So, apparently MOSTLY everyone needs to return their RTX 3080 / 3090 because of "cheap" components. I'm def returning my EVGA 3090.

Ocellaris

Fully [H]
Joined
Jan 1, 2008
Messages
18,941
On what grounds? All of the cards easily boost past their guaranteed boost clock. Read the specs. EVGA's highest end card only advertises 1800MHz boost clock.

The speculation, with zero evidence, is entire product ranges of cards are bad because of a board layout choice. Some may actually be bad with a specific configuration, but it’s the minority(and no EVGA cards). Now anything without “the best” possible layout if deemed poop. This is about the same as getting upset when you find out your board has cheap screws and not NASA grade hardware holding it together.

This witch hunt mostly being decided upon by people who just now realized every component in their PC is built to be profitable. Which was all started by an article which didn’t even use the correct terminology...
 

mnewxcv

Supreme [H]ardness
Joined
Mar 4, 2007
Messages
7,593
The speculation, with zero evidence, is entire product ranges of cards are bad because of a board layout choice. Some may actually be bad with a specific configuration, but it’s the minority(and no EVGA cards). Now anything without “the best” possible layout if deemed poop. This is about the same as getting upset when you find out your board has cheap screws and not NASA grade hardware holding it together.

This witch hunt mostly being decided upon by people who just now realized every component in their PC is built to be profitable. Which was all started by an article which didn’t even use the correct terminology...
Even if the cards we're crashing [if they had lower boost table set], the cards would effectively have 0 overclocking headroom before crashes develop. I wonder if the 3070 will be plagued as well, considering it has fewer cores and higher clocks.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
5,687
So are we going to see upgraded boards from the AIBs? How many will still have the questionable configuration, low end components in the next 2 to 3 months? Also how is availability affected with the needed changes for a number of 3080's? I can see Gigabyte just halting their 3080's until needed changes are done, Zotac may just keep producing and let the users figure it out. The real question is what is going to happen to the pricing?
 

Rizen

[H]F Junkie
Joined
Jul 16, 2000
Messages
9,363
Even if the cards we're crashing [if they had lower boost table set], the cards would effectively have 0 overclocking headroom before crashes develop. I wonder if the 3070 will be plagued as well, considering it has fewer cores and higher clocks.
So then they boost like Ryzen CPUs?

I know this is [H] and I love overclocking as much as anyone and I too would be disappointed if I couldn't push my GPU much further over stock. BUT. I wouldn't really feel it's fair to say it's a "bad product" because the boost algorithms get as much as possible out of the chips without user intervention.

But if that's the case, then the fuck up here is the very aggressive boosting behavior.

That said, GPU Boost has been pretty aggressive for at least the last couple of generations relative to advertised clocks, so I am not certain that's the problem here.
 

mnewxcv

Supreme [H]ardness
Joined
Mar 4, 2007
Messages
7,593
So then they boost like Ryzen CPUs?

I know this is [H] and I love overclocking as much as anyone and I too would be disappointed if I couldn't push my GPU much further over stock. BUT. I wouldn't really feel it's fair to say it's a "bad product" because the boost algorithms get as much as possible out of the chips without user intervention.

But if that's the case, then the fuck up here is the very aggressive boosting behavior.

That said, GPU Boost has been pretty aggressive for at least the last couple of generations relative to advertised clocks, so I am not certain that's the problem here.
yes, GPUs boost like CPUs. They are rated around 1740MHz as the boost clock, but can go higher based on power limit and temperature. It seems they just went too aggressive with it to me.
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,880
Looks like 3080 founders edition uses poscaps to me.
*** The FE 30xx (only 3080?) cards are the only ones using poscaps (so far). Everyone else is using sp-caps (or some other, not tantalum poly capacitor) afaik.
 

vjhawk

Limp Gawd
Joined
Sep 2, 2016
Messages
318
So are we going to see upgraded boards from the AIBs? How many will still have the questionable configuration, low end components in the next 2 to 3 months? Also how is availability affected with the needed changes for a number of 3080's? I can see Gigabyte just halting their 3080's until needed changes are done, Zotac may just keep producing and let the users figure it out. The real question is what is going to happen to the pricing?

