stryker7314

_Gir_
My rough measurements of the pads as I found them on the card was 2.5mm for everything except the piece on the backside of the GPU which was 2.0mm.

That's unusually thicc, anyway you can measure it?

Probably best to get the info from EVGA CS.  Since mine were already used and compressed, the edge might have flared out.
 

 
 

stryker7314

_Gir_

stryker7314

_Gir_
My rough measurements of the pads as I found them on the card was 2.5mm for everything except the piece on the backside of the GPU which was 2.0mm.

That's unusually thicc, anyway you can measure it?

Probably best to get the info from EVGA CS.  Since mine were already used and compressed, the edge might have flared out.
 

 
 

You're awesome, thank you so much!!!

No problemo.  Just saw your post in different forum--if you're using fujis which I've never used but heard are quite firm.  You may want to use thinner for the memory on the front next to the die.  Probably ~ 2mm?  There's no harm in taking yours apart now and evaluate what you want to do before putting down a chunk of money for expensive pads.  Not sure if you saw my edit from earlier but the stock evga pads are very soft and "give" easily.

Well it's impossible to find the 2mm 17 W/mk thermal pads so may just leave the stock stuff since I have seen posts that have decent temps with it.  Just going to crack it open to use liquid metal for the gpu and also liquid tape the dealios (capacitors?) around the chip for protection, not that I have ever seen liquid metal run but not gonna leave them exposed.
 
Also makes me wonder what EVGA uses for their stock thermal pads???

xxDarkxx
https://www.amazon.com/Thermal-Grizzly-Minus-High-Performance/dp/B00ZJS5M60/ref=sr_1_6?dchild=1&keywords=Thermal+Pad+2.5mm&qid=1612536025&s=electronics&sr=1-6
 
these pads are good but you will need to wait for restock for price to normalize
not the greatest thermal conductivity though. The fujypoly will cost too much at that thickness your better off waiting for the optimus block

Willing to pay but they aren't for sale anywhere even though they do exist, messaged fujypoly directly and also performancepcs to see if they can aquire them and sell them, let them know there's a market for them in modifying the nvidia 3090 series so there's incentive.
 
These seem like a higher performing alternative option at 12 W/mk, hope it's accurate, but in my experience Gelid is solid. : https://www.amazon.com/gp...W9QXMLY5QE3F&psc=1

Anyone ever have artifacts after long benchmarking runs back to back stress testing when power targets are max in X1? (Stock clocks) Seems like only port royal artifacts. I’m wondering if this is thermal related since it only happens during heatsoak. Are the power controllers poorly cooled? Or not at all? Temps 60's for vrm/mem/core during. PSU 850 G3. 5950x, strix x570-e.

I’m not at my computer right now but what I can tell you is when the doors open for the ship to come in the ray of sunlight beaming in seems to flicker. The computer rebooted after that.

I wonder if the memory on the backside of the card is getting too hot during repeated testing

alpharius194
I've definitely found these cards to be memory temp limited as far as stability goes, any time mine gets over 75C and I have an mem OC it will crash in cp2077 at some point. I dropped my daily memOC to +250 on stock voltage and it seems to never get above 70C. Try monitoring Mem2 temp, I use HWinfo64 so it saves the Max and average temps and this is what ive found to be my weak link. My question, though I don't think it's possible, but is there a way to get the classified tool to load a profile on start? I run my NV on auto for day to day use but my card acts up at some point if I have a memory overclock and forget to set FBV.

MEM1 is 71c and MEM2 is 68.
 
I just ran time spy extreme and port royal stress test back to back without issues. Then I ran a couple port royal bench runs and when the benchmark ended to the result screen the pc rebooted, no artifact noted prior. This was with stock clocks and power limit at max.

When I overloaded my 750w PSU by starting Cyberpunk at 450w on the gpu (5800x cpu), the psu just shut the machine off like I had held the power button down. 850w holds on at that level but I bet it doesn't have a lot of headroom. CP must cause a surge since I had successfully ran multiple 3dmark runs before that with no issue. The instant CP started in 3D mode the machine shut off.

talon951
When I overloaded my 750w PSU by starting Cyberpunk at 450w on the gpu (5800x cpu), the psu just shut the machine off like I had held the power button down. 850w holds on at that level but I bet it doesn't have a lot of headroom. CP must cause a surge since I had successfully ran multiple 3dmark runs before that with no issue. The instant CP started in 3D mode the machine shut off.

During the space between load screens completing & the game being displayed there are some pretty real spikes in the game, I've noticed them by "POWER" popping up on my MSI Afterburner OSD consistently during those times.  Also happens when going from menus or inventory back to the game on occasion.  Happened on both GPUs I've run the game on, so there's some definite spikeage going on, but with my PSUs I'm not gonna be seeing any shutdowns.
 

bmx045

Dabadger84
I don't even get artifacts that I've noticed while running Port Royal Stress Test or Time Spy Extreme Stress Test, on my OC/undervolt settings that I've tested to be "stable", definitely not at stock.  
What timestamps are the artifacts occurring at?  Are they consistent or do they change?  Is it flashing or triangles or what kind?

