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

This issue is caused by an incompatibility between the display and the bios. I spent extensive time (and money paying for bios updates) troubleshooting this with Svet when I swapped over to a 120hz panel from 4k and got a different revision of the 120hz panel. I was able to rig a fix to work with a newer driver set by tricking windows with the uid of the monitor via copying the old driver inf files that work into the newer set, but it requires you to have an external monitor hooked up every time you boot windows because it updates those files and reverts back to a black screen. I wouldn't swap out a working 120hz model for a 144hz and have to deal with that nonsense.

Prema may know how to make the two work and rig up an updated bios, but Svet could never figure it out after updating 8 different bios revisions with my model panel.

 

Excellent, the only issue is some games won't launch with those drivers but there is a work around but it requires disabling the NVCP. No issue since I think you can make changes with NVInspector app.

 

What really pisses me off is that Nvidia's older drivers are working just fine with his display meaning they did something to purposely break the displays unless you have the BIOS to say it's ok. That is ******** and total greed on part of the OEMs selling these laptops. Yet another reason why I'm about done with laptops all together. I love my Tornado F5, but love the open platform of my desktop more.

 

I agree man. I've had the same desktop for years and I upgrade the cpu every 2-3 years and the GPU yearly. It is nice not having to worry about temps either. I only use my F5 when I can't be in front of my desktop.

 

That's super disappointing. I far and away prefer my 4K panel for normal desktop work, but I figured if I was going to run games at 1080p to keep noise down I might as well go full 1080p to remove interpolation even though I'd rather not.

Funk that noise, not worth the trouble.

I wonder if it's possible to rig a display port to edp cable. Could just use the external display port on the back and run a cable internally and probably get any panel working that way maybe since it isn't using edp directly?

Possible this might give clues...
https://hackaday.io/project/4765-me...een-on-a-boom-arm/log/15538-part-2-hello-ipad

Also this?
-DisplayPort-Convertor-Controller/dp/B01L3A07TY" target="_blank" class="externalLink" rel="nofollow">https://www.amazon.com/Eztronics-Corp -DisplayPort-Convertor-Controller/dp/B01L3A07TY

This seems like a lot of hacking.. that is pure BS that any panel doesn't just work..

 

So I did some looking into this to see if it was possible to find a way to convert DisplayPort to eDP.

You'd need a whole LCD controller board. I can't even find one that would support 144Hz, let alone the fact that the better ones are large and expensive.

Be better if nvidia just supported the use of any panel if possible.

 

Well this is strange. I bought some conductonaut, delid tool, silicone conformal coating, and some high temp adhesive to relid. Planned on applying conductonaut today to try to drive the temps down to keep the fans quieter.

I reinstalled 3dmark to do some firestrike runs as a baseline so I could see some before and after temp graphs and saw that I was running 13500-14000 instead of the 17800 I was running back in October. I've triple checked CPU and GPU throttling. The GPU is running fine, so is the CPU. Clocks on both are where they should be during the runs. GPU is reaching 87C, it was around 80C with my uvolt but putting it back at stock didn't help but slightly.. CPU is never getting over low 80s. Both are maintaining good clocks through the entire run. G-Sync disabled. Vsync disabled.

I even blew away the drivers with DDU and went back to the same drivers I had installed back in oct. No change.

There's definitely something that dropped my scores across the board but I can't find what it is.

I didn't want to move forward with the LM application until I got my baseline back where it should be.

I thought maybe CPU cache was disabled or something, doesn't appear to be.

The one thing I did do was, not too too long ago playing with undervolting my CPU was reset prema bios back to defaults. I don't see any bios options set to tank my perf this much though. Admittedly I don't quite know what every option does. I figure if my clocks look good on both GPU and CPU it shouldn't be throttling though. Kind of stumped, has to be something simple.

Edit: I guess one other thing that changed is 3d Mark version, but losing 3K there? Not likely.

 

Others have been complaining in other make / model threads about the same drops, likely due to Nvidia / OS updates, some found dropping way back in Nvidia versions somewhat fixed it on newest forced Windows update.

 

I was afraid it might have been Spectre/meltdown fixes but the drop seemed too far for that. Well.. going back to the same driver that was giving me 17800 back in oct didn't help. At least I'm not alone haha.

