Official Clevo NH55JNNQ / NH5xJNxx 12th gen Socket 1700 LGA laptop thread

[win32asmguy]

5 hours ago, jaybee83 said:

this is weird. could be that the current bios only allows a parallel adjustment of core and cache voltage in unison? 

 

@electrosoftany input u can bring to the table with your binned 12900k simulated at 125W would be sweet 😁

Yeah I think its a bios bug. To clarify further if I set Core offset it does show the value as applied (and persists across warm reboots) but when running a benchmark there is no change in observed voltage, load core ratios, or score. If I set even -1mV to P Cache, it crashes, reboots and now shows lower voltage under load, higher ratios and higher score.

 

Oh, and just to rule out software I tried uninstalling XTU, and also tried applying the undervolt via the QuickCPU utility where it also crashed only applying P Cache offset.

[jaybee83]

hey guys, so seems like i was able to lure in the big wiz to crack his magic knuckles and get to work on the NH55 😁

plan is to start working together on it once my gf's system arrives sometime this week and we've assembled all the parts (including some pimping of the cooling system). let's see what we can coax out of this nice little machine!

only thing to consider here is that im not sure how "openly available" this will be to other users, but still, it'll be a nice showing of what the system in general could be capable of 🙂 will keep you guys updated!

[electrosoft]

44 minutes ago, jaybee83 said:

hey guys, so seems like i was able to lure in the big wiz to crack his magic knuckles and get to work on the NH55 😁

plan is to start working together on it once my gf's system arrives sometime this week and we've assembled all the parts (including some pimping of the cooling system). let's see what we can coax out of this nice little machine!

only thing to consider here is that im not sure how "openly available" this will be to other users, but still, it'll be a nice showing of what the system in general could be capable of 🙂 will keep you guys updated!

 

I have no plans on getting one till RPL where hopefully I'll just pick up a used/discounted unit and slide my 12900k right on down and then build out either an RPL or AMD 7000. Hopefully in the next 5-7 months a lot of potential and issues will be worked out and addressed to really extract as much power from this lil dynamo as possible. It is already beating 12900hk systems so we're already up.

 

@win32asmguy how are the fans when just using it normally? I know my X170SM-G is basically nearly silent and only comes to life when gaming or stressing.

 

 

[win32asmguy]

2 hours ago, electrosoft said:

I have no plans on getting one till RPL where hopefully I'll just pick up a used/discounted unit and slide my 12900k right on down and then build out either an RPL or AMD 7000. Hopefully in the next 5-7 months a lot of potential and issues will be worked out and addressed to really extract as much power from this lil dynamo as possible. It is already beating 12900hk systems so we're already up.

 

@win32asmguy how are the fans when just using it normally? I know my X170SM-G is basically nearly silent and only comes to life when gaming or stressing.

The fans are quieter than the X170KM. They do not "stair step" as they are ramping up and down. The CPU seems to always be running even with package temp under 40C at idle. I think the fans may be of higher quality too so pushing more CFM at lower noise. Undervolting helps even more but not really something viable for day to day use unless those e-cores are disabled. Also so far I do not think any of the Linux undervolting utilities have added Alder Lake support.

 

On ongoing issue I have not mentioned so far is that in Windows 10 the internal speakers are not so great. There is a supplemental package related to Sound Blaster Cinema 6 which won't install on Win10.

 

Any success with the H660 motherboard? I thought maybe about picking up a different 12900K from Micro Center next week to see if its any better than the one from Best Buy.

[electrosoft]

54 minutes ago, win32asmguy said:

The fans are quieter than the X170KM. They do not "stair step" as they are ramping up and down. The CPU seems to always be running even with package temp under 40C at idle. I think the fans may be of higher quality too so pushing more CFM at lower noise. Undervolting helps even more but not really something viable for day to day use unless those e-cores are disabled. Also so far I do not think any of the Linux undervolting utilities have added Alder Lake support.

