*** The Official 2019 MSI GS65 Stealth with RTX GPUs Owners and Discussions Lounge ***

3DMark scores look bang on the money. You probably can gain another 500pts on graphics if you disable ASPM in bios.

 

Can you run the aida64 benchmark with the first 3 boxes checked? So we can compare our scores and see if our LM repaste are showing the same temps. For the GPU I didn't find the courage the flash to 90W, I think the GPU is already good enough to have every game out there run at full specs with decent framerate. The only things I did was OC the GPU a bit with afterburner (which give me around 300 points boost in firestrike). My score in firestrike is: 18 965 for graphics, 15 513 for physics and 8 322 combined.

 

Nae, ebay in India is terrible. I guess i'll just have to live with this now and mstsc into the PC when this laptop can't handle it.

 

Will do once I get my driver issue resolved (I edited my post to remove the scores and focus on the drivers since that's a bigger issue at the moment). I can't get the damn thing to boot into Windows and actually let me use the 2080. At the moment, it looks like the problem might be on the Intel 630 side, which reads as broken every time I reboot. Still sorting it out.

 

Seems like simply uninstalling the Intel drivers did the trick. It would be nice if I could use integrated graphics from time to time but I'd rather have this thing playable.

 

Do you encounter this issue that I posted few days ago by any chances after reinstalling your GPU drivers?

I'm encountering a software/driver issue with my unit, I wonder if you had it too.
Basically, on IDLE my computer is using between 15 and 25% of CPU by the process "System". After some research, it seemed the culprit was a bad driver installation so I went and reinstalled them all until finding it.
I finally found it and it seems this CPU spike come from "Intel UHD Graphics 630". As soon as I go into the device manager and I disable then reenable it the issue is gone and my CPU usage goes down to 3-7%. I tried everything from reinstalling it from the drivers on the msi website, the drivers from windows update, and even using DDU in safe mode for a clean install. But each time I reboot the issue come back and the only solution is to disable/enable the device. Does any of you have experienced this issue? Do you have a fix? Thanks!

 

Here are the AIDA64 GPGPU Benchmark scores:

RTX 2080
Memory Read 11967 MB/s
Memory Write Unavailable
Memory Copy 321951 MB/s
Single-Precision FLOPS 9188 GFLOPS
Double-Precision FLOPS 237.9 GFLOPS
24-bit Integer IOPS 8090 GIOPS
32-bit Integer IOPS Unavailable
64-bit Integer IOPS 1668 GIOPS
AES-256 24397 MB/s
SHA-1 Hash 71533 MB/s
Single-Precision Julia 2363 FPS
Double-Precision Mandel Unavailable

UHD630
Memory Read 11408 MB/s
Memory Write Unavailable
Memory Copy 40674 MB/s
Single-Precision FLOPS 401.0 GFLOPS
Double-Precision FLOPS 100.3 GFLOPS
24-bit Integer IOPS 133.6 GIOPS
32-bit Integer IOPS Unavailable
64-bit Integer IOPS 15.88 GIOPS
AES-256 1114 MB/s
SHA-1 Hash 4223 MB/s
Single-Precision Julia 105.3 FPS
Double-Precision Mandel Unavailable

8750H
Memory Read Unavailable
Memory Write 40611 MB/s
Memory Copy 36714 MB/s
Single-Precision FLOPS Unavailable
Double-Precision FLOPS 372.9 GFLOPS
24-bit Integer IOPS 334.4 FIOPS
32-bit Integer IOPS 334.5 GIOPS
64-bit Integer IOPS 46.61 GIOPS
AES-256 26579 MB/s
SHA-1 Hash Unavailable
Single-Precision Julia Unavailable
Double-Precision Mandel 129.9 FPS

 

Yes, I was having exactly the same issue. Uninstalling the Intel 630 drivers in the device manager and restarting (note: no reinstall was necessary) seems to have resolved my issues.

 

Lol thanks, it seems it has worked for me too... The solution was so simple lol

 

Related: I'm trying to install the Windows 10 October 2018 Update (version 1809) so I can run Port Royal. When I start the install process, it says my CPU is less than 1 GHz so it can't install. I assume this relates to all my Falkentyne changes. What did you guys do--did you roll back your BIOS mods to do the install?