 

Can you post your 3DM FS scores (I want to see the physics scores). If you use different Cpu oc profiles then post all. Thanks

 

Oct vs Today, same nVidia drivers:
https://www.3dmark.com/compare/fs/13778386/fs/15694185

XTU:


Afterburner during the ~14K run CPU clocks looked to stay steady, GPU clock was mostly ~1700-1800 which was normal..it dipped to 1500 during the physics test.. During GPU tests clock was above 1800 nearly the whole time..

 

You run with G-stink enabled + lower cpu clock speed vs. previous FS tests. And same panel screen?

 

G-Sync was enabled, but I had forced it to be disabled for the 3dmark various executables back then. I disabled it outright this time to be sure it wasn't limiting frames. Enabling it did nothing for the scores. I also bumped up to 4700 to try to increase score, it helped slightly.. I have a bunch of 13200-13700 runs today as well. :/

 

4.7 should be more like +-15000 in physics. Weird. You are sure nothing running in the background? You could test with Cinebench R15. Aka pure Cpu benchmark. You should manage +1040cb with 4.7GHz.

 

@hfm Here can you download older Futuremark 3DMark + TimeSpy suite for debugging of your weird scores... Click show more versions.

 

975cb is low stock clock score. Aka with 4.4GHz all 4 cores for 7700K. Run Hwinfo next to and follow clocks in tests.

 

I couldn't run firestrike on the old version, I bought it on Steam.
But I do have timespy comparison from a year ago..

https://www.3dmark.com/compare/spy/3897614/spy/1544614#

 

Steam version shouldn't matter. No need to download and run from Steam. The Futuremark key follow your 3DM account. You should be able to use it. Unistall Futurmark from Steam and install the standalone package. Don't update the System info. Use what follow the version.

Remember you can see how clocks, temps, fps worked in "show details" after you have done the prefered Futuremark benchmarks.

 

Thermal throttling ahoy! But it did stay at 4400 most of the time... (got 890 this run)

 

Yeah, I could smell it. What load voltage? You use XTU and default voltage?

 

Upgraded the Tornado F5 today with 3000Mhz ram. Games like BF1 are running much smoother without the noticeable hitching that I had occasionally with 2400Mhz. I used Corsair Vengeance ram. XMP profile worked perfectly.

 

I just applied 80mv undervolt, didn't change much...

 

Yeah, it run hot. But take a look on @Mr. Fox TF5.OC.PDF.

 

I did another test with 80mv undervolt Maintained 4700 the entire test. 905 score. Something else is at play.

 

If you don’t use ThrottleStop, try it. Uninstall XTU OR disable with Autoruns.exe
Can’t run simultaneously. Click Limit reasons on main TS window (have both TS and limit reason up while testing). Remember click clear all settings in Limit reason before you start. Follow what’s happening while you test.

Post pict of TS settings after you start it up.

 

It didn't limit, the graph is dead solid on 4700 for the entire test. CPU hit 90C but didn't throttle.

 

Then it must be something who steal clock cycles in the background. Clean clocks should mean correct scores.

Edit. Read my and @Huniken replies before and after this post.

 

972 this time, arrows pointing to test run area on the graph. I went back to "better performance" mode in Win 10 to keep the clocks from running at 4700 constantly, hence the areas where it isn't 4700 surrounding the test timeframe.

 

This is from old scores... Even 20 point below these scores for each clock speed should mean it still works correct (depends on how much you have running into background).

Don’t forget the stock Hwinfo sensors polling frequency is slow(you can change it in settings). ThrottleStop has as well own log feature.

[email protected]


[email protected]


[email protected]

 

And they say RAM isn't important for gaming...

 

I definitely think something is holding it back that has nothing to do with CPU or GPU throttling. I see nothing in task manager taking up any CPU time of note though. (maybe 1-2% between some random system processes and Steam) It seems we're talking a 10-15% performance hit here on CPU in Cinebench, which is also nearly the same difference in Fire strike for the physics score (9.8%). But my graphics score is 23% lower, could be because of the CPU but that seems like a rather large drop in GFX score. I double checked my memory was still running in dual channel (seems like it would be physically impossible for a pair of sticks to suddenly go single-channel out of nowhere.. )

Stumped..

 

The other thing I find .. er.. mildly annoying is that every time I reset prema bios to defaults it sets the turbo power max to 1W, which as you can imagine ends up making the system crawl until I fix it permanently in XTU. Makes me wonder if there is some bios setting that is causing this but I just don't know what it is. Maybe @Prema can help?