 

On ongoing issue I have not mentioned so far is that in Windows 10 the internal speakers are not so great. There is a supplemental package related to Sound Blaster Cinema 6 which won't install on Win10.

 

Any success with the H660 motherboard? I thought maybe about picking up a different 12900K from Micro Center next week to see if its any better than the one from Best Buy.

 

 

I was wondering how the fans worked/sounded under normal use. The poor X170KM had so much against it from super hot 11900k chips to a less than optimized cooling system to a locked down BIOS and more. I said it in the other forums, but going from a Prema enabled, hardware modded X170SM-G to the KM was an exercise in frustration getting the 11900k to run full tilt which I did get going with no throttling but relying on CCC was nails on the chalkboard.

 

The 12900k locked in at PL 125w should/could produce a CB23 score of a low of ~18810 up to ~23000 if thermal constraints aren't an issue and depending on sample. This is from a review of budget boards from HWU and you can see the range of scores achieved (VRM temps pictured) after ~1hr CB23 so you are definitely in the ballpark for PL1 limits:

 

 

 

The board I have is the Asrock B660 Pro RS which definitely enforces 125w limits.

The air cooler installed is  this model with a 180w TDP. Right now it is keeping a 12100f(65w) ice cold: http://www.idcooling.com/Product/detail/id/276/name/SE-214-XT ARGB

I'm trying to get a window of time to swap it over and test. I had the 12100f system up for sale but I took it down just so I could run these tests before putting it back up

 

The only problem is reproducing any potential thermal constraints since at PL 125w you are seeing some thermal issues. If you do pick up another (unless they're near identical silicon) you will immediately see a potential difference using stock and then matching uV steppings.

 

 

 

 

 

[win32asmguy]

53 minutes ago, electrosoft said:

The only problem is reproducing any potential thermal constraints since at PL 125w you are seeing some thermal issues. If you do pick up another (unless they're near identical silicon) you will immediately see a potential difference using stock and then matching uV steppings.

I could also pick up a motherboard from Micro Center to try and better test these CPU's. I have all of the other parts for such a test bench although the NH-D15S may still be a limiter.

 

Tomorrow BGA Alder Lake HX CPU's should be announced as well, so we may have a couple tech influencer reviews showing how performance compares. I keep meaning to run Geekbench to see how it compared to a leaked Legion 7i with 12900hx. Although with the prices for 12900hk I would not be surprised to see most of these potential models to be over $3500 even with 3070ti.

[electrosoft]

1 hour ago, win32asmguy said:

I could also pick up a motherboard from Micro Center to try and better test these CPU's. I have all of the other parts for such a test bench although the NH-D15S may still be a limiter.

 

Tomorrow BGA Alder Lake HX CPU's should be announced as well, so we may have a couple tech influencer reviews showing how performance compares. I keep meaning to run Geekbench to see how it compared to a leaked Legion 7i with 12900hx. Although with the prices for 12900hk I would not be surprised to see most of these potential models to be over $3500 even with 3070ti.

 

For your thermal limits (PL125 / PL150) a NH-D15S is more than sufficient. I just finished swapping everything over to my mentioned hardware above to test and the 180w  single fan aluminum air cooler I listed above was more than sufficient for testing and never broke ~65c.

 

As for testing, locking in a PL of 125w and everything on auto no uV or LL calibrations resulted in a CB23 score of 24056 and a max temp of 65c which beats HWU 12900k same motherboard and test by ~1000 points as it is giving a little more punch at the same power threshold.

 

Playing a bit with LLC, since PL is locked you don't get a dynamic trade off and greater Vdroop = lower scores.

 

So it looks like at least a somewhat optimal score for a PL125 lock would be ~24k which is only ~3k off of a normal 12900k running stock. The cores were clocking in around 4.3-4.4 on average running @ PL125. 

 

 

 

[jaybee83]

8 hours ago, win32asmguy said:

The fans are quieter than the X170KM. They do not "stair step" as they are ramping up and down. The CPU seems to always be running even with package temp under 40C at idle. I think the fans may be of higher quality too so pushing more CFM at lower noise. Undervolting helps even more but not really something viable for day to day use unless those e-cores are disabled. Also so far I do not think any of the Linux undervolting utilities have added Alder Lake support.

