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

You are definitely right about that. That's why I have another U3 pad coming in and I'm going to reuse the fans/controller I got when I purchased @Mr. Fox U3 modded pad. It will be used for my everyday computing tasks. My U3E Pad will only be used for OC'ing sessions, I have already received complaints from my girlfriend and I can only use it during "normal" hours.

4.7 isn't bad at all for everyday use on a 6700K. That's on par for what I use in the Fox AK1 but the 6700K in that can go 4.8 for OC'ing purposes. What kind of cooler are you using? Air, AIO? Custom loop?

The 7700K do have an advantage but they run hot. That's the tradeoff. For example, without the U3E's aid, I hit 75 to 79C with silent fans on full load. Those temps approach my discomfort spot. Over 80C is unacceptable for me in regards to normal computing, that includes gaming.

 

That's good, so your cooler knocked off over 10degC from your load temperatures.

I'm using Noctua NH-D14 air cooler with Coollaboratory Liquid Ultra delidded CPU (Thermal Grizzly Kryonaut on interface between heatsink and IHS). Oh dear, just got a strange black screen after about an hour of OCCT @1.376V - have now upped voltage to 1.384V and retesting! 4.8Ghz looks Prime95 stable though, it's just I'm setting a higher bar for a rock stable overclock, so in the process of fine tuning this 4.7Ghz overclock.

I agree with what you said, I wouldn't really want to see over 80 degC for gaming in a laptop or desktop, and certainly not if overvolted, but if it was at stock voltage then I'd probably be ok seeing up to 85-89 degC (I'd still do everything in my power to lower that temp though without sacrificing performance). The worst combination is overvolting & high temperatures, that's why I delidded my CPU, so it gives me the low temperatures to run over 1.3V with confidence in the 'knowledge' that my CPU won't degrade overtime with the overclock.

 

Guys... looks like @Prema BIOS fixes 120Hz display brightness control on my machine as well.
@Donald@HIDevolution @Ted@HIDevolution @thattechgirl_viv @Zoltan@HIDevolution

 

And, no more black screens with newer drivers. I did a clean install of W10 Creators Abortion and let it install the driver from Windows Update and it's actually working fine.

 

So it took @Prema to fix this issue, whereas Eurocom couldn't? If he gets G-Sync working, I'm going to be angry while I give him a slow clap.

 

He did some tweaks on Clevo BIOS as well. Johnksss never had brightness controls on his aftermarket upgraded (same part as stock) 120Hz panel with non-G-sync video cards and now John has brightness controls. To be fair, Eurocom does not do firmware mods. They have always relied on Prema for this on the Clevos they sell, as they are a Prema Partner Shop.

Just installed this @j95 driver mod and no black screen. Can't hate that. Thanks @Prema!


Edit: Hmm... looks like Windows 10 night light is working, too. Need to disable that (I hate it). I always keep my screens at full brightness, but I am glad that is fixed in case someone else owns this machine someday.

 

yeah, to really push these CPU's I've gone as high as 1.8v to hit 5.3Ghz

Right but this is a MSI with the new bios system. You want/need SVET instead I think.

 

At voltages this high, I don't think it would be worth it even just to bench! What kind of temperatures were you getting (average) and what did you do to cool the processors at that frequency? Was this in the MSI barebones?

Is the Prema BIOS allowing voltage control over the gpu? Can we get beyond the 150 w limit?

 

I have that on another chip and it did not fix my 120Hz brightness controls.

I am working with @Prema now on the HIDevolution BIOS for their 16L-G-1080.

Wonder why that is? Poorly binned CPU? I only need like 1.300V for 5.2GHz.

The system BIOS has nothing to do with the GPU overclocking capabilities.

 

You need to go work directly with the manufacturers so they can fix this nonsense out of the box as it should be. Prema as well. He should be banking on the miracles he is responsible for.

 

I have to agree with that last part about @Prema for sure.

 

Woaaa, 1.8V, you're kidding me! That sounds like instant death to me unless you were running under LN2 or something? I'm not gonna go above 1.4V I don't think, perhaps for short tests I might take it to 1.45V for short benchmarks, but that's probably about it.

 

Our laptop motherboards (Clevo and MSI) cannot actually deliver that much voltage. This is something I, @Johnksss@iBUYPOWER, @Papusan @bloodhawk and others have observed. In fact, one reason they are overclock limited is because they cannot. Their limitations are such that you cannot give the CPU enough voltage to fry it because they can barely handle a range between 1.425V and 1.500V and any values higher than that are simply ignored. Otherwise, you would be seeing higher overclocks.

