• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Rtx 3080 lower quality capacitor Issue

Weird one for me. Gaming stability is perfect. I've thrown everything I can at the card to make it crash under stress and it doesn't.

But the driver crashes occasionally while playing back video in MPC-HC or MPV.
 
Well gigabyte cards dont have any problems since they use 470 rated SP caps. Why cant other manufacturers do the same? If they did then non of these cards wont have this problem when it boosts over 2ghz just like the gigabyte rtx 3080 eagle oc card. Most manufacturers are just cutting corners just to save money and sell a crappy product to a customer and charge more than MSRP. What a joke.
 
Last edited:
sorry im not getting this to much, should we cancel our orders for the 3000 series cards?

Ive got a Asus strix 3080 and 3090 ordered and will keep which ever comes 1st, but i wont keep either if there is a manufacturing defect on the 3000 series cards

Ive sold my 2080ti on the bay 2 days ago for 625 and it sold in 30 mins of listing, truth

I then cancelled it with the buyer, because of the delay and this defect issue

Can peoples advise if we should cancel and wait? rather than hope for the best when the order is received

I can imagine the anger and frustration of buyers that get faulty cards

Standing by.
 
Rroff said:
I'm still not really buying the capacitor side of the story - this use of capacitors is generally a cheap and reliable way of hugely increasing your stable operating margins and is usually massively overkill - it is possible the crammed PCB or some other factors make a difference here or someone has really screwed up the design and the voltage regulation is easily destabilised.

Either mixed silicon quality or a bug in the boost algorithm is more likely IMO.
Click to expand...

+1

It's looking like issue with Silicon, maybe BIOS, that's exacerbated by certain bypassing arrangements. Keep in mind we still haven't seen any measurements showing sizable deviation of the supplies, so this could be a combination of silicon instantaneous draw, silicon power distribution, PCB bypassing, VRM design, PCB design, etc.

Being the cynic that I am, what's the bet NVidia will just increase the voltage applied by +50mV to give more headroom, eating into your already power constrained performance.
 
Last edited:
pc-guy said:
Get the cards put under a scope isnt easy. You need the scope and open bench setup and know how.

no one has the circuit board design so it is a guess as to where you stab the contact. But buildzoid has identified a couple of contacts look like a good start.

anyway I think the reviewers don’t have expertise or don’t want to go into this level of technicality for whatever reasons ( maybe cheesing off their financial sponsors). Then the average Joes who have cards and just don’t have a clue. Or someone maybe have the card and does have a clue but frankly cba with it all as not wanting to void the warranty of the card.

Let’s face it, if you paid for the card yourself you don’t want to void that warranty by ripped your card apart and start probing it. Even if you are wealthy beyond imagination, it’s not like you can get another card readily atm if you damage it.
Click to expand...

It's really not. If the bulk caps in question are the cause, then you just need to put a single probe, ideally differential, over the capacitors' terminals. If it's not a diff probe, then figure out which side of the cap is GND using a multi-meter and use a single-ended probe with short GND springy thingy.
 
cutterslade said:
I don't think that anybody, as long s/he is not an hardware engineer, knows about power delivery networks and additionally about the 30x0, can properly comment on this topic and say what the best constellation is to avoid the problems we see.
Click to expand...

Nail on the head.

Only NVidia can realistically speak with confidence on this issue, as only they have total system visibility. Even board partners only know what's going on at the boundary of the GPU package, not even the die... I feel for the partners, where they'll have been given bypassing requirements, which they will have met.
 
Not sure if someone else had mentioned but Pauls Hardware did a pretty good video explaining at least his findings, wont link it just YouTube Pauls hardware
But it seems that the out of the box experience is fine, but when adding anything that pushes the card to far over 2ghz boost depending on the card you may see more boost or less but all cards including founders all crashed over 2ghz at one stage or another
benching Tomb raider 15 minuets game play.

You'll have to decided for yourselves if you think that is a good result or not,
I'm ok with it, the card I've bought will achieve the speed they sold.
 
whippyice said:
Not sure if someone else had mentioned but Pauls Hardware did a pretty good video explaining at least his findings, wont link it just YouTube Pauls hardware
But it seems that the out of the box experience is fine, but when adding anything that pushes the card to far over 2ghz boost depending on the card you may see more boost or less but all cards including founders all crashed over 2ghz at one stage or another
benching Tomb raider 15 minuets game play.

You'll have to decided for yourselves if you think that is a good result or not,
I'm ok with it, the card I've bought will achieve the speed they sold.
Click to expand...
Going from 2000 > 2100 is a gain of 2 fps so not a big deal although it could be an issue for those who want to water cool and will see higher clocks due to lower temps.
 
Joxeon said:
Going from 2000 > 2100 is a gain of 2 fps so not a big deal although it could be an issue for those who want to water cool and will see higher clocks due to lower temps.
Click to expand...

yeah agreed,
this situation has brushed up my gpu workings and information a fair bit at least thats good :D
 
TBH you could chase your tail speculating on this.
What seems to be a common factor is if the GPU clocks to above 2ghz it pushes into instability. You can downclock your card if you have an issue and it will work fine. I would surmise that the answer from nvidia or AIB's will be either a bios or driver release to sort.

I would say that is probably more down to the silicon lottery than anything else, but everything plays a part.
AFAIK even the highest clocked AIB's only specify a boost of around 1815 so technically you have what you paid for as long as it will do the specified boost reliably.

People asking should they cancel ... I would say not, just adjust your expectations of clock speeds. seems most cards will game around the 1900 area quite happily, and still be faster than what it replaced.
 
Vogon said:
TBH you could chase your tail speculating on this.
What seems to be a common factor is if the GPU clocks to above 2ghz it pushes into instability. You can downclock your card if you have an issue and it will work fine. I would surmise that the answer from nvidia or AIB's will be either a bios or driver release to sort.

I would say that is probably more down to the silicon lottery than anything else, but everything plays a part.
AFAIK even the highest clocked AIB's only specify a boost of around 1815 so technically you have what you paid for as long as it will do the specified boost reliably.

People asking should they cancel ... I would say not, just adjust your expectations of clock speeds. seems most cards will game around the 1900 area quite happily, and still be faster than what it replaced.
Click to expand...

Yeah this is why i am fine with the purchase I've made, my xc3 states on the box what i can do with it, and i wont be pushing it past that and then getting disappointed when it doesn't go past what they have advised, which is why they are standing behind what they have sold.
it seems like there needs to be more samples of crashing and what the clock speeds at the time was, because i think that would all of a sudden make everything clear.

but if i got the card and it wasn't even meeting base line expectations then we would have a problem
 
You must log in or register to reply here.
Back
Top Bottom