 

On ongoing issue I have not mentioned so far is that in Windows 10 the internal speakers are not so great. There is a supplemental package related to Sound Blaster Cinema 6 which won't install on Win10.

 

Any success with the H660 motherboard? I thought maybe about picking up a different 12900K from Micro Center next week to see if its any better than the one from Best Buy.

thats one advantage of the 12600 since its missing the e cores, so UV should be straightforward. unless of course it is blocked just like with the 12900 (ugh)

thats not good news with the sound driver being w11 exclusive... but clevo has always been a bit finicky when it comes to sound enhancement package installs....

fans seem to be great tho 🙂 cant wait!

 

@electrosoft this was the cherrypicked 12900k right? so now we have an "ideal reference" to orientate ourselves toward. roughly 90% of ideal stock on a desktop sounds pretty sweet to me 😁

[electrosoft]

8 hours ago, jaybee83 said:

thats one advantage of the 12600 since its missing the e cores, so UV should be straightforward. unless of course it is blocked just like with the 12900 (ugh)

thats not good news with the sound driver being w11 exclusive... but clevo has always been a bit finicky when it comes to sound enhancement package installs....

fans seem to be great tho 🙂 cant wait!

 

@electrosoft this was the cherrypicked 12900k right? so now we have an "ideal reference" to orientate ourselves toward. roughly 90% of ideal stock on a desktop sounds pretty sweet to me 😁

 

I suspect the 12600 will run into the exact same issues as the 12900 as a non-k bin

I would not be happy if Clevo isn't giving full W10 support but there might be a way to get it to work hopefully.

 

Yeah, I binned multiple chips. 1x Amazon, 1x Newegg, 3x Best Buy, 1x an Alienware OEM. All binned and tested in my Strix Z690 D4. The Newegg chip was the best of the bunch. It ran well enough that I even sold off an SP94 12900ks at cost to a buddy as on the top end for day to day it only ran 1 bin higher (5.2 vs 5.1). Both could do 5.3 all core (KS did it at ~30w less though) and both turned into uncontrollable monsters at 5.4 pulling 330w+ which overpowered my AIO.

 

This SP91 from Newegg was the clear winner hands down. Setting even the 12900ks to stock 12900k clocks, it couldn't touch this SP91 and required 175w when dialed in vs the 155-162w on this SP91 12900k. I  think I even mentioned it in the Desktop thread on the old forums when I was binning this one the first thing that popped into my head was, "Damn I wish Clevo was making a 12th gen laptop. This chip is golden for that form factor."

 

And I'll take 90% desktop performance in that form factor all day. In an X170 / P870 style chassis this chip would blaze and run full tilt with the 200w PL and heatsink in those monsters.

 

 

 

 

[win32asmguy]

40 minutes ago, electrosoft said:

 

I suspect the 12600 will run into the exact same issues as the 12900 as a non-k bin

I would not be happy if Clevo isn't giving full W10 support but there might be a way to get it to work hopefully.

 

Yeah, I binned multiple chips. 1x Amazon, 1x Newegg, 3x Best Buy, 1x an Alienware OEM. All binned and tested in my Strix Z690 D4. The Newegg chip was the best of the bunch. It ran well enough that I even sold off an SP94 12900ks at cost to a buddy as on the top end for day to day it only ran 1 bin higher (5.2 vs 5.1). Both could do 5.3 all core (KS did it at ~30w less though) and both turned into uncontrollable monsters at 5.4 pulling 330w+ which overpowered my AIO.

 

This SP91 from Newegg was the clear winner hands down. Setting even the 12900ks to stock 12900k clocks, it couldn't touch this SP91 and required 175w when dialed in vs the 155-162w on this SP91 12900k. I  think I even mentioned it in the Desktop thread on the old forums when I was binning this one the first thing that popped into my head was, "Damn I wish Clevo was making a 12th gen laptop. This chip is golden for that form factor."