 

That's cool, I'm still hoping to hear from Woodzstack, because I think he may be talking about desktop regarding his 1.8V, and 1.8V just sounds crazy!

 

I wouldn't bench with that voltage. Benches stress the CPU out, you would risk damaging the components.

When I run that voltage, and get it stable, I have little to no load, and a few things taken care of. temps are low because no load. But it can handle it, at 1% load.... I did it for shiz and giggles mostly. On a desktop you can run that and be fine if you can keep it cool - but do not bench it ! anything taking it past 75% load is a risk at that voltage on a desktop too.

 

Yeah, 1.8V just sounds crazy, not going there!

 

Woodz Yoo mean MSI laptop motherboards support-deliver 1.8V?

 

Actually many of the systems , the F5's I shipped out were over 1.5V when I tested it at 5.0Ghz.

No the F5 can go past 1.5V. At least the systems I've sold to my customers could. If you look, all the ones I sold have like 5-10% higher Physics scores in firestrike for instance. Though we do have one guy hitting pass 20K with the 1070 in firestrike which is around where the 1080 starts - so might be the whole config is different too. IDK>
All I can say is our bios is not the prema bios Mr.Fox uses. Maybe his is limited or different than mine. Or maybe the systems I have sold have all been different, who is to say. But I know for a fact the voltage was always reported higher than 1.5V when I when I went nuts on them - if the voltage is being reported accurately I can not say , not my field of expertise with these, I didn't have complicated equipment running to and from it everywhere to know 100% for certain how accurate it was, I was relying on software.

Anyways - you can not always give into marketting. Got to do your home work. While I could go past 1.5v on this laptop I don't recommend it, I actually try not and post any of the crazy benches I've seen of done, I usually delete them from my 3dmark account too, because anyone could just get the link see the results and duplicate it, and sometimes I'm testing in an environment that is not exactly evident by looking at the numbers, and is actually my safety net.

While I laugh about and joke about the 1.8v thingy, I never held a load and didn't try, and it was 1.8v with the offset as a total, something like 1.475 +275 offset so close enough. however I did and can clock my 6950X to that 1,8V so you are also right about the desktop comment.

 

my bios had 1.65v directly without offset, so the 1.5v max thing is not true, or maybe different version or something is off.
also you can add an offset, and I usually do. Also I do not clock my laptops the common way you guys do it, I also use my blk to it's fullest and my offset and my core cache multipliers.
Hope that helps. But you can see that in my benches if you ever seen them.

however mr.fox said anything past 1.5v is not making a difference, or something along those lines, but I do not know if that is true or not, so maybe it makes no difference - but I confirm I've gone past 1.5v. I always tested at 1.65v after to see how thorough my deliddings were. It was pretty much habit and the sweet spot to getting okay performance at 4.8Ghz-5.0ghz (if a 7700K) or like 4.5Ghz if a 6700K

 

people forget this...

anyways heres a basic test result, what to expect at like 1.5v or so without really going past 4.5Ghz even.
http://www.3dmark.com/fs/12152609

Matt's beats mine, but his laptop was from me. could be that we use Svet though, instead. Svet is a moderator on MSI forums, he does MSI bioses all day everyday the way Prema could be considered that for Clevo (though honestly Svet does this much more often, trust me.)
This laptop is a MSI laptop after all. However - I personally do not know the distinct differences between them other than that.

 

oops sorry wrong one, that was the usual 4.9Ghz test I do at 1.65v.... my bad.

 

if you go and do a search my laptops fill the top 100.... but no Prema bios ones from HID though... so this could be a big difference, and it could be the vbios that makes the difference.

since now the vendors use different vbioses but everyone is suggesting theirs is "special" and the bottom line is we're looking for performance and top dollar value.

So results speak volumes. But these scores are not with 1.4v... and not with an HID/EVOC version or Prema version, these are with my services, and special thanks to Svet for his bioses, and thanks to the creator of this thread - who actually sent me to Svet in the first place ! (I think it was him... @Diversion was you that sent me to Svet right?)

