Manufacturers on RTX 3080 CTD issues, Inno3D Offers a Shocking Response

XenthorX

2020-09-29 22:49

We're merely one week after release and suddenly the internet is filled with electrical/electronical engineers pointing fingers at AIB partners, and affirming without the slighest doubt that their design is faulty, jumping to conclusions.

Neo Cyrus:

"Your $1500 USD video card has to be throttled by software to prevent it from crashing? No issues, working as intended. Carry on, peons." That's a dogshit response. If anything he's underreacting. You don't sell high end products at stupid price tags, then say there's nothing wrong when everyone knows there is, when it's YOUR FAULT for using cheapass parts on an incredibly expensive item. "2.1GHz+? Just settle for 1.9GHz, it's not like you paid $1500 for it, there's nothing wrong, stop complaining, shut up." That attitude needs to be shot straight into the deepest depths of hell and burn there for all eternity. Companies are horrible enough with people complaining, if everyone became even more complacent it's not going to stay the same, it gets WORSE.

So new hardware should have perfectly working drivers on day one? One month ago AIB couldn't even run real world applications, just Nvidia testing suite. Again there's no "cheaping out" here, those capacitors are different product period, there's not even marginal cost difference here. (MLCC are actually cheap ! They're the most common!) https://imgur.com/3HH4dkX.jpg PosCaps red line https://i.imgur.com/MEFZ1Ay.png https://i.imgur.com/QG0LtLJ.png When you see those graphs the full PosCaps configuration doesn't look silly at all.

#5836705

tsunami231

2020-09-29 23:06

interesting read not sure i buy that response as truth, A PR made response to downplay it more like it. I sure they will "fix" threw Firmware/drivers then in later revisions of the card they will charge hardware configuration

#5836724

cowie

2020-09-29 23:36

ViperAnaf:

so the "fix" is to downgrade the promised performance lol

I highly dought 20 to 40 mgz would any difference in performance at all if you are talking a few hundred then maybe 2-5 fps so are the drivers working to fix the ctd?

#5836747

TheDeeGee

2020-09-30 00:11

cowie:

I highly dought 20 to 40 mgz would any difference in performance at all if you are talking a few hundred then maybe 2-5 fps so are the drivers working to fix the ctd?

I believe it's been fixed for "most" people, some still get crashes and have to underclock, but that could be because of a factory OC. With the standard FE settings the issues are gone i believe.

#5836782

Neo Cyrus

2020-09-30 01:43

XenthorX:

We're merely one week after release and suddenly the internet is filled with electrical/electronical engineers pointing fingers at AIB partners, and affirming without the slighest doubt that their design is faulty, jumping to conclusions. So new hardware should have perfectly working drivers on day one? One month ago AIB couldn't even run real world applications, just Nvidia testing suite. Again there's no "cheaping out" here, those capacitors are different product period, there's not even marginal cost difference here. (MLCC are actually cheap ! They're the most common!) When you see those graphs the full PosCaps configuration doesn't look silly at all.

You do realize what you just showed doesn't prove anything regardless of what it is, and that it's ambiguous right? I'm not even entirely sure what you're showing; not enough information. A 10 MLCC array costs more than a single poscap which is what TEN of those replace. You do know that's where manufacturers cut costs and do that for no reason other than to cut costs... right? Not only does it cost more, supposedly, according to everyone who's reported on it it's: "SIGNIFICANTLY MORE COSTLY!" - Everyone Whatever the hell that means. Even a ton of them doesn't cost much but with companies pinching every penny per 1000 units on $1500 items, it's to be expected I suppose. And that's still dodging the issue of Inno3D being turds pretending there is nothing wrong when it's their fault they're selling a faulty product and now are shrugging their shoulders saying nothign is wrong. I'd be furious if I was missing 200MHz+ on something that's clocked at only 1.9GHz and scales heavily with frequency. I can't be the only one here who OCs my video cards any way I can to squeeze out every last drop.