 

And I'll take 90% desktop performance in that form factor all day. In an X170 / P870 style chassis this chip would blaze and run full tilt with the 200w PL and heatsink in those monsters.

 

 

 

 

Thanks for all the testing and insight.

 

Sounds like it will probably take going through many chips to find a good one. It's too bad the 12900ks does not help the chances, I would happily pay $200 more for one if it could be a good performer at 125W. I am also worried that a higher percentage of good 12900k chips are destined to be sold as 12900ks now so Intel can profit more from them...

[jaybee83]

yeah thats the tricky part with the binning though: you can bin for different things. all the silicon going towards the 12900KS is only the best at reaching high clocks stable without blowing up. doesnt mean that its also frugal at lower clocks and voltages in terms of wattage. so i wouldnt be too sad about "good silicon" for our specific laptop purposes are all heading towards the KS sku 🙂 

[win32asmguy]

2 hours ago, jaybee83 said:

yeah thats the tricky part with the binning though: you can bin for different things. all the silicon going towards the 12900KS is only the best at reaching high clocks stable without blowing up. doesnt mean that its also frugal at lower clocks and voltages in terms of wattage. so i wouldnt be too sad about "good silicon" for our specific laptop purposes are all heading towards the KS sku 🙂 

Yep. My other thought was just to send my locked 12900 over to rockitcool for a copper relid for $70, then just post a perpetual WTB binned 12900k on some forums in hopes that someone might want to move to a KS or in preparation for RPL.

[electrosoft]

48 minutes ago, win32asmguy said:

Yep. My other thought was just to send my locked 12900 over to rockitcool for a copper relid for $70, then just post a perpetual WTB binned 12900k on some forums in hopes that someone might want to move to a KS or in preparation for RPL.

 

The problem is your binning criteria is going to be different than a traditional binning process so you will need to ask potential sellers if they can at least do a pure auto run at stock or if it is a 12900ks seller adjust their core clocks down to match a 12900k. Maybe see if they are willing to lock in long and short 125w PLs which will automatically lock in and limit both 12900k and 12900ks chips and give them a CB23 run to get an idea of their scores at that 125w PL. Thermal characteristics vary by cooling mechanisms but anyone who has a 12900k on their desktop will be able to cool it easily at PL125w taking heat out of the equation.

 

You are looking for a 12900k/12900ks (At PL125w they are basically equal in regards to this testing criteria) that runs exceptionally low pull and less leaky in a specific PL125 window. This means normal binning procedures may not apply.  Like the 12900ks I had which was an SP94 and was definitely better than my SP91 12900k >=5.3 but couldn't touch it when you were looking for a low pull / less heat. No matter what I did ~175w was the lowest it would go at 4.9.

 

Leaky chips also come into play as some will throw out more heat than others at that PL125w window. To put my 12900k in perspective. That same 180w cooler when running CB23 with a 12100f heated up to ~60c pulling 65w yet my 12900k only heated up to 65c pulling 125w with 2x the P cores and 8 e cores but it is well known 12100f's are garbage bins because they can be because they're  4/8 P-cores only and even bad bins can be cooled by ok air coolers.

 

Luckily, you already have the laptop which is the ultimate test bed. Pick up a few more 12900k's and pop them in and start differentiating one 12900k from the others and documenting the differences. Start with pure stock and jump to a -0.050uV and continue down in -0.020uV intervals. When it finally fails/BSODs back off to the previous interval and have it pass multiple runs of CB23 and other stressors to lock it in.