 

The highest one from Mr.Fox I could find on the seach was still an F5 from Eurocom not HID/EVOC :
(this one with a 1080 instead in the search)

http://www.3dmark.com/fs/11775202

or 11th place:


Note the CPU score still lower, yet 5.0Ghz, so seems like almost two completely different laptops , but not really.
If I could get a copy of this Prema bios I would love to test the difference.

 

Interesting, I can imagine Mr Fox wanting to get some more info from you on the 1.5V+ capabilities of your systems!

 

it might be just the bios, or it might be everyone can do it, and it simply is ignored as in makes no difference, but I can not test until I have another one in my hands. Since that means the next time I have buyer, and with thier permission only, who knows. It's been like a week since I touched one.

I'd like to have this Prema bios and test it for myself, for all I know I'm taking everyones word on it...
see if there is a difference compared to Svet's bios and what it is.

 

idle voltage > load voltage

 

@woodzstack

Have you ever used a @Prema BIOS? He does way more than opening up the BIOS and giving it to a user. I mean that in no disrespect towards Svet, as I've employed his services to unlock the BIOS in my Tornado and I have enjoyed the results so far. However, Prema WAS able to enable the brightness controls for the 120hz panel whereas Svet's BIOS did not. An example of Prema's extensive abilities.

As I've mentioned in the past in this thread, I've used BIOS (Svet's in the MSI 16L and Prema's in my Clevos) from both gentlemen and if I had to choose, I'd pick Prema's. Congratulations to the buyers of HID's EVOC MSI 16L. Your machines will be awakened by a deity.

 

HIDevolution buys their product from Eurocom. Nobody but me has a Prema BIOS at this time. I worked with Svet to refine his BIOS and make it better, just as I am doing with @Prema now. They are both good BIOS, but MSI needs to fix some messes in their underlying code that seriously hinder performance with artificial throttling of the CPU. Svet BIOS has the same issues, as @Rage Set can confirm. With maybe a couple of random exceptions, all of my record-holding benchmark scores for this product are while using a Prema BIOS. I can switch between them at will, since my BIOS chips are socketed.

The settings that populate menus will, in some cases, go beyond what the hardware is capable of applying. In some cases the setting can be applied but is of no effect past a certain point. Tweaking firmware to make settings visible certainly makes a big difference, but it does not change anything at a hardware level (i.e. VRM, MOSFET capacity, etc.) or at the hex code level. Owning the best hardware samples and having the knowledge and experience to use it in overclocking and benching is never enough to bypass the harmful effects of poorly written or maliciously crafted firmware. The latter will be a perma-gimp that no amount of overclocking experience, cooling or adjustment of settings can fix.

You can have the best BIOS available and a few pieces of cancer code in the EC or buggy ME can marginalize or even totally negate the benefits of the BIOS mod. As we have seen many times, simply using a different EC can mean the difference between horrible and amazing results. This can make or break the outcome, and both Clevo and MSI have released some pretty messed up firmware. Sometimes the messes are intentional. The same can be said of other brands, but it doesn't matter since all of those are BGA that isn't worthy of mention.

 

That is way too high. And pretty much impossible on laptop motherboards.

THIS -

 

care to explain the last quote? :|

kinda didn't understand that

 

On the laptop motherboards, the processors idle at way higher voltage. The moment you start stressing them they will drop to the proper required voltage in most cases. And the voltage might vary depending on that load.

Example - TS Bench 32M @ 4.8Ghz stabilizes voltage at - 1.190V , however the same 4.8Ghz running wPrime 1024M or Cinebench will need 1.20-1.215V (at least) (depending on the how good the chip is, temperature etc.)
However when idle, the same processor sits at about 1.25-1.275V sometimes, and jumps around.

 

lower idle voltage somewhat indicates that the cpu is somewhat better?

 

Does anyone have a video of what the fan noises like at stock CPU clocks and liquid metal re-pasted while running firestrike


Sent from my iPhone using Tapatalk

 

I don't believe so. Since that depends on the binned/baked in voltage for that processor or even the BIOS.
Only way to k ow exactly how good the chip is to find it base voltage and then push it to as far as it will overclock AND how much of an undervolt it can take for stock speeds.
In most cases the latter is pretty much around the same point, from what I have observed.

 

The 16l allows the fans to be at max speed even when the system isn't stressed. @Mr. Fox has a video showing the decibels which, I believe, came in around 54 decibels. Not bad. I would gladly trade out my fans for louder fans if they moved enough air to cool this beast a bit better.

