Yes I think this is the case. I have written him another email and explained again what I meant. Should hopefully be sorted
Nova 15.6 Overheating CPU
- Thread starter demon28
- Start date
SpyderTracks
We love you Ukraine
Yeah, let me know what he says and if necessary I’ll get it clarified by the admins for you.Yes I think this is the case. I have written him another email and explained again what I meant. Should hopefully be sorted
I've emailed AMD as I think I've discovered the source of all the issues . ( Could be bios related but I'll find out from AMD to be sure)
CCD0 draws 15-20% more power for a given clock compared to ccd1.
So for example to get them to draw the same amount of power at 1.05v I have to run ccd0 at 3ghz and ccd1 at 3.7ghz .
That is a crazy amount of power difference even if you had badly matched binned CCDs I'd imagine they would be within 5% tolerance of each other.
Due to this ccd0 runs much much hotter at stock speeds and at static voltages.
This needs to be addressed.
CCD0 draws 15-20% more power for a given clock compared to ccd1.
So for example to get them to draw the same amount of power at 1.05v I have to run ccd0 at 3ghz and ccd1 at 3.7ghz .
That is a crazy amount of power difference even if you had badly matched binned CCDs I'd imagine they would be within 5% tolerance of each other.
Due to this ccd0 runs much much hotter at stock speeds and at static voltages.
This needs to be addressed.
Interesting observation. Ill have a look to see if mine does too . Have you tried different chipset drivers?
Just what came on the tech support section and the latest from AMD ( latest a week ago anyway)
I utilised open hardware monitor as this seems to be the most accurate and comprehensive sensor software for our laptop.
See photographs below of both CCDs set to 1.05v at 3.6ghx during a cinebench run where you can see the power and temperature difference.
Running it at stock settings or entertainment mode, power save and quiet mode results in the same behaviour just with smaller /larger numbers.
With these results in mind if this problem is resolved we should be able to shave 15c off maximum temperatures.
Last edited:
Hey @SpyderTracks I finally got a reply today. I think me and the PC'S representative are on the same page about what I was asking. The answer still seems to be that I wont be covered by Warranty. I am attaching the reply I got.Yes I think this is the case. I have written him another email and explained again what I meant. Should hopefully be sorted
I am a bit disappointed to be honest.
Attachments
SpyderTracks
We love you Ukraine
I just don't accept that, that's not ok.
You are asking for a PCS BIOS rather than a custom one? What exactly did you ask?
Who was it that replied to you?
Direct them here https://www.pcspecialist.co.uk/forums/threads/bios-updates.58007/
Having recently done a bios update I can tell you that once pc specialist send you the bios and update tool from a warranty/legal perspective you become authorised personnel. As they have given you the tools and instructions to carry out works.
Not only do they authorise they furnish you with the tools to carry out the job.
I'm pretty certain that's why they are so withheld when it comes to bios information because it is covered by warranty and it costs them money if it goes wrong.
Last edited:
That makes sense, I wouldn't imagine they would encourage BIOS updates. I am waiting for them to get back to me, Ill keep chasing this and see what they say!
SpyderTracks
We love you Ukraine
I'm going to chase them up also.
Be sure to find that section in their warranty and refer them too it
SpyderTracks
We love you Ukraine
Just to confirm for all, we've had confirmation from Mustafa who is very senior in the Customer Services department and his word is gold, it is as we suspected, that previous advice you received is incorrect.
If you flash a BIOS under PCS guidance and it fails, your warranty will remain fully intact. It's only if you flash a BIOS without PCS prior agreement that your warranty would become invalid.
Mustafa's response below:
"Hi,
Thank you for waiting, sorry about the delay, @demon28 was updated;
Just to confirm, if a customer updates a system BIOS based on our advice and following instructions provided by us and the update fails, this will not void the customers warranty. The warranty will only become void if the customer willingly updates their system BIOS without authorization from us and this contributes to a system failure."
If you flash a BIOS under PCS guidance and it fails, your warranty will remain fully intact. It's only if you flash a BIOS without PCS prior agreement that your warranty would become invalid.
Mustafa's response below:
"Hi,
Thank you for waiting, sorry about the delay, @demon28 was updated;
Just to confirm, if a customer updates a system BIOS based on our advice and following instructions provided by us and the update fails, this will not void the customers warranty. The warranty will only become void if the customer willingly updates their system BIOS without authorization from us and this contributes to a system failure."