 

If you're going to set up a desktop test bed I'd suggest at least an Asus Z690-A Prime or Strix D4 at the min so you can at least get the SP rating to go along with your testing. You can further simulate a limited environment by going with a cooler like this which is rated right around 125w (ish) or similar:

 

https://www.amazon.com/Noctua-NH-L9x65-Premium-Low-Profile-Cooler/dp/B00VB3Y89E/ref=asc_df_B00VB3Y89E/?tag=hyprod-20&linkCode=df0&hvadid=309776868400&hvpos=&hvnetw=g&hvrand=1157509063284803367&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9003918&hvtargid=pla-386303669508&psc=1

 

 

[win32asmguy]

4 hours ago, electrosoft said:

 

The problem is your binning criteria is going to be different than a traditional binning process so you will need to ask potential sellers if they can at least do a pure auto run at stock or if it is a 12900ks seller adjust their core clocks down to match a 12900k. Maybe see if they are willing to lock in long and short 125w PLs which will automatically lock in and limit both 12900k and 12900ks chips and give them a CB23 run to get an idea of their scores at that 125w PL. Thermal characteristics vary by cooling mechanisms but anyone who has a 12900k on their desktop will be able to cool it easily at PL125w taking heat out of the equation.

 

You are looking for a 12900k/12900ks (At PL125w they are basically equal in regards to this testing criteria) that runs exceptionally low pull and less leaky in a specific PL125 window. This means normal binning procedures may not apply.  Like the 12900ks I had which was an SP94 and was definitely better than my SP91 12900k >=5.3 but couldn't touch it when you were looking for a low pull / less heat. No matter what I did ~175w was the lowest it would go at 4.9.

 

Leaky chips also come into play as some will throw out more heat than others at that PL125w window. To put my 12900k in perspective. That same 180w cooler when running CB23 with a 12100f heated up to ~60c pulling 65w yet my 12900k only heated up to 65c pulling 125w with 2x the P cores and 8 e cores but it is well known 12100f's are garbage bins because they can be because they're  4/8 P-cores only and even bad bins can be cooled by ok air coolers.

 

Luckily, you already have the laptop which is the ultimate test bed. Pick up a few more 12900k's and pop them in and start differentiating one 12900k from the others and documenting the differences. Start with pure stock and jump to a -0.050uV and continue down in -0.020uV intervals. When it finally fails/BSODs back off to the previous interval and have it pass multiple runs of CB23 and other stressors to lock it in.

 

If you're going to set up a desktop test bed I'd suggest at least an Asus Z690-A Prime or Strix D4 at the min so you can at least get the SP rating to go along with your testing. You can further simulate a limited environment by going with a cooler like this which is rated right around 125w (ish) or similar:

 

https://www.amazon.com/Noctua-NH-L9x65-Premium-Low-Profile-Cooler/dp/B00VB3Y89E/ref=asc_df_B00VB3Y89E/?tag=hyprod-20&linkCode=df0&hvadid=309776868400&hvpos=&hvnetw=g&hvrand=1157509063284803367&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9003918&hvtargid=pla-386303669508&psc=1

 

 

Yeah, I will try a 12900k from Micro Center next week and go from there. Still undecided on the Z690 board as I need to figure out what I would want to do with it after the testing.

[jaybee83]

got the machine in today! very sleek, got a nice feel to it. and amazing to see how much smaller the footprint is, including the 230W brick which is also way smaller than the 230W brick i originally got with my P751DM. good to see that psu tech also made some progression!

next steps: compile driver package, secure erase ssd via linux, prep windows install usb, build system with gf, then install everything and get to tuning 😁

[electrosoft]

22 hours ago, jaybee83 said:

got the machine in today! very sleek, got a nice feel to it. and amazing to see how much smaller the footprint is, including the 230W brick which is also way smaller than the 230W brick i originally got with my P751DM. good to see that psu tech also made some progression!

next steps: compile driver package, secure erase ssd via linux, prep windows install usb, build system with gf, then install everything and get to tuning 😁

 

Awesome! Glad to see it made it safe and sound. Make sure to post some pics and results bro. 💪

 

 

[jaybee83]

57 minutes ago, electrosoft said:

 

Awesome! Glad to see it made it safe and sound. Make sure to post some pics and results bro. 💪

 

 

 

for sure, u got it 🤠

[GT3000]

Still waiting on the QHD screen from Eurocom. Not their fault mind you, COVID is holding everything up. Any day now...

