Sager NP8454 (Clevo PB51RF-G) early impression

Thank You for answer.
Manual is quite confusing (page 17 - service manual)
"
M.2 Slots
Slot 1 for Combo WLAN and Bluetooth Module
Slot 2 for SATA or PCIe Gen3 x4 SSD
Slot 3 for PCIe Gen3 x4 SSD
"
It is good that it works in Your configuration.
I am wondering if Slot 2 nvme will share the PCI bus with SATA drive and slow it down under full load.

 

Could you please restate your question? If you place an NVMe module into Slot 2, it won't be connecting to the M.2 SATA pins, so I don't see how there's loading problems going on with the 2.5" SATA bay. Now, if you're asking if a SATA module is installed into Slot 2 (why would you do that?), it would be up to @Meaker@Sager as to whether the SATA pins are shared between these two modules?

In any case, I'm beating up these interfaces to the max, and definitely have not found any problems with loading in this fully-loaded system!

 

The south bridge handles all i/o except for the GPU on intel mainstream platforms. All other i/o is therefore capped at a total 3.5GB/sec no matter the interface.

 

This phrase is confusing for me "Slot 2 for SATA or PCIe Gen3 x4 SSD". I don't know exact internals, but word `or` at manual sounds like the SATA and PCIe exclude each other(or slow down). So if you have SATA drive at bay You second M.2 PCIe drive will be somehow influenced.
I am still thinking about upgrade and want the best long term solution for drives and don't want surprises

 

It means slot 2 can take either card type of SATA or PCI-e which are electrically different.

Slot 3 is PCI-e only and wont work with SATA based devices.

 

What that phrase means is that you can either have a SATA SSD in that slot, or you can have a PCIe/NVMe SSD in that slot. It refers to that M.2 slot only, not to anything in the drive bay. In case you didn't know, there are (older) M.2 SATA drives. Does that make more sense now?

Sent from my SM-G975U using Tapatalk

 

would this unit handle the i9 9980hk well enough? I've been impressed with the laptop from reading posts from this thread so far but one one mentioned the i9 sku. i'm looking at the i9 + 4k oled option for cad drawing + video editing but not sure if this laptop faces throttling issue like other thin 15" with 8 cores

 

I've got the 9980HK and it handles it great.

At max fans I can run it at 4.6ghz all-core at around 150W which is absolutely bonkers. As impractical as it is, it's nipping at the heels of the 9900K at those settings.
At stock it tends to stick to a more reasonable ~80W limit which puts it around 3.9ghz all-core which is still far beyond the base clock of 2.4ghz (which is what's required to stick to the absurd 45W listed TDP).

For day to day use and gaming I run mine at a "mere" 3.6ghz all-core which maxes out around 68W and 80C without lifting the laptop rear up. With a cooling pad or something to prop it up, you'd be looking at 75C on the cpu core which is very good.

I've never seen it hit the 5ghz single core turbo though. I don't think it's even possible due to the Windows scheduler always allocating work to more than 1 thread.

 

How long can it do this before overheating, enough for a short CineBench run?

 

It might hit the 5GHz mark but only for a millisecond.

 

Would you mind reporting the PL on your unit?
The 9750h config is limited to 55W in Performance mode, and that is enough to push it to almost 90deg while gaming (i.e. with gpu pulling 115W).
With cpu-only load I think it is settling at 80deg with fans at medium speed (it is way too noisy at max speed!), I need to doublecheck my data.

I wouldn't expect the cooling system to be able to handle 150W on the cpu

Does the 9980hk config have a beefier CPU heatsink?

Based on my experience (9750h 2070 config) , I find this model has great GPU cooling, but not-so-good CPU cooling.

 

M.2 SATA(older) drives. That makes sens. Thanks.