 

Tried a NVRAM reset? A Clean up could help.

 

Sometimes NVRAM can't help it, when I changed the IMON value to negative -31999 for benchmarking I couldn't see the machine making a difference when I wanted to disable it, NVRAM doesn't cut it, had to disconnect CMOS battery to reset BIOS....CMOS battery mode was a really essential mode by @Mr. Fox .

 

Hey there,

I have recently bought a Eurocom Tornado F5 and love the machine
However, after reading several hundred pages of this thread and still not nearing the end of the tunnel, I'd like to ask for your advice:

1. I bought my machine from Skylake_ and he did the fan mod, removing the bottom chassis directly below the fans.
After doing some benchmarks, I noticed that my PCH would hover around the high 80s to low 90s - which is a bit much, in my opinion. (it is summer over here and the room did have a temperature just shy of 30°C)
Several search attempts in this thread seem to indicate that this increase in temperature is due to the changed air flow.
But one attempt with copper fins aside, I have not seen any solutions to this.
Is there any best practice yet, on how to keep the PCH cool, when the laptop has undergone this modification and is not being used on external fans?
(So far, I have put some shock absorber for speakers below the back, so that the fans are raised by about 2cm, and have undervolted the CPU core / cache as well as the system agent)

2. My device gives off a strange sound / noise every now and then (once or twice per day?).
The rough location seems to be the left fan (sitting at the keyboard, it appears to originate from the left back of the device) or GPU.
And it sounds like wood that is still setting or a relais that switches, so quite strange :-D
Has anyone experienced something similar?

Thank you very much for any support or hints

Cheers!

 

1. PCH mod... http://forum.notebookreview.com/thr...6-owners-lounge.797128/page-722#post-10570737

http://forum.notebookreview.com/thr...6-owners-lounge.797128/page-686#post-10556803

Latest post means latest mod.

2. Buy spare fans. The fans MSI used for this and some other models isn’t best quality. They don’t last very long.

 

How do you download the drivers for the built in camera? I don't think any of these drivers are for the camera of the Tornado F5 after looking at the following. These are all the drivers they list for the Tornado F5.
Audio download download download
CardReader download download download
Chipset download download download
Hotkey download download download
IME download download download
IntelWLAN download download download
IRST download download download
KillerWLAN download download download
LAN download download download
NVME download download download
RAID download download download
SBX download download download
Thunderbolt download download download
TouchPad download download download
USB3 download download
Video_NVIDIA_10xx download download download
Video_NVIDIA_9xxM download download download
VideoQUADRO download download download

Also, my fan on the right side (backspace button side) is not turning on ever. Is there an easy fix to this? Thank you for your time and input.

 

Fans on these are extremely poor quality.

Hmm. Not sure which would be the web cam...

 

I've run out of options to try, I've fiddled with so many settings at this point, still 3K points down in Firestrike. I've got the CPU running at 4700 solid, finally scoring over 1000 in Cinebench but still slightly under @Papusan's 4700 score but it's within acceptable range. It looks like strangely that maybe the GPU is underutilized. Perhaps pointing at some power issue, but it's unclear. Nothing seems to be unstable at all, just slower than it should be. GPU clocks look ok, CPU clocks look ok. Pretty much every benchmark I throw at it has it coming in way under where a 7700K+1080 should be sitting. Highest I've been able to muster in Firestrike is around 14400, which is 1070 land.. I don't see any other process taking up any statistically meaningful CPU time.

I've even poured of the nv cpl settings to make sure something there isn't set correctly. Nada.

Here's a shot of hwinfo while heaven is running. Looks ok outside of temp. I could just do the LM repaste, but I feel like it'll be cooler with the same perf. I'm never actually hitting 90C when doing a firestrike run.

 

You know the reason you're throttling is *BECAUSE* your GPU is throttling right?
You're complaining about lower scores.
Look at your GPU!
Look at HWINFO
It clearly says "POWER LIMIT- THERMAL" is FLAGGED ON.
I honestly don't understand.
You see that, but you're complaining about lower scores.
It's right in front of your eyes. Not trying to insult you but come on...you answered your own question.

Repaste your GPU. The old paste is clearly dried out and not working at all.

