*** Official Clevo P177SM-P170SM / Sager NP8290-8270 Owner's Lounge ***

Hi Meaker, thanks for the response.

I ran another bench today doing a small oc to GPU clock 150mhz core and 350mhz to mem. I did not notice any stuttering during the bench, however I did notice cpu core clock dipping down to around 2700mhz for split seconds a few times during graphical test 1 and a small portion of graphical test 2. Look at the purple line. I am not sure why this is happening, temp seems to be fine so it shouldnt be throttle right?

Anyhow
My new bench score (also my best to date): http://www.3dmark.com/fs/26283496
and bench I am comparing with: NVIDIA GeForce GTX 1060 (Notebook) video card benchmark result - Intel Core i7-4800MQ Processor,Micro-Star International Co., Ltd. MS-1763 (3dmark.com)

Here is a the leaderboard for people benching with 4800mq and 1060gtx: Search (3dmark.com)

All the benches with the swedish flag are<my benches. I have the worst Graphical test score between 10000-10600. Everyone else scores 11000+ on graphical test.


EDIT:
Okay it seems like Ive managed to flatten out the cpu clock curve by editing PP0 Current limit.
Got a new highscore: http://www.3dmark.com/fs/26284162
The score improvements seems to be mainly from Physics test score, so Graphics test 1&2 score still garbage . Any idea?

 

Remember the board's can have some highly tweaked numbers in them.

Have you tweaked your LOD in your driver's? Your CPU and combined numbers look good. It's the individual first two tests that are lower.

 

Yeah driver settings has been tweaked for optimal performance. My cpu is basically carrying the bench score…My 1060 is getting the worst graphics score out of all the cards which kinda sucks. I just ordered a programmer so I can tweak the cards tdp. Hopefully that helps

 

If it's restricting the clock via power limits then that would help.

 

I do get powerlimit warning in various softwares like GPU-Z and Aida64. It doesn't seem to affect the clockspeed that much though...still running at 1,873mhz average which is normal for this gpu.
Even then my graphics score is bad. I have seen people run firestrike with average 1,600mhz gpu clock speed on the same gpu with better graphics score. How is it possible that a higher clock speed yields a worse score? Does powerlimit throttling automatically lower clock speed or is there any other way it can make the gpu run worse despite seemingly normal clockspeed reads?
I am so confused

 

Technically under panic thermal conditions it can skip cycles (clocks but does no work) but that would not apply. If you have tuned everything try stripping out the GPU drivers and re-installing them.

 

Ahh okay, so thats a possibility. Upgrading gpu from 780m to 1060, i’ve reinstalled gpu drivers multiple times. Always removing with DDU in safemode, then installing modded inf driver file with signature enforcement off. Idk if it matters which driver section you use in the inf file, but i’ve tried out a few with no difference in performance. I’ve also tried with a little older driver 461.92 and 417.71. My gpu was running hottest at around 65c during the bench. Is that really hot enough to cause thermal throttling?

 

I only tried TimeSpy and I get a 3600 GPU score in it.

 

It will be interesting what a higher power limit would do, the clocks are still somewhat variable. Have you used a more advanced program to track the clocks?

 

Hi Blacky. What specs and drivers did you use? I just ran a timespy bench and it did surprisingly well in terms of graphics score there 3790. I wonder why it runs so bad in firestrike :/

Edit: hmmmm now when I think about it. Timespy uses directx12 and firestrike uses directx11. Ive been receiving bad performance on games like sekiro that uses directx11. I wonder if there is any connection or merely a coincidence. Need to investigate this.

 

I am not sure what would be considered advanced programs. I use gpu-z to track gpu and throttlestop for cpu. I use msi afterburner and rtss to monitor clocks while benching. I noticed a weird behavior with my cpu occasionally going back to base clock from time to time during bench when cpu usage is low and it draws less power. When i oc all my cores from 37x to 39x this issue stopped.
I also logged gpu clocks and I noticed the huge clock variants. It moves from 2000mhz and ocassionaly to 1400mhz (although this probably between loading screens when gpu load is 0) but most of the time stays between 1700-1850. Temp stays at constant 62-65c.

