to navigate the menu's there is a thing... for lack of remembering the specifics... a boot loader for it, forget what it's called csm or something (or am I getting this mixed up). It is not under the boot menu though.

 

the interesting part, that I cannot see under windows either... I try to find this csm... thx

 

You have to enable the drive in the BIOS.

 

the big question, where?

 

I think in the second page depending on your bios, there is a long list of menu's that have thier own page, some even repeat themselves, there should be one that has stuff related to the NVMe's

 

Well, what BIOS do you have? I can tell you where mine is, but that won't fix your problem if your BIOS isn't the same... it is clearly labeled in my PREMA powered ultra book.

 

Maybe you don't deal much with soldered hardware....

 

I have unlocked bios prepared by svet 1 month ago. Under nvme menu i cannot set anything

 

Sorry, I am not familiar with that BIOS.

 

Thanks for the reply! I'm suspecting it could be a cable issue. I've checked the cable and ensured it wasn't being pinched as well and examined the GPU without finding any issues. Here's the issue I'm encountering:

 

yeah that's cable, just remove it and re-install it, could be loose, first try where it's easier access, near the fan and wifi.

If that doesn;t fix it, then you will have to remove the bezel on the panel and unscrew the panel and carefully check the cable there, if the cable is pinched or damaged it is most likely where the cable goes through the corner of the panel and into the laptop, so check there last too.
Might be the cable but hard to say if it is damaged, you;d see signs of it most likely, and normally the issues on the panel would be changing colors so this might be the circuit on the back too. Just check everything.
If none of that works, it could be the bridge that connects the panel to the motherboard, thats on the back of the panel, in which case you might just as well get a new panel, I can help you start an RMA if you need help with whomever, either MSI, EUROCOM or wherever or even help you find a second hand panel for dirt cheap or something.

 

it's not under an NVMe menu... argh I do not have one of these infront of me, but I can check tomorrow or thursday if needed if you still have the issue.

 

haha actually your right I do not, and I've been particularly critical of any CPU forever and done my best in any of my installations, so I never ran into an issue on one of my BGA laptops I've sold - however the BGA lineups I have were the last to be added to my lineup, I largely avoided them for a few years. So you got me there.

ALSO this was apparently from GENTECH and was an ASUS, so totally not me, nd might not even be GENTECHs fault either, most likely ASUS's of whomever assembled that laptop and it's QA

 

Hahaha, I think you may have quoted the wrong post. I am set with my BIOS options. I was attempting to help @matyee.