 

That's the Hackness changes. You have to set it back to 2.1GHz from 8. Or whatever the minimum is.

It's the unlocking the i-8750 thread. @atxtonyc

Sent from my BLA-L09 using Tapatalk

 

Thanks! But I meant the stability tests actually, the ones that you get by clicking on the little graph icon. Then check the "Stress CPU, Stress FPU and Stress Cache". I'll remove my cooling pad and post my results running this for 10min

 

CPU stabilizes around 77C. GPU while stressing the CPU seems to idle at 53C. No throttling.

 

Here, no cooling pad, no fancy fan curve, nothing fancy except CPU undervolting of -0.050mv:

 

Here's mine after implementing Hackness and Falkentyne. These are definitely lower temps than I was seeing before, though.

 

Well this is beginning to really grate on me. I'm still trying to stress test my CPU to find the most stable undervolt. No matter what stress test I run (AIDA64, XTU) my underclocks almost always pass at ridiculously high numbers, and then it proceeds to crash the game when I load up Anthem. For example I've been running an -80 underclock on the CPU WITH the bios mod that removes overvolting, with the CPU running at full Mhz, and it passes all these tests no problem. Others have their range around 40 to be stable and I know for a fact **** is gonna crash in Anthem, as it has already. There's got to be a better way to determine my undervolt other than loading up Anthem for a BSOD, as that will take forever.....

 

TS BENCH does not report errors?
@falldragon

Sent from my BLA-L09 using Tapatalk

 

Experienced a similar situation. I've undervolted to -0.080 with all tests in TS BENCH passing and when I launched For Honor I would get crashes from time to time with my controller being disconnected. Then I changed the undervolt to -0.050, uninstalled dragon center and everything is fine since then.

 

Didn't know about this! Just tried it, got a few errors, lowered to 70 undervolt, no errors on any of the tests. But I'm pretty sure 70 will crash all the same.

 

Btw guys I dunno why but my power light is always in the ember color, even when no process is running on my nvidia GPU (checked with the NVIDIA GPU activity), I also tried to deactivate the card in device management but I still get that amber light activated. Also my battery longevity sucks, I get about 2h by just browsing the web and when I unplug my device it's at 92% instead of being at 100%. I also did a battery calibration from dragon center but I still get a crappy battery (same story on linux). Any ideas?

 

Run it maybe 5 times. Its normal for it to be stable and then crash on idle 2 weeks down the line. Just lower UV if that happens.

Sent from my BLA-L09 using Tapatalk

 

Have you tried the program "Process monitor"? You can sort by GPU.

Sent from my BLA-L09 using Tapatalk

 

I dunno for how long the fix lasted on your side but for me the issue back again lol. It seems to be even worse actually lol

 

After many tests... down to -60. Seems more reasonable. If I get any more crashes I'll just lower another 10 like you say. I also have an underclock 37-39 running without error at -80. Thanks for the help as it as somewhat eased my mind. Now to decide whether to fiddle with the GPU curve in afterburner again as after an uninstall and reinstall it's only taking up the normal .1% processing load rather than the 10% it was glitched to prior. I wonder if creating a curve puts a bigger load on the processor... even if it did I can't imagine it'd be 10%

EDIT: nevermind was looking at the wrong number. Afterburner is still pushing my "System" CPU load up to 10%. But after unchecking all the monitoring systems and setting my "polling period" to 60000 System use is back down to less than a percent. Guh so many ins and outs. Seems like polling period was the biggest offender even when all the monitoring systems had been unchecked.

 

I quite like the look of the silver P65 coming out end of april. However, I owned the first GS65 for all of 2 days before I realized I couldn't fix the filmsy feel of the chassis that creaked when I would pick it up or even type on it.

Assuming they use the same 2nd gen chassis for the P65, can anyone with experience with both Pascal and Turing GS65 generations tell me whether the creaking is totally gone or not? Thanks!

 

Palm creak is GONE. On the RTX versions. A lot sturdier, I love it.

Sent from my BLA-L09 using Tapatalk

 

Awesome, thank you. That and the really bad uneven backlighting on my model made me loathe it. Would get the GS65 now but I like the fingerprint sensor on the P65 and the less flashy colors.