 

Did you find any 3rd party reviews running 3dmark?

 

3rd party review as in review websites like Notebookcheck? They listed TimeSpy graphics score lowest at 3396 and highest at 3790 (I got 3790 here so all is good).
In firestrike however, lowest is 10708 and highest 12298 (I got around 10600, so its even lower than the lowest. And this is with the clocks tinkered, otherwise I got 10000). I don't know what is different between the two tests besides DirectX versions :/

Pictures for reference:
TimeSpy graphics score:

lowest:

highest:


Firestrike graphics score:

lowest:


highest:

 

I have 4940MX with GTX 1060 and 32Gb of RAM. It may also depend on the drivers, I don't know.

 

Driver versions can have an impact, you could try copying the other bench.

 

Okay, thanks for sharing your bench scores. At least now I feel a little calmer knowing that timespy bench seems fine so hardware is not faulty. Just have to figure out why firestrike does so badly.
Do you think you could do a firestrike run and share your score and driver version? I only have 16gb ram. I wonder if upgrading it to 32gb could make a difference hmmm

 

Yep ive done just that. Looked through the top benchers and downloaded the same driver version as them. 3dmark seems to have an issue with older drivers though. Refuses to open as it gets stuck with message ”initializing”.

 

Are you running the steam version or standalone?

 

The steam version. I wasnt aware there were a standalone version xD

 

Try that version instead.

I was using 3dmark back in the 3dmark 2000 era....

 

I noticed that my card was running on pci express x1 3.0 instead of x16 3.0 during rendering and bench. I googled the problem, and a lot of people suggested to reseat the card and clean the pins. I did just that…and then i friggin dropped my heatsink and it bent and got dents(had to bend it back so it would fit on gpu die…probably made things worse). Now when I run anything, I get 90c. I probably f****ed up my heatsink. Card still runs x1 3.0 tho. Oh well, any idea where to buy p170sm heatsinks in europe and do you know if the pcie x1 is normal behavior?

 

A local reseller or the likes of aliexpress for heatsinks. PCI-E 1x is a power saving feature when the bus is not being used much. Could cause some slowdown.

 

I turned everything to performance mode, both in windows, clevo software and nvidia, but it still refused to switch to x16. Thanks ill check if there are any local reseller

 

Did you use gpu-z to test the link speed? It has a render test and good monitoring to make sure it's not a big in detection.

 

Hi, yes. When i open GPU-z it says Pci-e x1 1.1. When i run the render test, it switches to x1 3.0. But never x16 and I have no clue why I tried removing the card and cleaning the pci slot then reseating card, no luck. I also tried the gpu bench in aida64 and got only 700mb/s read speed which sounds like x1 speed so I dont think its a bug in detection. I will try to reinstall os and reflash bios next :/

edit: I reinstalled my 780m just to see if it was a faulty card. Seems like I get x1 there too. I hope the slot or mobo is not damaged somehow…

 

That would explain the results. What BIOS have you got on the machine?

 

Hi Meaker, I am running Prema Bios 1.03.04PM v2. Today i tried to install fresh os of windows 7 (I used 10 previously) to see if it was because of chipset drivers incompatability with win10. I also tried to remove cmos battery. Still stuck at x1 urghhh

 

The Prema BIOS does have a lot of options, how much have you tweaked?

 

I tried to look for linkwidth adjust in bios but I couldnt find any. Under chipset tab, it allows me to switch between gen 1,2,3 in dgpu but nothing more than that. Tbh Im not really sure what I should tinker there. I am not very good with tech stuff so I just learn as I go and ofc relying on ppl like you on tech forums Any suggestion what I could look out for in bios?

 

I never had an SM series machine so I never explored that BIOS, I was a P570WM guy in that generation.

