*** Official Sager NP9175 / Clevo P775TM Owner's Lounge! ***

if you read through this thread you’ see that prema bios is only available to prema partnered suppliers, therefore to install it, go buy the machine from a prema partner

 

You can't get the bios unless you buy a system from a Prema partner unfortunately.

 

Despite this the aim is to reduce operating temperature. In the end is it true that permanent undervolt of Pascal GPUs is impossible?

 

I received some tips from Prema how to change long term power limits and now it is stable like a rock , no fluctuations. If You have Prema bios I can guide ypu what changes you need to do.

 

Prema has sent me the same, but im still getting the fluctuations.

 

You can undervolt, but like you said the curve can't be moved left or right so you have to work around that limitation. You have to move the whole curve up using an OC offset in the main interface or Shift+drag, then flatten out the curve after your desired voltage. Go from this to this.

Permanent as in done in the vBIOS? You need an SPI programmer to flash back a modded vBIOS on Pascal. Undervolting using the curve in Afterburner or Precision is done via the driver/NvAPI so it's not "permanent".

 

Pick a voltage you want to keep the core at, say 0.912v then move all the points after that up to the same core clock. Start off with like 1886 and try 3DMark Fire Strike Ultra Stress Test abd see how it goes.

Right now im trying 1873 at 0.912v and hopefully i stay at 85C

 

This is what im facing with CPU temps. https://imgur.com/Lxy0mjL
And that's just with CPU-Z stress test

 

Can you share these tips/settings?

Regarding my gpu (1070) i found it "factory boosted" to 2ghz (!!!!). Restoring on its default frequencies (1750~ boosted) with and heavy undervolt (0.825) by msi ab, now it works lovely @70° max (65/67avg) with lower fan speed (and noise level!). This machine is really noisy under load, but i had to say that it works beatiful.

Now i'm going to find out how cpu freq/voltages are working...launching an intel xtu stress test i can see the "Power Limit Throttling" cut out my frequency

 

If you can I would use throttlestop. CCC & XTU have a history of conflict... Just set cpu and uncore to -100mv and see how she fares after that. you can slowly work way down further to see how far you can push it but 100 is a good place to start.

 

Make sure uncore is linked to cpu cache voltage and not to "I/O analog voltage" though. Bug in throttlestop? :/
No hate on my excellent paint skills, brah!

 

I think that your problem consists of several factors.

The most important is uV potential of your CPU. My previous 6700K could handle -180mV, it was awsame, cool and quiet. Now i have 8600k and max stable uV is -120. This is a lottery.

If i put 4,7Ghz at all cores I also have thermal throtling in stress tests but in games its good but it is warm and loud anyway. Thats why for my every day gaming i use 4,3Ghz -120mV and its stays near 60-65C (in games, remember ).

Second factor, in my opinion, is separated heatsink. I think that combined heatsink from version DM/DM1/DM2/DM3 was more effective in stres tests CPU, because whole heatsink with both fans was cooling CPU (in 6600K without delid nad had 55-60 temps and fans was
barely audible, I have video from that time

)

I am more disappointed of cooling capacity gtx 1060. In P775DM I had 980M with prema OC and I thought that the gtx 1060 would be easier to cool down. I was wrong . I think that combine heatsink was better, thats why (for example) Rog handle gtx 1070 with very quiet way.

As soon as I have free funds I will buy this https://www.ebay.com/itm/new-for-Cl...676641?hash=item3d54db8121:g:eNQAAOSw~vpaTFqV and I will share test results.

 

I give up lol, I've spent 4 days on this and im no where near to getting it near what i want.

 

What "Cpu package power" hwinfo show on stres test?

 

Please contact our Tech Support Team for help with this.

If you should ever need tech support or warranty service, please understand that I am not a technician, so it would be better to contact our Tech Support Team by phone at 011+1-562-485-6541 Extension 3, between 4 PM and Midnight GMT, Monday through Friday, by email at [email protected], or by Online Support Ticket at http://www.hidevolution.com/contacts/ (in the department dropdown menu select Zoltan S – Tech Support).

Please let me know if they don’t resolve your issue.

 

It shows 123W.

Yep, I'm in conversation with Zoltan.

 

I keep seeing people posting -xxxmV undervolting etc. You guys really need to use override (static) voltage mode instead of adaptive.

 

Yep, I've already switched to override to 1125 in bios.

 

At what clocks?

 

4.7GHz

 

Your chips seems better than mine. I need 1175mV override to be stable at 4.7GHz.

You get 123W package power running CPU-Z stress test? I get less than 100W at higher voltage than you. 123W is almost like Prime95 for me. Does Hyper-Threading really increase power draw by 30%? Doesn't seem right.

 

I do think that HT increases power draw. I don't think my system is stable though. I am able to run cinebench and CPUZ stress test, I haven't tried any games yet. This is my testing that has been done today. Any other stress test software makes my clocks fluctuate like mad and that doesn't seem right.

 

my volts running cinebench. CPU package power is 126W.
https://imgur.com/dIs8ves

running CPUZ stress test, my volts are 1.18v

 

11min CPUZ stress test stable it seems 89C max temp

 

You don't get anything for free, if you are increasing the time spent processing you are increasing the power draw.

