rtx 3080 crashing possibly due to capacitor choice

[spartaman64]

Quote

Igor's Lab has posted an interesting investigative article where he advances a possible reason for the recent crash to desktop problems for RTX 3080 owners. For one, Igor mentions how the launch timings were much tighter than usual, with NVIDIA AIB partners having much less time than would be adequate to prepare and thoroughly test their designs. One of the reasons this apparently happened was that NVIDIA released the compatible driver stack much later than usual for AIB partners; this meant that their actual testing and QA for produced RTX 3080 graphics cards was mostly limited to power on and voltage stability testing, other than actual gaming/graphics workload testing, which might have allowed for some less-than-stellar chip samples to be employed on some of the companies' OC products (which, with higher operating frequencies and consequent broadband frequency mixtures, hit the apparent 2 GHz frequency wall that produces the crash to desktop).

Quote

Another reason for this, according to Igor, is the actual "reference board" PG132 design, which is used as a reference, "Base Design" for partners to architecture their custom cards around. The thing here is that apparently NVIDIA's BOM left open choices in terms of power cleanup and regulation in the mounted capacitors. The Base Design features six mandatory capacitors for filtering high frequencies on the voltage rails (NVVDD and MSVDD). There are a number of choices for capacitors to be installed here, with varying levels of capability. POSCAPs (Conductive Polymer Tantalum Solid Capacitors) are generally worse than SP-CAPs (Conductive Polymer-Aluminium-Electrolytic-Capacitors) which are superseded in quality by MLCCs (Multilayer Ceramic Chip Capacitor, which have to be deployed in groups). Below is the circuitry arrangement employed below the BGA array where NVIDIA's GA-102 chip is seated, which corresponds to the central area on the back of the PCB.

Quote

It's likely that the crash to desktop problems are related to both these issues - and this would also justify why some cards cease crashing when underclocked by 50-100 MHz, since at lower frequencies (and this will generally lead boost frequencies to stay below the 2 GHz mark) there is lesser broadband frequency mixture happening, which means POSCAP solutions can do their job - even if just barely.

IMPORTANT NOTE: apparently the evga 3080 ftw3 has 2 mlcc and the picture on the EVGA website is a preproduction model

https://imgur.com/a/IMVFdTP

update 2: apparently some evga 3080 ftw3 have 6 spcaps and some have 4 spcaps 2 mlcc

update from jacob from evga

Quote

During our mass production QC testing we discovered a full 6 POSCAPs solution cannot pass the real world applications testing. It took almost a week of R&D effort to find the cause and reduce the POSCAPs to 4 and add 20 MLCC caps prior to shipping production boards, this is why the EVGA GeForce RTX 3080 FTW3 series was delayed at launch. There were no 6 POSCAP production EVGA GeForce RTX 3080 FTW3 boards shipped.
 
But, due to the time crunch, some of the reviewers were sent a pre-production version with 6 POSCAP’s, we are working with those reviewers directly to replace their boards with production versions.
EVGA GeForce RTX 3080 XC3 series with 5 POSCAPs + 10 MLCC solution is matched with the XC3 spec without issues.
 
Also note that we have updated the product pictures at EVGA.com to reflect the production components that shipped to gamers and enthusiasts since day 1 of product launch.
Once you receive the card you can compare for yourself, EVGA stands behind its products!

 

https://forums.evga.com/Message-about-EVGA-GeForce-RTX-3080-POSCAPs-m3095238.aspx

In the buildzoid video he said they are not POSCAPs which are a specific pansonic product line they are SPCAPs which the higher end ones perform better than POSCAPs. But he agrees the mlcc caps are better for filtering and having multiple of them in parallel makes them even better. And he said he thinks we should blame nvidia instead of the board partners because nvidia has to approve all of the boards and they allowed the board partners to do this. Also he noticed some of the cards with mlccs cheaped out on the other SPCAPs.

 

Der8auer replaced the spcaps with mlccs and it solved the crashing problem which confirms that it is at least a major contributing factor.

 

source: https://www.techpowerup.com/272591/rtx-3080-crash-to-desktop-problems-likely-connected-to-aib-designed-capacitor-choice

 

https://www.igorslab.de/en/what-real-what-can-be-investigative-within-the-crashes-and-instabilities-of-the-force-rtx-3080-andrtx-3090/

 

Yikes this launch is sort of a mess. If you are experiencing crashing try underclocking a little and if you are buying a card look out for the ones with bad cap layouts. Idk why nvidia felt the need to rush this launch by this extent. Also sort of strange manufacturers would put low quality caps in a high end gpu and how much do they even save on that?