 

Okay, thanks for all the help so far. I reverted back from windows 7 to 10 again. 7 was buggy as hell with the upgraded gpu. I will try reflashing the Bios today, and if that doesn’t help, ill have to admit defeat and acknowledge that the pcie port is probably busted.

 

Fingers crossed it works for you.

 

Thanks Meaker! The reflash didn't help at all, so mobo is probably faulty. Do you have any experience with switching motherboards? I have been looking for a replacement mobo for my laptop but I am not sure how I can make sure that the mobo I buy will work in my laptop. A lot of these pages states that I should look for "Part number". I found matching part numbers of my mobo with some mobos in aliexpress...but is that the only requirements for me to change the mobos? I noticed that my current mobo got Intel hm87 as chipset, but these mobos in aliexpress got intel hm86 as chipset instead.

 

The HM86 does not support raid for one thing. Ideally you want an exact replacement.

 

Damn ok, thats disappointing. The version with hm87 chipset seems really rare. Can only find hm86 and A55 chipsets.

 

I think it's the difference between the P170SM and P170SM-a.

 

I thought P170SM where distinguished by the partnumber 6-77-P170SMA0 while regular ones are 6-77-P170SM00? This one got 6-77-P170SM00 as partnumber, but hm86 chipset somehow

 

Maybe a recycle repair?

 

.

 

I need help please, i bought 8970m for my p170sm clevo. My old 8970m died, started having artefacts and thats reason i bought new one. But it seems i got wrong card becouse my laptop starts to make alarm sounds and caps lock, num pad and leds near power button start to blink, after maybe one minute of blinking and sounds laptop turns off.Fans are spinning at max speed also. I have prema bios. I tryed few bios for new graphic card but no help also. I even tryed bios from my old card but nothing. When i put my old dying car back, there is no problem anymore(except i cant use 3d apps becouse card is dying) and i can use pc normaly.Is there anything i can do or try becouse i have waited almost 2 months for card to come, and i really dont want to return it and wait again... This is the card i bought https://www.amazon.com/gp/aw/d/B081D5KG8X?ref=ppx_pt2_mob_b_prod_image
As i see by the other posts on forum it seems that either my card is not compatible (it should be), or temperature sensor is faulty (but it works with old card so probably it isnt). I tryed ecu reset also, all things i could find on forums and from people that had similar problem.Graphic card ia recognized in windows and in bios also. I really dont know what else to do and try. I have video of problem if needed i can upload it and put link here. Hope to find some kind of solution, im still a believer haha

Thanks for answers

 

@Improwizz Something's probably faulty on your card indeed. The sensor comes with the card, so yes that could very well be it... or maybe the sensor is OK but isn't connected somehow, and the EC panics since it doesn't get a signal.

That being said, I see you bought from a chinese seller, which is a very bad idea. While their prices are very attractive, there's a reason for that : the components are generally of bad quality and don't last long. Not long ago, I've seen a Youtube video of someone who got a chinese-sold GPU, there were markings in chinese on the card and a viewer who can read chinese said those were meaning "Doesn't work"... so sometimes they don't even bother removing the markings ! Why would they ? People are still buying after all, because it's "cheap". Cheap, really ?

My advice is : get refunded asap. And I know what I'm talking about, since one year ago I made the same mistake. The price was attractive, the seller even said there was a six month warranty... WHAT COULD GO WRONG ? So I got my card, the first day it worked well, but when I powered up my laptop the next day, it was dead. I spent weeks trying to get it working, en pure perte as we say in France, i.e. for no result. Then I asked for a replacement, so the seller asked me to return the card, but strangely he always was unable to give me a return address... there was always a funny reason for that, like the last one I remember "It's the spring festival right now, I'll give you the address once it's over". Oh really ? So I asked for a refund, but never got any answer from that point. So basically I paid a hefty price for nothing, heck, even far worse than that, considering the time I spent ! In case it may be of some use, the seller's name is "HT Sunshine", but I'm pretty sure he's far from alone, there's a whole scamming industry down there.

TL;DR : don't lose your time. REFUND (if you still can).

 

Sounds like a defective card.