 

Anyone have any ill effects from the 90W GPU mod (in conjunction with unlimited CPU TDP)? I've been getting stutters in Apex (and Firewatch, and GTA, but those seem to always be there...) and I don't recall them being there when I got the laptop, but maybe I'm misremembering. My temps are solid and my framerate is too but I've getting tiny pauses...

 

Found another cause of the process "System" eating all CPU: Afterburner. Just kill it and the System process will stop eating all your CPU.

 

Hello, guys!
Thank you for really useful information and tips in this thread.
I’m one step away from buying gs65 2070m, so far I have read a ton of info about rtx max q and different benchmarks. In this thread users suggest 2060 over 2070, but can someone provide please real life compression (numbers) of these two gs65 gpus (stock)? Also, so far I could find in web gs65 2070 has 90w tdp?

 

For those that have done a fresh OS install; I highly recommend you update the Intel RST drivers, replacing the MS Standard NVMe drivers. By doing so, this fixes the problem where the CPU package would not enter C8 state. This makes about a 4W difference when idle.

 

Along with the above I also ran a Firestrike run after updating all my drivers. Don't know if it was a fluke but I got my highest score yet!

 

Maybe I'm being very dense, but how? I don't see any option to sort by integrated/dGPU.

 

I didn't try that yet but I found what program was using my dedicated GPU: MSI Afterburner. I believe it doesn't use it directly but just monitoring the GPU is sufficient to make the light ember. Afterburner was also the cause of my 15% CPU usage increase when IDLE. I just removed it, installed dragon center again and I now use Rivatuner + HWMonitor for in-game statistics. At least I don't have any problem with that configuration. It's just my battery which sucks very hard even on battery saver profile and after having that white light on the power button lol: https://www.3dmark.com/pcm8wk2/108341

 

Any idea how to prevent silent option from running on the dGPU? I manually added it to the NVIDIA Control Panel and told it to use integrated graphics, but the minute I turn it on the orange light goes right on. On restart, the NVIDIA Control Panel shows that silent option is using the dGPU.

 

Results looks much better with Afterburner + dragon center closed

 

What do you mean by "silent option"? It's in dragon center or something?

 

No, it's the separate fan software that allows you to control the fans without dragon center.

 

Uh? I'm not aware of this, what is the name of the program?

 

It's literally called Silent Option: https://forum-en.msi.com/index.php?topic=255972.0

You can download the Skylake one if you want to give it a try.

 

Oh ok sorry I didn't understand, I thought it was an option in a third party software or something lol

 

I'm not particularly knowledgeable about this. But the Intel RST drivers can be used for the samsung nvme ssd? Does it give similar performance? The MSI support page lists the normal RST driver and RST F6 driver. Which one to install?

 

I have since tried this. It doesn't always work. It's called process explorer not monitor. You go to view, select columns and select GPU check boxes.

Sent from my BLA-L09 using Tapatalk

 

http://forum.notebookreview.com/index.php?posts/10812428

@hackness wrote this guide. Just follow the layout. You will get there.

Sent from my BLA-L09 using Tapatalk

 

Was going to flash my RTX 2060 80W with the Asus 90W; however subvendor and memory are different so holding off for now. If someone could provide feedback (for or against) it would be greatly appreciated.

Sent from my BLA-L09 using Tapatalk

 

Hey everyone....had to duck out for a while....(had to get some real work done).

Anyway, I ended up working with Zoltan from HID and he helped me through some testing. Here were his recommendations:

1) Use BIOS undervolting. Do not use throttlestop because it may conflict. If you really want to use throttlestop, then set the undervolting in BIOS to 0. I set mine to -64mv. 65 was the most Zoltan has seen reliably on the new GS65, and I did try -65 and -70 but had issues. I seem to be running stable at -64.

2) Don't run CPU and GPU test at same time in Aida. Yes, you can certainly test them at the same time to see how high you can get, but he said that is not realistic when gaming. You might see some games use one more than another, but rarely do they ever max out both. You will see 90+ temps when you do that because the heat pipes are trying to share the load.