 

I'm not throttling during firestrike though.. sometimes vrel but that's common.. notice clocks for GPU are still 1750.. it's just not 1835..

I can undervolt and underclock slightly to get rid of those, nowhere near 25% underclock, maybe 3%, where it's not exhibiting any perfcap.. and I still get 13500-14200 where it should be something like 16800-17300.. unless that 17800 I scored back in oct was a fluke.. but I recall pulling a lot of high 16000 and low 17000 while doing a small amount of tweaking. There's nothing here that shows me why every benchmark I run is telling me 25-30% performance loss. Even that throttling above the card is still running pretty high clocks. Is there something intrinsic in the throttle where it can appear the metal is running close to where it should be but it's deceiving?

I can do the undervolt again and get rid of all those throttle indicators and it'll still be a firestrike drop of 3K or around 25% at least.. I tried other benchmarks like superposition and the same, way lower than a 7700K+1080 should be. I would think the clocks that seem close to where a 1080 should run don't lie, but maybe there's something else afoot.

I'll see if I can apply the TIM this weekend and see what that does, but I don't have high hopes for it given the undervolt to keep it below 90 didn't help. If that doesn't help I have an m.2 ssd laying around, I'll throw that in and use the system restore USB I bought from hidevolution and see if there's something Windows related going on.

 

Are these F5 heat sinks made out of aluminum and just copper plated? I've been working on this with alcohol pads for an hour... IC diamond was on here previously I think..

If so I'm screwed, can't put LM on this garbage

EDIT: bah.. did it anyway.... i'll buy a new heatsink if it comes down to I ruined it.

 

well... still getting Power and Reliability Voltage "throttling" .. if running at 1800+ for the whole test is throttling...
81C max now with LM ... re-delidded and LM'd the CPU .. that maxed at 82... It was LM under the IHS previously so it stands to reason that didn't get a huge benefit..

https://www.3dmark.com/compare/fs/15725194/fs/13778386

 

Hi,
The heatsinks are NOT copper plated. It's a full copper block. it's just aluminum around the edges (or another material, i don't know if it's aluminum or another metal). The copper section is pure copper.

Reliability voltage "limits" is normal, even desktop cards have that. It just means that the card can't boost any higher because the current voltage is limiting the clock speed or otherwise it would crash instantly.

Power Limit on an unmodded 1080 is normal. It means you bypassed 200W TDP so you hit the power limit. I actually thought the power limit of a 16L13 1080 was 150W. Does that card have an external power connector?

 

Try to use MSI GT62VR camera drivers. That worked for me in the past

 

Looks like mxm 3.0 with no other connectors, I didn't take the card out or study it closely though.

I guess if reinstalling Windows doesn't help I'm down to the super mystery performance drop. Clocks look good (they always did, I didn't expect reapplying TIM to help much).

 

If you read my old post i was with the same trouble, i still using my 4k display B156ZAN02.3 that looking perfect, but i still can't update to newer nvidia drivers, however if you will find a solution let us know.

 

@hfm , just been reading your posts about CPU/GPU performance issues. Looks like you've got your CPU issue sorted - good. Low GPU scores you're seeing in Firestrike - are you using the Ctrl-F Afterburner curve and locking a frequency & voltage? I tried doing that on my desktop GPU, and it's possible that if you choose a frequency that is way too high for the voltage you're setting then it will not crash but instead show really poor performance (close to stock). I saw this happening on my GPU when I locked something silly like 2100Mhz at 1.0V. If you are using the Ctrl-F Afterburner curve, then I suggest as a test that you set everything in Afterburner to stock and then just do a traditional offset overclock - e.g. +100Mhz on the core or something (whatever is stable) - then take a look at your scores. Pascal GPUs sometimes behave very stangely with the Ctrl-F Afterburner curve when it's not set up right - I'm personally not a fan of that curve - just try a traditional offset Mhz overclock instead as a test (you know...using the Mhz slider rather than the Ctrl-F curve).

 

Even if you're not overclocking, setting a custom voltage curve can do wonders. At stock frequencies and voltage, my 1080 was seeing slightly less than expected performance. The system wouldn't pass the 3DMark stress test, with ratings in the 93-94 percent range. After more time than I care to admit, I eventually got the curve set in a satisfactory way. It achieved greater than 97 percent in the stress test, and it got higher, more consistent frame rates.