 

Have a read of this.

https://www.google.com/amp/s/www.on...3-vs-m-2-vs-nvme-overview-and-comparison/amp/

 

I was so focused at best M.2 NVME that I forgot that M.2 SATA exists

 

Woah, imagine this 15" could handle 150W from CPU. What a beast. Whats your ambient temp? and whats the fan setting + temps for 4Ghz @ 80W.

Also, from the reviews I read this laptop has a slight problem with airflow and need to be propped up, right?

 

Hi, @Le Quan! For all the heavy duty benchmarking I throw at my NP8454, I'm absolutely delighted with the TeckNet Laptop Cooling Pad. It's not silent, but it is very quiet, and keeps the system cool. It's a perfect size for the machine, and gives a nice sloped typing angle. It doesn't even need a USB port, since it includes a USB hub port where you can plug in any external keyboard/mouse receiver. There's no setup or maintenance; just plug it in to the rear USB Type A connector on the back of the machine using the included cable. No fiddling with switching it on or off, since it follows the system's power. It's also just $22. Highly recommended!

 

It lasts long enough for a Cinebench run and a FireStrike run.
Firestrike is here: https://www.3dmark.com/fs/19676402
Cinebench:

I suspect it could do 4.7 or even 4.8ghz with a cooling pad and liquid metal.

I have no idea what the absolute Power Limit is (other than being limited by the power adapter). HWInfo reports Unlimited for max PL. At stock PL1 and PL2 are set to 55W and 107W respectively.

Not that I know of.

It does indeed. On my overclocking runs I run it propped up.

Ambient temps during testing is ~22C.
At 4ghz all-core, max temp is 77C with automatic fans (propped up and ~70% fanspeed by my estimation). This is under a CPU only test though.

 

Sounds about right. CB and FS Physics Test last about the same duration from what I recall.

 

Sata versions can be had pretty cheap and still offer very good performance.

 

Nice one. I saw a review of the new Aero 17 (9980HK) on youtube and it maintains above 4GHz under extended combined CPU+GPU stress test with liquid metal and reaches 84C max. This clevo might do slightly better with LM repaste. The Aero beats this in appearance by far, though comes at a more expensive price tag.

The Aero can also hold 4.6GHz under CPU only test (Aida64).

 

Hi fellas,

Pretty stoked to get my new PB51RF-G . Mine is configured with the i9 9980HK + 2070, and the 1080p WHVA 144Hz Gsync display.

Overall, I'm pretty impressed with the build quality, chassis feels solid enough. However I feel the screen lid is a little on the flexible side. I guess that's a trade off for having a thinner lid.

Clevo Control Center 3.0 functions well so far, but I'm currently still on Win 10 1803. The fans will ramp up aggressively in "Performance" mode, and with a 6% offset on Automatic, they will reach max speed well before 80 deg C on either CPU or GPU. When the CPU is loaded the GPU fan will also ramp up to help even if the GPU is idling around 10W on the desktop at 50 odd deg C (only tested in discrete so far).

Still this didn't save my 9980HK from thermal throttling when it wasn't propped up, with the stock 4.2 GHz at 95W all 16 threads doing 7-Zip, albeit with 25 deg C ambient and stock paste. I feel that the CPU's auxillary heatpipe leading to the heatsink on the left side is inadequate, the air that came out there felt slightly less hot than that of the two rear exhausts. With the CPU loaded, it also struggled to cool the 2070 under load with the GPU creeping up to 87 degrees C where it started to thermal throttle.

According to HWiNFO the combined system power draw plugged in seems to be limited to around 220 W.

Got a few questions now,

1. With the current 9th gen laptop CPUs, what core voltages are safe? When a core boosts > 4.6 GHz, it appears to reach a max VCore of 1.396V! Is this normal? To start with, I've currently set an undervolt offset of -100mV, with a 4.2 GHz max boost ratio and it still does max 1.274V under load. Do these laptop CPUs run at a higher voltage?