As Jay suggested, the AIBs with the cheaper layouts will just rollout a firmware update that underclocks the video cards to frequencies where they will be more stable. That's the cheapest fix, so that's probably what they will do. Don't hold your breath on manufacturers revamping their product lines when it will cost them money. The only AIB who has actually done that is EVGA. So Kudos to them for catching the instability and revising their boards to overcome this issue.

Cap-Gate indeed! :p And yes I am now aware that they are actually called SP-Caps rather than Pos-Caps after watching Buildzoid's rant video.
 
  • Like
Reactions: noko
like this

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
18,668
I'm still waiting to see what happens before I do anything. I re-installed everything and physically made sure everything was plugged in and seated ideally. I haven't had any crashes or issues since Friday.
The better manufacturers like EVGA have a stout warranty so it's not like anyone needs to RMA anything today or tomorrow.
 

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
18,668
I figured a new driver would be incoming. I'll put these through their paces in a bit and see how they do. Will also give 3DMark a go and see if scores are affected, too.
 

c3k

2[H]4U
Joined
Sep 8, 2007
Messages
2,238
The capacitor issue is being examined. Great.

Sucks for someone to be an early adopter, but, well...thanks, guys.

To me, this is the real question: What OTHER components did Nvidia allow to be "downgraded"?
 

mnewxcv

Supreme [H]ardness
Joined
Mar 4, 2007
Messages
7,593
The capacitor issue is being examined. Great.

Sucks for someone to be an early adopter, but, well...thanks, guys.

To me, this is the real question: What OTHER components did Nvidia allow to be "downgraded"?
what is your concern? If the card works to spec and doesn't die in warranty, as far as I'm concerned it's a non issue.
 

BassTek

Supreme [H]ardness
Joined
Jul 13, 2002
Messages
5,777
The capacitor issue is being examined. Great.

Sucks for someone to be an early adopter, but, well...thanks, guys.

To me, this is the real question: What OTHER components did Nvidia allow to be "downgraded"?

Watch a few of the Buildzoid videos he goes over the FE, MSI and Zotac PCBs. Main differences are power stages, different VRMs, capacitors, layout.

The FE is built up a lot more than the AIB models you will probably need to wait for the second wave and higher tier AIB models to match it.
 

Rizen

[H]F Junkie
Joined
Jul 16, 2000
Messages
9,363
Only if you are NVidia, customers have other genuine concerns.
What other genuine concern could you have, other than

A) The product working as advertised
B) The product not dying prematurely

I'm asking legitimately.
 

TaintedSquirrel

[H]F Junkie
Joined
Aug 5, 2013
Messages
10,114
Nvidia statement

https://www.nvidia.com/en-us/geforc...x-3080-board-stability-new-driver-capacitors/

NVIDIA posted a driver this morning that improves stability. Regarding partner board designs, our partners regularly customize their designs and we work closely with them in the process. The appropriate number of POSCAP vs. MLCC groupings can vary depending on the design and is not necessarily indicative of quality.
 

Teenyman45

2[H]4U
Joined
Nov 29, 2010
Messages
2,570
So the early feedback being posted online is that the driver "fixes" the issue by downclocking the video cards. From the early numbers being tossed around, it seems like you can expect greater stability by sacrificing about 1.5%-5% performance on a 3080 (and maybe a 3090) due to the change in the way boost is being handled. Cheapening out on components can be solved by reducing performance. I don't perceive this as a fix unless the hardware websites change their names from an overclock naming theme to an "underclock" one.

What other genuine concern could you have, other than

A) The product working as advertised
B) The product not dying prematurely

I'm asking legitimately.

There are two separate aspects to "as advertised." I've posted previously, but Nvidia advertised base spec is exceptionally conservative at 1,440MHz. On that basis, anything over 1,440 is a "bonus." However, since the reviewing and the rest of the marketing is based on Nvidia's boost algorithms sending the frequency to the high 1800's if not 1900's with spurts past 2GHz, then if the consumer can't reliably have the same performance then all advertising linked to engendering those reasonable expectations would be false or misleading.
 