MathewEvan449
Active member
I can confirm I had to update the BIOS. and yes the thermals are better. Id still recommend slightly under-clocking if you push the system to hard it can still go into heat protection mode and turn off. But frankly the issues i did have are now gone.
SpyderTracks
We love you Ukraine
If you're having to underclock or undervolt then there's a problem that needs addressing. The chassis works absolutely fine under full throttle so long as it's properly pasted and heatsink is fitted properly.
If you're getting any kind of shutdowns or high temps then something is wrong.
May I ask, which CPU do you have?
Yes it reduced temperatures but not enough on my system. Combined loads still cause thermal shut down.
My particular 3900 has a CCD that draws 20% more power than the other which is causing the issues for me. So during CPU only load one will max out at 80-81 whilst the other will 96-97 . ( So for example if ccd0 draws 6.5 watts per core during cinebench r20 multicore ccd1 will draw 5.3w per core)
During combined loads the hot one reaches shut off temperatures
Currently waiting for Dana to get back to me about a CPU swap, AMD offered me an RMA after describing my issues so pcspecialist will presumably be able to return the CPU to their supplier or AMD so I'm hoping they're fine with it.
My particular 3900 has a CCD that draws 20% more power than the other which is causing the issues for me. So during CPU only load one will max out at 80-81 whilst the other will 96-97 . ( So for example if ccd0 draws 6.5 watts per core during cinebench r20 multicore ccd1 will draw 5.3w per core)
During combined loads the hot one reaches shut off temperatures
Currently waiting for Dana to get back to me about a CPU swap, AMD offered me an RMA after describing my issues so pcspecialist will presumably be able to return the CPU to their supplier or AMD so I'm hoping they're fine with it.
Last edited:
Are using a cooler? I wrote a review about my findings yesterday. I found out that the CPU fan on this laptop gets suffocated unless the laptop is significantly lifted up. I initially used two books to lift the back up by 5cm and my fan was still struggling to suck enough air. Now I am using an aluminium stand that lifts my laptop up by around 15cm and the temps differences are very significant. My cpu now idles at 48 degrees and doesn't exceed 80(when using manual OC, PBO will get me to 83-85 but sometimes with temp spikes). I know my CPU is much weaker and less power hungry than yours, there is a massive 60+ page thread over at notebook review about owners of this laptop with 3950x-3900x(and 3900) processors. If you haven't seen it already there are a lot of findings that could maybe help you. I can pm you the linkYes it reduced temperatures but not enough on my system. Combined loads still cause thermal shut down.
My particular 3900 has a CCD that draws 20% more power than the other which is causing the issues for me. So during CPU only load one will max out at 80-81 whilst the other will 96-97 . ( So for example if ccd0 draws 6.5 watts during cinebench r20 multicore ccd1 will draw 5.3)
During combined loads the hot one reaches shut off temperatures
Currently waiting for Dana to get back to me about a CPU swap, AMD offered me an RMA after describing my issues so pcspecialist will presumably be able to return the CPU to their supplier or AMD so I'm hoping they're fine with it.
Last edited:
I use a laptop cooling pad that also lifts the laptop a fair amount.
My ccd1 temperatures are more than fine it's this powedraw discrepancies that causes the issue.
I'd be 10c below thermal cut off otherwise which is toasty but acceptable given the transistor density of this CPU.
Haven spoken with other people with the same CPU their reported power usage per CCD is equal or within margin of error.
I've also used every power option in Windows, CCC etc and ccd0 always draws 20% more power under full load.
My ccd1 temperatures are more than fine it's this powedraw discrepancies that causes the issue.
I'd be 10c below thermal cut off otherwise which is toasty but acceptable given the transistor density of this CPU.
Haven spoken with other people with the same CPU their reported power usage per CCD is equal or within margin of error.
I've also used every power option in Windows, CCC etc and ccd0 always draws 20% more power under full load.
Its possible that you got really unlucky with your CPU silicon. I could be wrong about this but I read somewhere that to get lucky with the higher end Ryzen 3900+ CPUs you need to have two decent CCD to get lucky with the silicon. Have you thought about manually OCing each CCD through Ryzen master? Run CCD0 a tiny but underlocked and CCD1 possibly a tiny bit overclocked? I don't know if its a great idea but I had heard a few tech people suggesting to manual OC each individual core of the Ryzen CPU's since each core is different. Hopefully a replasement CPU will fix all of these