2. Does Intel XTU conflict with Clevo Control Center? After increasing the PL1, PL2, Core ICCMax values, setting undervolt and changing the clock ratios in XTU 6.5.1.360, the system was slightly unstable and wouldn't reboot properly. It'll always do a "false start", that is without POST and then turn itself off and on again before successfully booting, with all settings reset in XTU. Even the saved profile in XTU was lost if it wasn't exported. Note that this was with Clevo Control Center left in Entertainment mode.
Setting those same values in Clevo Control Center instead with the Performance mode active was ok, with the system stable and rebooting properly. I haven't tried changing values in XTU with Performance mode active.

3. My unit has some audible coil whine, especially from the GPU power supply when it is under heavy load. Does anyone else experience any? I'm not too fussed as the fans drown it out anyway.

4. What PCH / South Bridge temperatures do you get? Mine once went up to 85 deg C, so I put a couple of thermal pads between it and the GPU heatpipe above. Now the PCH idles around 55 deg C but still maxes out at 77 deg C when the GPU heats up.

Cheers!

 

Those peak voltages are pretty normal.

 

Just to let folks know what a difference a cooling pad can make, on my MacBook Pro/Bootcamp (i.e. not an NP8454) , CineBench R20 scores jumped from 427 to 714! It's really worth the $20...

 

Oh ok, thanks. Just seems a tad high compared to my old Ivy Bridge laptop. I was reading up on the ASUS ROG G703GX review on UltraBookReview.com and it seems like their core voltage reached 1.5 volts whilst pushing 5 GHz!

I guess these mobile CPUs tend to use more volts than amps. Any ideas of the safe maximum voltage?

 

You will be thermally limited before voltage. Remember boost is now more aggressive per instant and they can get towards your old 5Ghz.

Plus frequency scaling is different.

 

Hi Mark.
Can You estimate what would be difference in speed between your RAM settings(tuned) and Kingston DDR2666 CL19(currently in my lap)?
What I am interested is speed in software applications(apps), not so much in games nor benchmarks.

Does the difference is noticeable?

 

Intel i7-8700K Coffee Lake Memory Benchmark Analysis

 

Those games with a difference are likely seeing an improvement in minimum frame rates but it's certainly usually not a big difference.

 

Hi, @DkmCd! It's interesting to note that your current memory is considerably slower than the slowest memory they tested - CL19 vs CL14 at the DDR2666 speed. They also emphasized that CAS latency is particularly important, so this makes a difference. The question is, will you notice a worst-case application delta that I'd estimate would approach 10%?

I'll do some actual tests to verify this conclusion when I can, but it appears that my CPU heatsink is falling off at the moment! I'm seeing a 17° C delta on my core temps, and will be fixing this today (with heatsink tweaks and IC Diamond paste).

 

True the latency is pretty high.

 

WOW. I just check IC Diamond paste out of curiosity. It has 2400 W/mk !!! thermal conductivity. I thought that my liquid metal is the best with 73 W/mk

 

That’s the conductivity of pure diamond. ICD has a rating of 4.5 W/mK.

 

So it is just good marketing. LOL
"IC Diamond 7 Carat Thermal Compound maximizes thermal heat transfer between the CPU core and heatsink by taking advantage of diamond's superior thermal conductivity Purified synthetic diamond has a thermal conductivity of 2,000-2,500 W/mK compared to 406-429 W/mK for pure silver."
https://www.amazon.com/Innovation-C...1_3?keywords=IC+Diamond&qid=1572029344&sr=8-3

Paste
Thermal Conductance: 4.5 W/m-K (data acquired with an ASTM D - 5470 thermal interface test instrument)

 

Is it possible to change at Control Center, Fan speed the lowest temperature (first dot)? Is there other software to control fan speed for this clevo?

 

Whoops! Thanks for the heads-up on IC Diamond; back it goes.

 

Yeah the thermal conductivity of a solid and then particles of it floating in a solution are two very different things.