Rizen

[H]F Junkie
Joined
Jul 16, 2000
Messages
9,363
There are two separate aspects to "as advertised." I've posted previously, but Nvidia advertised base spec is exceptionally conservative at 1,440MHz. On that basis, anything over 1,440 is a "bonus." However, since the reviewing and the rest of the marketing is based on Nvidia's boost algorithms sending the frequency to the high 1800's if not 1900's with spurts past 2GHz, then if the consumer can't reliably have the same performance then all advertising linked to engendering those reasonable expectations would be false or misleading.
Which is interesting in light of the claims that the MLID video made ahead of the release, which is that the FE cards were going to be intentionally misleading in performance and that partner cards would not perform the same. Hm.

That said, the boost clocks are highly variable and always have been. From the reports I've seen, people have lost 5-30MHz off peak boost frequency. That's not going to translate into meaningful performance. Has anyone reported consistent frequency losses more than that?
 

Teenyman45

2[H]4U
Joined
Nov 29, 2010
Messages
2,570
Some that I've seen are as much as 75-100MHz. Silicon lottery's gonna be painful here.

The Founder's Edition cards have a 4 and 2 capacitor arrangement and slightly higher power limits than some of the AiB's cards. An extra 20-ish watts can mean those extra few percent.
 

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
18,668
My 3DMark score didn't change at all with the new drivers (EVGA 3090 FTW normal). After 3 rounds of the old and new drivers, the scores alternated higher and lower and the scores were only separated by 20'ish points.
 

TheHig

Gawd
Joined
Apr 9, 2016
Messages
703
I would wager that there will be some design tweaks and perhaps some revisions to AIB cards following all this. Imagine two years from now trying to sell your 1st wave 3080 10GB. Is anyone going to want it? Will the hold any value? Personally I’m not buying one new or used until the teething issues are solved. For these reasons if I had purchased a launch 3080 I would return it.

My 2c.
 

Domingo

Fully [H]
Joined
Jul 30, 2004
Messages
18,668
People still buy day 1 2080Ti's. Around that launch people were making wild claims about a quarter of them being Space Invaders emulators.
 

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
9,671
People still buy day 1 2080Ti's. Around that launch people were making wild claims about a quarter of them being Space Invaders emulators.
I have one release day 2080 Ti. No space invaders here. It happened but didn't happen to me. Happily gaming since 2018!
 

polonyc2

Fully [H]
Joined
Oct 25, 2004
Messages
19,024
so it'll require a new capacitor layout for optimal performance?...Revision 2.0 of the cards?...how many of the current cards are already in production/circulation?...how long will it take for AIB partners to release new 2.0 cards?...I'm sure they will want to clear out their inventory of the older cards first

the people that are calling the defective cards 'crazy rumors' are the ones that bought a 3080/3090 and are trying to justify their purchase (praying that their card does not have any issues)
 

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
9,671
so it'll require a new capacitor layout for optimal performance?...Revision 2.0 of the cards?...how many of the current cards are already in production/circulation?...how long will it take for AIB partners to release new 2.0 cards?...I'm sure they will want to clear out their inventory of the older cards first

the people that are calling the defective cards 'crazy rumors' are the ones that bought a 3080/3090 and are trying to justify their purchase (praying that their card does not have any issues)

LOL, you go girl.
 

BassTek

Supreme [H]ardness
Joined
Jul 13, 2002
Messages
5,777
so it'll require a new capacitor layout for optimal performance?...Revision 2.0 of the cards?...how many of the current cards are already in production/circulation?...how long will it take for AIB partners to release new 2.0 cards?...I'm sure they will want to clear out their inventory of the older cards first

the people that are calling the defective cards 'crazy rumors' are the ones that bought a 3080/3090 and are trying to justify their purchase (pray that they're card does not have any issues)

Other than Zotac and Gigabyte pretty much all the other vendors have at least one MLCC array and should be fine. I don't think there will be major hardware revisions.

They might tweak drivers and firmware a bit more to stabilize boost clocks if there are still issues after the driver update.
 

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
9,671
Other than Zotac and Gigabyte pretty much all the other vendors have at least one MLCC array and should be fine. I don't think there will be major hardware revisions.

They might tweak drivers and firmware a bit more to stabilize boost clocks if there are still issues after the driver update.