#5836789

Clawedge

2020-09-30 02:35

The only mooncake i will accept https://cdn140.picsart.com/320726491202211.png?type=webp&to=min&r=240

#5836790

Agonist

2020-09-30 02:36

Neo Cyrus:

"Your $1500 USD video card has to be throttled by software to prevent it from crashing? No issues, working as intended. Carry on, peons." That's a dogshit response. If anything he's underreacting. You don't sell high end products at stupid price tags, then say there's nothing wrong when everyone knows there is, when it's YOUR FAULT for using cheapass parts on an incredibly expensive item. "2.1GHz+? Just settle for 1.9GHz, it's not like you paid $1500 for it, there's nothing wrong, stop complaining, shut up." That attitude needs to be shot straight into the deepest depths of hell and burn there for all eternity. Companies are horrible enough with people complaining, if everyone became even more complacent it's not going to stay the same, it gets WORSE.

Nvidia fanboys will do what ever to not let them look like shit. Every time there is driver problems, card problems, GTX 970, Fermi degradation, Gameworks bullshit, most just act like they do no wrong. I can not stand i with the Nvidia double standard. AMD gets the shit show they deserve when their PR truly screws up. The Fury X and FX chips debacle was terribly deserved.

#5836791

XenthorX

2020-09-30 02:38

I really don't know what you're talking about @Neo Cyrus , Derbauer replaced 2 PosCaps by 20 MLCC, the difference was 20MHz. On a Gigabyte 3090 Gaming OC with originally 6 PosCaps. 200MHz is ludicrous and out of nowhere. [youtube=ud6NrbJllzk]

#5836802

Neo Cyrus

2020-09-30 03:46

@XenthorX That's just 2 and on 1 specific card. I'm hearing reports of having to stop boosting much above 1900MHz because of crashes and especially at 2.1GHz. 2.1GHz seems to be the hard wall. Do you realize how, for a lack of better words, narrow in scope and meaningless all your examples are? On it's own it's so generic and lacking in total meaning even in just the narrow context of ONLY discussing those 6 caps... but do I really need to spell it out that speaking of these poscaps puts them in a relative setting, as in we should be discussing the entire cards being cheaped out on, not just those 6 fucking caps which should have been 60 of something else. No matter how you want to put it, no matter what esoteric examples you want to give, these companies cheaped out. CHEAPED OUT HARD, according to literally anyone who knows about the pricing of these things who have reported on it. So either all of them are lying to us, or what I believe, that it's true. And just for the record we're all using the wrong term according to Buildzoid, they're not actually poscaps, those are specific types made by Panasonic. Whatever.

#5836816

Spets

2020-09-30 04:53

What are you on about? These cards aren't meant to hit 2.1ghz, it can't be done without a shunt mod or extreme cooling. Every review card oc'd was barely close to 2.1. The problem was the cards would run at their proper speeds then in certain scenarios would go a boost step above 2.1 which would create an instant crash. I really don't understand why your claiming these cards should be doing 2.1 at stock oc speeds.

#5836820

tsunami231

2020-09-30 05:13

Clawedge:

The only mooncake i will accept https://cdn140.picsart.com/320726491202211.png?type=webp&to=min&r=240

Mooncake is only thing that makes sense anymore

#5836826

Astyanax

2020-09-30 05:31

XenthorX:

We're merely one week after release and suddenly the internet is filled with electrical/electronical engineers pointing fingers at AIB partners, and affirming without the slighest doubt that their design is faulty, jumping to conclusions. So new hardware should have perfectly working drivers on day one? One month ago AIB couldn't even run real world applications, just Nvidia testing suite. Again there's no "cheaping out" here, those capacitors are different product period, there's not even marginal cost difference here. (MLCC are actually cheap ! They're the most common!) https://imgur.com/3HH4dkX.jpg PosCaps red line https://i.imgur.com/MEFZ1Ay.png https://i.imgur.com/QG0LtLJ.png When you see those graphs the full PosCaps configuration doesn't look silly at all.

When you understand that the spcaps are noisy and the gpu requires filtering to achieve stable high clock operations, that capacitance looks stupid.