 

[jaybee83]

1 hour ago, GT3000 said:

Still waiting on the QHD screen from Eurocom. Not their fault mind you, COVID is holding everything up. Any day now...

 

haha i just decided its not worth the wait. if the need / want arises down the line, ill just get the screen from aliexpress / ebay and do the upgrade myself 😉

 

@win32asmguy hey man, which version of control center do u have? i found 3.70 as the latest, u got anything newer? currently driver hunting here... also does your machine have a fingerprint sensor?

[win32asmguy]

2 hours ago, jaybee83 said:

haha i just decided its not worth the wait. if the need / want arises down the line, ill just get the screen from aliexpress / ebay and do the upgrade myself 😉

 

@win32asmguy hey man, which version of control center do u have? i found 3.70 as the latest, u got anything newer? currently driver hunting here... also does your machine have a fingerprint sensor?

Yeah just 3.70. Did you find an issue with it? I only really used it to set the keyboard backlighting then uninstalled. I did notice that the profile switcher always thought the system was in entertainment mode.

 

No fingerprint sensor.

 

I took mine with me on a flight with two connections yesterday. Way easier to travel with than the X170 and get through security. You can't really even tell it has an LGA CPU.

[jaybee83]

23 hours ago, win32asmguy said:

Yeah just 3.70. Did you find an issue with it? I only really used it to set the keyboard backlighting then uninstalled. I did notice that the profile switcher always thought the system was in entertainment mode.

 

No fingerprint sensor.

 

I took mine with me on a flight with two connections yesterday. Way easier to travel with than the X170 and get through security. You can't really even tell it has an LGA CPU.

nah no issue, just a habit of mine to hunt down the most updated driver packages for a fresh install 🙂 thx for confirming.

yeah ikr @cant tell it sports desktop socket, its significantly smaller and lighter than my P751DM, pretty amazing!

 

ok so driver hunting is done and i also built the system together with my lady, just finished repadding the whole thing, might as well do it while its disassembled (we all know how crappy stock thermal pads are, both in terms of thermal conductivity and also placement!)

i used fujipoly extreme, arctic and phobya extreme pads for replacement. thicknesses required were 1, 2 and 4 mm.

 

also took some pics and will detail the position and thickness of all the thermal pads, might help other users here if they decide to do some performance maintenance 🙂

 

one weird thing i encountered: i took two of my ddr4-3800 sticks out of my machine to give them to my gf (thus 16GB each), but now with two sticks my machine doesnt wanna boot the memory profile anymore lol. in theory, two sticks should actually run better than four, but my system only posts the custom memory profile with all four sticks, at 2 it shows blackscreen 😄 will need to figure out a new profile i guess...

 

but now time to hit the sack, cheers! 

[jaybee83]

aight, i figured before i forget all the thicknesses lets do this real quick 😛

here we go (numbers indicate mm thickness):

 

1 and 4 mm are "native" pads, whereas for 2mm i just stacked 1mm pads, works just fine 🙂 blues are arctic, light grey is fujipoly extreme 17W/mK and dark grey is Phobya Ultra. in case youre wondering, i prioritized the leftover fujipoly extreme pads for the gpu vRAM and then switched to arctic (performance wise, arctic lies between fujipoly 11 and 14 W/mK) once i ran out 😄

 

here some additional pics, first up the overview after disassembly and cleaning:

unfortunately, tri-pod assembly for the gpu cold plate 🙄 but the cooling system itself looks quite robust and capable, looking forward to stress testing it!

 

here a close up of GPU, CPU and power delivery:

shiny! 😁

 

lastly, the stock arrangement of thermal pads, cheap-ass stuff which wasnt even properly covering all compoents its supposed to cool:

 

rule of thumb for proper thermal pad application: DO NOT cut and place the thermal pads based on the cooling plate but rather on the ACTUAL COMPONENTS being cooled! most times the cold plate is a bit shorter than the component and thus doesnt cover it whole, so best to at least cut and place the pads in such a way that they cover the whole component. better than parts of them being left "naked" and not being heatsinked at all.

example: look at the image directly above, notice how the thermal pad on the lower right corner (behind the stamped date) is only covering "one and a half" vRAM modules? yep, so basically half of one vRAM module was left completely naked with that setup. compare that to the equivalent vRAM chips shown on the first picture after repadding, theyre all properly and completely covered.