I just ran time spy extreme stress and pr stress back to back without problems. It seems to be very inconsistent. I'm wondering if an 850w PSU is enough for 5950x (PBO on, otherwise stock), kingpin w/ high power limits, 8 fans and cpu aio....

It does sound like heatsoak or power issues indeed then.  But the artifacts you've talked about could be overheating or slight instability.
I doubt the backside vRAM is getting too hot, even before I added heatsinks to my backplate & a fan, I was only seeing low 70s on the hottest iCX Mem temp at the most, and that was in game not in benchmarks, you'd need some really bad case flow to be overheating I think.  The backplate of the Kingpin may just be a mostly-solid metal "block" but it does a decent job of spreading that heat out from those thermal pads.
 

menko2
I still don't get it.

When I had my 3090 Strix and the Vbiod power limit was 480W, the card used around 480W aprox as expected. All the cards use until their max power in Vbios.

Here with the 3090 Kingpin I use the LN2 bios and it doesn't go higher than 465-470W unless I use the Classified Tool that i don't want to use as precaution-safety.

Is there any way I can use all the power that the LN2 has like with the Strix?

It's hard to push a 3090 to 520W+ without voltage bumps, keep in mind, the stock/reference/FE power spec for a 3090 is 350W.  So getting to 450-500W of draw is pushing things a fair bit already.  I haven't seen draw get to 520+W without a hefty OC plugged in or some voltage bumps.  Don't think I've seen more than around 480W on stock voltage with tickles up to 500W with pretty high clocks.
Running my 2160MHz @ 1068mV with +1000 vRAM clocks in Cyberpunk only gets up to the 450-480W range most of the time, and that's "stable" (still testing, but pretty sure it's actually stable because it's passed hours of gaming/testing) clocks that some people can't even run consistently.

menko2
It's so strange that in the same benchs and games the Strix pulled more watts than the Kingpin LN2 bios.

Even considering the temps with the Strix being at 75°C and the Kingpin at 50°C (fans at 80%).

I keep seeing the power limit in afterburner when the card is around 460-465W.
The Strix at around 480W.

Something is not quite there with this bios.

I agree. I also think EVGA should release a 600-650W bios that doesn't void warranty, I hit 520W power limit during some games (Quake II RTX comes to mind, among others).

bmx045
 
 
MEM1 is 71c and MEM2 is 68.
 
I just ran time spy extreme and port royal stress test back to back without issues. Then I ran a couple port royal bench runs and when the benchmark ended to the result screen the pc rebooted, no artifact noted prior. This was with stock clocks and power limit at max.

Try running other benchmarks and see if these inconsistencies hold with the stock clocks.  Than test with downclocking and if the GPU stabilizes, I would get a hold of EVGA CS.

menko2
It's so strange that in the same benchs and games the Strix pulled more watts than the Kingpin LN2 bios.

Even considering the temps with the Strix being at 75°C and the Kingpin at 50°C (fans at 80%).

I keep seeing the power limit in afterburner when the card is around 460-465W.
The Strix at around 480W.

Something is not quite there with this bios.

You can try enabling both NVVDD/MSSVVDD dipswitches and it should pull more wattage. I believe the 520W BIOS has some sort of limit when you reach 500W (maybe it reserves the extra 20W for the fans and pump). With the dipswitches on, it can easily pull 500-524W. 
 
Without dipswitches and card at stock with the OXC BIOS it can pull 640W in RTX Quake 2 so the 520W BIOS has some limitations.  

sparetimepc
I see they only have 1 kingpin listed on eBay right now, I wonder if they got bought at the outrageous prices or sellers decided the high prices were not getting them sold.

All the drops from ~ 2 weeks ago that got scalped already sold, at least in the U.S. anyway, I know some stores in other regions are apparently selling them according to some folksO_o
 

 
Looks like they're going up, as will others, with the Lunar New Year & supply shortages combining to raises prices right before we get slapped with the full boar of the Tariff exemption expirations fully taking effect in April.

alpharius194

bmx045

Dabadger84
I don't even get artifacts that I've noticed while running Port Royal Stress Test or Time Spy Extreme Stress Test, on my OC/undervolt settings that I've tested to be "stable", definitely not at stock.  
What timestamps are the artifacts occurring at?  Are they consistent or do they change?  Is it flashing or triangles or what kind?

I just ran time spy extreme stress and pr stress back to back without problems. It seems to be very inconsistent. I'm wondering if an 850w PSU is enough for 5950x (PBO on, otherwise stock), kingpin w/ high power limits, 8 fans and cpu aio....

When I was looking into the 5950x (could only get my hands on a 5900x) I saw that quite a few people listed power draw with PBO to be above 300w and manual OC to be above 350w. Not saying its the issue on stock bios but to run the LN2 bios and the 5950x with PBO you may run into the limit of what your PSU can provide. Try backing the 5950x to stock and see what happens.
https://www.kitguru.net/c...%20for%20the%20system.

O_O 16 cores really do be like that I guess.  I'm not surprised it's power draw is that high though, that's a lot of cores, and a lot of very powerful cores, in that CPU's case.  The 5950x is a beast, I pity the AIO tryin' to keep that thing cool lol