Who knows? Only time will tell.

I think there's something in between these two...
1601335967013.png
1601336023701.png
 

polonyc2

Fully [H]
Joined
Oct 25, 2004
Messages
19,024
Other than Zotac and Gigabyte pretty much all the other vendors have at least one MLCC array and should be fine...

those are 2 of the biggest AIB partners...I've personally gone with Gigabyte G1 Gaming cards for my last 3 GPU upgrades
 

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
9,671
Well maybe they will add a bunch of MLCC to newer cards since everyone is an electrical engineer and wants only the best. It could be a good way to market things.

True enough. So many here are EE and GPU PCB designers...

Let's be honest. There are so many that are perpetual complainers and fan boys that stir the pot no matter what. Same thing will happen when AMD releases. Someone will cry out and the mob will pounce again.

Not saying there isn't a problem, there is. However, if it's fixed tomorrow you'll still have the same folks pumping it up for weeks. Happens all the time.
 

BassTek

Supreme [H]ardness
Joined
Jul 13, 2002
Messages
5,777
True enough. So many here are EE and GPU PCB designers...

Let's be honest. There are so many that are perpetual complainers and fan boys that stir the pot no matter what. Same thing will happen when AMD releases. Someone will cry out and the mob will pounce again.

Not saying there isn't a problem, there is. However, if it's fixed tomorrow you'll still have the same folks pumping it up for weeks. Happens all the time.

True there is a problem, if it is the boost clock and cutting by 50-100mhz solves it then I think it is minor. If cards had two weeks to a month extra to test they probably would have been limited from the factory and no one would have been wiser. Base cards boosting above 2ghz does seem to be a bit much, but I'm not an expert so what do I know.

Now if it doesn't get solved by driver/firmware then yes there are bigger issues. Too early to say yet.
 

Bman123

Supreme [H]ardness
Joined
Dec 19, 2007
Messages
4,933
So the ampere cards are like ryzen cpus and everyone is throwing a fit about it now? This is ridiculous

They could of set a hard limit in the bios and no one would of had a fucking clue and we wouldn’t be doing this right now. For some reason everyone has their panties in a bunch and wants a certain layout on their board.

Your not guaranteed 2000mhz, the boost algorithm is too aggressive and needs tweaked. That’s the issue other then people crying about it online everywhere
 

tunatime

Well...OK
Joined
Sep 15, 2011
Messages
4,138
So the ampere cards are like ryzen cpus and everyone is throwing a fit about it now? This is ridiculous

They could of set a hard limit in the bios and no one would of had a fucking clue and we wouldn’t be doing this right now. For some reason everyone has their panties in a bunch and wants a certain layout on their board.

Your not guaranteed 2000mhz, the boost algorithm is too aggressive and needs tweaked. That’s the issue other then people crying about it online everywhere
Nah the issue is Nvidia not giving time for aib to test/know what the chips can do... nvidia sets the clocks/gives them the data to build cards ect. Now it is and issue because we don't know where the floor will be set and when you look at day one reviews vs a card you buy it's might not be the same.....this could easily be seen as nvidia knew the cards couldn't take the clocks but wanted the reviews to look better then if they lowered the clocks to what is 100% stable....or it shows they are having problems with the die working the way they thought it would
 

Ebernanut

[H]ard|Gawd
Joined
Dec 15, 2010
Messages
1,350
So the ampere cards are like ryzen cpus and everyone is throwing a fit about it now? This is ridiculous

They could of set a hard limit in the bios and no one would of had a fucking clue and we wouldn’t be doing this right now. For some reason everyone has their panties in a bunch and wants a certain layout on their board.

Your not guaranteed 2000mhz, the boost algorithm is too aggressive and needs tweaked. That’s the issue other then people crying about it online everywhere

No many cards are exhibiting instability it has nothing to do with Ryzen or the fact that it boosts. It also appears that they rushed the launch and reference specs were half baked since they didn't realize this was an issue(good reason to have FE be reference but Nvidia decided to get cute with pricing/performance with the FE).

The fix might be to lower the boost which wouldn't be an issue if they had found the issue and lowered it before launch but now we have to question how accurate launch reviews are and find out what the potential performance hit is.
 
Top