 

aight, nuff said for now, off to dreamland!

 

cheers guys

[electrosoft]

18 hours ago, jaybee83 said:

aight, i figured before i forget all the thicknesses lets do this real quick 😛

here we go (numbers indicate mm thickness):

 

1 and 4 mm are "native" pads, whereas for 2mm i just stacked 1mm pads, works just fine 🙂 blues are arctic, light grey is fujipoly extreme 17W/mK and dark grey is Phobya Ultra. in case youre wondering, i prioritized the leftover fujipoly extreme pads for the gpu vRAM and then switched to arctic (performance wise, arctic lies between fujipoly 11 and 14 W/mK) once i ran out 😄

 

here some additional pics, first up the overview after disassembly and cleaning:

unfortunately, tri-pod assembly for the gpu cold plate 🙄 but the cooling system itself looks quite robust and capable, looking forward to stress testing it!

 

here a close up of GPU, CPU and power delivery:

shiny! 😁

 

lastly, the stock arrangement of thermal pads, cheap-ass stuff which wasnt even properly covering all compoents its supposed to cool:

 

rule of thumb for proper thermal pad application: DO NOT cut and place the thermal pads based on the cooling plate but rather on the ACTUAL COMPONENTS being cooled! most times the cold plate is a bit shorter than the component and thus doesnt cover it whole, so best to at least cut and place the pads in such a way that they cover the whole component. better than parts of them being left "naked" and not being heatsinked at all.

example: look at the image directly above, notice how the thermal pad on the lower right corner (behind the stamped date) is only covering "one and a half" vRAM modules? yep, so basically half of one vRAM module was left completely naked with that setup. compare that to the equivalent vRAM chips shown on the first picture after repadding, theyre all properly and completely covered.

 

aight, nuff said for now, off to dreamland!

 

cheers guys

 

Absolutely fantastic post bro! Definitely some classic Clevo inconsistencies as always. Looking forward to the thermal testing and scores both CPU and GPU.

 

 

[jaybee83]

3 hours ago, electrosoft said:

 

Absolutely fantastic post bro! Definitely some classic Clevo inconsistencies as always. Looking forward to the thermal testing and scores both CPU and GPU.

 

 

thanks man, appreciate the motivational booster, makes it fun 🙂 next step is to get my system stable with two RAM sticks first, since ill rely on it to reset the SSD and do troubleshooting whenever required. after that ill get to installing windows / driver / programs / games on the new machine and do baseline benches and thermal testing. THEN i can ping brother Prema and we can get this show on the road 🤠

[jaybee83]

alrighty, took me a good day and a half to sort the stuff with my machine not properly booting with only two ram sticks installed. i finally decided to do a cmos pull, at which point i was greeted with a black screen. weird enough, external screen DID actually show an output. in the end it turned out that either cpu or gpu werent properly reinstalled after i had done some maintenance and repasting, so did it all over again and now everything's back to normal.

 

aaaaaaaaaanyhow, back to topic: making good progress in setting up my lady's machine, done with windows, drivers and most of the programs & games. what ive noticed though: holy guacamole 12th gen runs HOT! 😱 like, wtf? the 12600 shes got installed is at PL1 90W and PL2 150W in perf. mode (full stock for now, no tuning yet via TS), and easily hits 90C+ with thermal throttling under load with clocks around 4.2-4.4 ghz. really wondering if i messed up the repaste with kryonaut or if theres an issue with cold plate contact, hmmm....lets finish the setup first, then ill check what can be done. in any case, the fans get LOUD once the cpu hits full bore, my lady aint gonna like that 😅