 

I know HT increases the package power, but based on my tests with HT on vs. off on i7s in the past, it's like 10-15% at most, not 30%.

 

What are your temps for both CPU and GPU?

 

82C/83C max on CPU/GPU when stressed individually in Prime95 and PlanetSide 2. Haven't tried them at the same time, but I can't imagine 330W of combined load ending well with a unified heatsink.

 

Wow, i.m shocked and you're running the little brother too.

I can't even run prime95 and get all cores to stick at 4.7GHz, they go all over the place.
I wonder whats up with my system.

Btw, how are you getting voltage monitoring to show up in AB, i have it checked in general, but i cant find it in monitoring.

 

Open Limit Reasons in ThrottleStop like I showed in my screenshot and check what it says when the clocks bounce around during Prime95.

The GPU voltage and power, and CPU per core VIDs, in the OSD are from HWiNFO64.

To enable the HWiNFO plugin in Afterburner, you have to click the three dots:

Spoiler

Then tick the checkmark by the plugin and click setup:

Spoiler

Add or remove the data sources you want (HWiNFO needs to be running):

Spoiler

By default the source values are in integer format, so if you want voltages displayed in floating-point you have to change format to %.3f (for 3 decimal places) in the source settings, and also change group to GPU so it's grouped in the same line as AB's other GPU data:

Spoiler

For GPU power if you want 1 decimal place precision you change format to %.1f:

Spoiler

For the CPU VIDs it's a little different. You have to change group to CPU%d to group it with AB's CPU data, then use the instance number to change which line it appears on. CPU1 is instance 0, CPU2 is instance 1, CPU3 is instance 2, etc.:

Spoiler

If you want to change the order of data within a group you just drag and drop the graphs within the list in the monitoring tab to rearrange the order.

 

I don't have the setup button

 

Are you on the latest versions of Afterburner (4.5.0) and RTSS (7.1.0)?

 

doh, i'm on 4.4.2. Didn't know there was an update.

TrottleStop shows limit as EDP

 

EDP under what categories, Core and Ring?

Try increasing the current limit in BIOS.

 

Thank you so much! I managed to do it.
https://i.imgur.com/qu3uNtO.jpg

 

Yep both core and ring. I have increased power limit 1 to 200 and power limit 2 to 250.

 

Looks good. Just keep in mind that CPU2 is actually the Hyper-Thread belonging to CPU1, CPU4 is the Hyper-Thread belonging to CPU3, etc. So you might want to move the VIDs to the odd numbered CPUs (these are the physical cores) for accuracy's sake.

What about current limit? Try setting VR current limit to 1104.

 

Hmm... Right i forgot about that, i might change it up then.

Where would the VR Current Limit be?

 

Should be in the same menu as your power limits but you may have to enable custom VR setting first before it shows up.

 

Alright ill take a look again.
Edit: i saw the VR current limit. According to Prema he said to set it to 900.
Edit2: I change the VR current limit to 1600 which is 200W, still core clocks fluctuating.

I'm having trouble with the instance for the VID, cant seem to get it right.
0, 2, 4, 8, 10 is the correct instances.
Also, how did you get the OSD to show CPU1 - 6, mines shows 1, 3, 5, 7, 9, 11

 

Maybe there's some other limits you have to raise, but without knowing what additional options your BIOS exposes compared to my stock BIOS, I can't be of much further help I'm afraid.

Your OSD in that screenshot looks fine, just change the instance to 0, 2, 4, 6, 8, 10 like you said for each VID. Mine only goes up to CPU6 because remember I have an i5.

 

Arr right I forgot you only have a i5. Thanks anyways for all your help!

 

@Donald@HIDevolution Could you or Zoltan please send me the instruction on how to undervolt and disable Turbo Boost in the BIOS interface please? I think I disabled overclocking entirely and that also disabled undervolt settings. Still haven't installed Windows yet but I want to tweak BIOS to my liking first.

Thanks.

 

If you should ever need tech support or warranty service, please understand that I am not a technician, so it would be better to contact our Tech Support Team by phone at 011+1-562-485-6541 Extension 3, between 4 PM and Midnight GMT, Monday through Friday, by email at [email protected], or by Online Support Ticket at http://www.hidevolution.com/contacts/ (in the department dropdown menu select Zoltan S – Tech Support).

Please let me know if they don’t resolve your issue.

 

I wouldn't tweak the bios settings before installing windows. Settings that POST would sometimes not load windows or even BSOD.

 

You need to run the system stock to ensure all is well for a while too.

 

Right so, i woke up today to find my CPU fan is stuck on full fan speed. Nothing I do will get it to go back to normal. Any ideas? Seems like CPU GT Cores is running @ 75C ?

 

Checked temp and Cpu/Gpu load? Test on/off max speed fans with Fn+1

 

Yep did that. Idle on the desktop. Hwinfo shows CPU GT Cores at 75C. As i understand the system doesn't even use the integrated gpu, so why would it be even used.

https://imgur.com/B1mhKm2

 

iGPU is physical still in Cpu die. Hence you will see temp reading. But it’s strange the Package temp is high while cpu cores show lower and more normal (idle) temp. Check as well with another tool.


Edit. Check (performance - running processes) in task manager.