2024/02/20 21:15:21
MhzManiac
Congrats on making some progress!
2024/02/23 14:24:13
dmann304
This is very interesting, so: clear UEFI in gui? or hard clear UEFI on button?  I heard this has some interesting effects.
I seem to be ok with mine, with a pair of cores at 48x, 6 of them at 46x, and rest at 45x, with 1.275 on 48, 1.2 on 46
and 1.175 on 45x, and seems balanced and over 10k in R23. Could do better, but seems happy now
2024/02/23 16:10:54
ZoranC
polykoma1I ordered some more kryonaut to do a repaste to safely exclude this aswell per previous suggestion.

FWIW, I've had very bad experiences with Kryonaut. To the point where I believe it is more of a hype than anything. Best experiences I had were with Prolymatech PK-3. Second best with Noctua NT-H2.
2024/02/25 07:54:51
polykoma1
Hey everyone! Thanks for all the additional Input.
 
I repasted my cpu (honestly i was trying X shape the last couple times since X shape is the new spread? anyway i had overall more bad than good results with it at this point)
 
I switched back to spreading and my CPU instantly ran 10C cooler. Also I resetted the CMOS once more after it.
I then proceeded with OC Robot 3.0 and targeted 65 Celsius and did NOT had a crash since thursday.
I will make an update down the road.
 
So the only changes basically I made are 1. Repasting 2. Resetting CMOS twice 3. HT enabled
 
Big thank you again to point me towards a repaste!
(Funnily my AiO screws had loosened 3 full twists after cranking it hand tight with my screwdriver after a week and plenty bluescreens - which seemed kinda odd but I kinda think the x spread is just **** at this point. I had experienced this before but not to that extent. The H150i likes to unscrew itself when you dont do it very tight out of my experiences with it over the years)
 
 
BTW  I was repairing/replacing my fans on my 2080 Super (which had a fan controller issue since day1 but my dumbarse didnt RMAd it) with EXACT identical fans according to brand and Item code. Only difference is its using another type bearing.
Anyway after installing them my RPM readings where across the board so I had to basically switch them back to the OG fans.
Here some gifs of the issue https:// giphy.com/channel/polykoma/evga
 
After reverting to my OG fans my RPM readings were normal again.
The issue I had with the card is, when fans start spinning they seem to get excessive voltage, they make a rrrtschhh  noise when initially starting. I dodged it by just letting them spin always.
The middle fan lately started making rattling noises (but ony sometimes - not all the time) so I ordered replacements in first place.
 
Funnily after reverting to OG fans the  starting noise occurs much more rarely, and is a lil softer too. Also it didnt started to rattle again (yet).
But now I gotta return those replacement fans.
 
Has there been any concensus on the faulty fan controller issues? I heard you can reflash the gpu bios?
Maybe you can swap an IC or whatever on the GPU PCB?
 
I dont think asking for a 2080Super repair at EVGAs side makes monetarily sense at this point as its only inconveniences which can be worked around and the card still does its job.
 
 
 
 
LONG STORY SHORT.
Bought a used
EVGA 3090 FTW3
at a bargain since the 2080 issue kinda pissed me off
 
:D
2024/02/27 13:58:25
dmann304
Mine, seems good now.  These chiips are really sensitive to tunning.  You really need
lots of time to find the sweet spot.  2 weeks of core to core testing and with hwinfo, watching
'each core, and temp, and adjusting vcore for each, and balance of cores, and just noticed, that
in hwinfo64, there starts near 4 different cores, i assume their the world threads, or which would
be the ones that matter most. staying under tjmax, even with push pull 420 on artic 4 paste, is
still a challenge, but 27,101 in R23 is ok for me.  
2024/02/27 15:48:16
ZoranC
dmann304These chips are really sensitive to tunning. You really need lots of time to find the sweet spot. 2 weeks of core to core testing and ... 27,101 in R23 is ok for me.

What was the score you were getting before tuning, with multipliers set to fused in ones and voltages left on auto, please?
 
2024/02/27 16:57:43
dmann304
I believe i was getting in R23, around 20k, and R20 single around 455. 
Seems when it was fused with auto, it would not go past 3800, or turbo to 48x, even 
with settings to auto on cstate, and turboboost 3 and eist on. but with mixed cores 45,46,48
I am able to get most in what i mentioned in previous post.
2024/02/27 17:21:50
ZoranC
dmann304I believe i was getting in R23 around 20k

Thank you!
2024/02/28 20:34:49
polykoma1
Im getting 23200 in R23 right now Multi Core.
 
I used OC Robot 3.0 Target 65 Celsius So my cores run 41x-44x
 
Temps are like 40C in idle. Pretty nice now :)
2024/02/28 21:08:17
polykoma1
dmann304
Mine, seems good now.  These chiips are really sensitive to tunning.  You really need
lots of time to find the sweet spot.  2 weeks of core to core testing and with hwinfo, watching
'each core, and temp, and adjusting vcore for each, and balance of cores, and just noticed, that
in hwinfo64, there starts near 4 different cores, i assume their the world threads, or which would
be the ones that matter most. staying under tjmax, even with push pull 420 on artic 4 paste, is
still a challenge, but 27,101 in R23 is ok for me.  


would love to see a video of core temp tool when you do the r23 run to compare it to my 360mm push pull kryonaut.
i would post mine too. :>

Use My Existing Forum Account

Use My Social Media Account