 

Edit: I looked up how much they cost and apparently they cost $0.00293 per unit if you buy them in bulk of 10,000+ ...

so you need 10 to replace an SPCAP and theres 6 SPCAPS so thats 60*$0.00293= $0.1758 ... at most they are saving 1.758* cents per card WTF

 

and actually thats if they replaced all the SPCAPs with MLCCs they only need 2 so they are saving $0.0058

 

[PCGuy_5960]

Huh, looks like the supply wasn't the biggest problem with the RTX3080 launch.

[Slottr]

Just now, PCGuy_5960 said:

Huh, looks like the supply wasn't the biggest problem with the RTX3080 launch.

Wait till we find out Nvidia rounded 6.5GB of vram to 10

[Moonzy]

so it is a hardware issue rip

incoming firmware patch to nerf all affected card to below 2000MHz

 

interested to see if 3070 is affected, and what caps are used in specific models.

[TempestCatto]

Maybe this is all a ploy by AMD so people flock to them instead.

[WereCat]

NVIDIA Recruiting Beta testers!

 

"We are offering you to pay us up to $1500 for testing our product!"

[PCGuy_5960]

Just now, TempestCatto said:

Maybe this is all a ploy by AMD so people flock to them instead.

Well first they have to make a GPU that can match the performance of nvidia

[WereCat]

Just now, PCGuy_5960 said:

Well first they have to make a GPU that can match the performance of nvidia

ez, just make a new benchmark in which your product is faster than the competition like an unnamed blue corporation did

[PCGuy_5960]

Just now, WereCat said:

ez, just make a new benchmark in which your product is faster than the competition like an unnamed blue corporation did

Or like Nvidia's "8K"

[Spotty]

Here's the original source article from Igor's lab. https://www.igorslab.de/en/what-real-what-can-be-investigative-within-the-crashes-and-instabilities-of-the-force-rtx-3080-andrtx-3090/

 

[Shreyas1]

This is why you never pay to become a beta tester buy a new product at launch

[AnonymousGuy]

Some guy has some giant fucking balls replacing one of the caps with the mlcc ones:

 

 

Probably had to cover all the other areas with kapton tape to not blow them away with hot air rework.  Even then it's probably dicey.

[Moonzy]

31 minutes ago, AnonymousGuy said:

Probably had to cover all the other areas with kapton tape to not blow them away with hot air rework.

those tiny SMD gives me anxiety when im handling the card

one accidental scratch and it'll probably come off

[AnonymousGuy]

6 minutes ago, Moonzy said:

those tiny SMD gives me anxiety when im handling the card

one accidental scratch and it'll probably come off

I'd be really concerned about damaging the metal layers on the other side connecting the PCB to the GPU.  NV spends a *shitload* of R&D optimizing that solder process with exact heating and cooling durations.  Get it wrong and you have failed solder joints eventually after some heat cycles.

[RejZoR]

Checked the card I've "ordered" from Inno3D, the iChill X4, in official promo it has 4x2 order and watching some guys video, it has 5x1. From what I've gathered, 5x1 is reference spec, 6x0 is not. And those seem to have problems. iChill X4 being 5x1 should be fine then? Right?

[GoodBytes]

So the affectedtd AIBs are going BELLOW specs from Nvidia. I would say: Return the card now!

No buts! The card was improperly built to specification, you, as a consumer, been had. If they cut there, they probably cut at other places too. 

Return it, and buy from a AIB that doesn't cheat its customers. If you don't, you essentially agree that it is OK for manufactures to sales you overpriced crap.

 

For most people in the world, people are still in their return window. There will NOT be a recall. Cost too much. The affected manufacture will release a vBIOS which will gimp the GPU abilities to 'solve' the problem.

 

[GoodBytes]

5 hours ago, Shreyas1 said:

This is why you never pay to become a beta tester buy a new product at launch

Well, in this case, is the manufacture of the card that went bellow specs to try a save a buck.