On the topic of cooling... I typically see average temps of 71-75 on my cpu maxing at 85-87, and about 75 average gpu temps maxing around 90 C. Should I be concerned about those max values?

 

I decided to experiment with a slightly more aggressive undervolt, leaving my clock speeds at 4.7 GHz. I'm now running 1.140 V instead of 1.150 V (static, in both cases), and after running a bunch of back-to-back Cinemark CPU benchmarks plus a Unigine Heaven 4K, it appears my temps dropped from 91C peak to 88C peak.

It turns out my occlusion map baking, which is another Unity batch job I frequently run, is a really good CPU stress test. If I don't manually max the fans, that gets me to 95C CPU temp even with the 1.140 undervolt. Fans maxed keeps the temps at 91C peak. It's still higher than I'd like, but fortunately this job runs only for about 2 minutes, not several hours like the lightmap bakes.

I might like to try even more undervolt, but I'm going to let this run for a couple of days to see if it's stable. I have plenty of time to be very slow and careful with tuning now.

 

Is your CPU delidded? Might be worth it, my temperatures on my desktop dropped 10-15 degC with a Liquid Metal delid on my 6700K.

 

ha, got it upside down: apparently you have a very good chip! how did u check stability at those clocks and voltages? working through my 7700k sample atm, can validate it at 5.3 and run cinebench 15 loops at 5.1/5.2. now its prime95 1344k time, to check and see what 24/7 clocks would be....

Sent from my HUAWEI NXT-AL10 using Tapatalk

 

Indeed it is, by the fine folks at HID Evolution.

 

I want it to be quiet lol


Sent from my iPhone using Tapatalk

 

If that is not on top of an AC and in the laptop, then that is one hell of a chip!

Most of us are doing wPrime 1024/Cinebench/Realbench to test 24/7 stable clocks.

 

that's great man, i have to downclock or increase volt for occulusion baking. i can do light baking with -155mv or 1.020 static but for occulusion it crashes with watch_dog error. have to leave voltage at stock for occulusion and it shoot temps to like 90c even on 4.0ghz.

 

you can control the fan curves, so even at high temp it will be quit but i wont recommend it. i hardly ever noticed fans when gaming or benching graphics. it do get noisy when i do unity Light baking. but that after many hours of running. i think its prob similar to running prime95. so you shoudnt worry about noise. its prob among quietest laptops.
edit: i have 1070GTX card so a 1080GTX version may be louder.

 

Wow really good to know .. thanks fro the info


Sent from my iPhone using Tapatalk

 

I've never had stability issues in occlusion or lightmap baking; it's just a battle to bring down my temps without downclocking. My clocks are set at 4.7 GHz turbo, 4.2 GHz non-turbo. During occlusion baking, it's borderline whether it can finish within the 90 seconds that is my turbo duration window.

 

Has anyone had an issue where Firestrike hangs at the "Gathering System Info" stage, before even loading the benchmark?

This is happening for me even after a reboot. I think (but am not sure) that it may have started with the Win 10 Creators update. Other Steam apps and games run fine, and it's just FS that is hanging -- the rest of the system moves along normally.

 

i see, i have prob with occlusion since i bake pretty huge environments with thousands of objects, it can take as much as 40+ min it doesnt use much of the cpu but crashes if im on -155mv. i have not crash once on any thing else thn occulusion or when i have final gather on. so def these 2 dont like low voltage.

 

I expect the unified heatsink in P751 will help the cpu if the Gpu is idle. The difference will come with max load on both. Less thermal headroom for processor if you fire up the Gpu to run fully tilt.

 

No, I've asked Prema a few times for stuff. Gone to his site, and he always says he is out fishing or travelling, or never gets back to me. Tried a few times every year for the last few years. I'd have loved to see if there is a difference or what that is, or if it makes a difference.

On GPU's his vbioses turned out to be all hype and caused more issues than fix for my clients, however - I think it would be different had they purchased from an unknown vendor - in which case I think having access to someone who has a few different sets of vbioses to make these work in multiple laptops is a good thing. Like if you buy from some unknown seller on some chinese website and the vbios is in some language you can not read and doesn't work well.
I've only see and tested those, because cards were returned to me, as if they were broken with his vbios - some people flashed to his vbios right away before even testing the ones we gave them, they admitted. This has caused soooo many headaches, still get the issue once a week sometimes.

 

Those vBIOS's need to be used properly and in the right systems.