3) Temps in High 80s (even 90/91) is OK. Today's CPUs are better at handling higher temps than before. I personally would prefer to keep mine to the mid to low 80s rather than the upper 80s. His suggestion was a CoolerMaster Note Plus U3 if I want to keep the temps down further. For now, mine is just elevated.

Now, what I have found is interesting. Some games like Witcher and Assassins Creed seem to raise my CPU temps more than my other games (Grimm Dawn, For Honor, Dark Souls III, Dragon's Dogma). I would hit 91C-92C during normal gameplay. But...what I found was that if I turned off Turbo, I would loose maybe 3-5 FPS but stay in the mid to upper 70s. All the rest of the games would stay in the 70s or maybe lower 80s with Turbo on.

So, I uninstalled ThrottleStop, Silent Option and AfterBurner and re-installed Dragon Center. I did this so that I could try out the different performance modes and see if any of them disabled Turbo. I also used it to set my fan speeds to 0, 60, 70, 85, 90, 100. Unfortunately, none of the performance modes (except ECO) disable Turbo in CPU. This was not an option.

After a bit more research, I discovered that adjusting your power plan settings in Windows to something other than 100% (i.e. 98%) would prevent Turbo from enabling. Bingo! Found what I needed.

So I created two new power plans in windows:

• Boost - Duplicate Performance leave CPU at 100%
• No Boost - Duplicate Performance, change CPU to 98%

Now, I switch to No Boost for Witcher and Assasins Creed and can game all day and keep temps low, with no throttling.

Here is my No Boost power plan:




Oh, I did end up leaving Dragon Center installed, but only use it now for controlling the Fan speeds.

In summary:

• Adjust undervolting in BIOS (in my case -64mv)
• Set fan speeds to 0, 60, 70, 85, 90, 100
• User Windows power plans to enable/disable Turbo
• Elevate/Fan assist laptop cooling

Now my GS65 is cool and fast with no throttling! Max temp during intensive gaming...87C...with a nominal temp of 77-79C. Fans are not roaring either.

Now...all I have left is to figure out how to maximize battery....here we go again!

 

Oh, one other very useful utility.

Check out Quick CPU ( https://coderbag.com/product/quickcpu)

It is great because it shows you exactly what you need to know about what CPUs are running, what power plan is operating and even lets you adjust your Turbo Index and Core Parking Index (which is essential for shutting down cores in battery mode).

I am not sure if this is a complete solution in maximizing battery, but it certainly is helping identify where everything is at in a single easy to read screen.

I did some gaming for a while just to confirm. Here is what it looks like for my current config.



Here is a screenshot after playing Assasins Creed with No Boost. Approx 56FPS on Max settings. Only hit 70C. On Turbo, I only get about 60-61 FPS, so it is not a huge difference; just the temps hit 91-92C.

 

Thanks for sharing this. Very helpful! Did you do the adjustments in BIOS for AC/DC load lines, etc.?

 

My 2 cents...
It's been a week since i bought this little beast.
Starved the CPU as far as i could.
Maximum fully stable undervolt! (tested with stress test and 3d benchmarks)
It looks like the CPU temps are just under the TT limit under full load for 5 minutes. (though tested it for long hours)
@Neothane i think you can push it a little harder!
And i know BIOS voltage changes are a scary thing. (Did not even touched the FAN curves yet myself!)

Anyway, in my use case (AI Object Detection On-The-Go) this machine is a great powerful working tool!
As always 10X!... All of you!
This thread helped choose my first (REAL!) AI mobile machine.

 

- Works fine for Samsung NMVE SSD. The Intel driver you're installing is just the host controller on the motherboard which is an Intel spec.

- Performance should be similar, I'm told the Intel one should be better but I never benched the SSD.

- I couldn't install the "normal" RST driver, running the setup .exe file would error out saying "hardware does not match manufacturer's ID" or something along those lines. Manually installing the F6 one worked fine. I think they're the same except one is in a .exe format and the other is in a .zip format that you have to install manually in device manager. If you look at the Intel download website for "Intel® Rapid Storage Technology (Intel® RST) User Interface and Driver" it's offered in .zip or .exe and the .zip format has the exact files as the F6 version on MSI's support page. Also (bizarrely), the version number offered on the MSI website (v17) is higher than the one offered on the official Intel website (v16.8), so I downloaded the version from the MSI support site.