#5837115

Neo500

2020-09-30 16:23

I was planning to get a 3080 when it arrives here in Denmark next summer .... bur how can i be sure that i not get a faulty card ?

#5837117

Astyanax

2020-09-30 16:24

Neo500:

I was planning to get a 3080 when it arrives here in Denmark next summer .... bur how can i be sure that i not get a faulty card ?

You can't be sure, never go into any product purchase completely sure it'll run fine, thats false confidence and clouds your judgement when issues that are obviously linked to the part arise.

#5837283

JAMVA

2020-09-30 23:50

Neo Cyrus:

You do realize what you just showed doesn't prove anything regardless of what it is, and that it's ambiguous right? I'm not even entirely sure what you're showing; not enough information. A 10 MLCC array costs more than a single poscap which is what TEN of those replace. You do know that's where manufacturers cut costs and do that for no reason other than to cut costs... right? Not only does it cost more, supposedly, according to everyone who's reported on it it's: "SIGNIFICANTLY MORE COSTLY!" - Everyone Whatever the hell that means. Even a ton of them doesn't cost much but with companies pinching every penny per 1000 units on $1500 items, it's to be expected I suppose. And that's still dodging the issue of Inno3D being turds pretending there is nothing wrong when it's their fault they're selling a faulty product and now are shrugging their shoulders saying nothign is wrong. I'd be furious if I was missing 200MHz+ on something that's clocked at only 1.9GHz and scales heavily with frequency. I can't be the only one here who OCs my video cards any way I can to squeeze out every last drop.

In this Buildzoid video he shows at the 17 min mark that the MLCC are $0.01 to $0.03 cent each so $0.10 to $0.30 for 10 , the SPCAPS are $0.61 to $1.67 each. [youtube=T1I-kS3lZUc]

#5837293

KissSh0t

2020-10-01 01:25

Astyanax:

You can't be sure, never go into any product purchase completely sure it'll run fine, thats false confidence and clouds your judgement when issues that are obviously linked to the part arise.

Maybe the board partners will just do nothing and push out all the defective stock onto unsuspecting buyers and by then there will be newer revisions of the cards available? It's what they usually do..

#5837409

Rich_Guy

2020-10-01 11:16

[youtube=lhyCdraz54s]

#5837429

alanm

2020-10-01 12:22

Rich_Guy:

[youtube=lhyCdraz54s]

Funny this guy now sees it as entirely a driver issue. I hope he's correct. And no performance hit resulting from the updated driver. But poor Evga, saw it differently and had to delay 3080 shipments to change the caps layout on their cards.

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.

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

#5837733

illrigger

2020-10-02 05:00

The Gigabyte statement was really interesting to me. I've had a Gigabyte 3080 Gaming 10G since the 18th, and have had zero crashes, with dozens of hours of gaming (mostly World of Warships, my current addiction) done on it. At 4k with all options cranked, it gives the card a very hefty workout. I've also dinked with Shadow of the Tomb Raider, Control, Wolfenstein Youngblood, and Borderlands 3 a couple hours each, as well as some other stuff, all cranked up to max settings. I was a little worried when I first looked at what they said, but they are being pretty clear: there are different specs of SP-CAPs, and just looking at the back of the board and basing your assessment on what the caps look like isn't really doing enough research.

#5837749

kapu

2020-10-02 08:12

illrigger:

The Gigabyte statement was really interesting to me. I've had a Gigabyte 3080 Gaming 10G since the 18th, and have had zero crashes, with dozens of hours of gaming (mostly World of Warships, my current addiction) done on it. At 4k with all options cranked, it gives the card a very hefty workout. I've also dinked with Shadow of the Tomb Raider, Control, Wolfenstein Youngblood, and Borderlands 3 a couple hours each, as well as some other stuff, all cranked up to max settings. I was a little worried when I first looked at what they said, but they are being pretty clear: there are different specs of SP-CAPs, and just looking at the back of the board and basing your assessment on what the caps look like isn't really doing enough research.

WHOLE situation was blewn out of proportions , it is hard to say how many % of cards were affected but vast majority was not.