 

Hi, Gang!

This is the first time I'm taking off the heatpipes, and seem to be missing something. I've removed the four fan screws, and loosened the six heatpipe screws, but something else is holding it down towards the center, across the width of the pipe assembly, closest to the battery. Might I ask what the next step is to remove the assembly?

If you're willing to pass on a service manual, that would be great, too! I'd greatly appreciate your assistance in any way.

Thank you!
Mark

 

You can find service manual here: https://premamod.wordpress.com/

 

@yrekabakery: I sincerely appreciate the service manual link. Thank you very much!

While the service manual didn't talk about removing the heat sink assembly, I finally figured it out; there were two screws hiding under the keyboard that hold the assembly in place. There's the 9750H...



Interestingly enough, it had looked like the chip was covered in thermal paste, but there was about 1/4th of it that apparently had an air bubble, which no doubt led to my weird temps.

I've decided to use graphite thermal pads for my install. I'll let you know how it goes!

 

Because of the 4.5 W/mK rating? Low W/mK doesn't mean it perform worse than thermal paste with higher numbers.

 

Now this is more like it. Instead of Cinebench R20 topping out at 95°C, it tops out at 83 °C, and the delta between core temps is just 3°, instead of 17°. I'd call this Graphite Thermal Pad a great success!

@Papusan: Sure, it's clear that different pastes have some variability in real installs, but a factor of 8X definitely starts to get my attention. That's the delta in the numbers between IC Diamond, and this thermal pad. I'm glad that I made the change.

 

Nice alternative for liquid metal with that pad(35W/mk). Definitely cleaner, less risky in application and REUSABLE.

 

Till now I didn't install Intel® Management Engine Driver. I know that there is "evil" Minix OS inside every intel chip .
Is there any performance/gain benefit from installing MEI drivers for this laptop?

 

Some reading...

 

One interesting consequence of the graphite thermal pads is that I no longer see any thermal throttling in benchmarks. Instead I see PL1 in CORE, and EDP OTHER in RING. This doesn't happen until the CPU warms up in wPrime; perhaps 45 seconds into a 1024M run, and continues until the threads start completing. I've been pretty aggressive in ThrottleStop's TPL. Should I increase the amounts even further there, or am I just running into hard-wired 9750H limitations? Thanks for your input!

 

Ahh, a simple fix for ThrotteStop anyways. Just enabling "Disable and Lock Turbo Power Limits" seems to have stopped the PL1 and EDP errors.

 

Hi Mark!

I have some confusion with bootable drives. Currently I have 3 disks: M.2 Nvme(Windows 10) , M.2 SATA(Windows 10), SATA 2.5 bay (Linux)
I have moved my 2.5 SATA from other computer and I would like to make it as BOOTABLE backup(as You suggest) and emergency drive. I look at BIOS but there is no possibility to make 2.5 SATA BOOTABLE. Only options at BIOS BOOT are: M.2 SATA and USB Boot if present. Do I have to make a GRUB bootlaoder to make it bootable or there is other trick?

 

I have made some simple tests with the same settings as Your at Throttlestop including undervolting: Offset voltage -230mV(CPU Core) and Offset voltage -99.6mV(CPU Cache). System looks stable at AIDA64 system stability test. Max temp for CPU only test is 73 deg. C. Max temp. for all components test is 83 deg. C. No throttling.

Cinebench 20: 2634 points.

Liquid metal is applied at CPU.

 

To become bootable, the BIOS just looks for directories called /EFI, /EFI/BOOT, and executable EFI files on a drive (such as /EFI/BOOT/Bootx64.efi). It won't boot into legacy (non-EFI) drives. I'm not sure if Grub is necessary, since I haven't run Linux in the EFI era. One good option to make it easy to change O.S. at bootup is rEFInd...

 

For what it's worth, my Cinebench R20 peak score is 3129 cb, but my peak CPU temp is 83°C.