This is like you buy a system, that cost 3k, with 3k worth of hardware, but you end up with RAM timings in the 250 instead of 16's, no screws anywhere, hot glue everything, even the motherboard. No chipset heatsink when it should have one, USB 1.1 instead of 3 let alone 2. They shoehorn DDR1 version of the RAM (as they never mentioned the type and speed) of the mentioned capacity instead of DDR4, single channel too, no antennas on the wireless card that comes included, and no place to add some so you have 0 reception due to the metal case. 

 

I say, do the right thing, and return it. Yes it sucks, but never buy from that company again. There is no shame in returning it! There is shame in buyer remorse. This is one of the main reason we have return windows.

[imreloadin]

*laughs in 1080 Ti*

[Shreyas1]

2 minutes ago, GoodBytes said:

Well, in this case, is the manufacture of the card that went bellow specs to try a save a buck.

This is like you buy a system, that cost 3k, with 3k worth of hardware, but you end up with RAM timings in the 250 instead of 16's, no screws anywhere, hot glue everything, even the motherboard. No chipset heatsink when it should have one, USB 1.1 instead of 3 let alone 2. They shoehorn DDR1 version of the RAM (as they never mentioned the type and speed) of the mentioned capacity instead of DDR4, single channel too, no antennas on the wireless card that comes included, and no place to add some so you have 0 reception due to the metal case. 

 

I say, do the right thing, and return it. Yes it sucks, but never buy from that company again. There is no shame in returning it! There is shame in buyer remorse. This is one of the main reason we have return windows.

I mean, in general nowadays I've seen so many products (tech and not tech) that come out and have problems immediately at launch. Problems that could be avoided if you just wait for  a month or so after launch for the company to fix their problems. 

[tim0901]

8 minutes ago, GoodBytes said:

So the affectedtd AIBs are going BELLOW specs from Nvidia. I would say: Return the card now!

No buts! The card was improperly built to specification, you, as a consumer, been had. If they cut there, they probably cut at other places too. 

Return it, and buy from a AIB that doesn't cheat its customers. If you don't, you essentially agree that it is OK for manufactures to sales you overpriced crap.

 

For most people in the world, people are still in their return window. There will NOT be a recall. Cost too much. The affected manufacture will release a vBIOS which will gimp the GPU abilities to 'solve' the problem.

 

Suddenly Nvidia's decision to make their own cards better makes a lot more sense... maybe relying on AIBs to sell your product isn't the best idea after all!

[Maticks]

Jayztwocents put up a video on this.
I was surprised that EVGA FTW cheaped out on power filtering.
I held my breath since I pre-ordered a Asus Tuf but looks like I picked well.
Only because ekwb only has the Asus Tuf/strix and founders announced for water blocks. I wonder if this issue has something to do with it. Why put a water block on a card that can't boost.. very interesting.

[Serin]

Feelin' pretty good about my 3090 ftw3 preorder... Fuck

[spartaman64]

7 minutes ago, Serin said:

Feelin' pretty good about my 3090 ftw3 preorder... Fuck

i mean 2 mlcc is up to spec so you are fine and apparently the 3080 ftw3 has 2 mlcc also just the picture on the evga website is outdated

[spartaman64]

17 minutes ago, Maticks said:

Jayztwocents put up a video on this.
I was surprised that EVGA FTW cheaped out on power filtering.
I held my breath since I pre-ordered a Asus Tuf but looks like I picked well.
Only because ekwb only has the Asus Tuf/strix and founders announced for water blocks. I wonder if this issue has something to do with it. Why put a water block on a card that can't boost.. very interesting.

they apparently didnt the picture on their website is a preproduction model apparently since someone who got a 3080 ftw3 has pictures of it with 2 mlcc

edit:

 

nvm apparently some have 6 poscap and some have 2 mlcc @Serin

[Maticks]

2 minutes ago, Serin said:

Feelin' pretty good about my 3090 ftw3 preorder... Fuck

I've only bought EVGA since I can remember but something was strange when ekwb didn't announce they had a water block in works for it. Which is why I reluctantly ordered the Asus TUF. Gigabyte apart from looking like the ugly duckling only a mother could love has also the worst power delivery and colourful, zotach. the others are just border line to spec. Probably boost to 1955Mhz but forget every getting into the 2000mhz even with voltage vbios mods with noise.

Looks like the strix, Tuf and founders might but they Will have good power filtering that doesn't mean that other problems will not prevent clocking higher. Let's see what the techtuber community tests. We are fortunate to have these guys who will test all this.