Zappcatt

javelina1

Zappcatt

lasttimei
really want the 3080 but i'll probably have to take the 3751 , for now hope it drops today, since the last 3751 was "3070 XC3 Black 3751 10/29/2020 06:09:41".
 
10/29/2020 6:10:02 AM
 

HOLY SMOKES, you were so close!
I hope they catch you with todays drop!

first movement on that SKU in over several weeks.  good to see.    I'm still a minute out.... which most likely will be another couple months or more.     But hang tight!
 
Mine: 
10/29/2020 6:11:24 AM PTNo

Wow, so you did not catch the movement today. Bummer. It looks like lasttimei might have been the cutoff.
I am new here, so might be totally stupid, but I am hoping that the queue will start moving much faster due to 2 issues.
1) Much greater supply as the COVID delays and uncertainty get further in the past.
2) People who signed up to the Queue receive product from other vendors.

Fingers crossed.
P.s.  A local reseller has a few 3070s in stock, but they are $80 above the NEW prices. They say their distributor is jacking up the prices, but they did not mind jacking them up again when the supply chain prices increased. They had 3070s and 3080s for $90/$100 over MSRP. If I had paid attention and knew of the increase in price coming up, I would have probably grabbed one :-(

 
.


 
I did receive the notification 1/13/21
08G-P5-3751-KR10/29/2020 6:10:02 AM PTYes
 

Oh also: Thank you to whoever used my Associate's code to buy a 3080, I appreciate it :-D You are funding what will likely be a Z590 motherboard purchase if I decide to go Intel instead of AMD with my next upgrade... which, given I can get a eVGA motherboard using eVGA Bucks, chances are I'll go the 11900K route because it'll end up costing me less... thusly making the resale of my 9900K/Z390 combo more... well, less "unprofitable" at least. lol
 

Goodnight to everyone. and good luck tomorrow. I should receive my 3080 tom. Wait is finally over! now If i get a noty for the 3080 hybrid Idk, damn tuff one. I will see.

philipma1957

Any one waiting for the  3060ti.  3667 model get an email today.  

 

I am 2 hours and 15 minutes at 910 am

 

ablatman you were at 7:01

We're getting to the point where there aren't enough users reporting back on the queues to give the kind of transparency that we've had to this point. The 3667 definitely appears to be the first card that it's happening to--last week it dropped on Thursday (~1 minute of verified movement) and again on Friday (3+ minutes of verified movement). Today it moved again, presumably from the same drop (3+ minutes of verified movement).

 

Usually we see the biggest movement on the first day of a drop, and then progressively less as the drop goes on over the next several days. The movement that we've seen on this queue makes it clear that we don't have enough datapoints to reliably guess how far the queue is moving on any given day.

 

Dabadger84My brain does not comprehend how it's possible that a month of production resulted in 8-15 units, even if they're holding that many back for RMAs... 30 in a month?  There's no way Kingpin yield is THAT low, that'd be ridiculous... not to mention the first two drops were over 100 cards. Each.

The Kingpin is an uber prestige variant of a prestige product. They likely only use the highest bin of GPU chip, along with the best RAM they can get, for that product. The ~200 units from the initial drops makes it sound like they'd been holding those chips for Kingpin production from basically the beginning, and then shipped months worth of production over in the first two shipments.

 

Also, it's possible that there are a lot of people in the queue who aren't reporting their positions, and/or don't actually intend to buy the card if they get notified.

Dabadger84

gvarv

Great stuff! I think what weirds me out with the 3080 PCB is that there appears to be two empty solder points for RAM, which makes me think (possible incorrectly) that the card could support 12 GB. Oh well, idly musings about cards I can't get without selling out to some scalper.

The 3080 & 3090 PCB for non-FE cards are usually identical, which means the 3080 has all the vRAM spots the 3090 needs, they're just empty... so they definitely could've put more vRAM on it pretty easily.  They just didn't.

No, it has to do with how they're made/designed. The really short version is that the 3080 is a 3090 with a bunch of bits that have been permanently disabled because they either don't work, or because they perform sigificantly worse than the rest of the GPU.

There are two ways to handle that: Either they design specific board revisions for every single possible combination of disabled components (so that the resulting boards will always be fully populated), or they could design a single board revision that's compatible with every GPU die (so that the boards will always have some areas that are unpopulated).

Nvidia and their board partners went the latter route, because it's significantly cheaper and more efficient. That means that every 3080 has 1/6 of it's VRAM attachement points left empty, because the parts of the GPU die that would talk to those memory chips have been permanently disabled.

ZappcattWow, so you did not catch the movement today. Bummer. It looks like lasttimei might have been the cutoff.
I am new here, so might be totally stupid, but I am hoping that the queue will start moving much faster due to 2 issues.
1) Much greater supply as the COVID delays and uncertainty get further in the past.
2) People who signed up to the Queue receive product from other vendors. <snip>

2 is definitely already happening. Early on, drops would get swallowed up almost whole on the first day, with no movement beyond the stragglers on the second day. Now drops are regularly taking 4-6 days to work through the available stock.


1 isn't really happening yet though, since a big part of what's driving the demand side is the increased number of people that need hardware in order to work and/or learn remotely (and these numbers aren't likely to go back down quickly). Supply side increases are reportedly in the works, but will likely take a few more months before we really start to feel their effects. It's likely that we'll see both halves of the current dynamic persist until around when schools start letting out in a few months.

Dabadger84<snip>& Quality DLSS @ 3440 x 1440</snip>

If you're using DLSS, you're not actually rendering at the display resolution. If you run DLSS in all games that support it, that's likely a big part of why the RAM usage you're seeing is as low as it is.

Adiyaaa1234567
me seeing that the hybrid 3090 que progressed an hour and a half.
me who is 1 minute and 30 seconds away for the 3080 hybrid:( ͡° ͜ʖ ͡°)

That's not really saying much. The 3090 hybrid variants didn't have a thick queue.

https://docs.google.com/s...6MaACCpobW4E/htmlview#

If you look at the 3080 hybrids you will see a huge difference in the number of cards.

lasttimei
 
 
I did receive the notification 1/13/21
10/29/2020 6:10:02 AM PTYes
 

Awesome, grats!

Brin_

Dabadger84

gvarv
Great stuff! I think what weirds me out with the 3080 PCB is that there appears to be two empty solder points for RAM, which makes me think (possible incorrectly) that the card could support 12 GB. Oh well, idly musings about cards I can't get without selling out to some scalper.

The 3080 & 3090 PCB for non-FE cards are usually identical, which means the 3080 has all the vRAM spots the 3090 needs, they're just empty... so they definitely could've put more vRAM on it pretty easily.  They just didn't.

No, it has to do with how they're made/designed. The really short version is that the 3080 is a 3090 with a bunch of bits that have been permanently disabled because they either don't work, or because they perform sigificantly worse than the rest of the GPU.

There are two ways to handle that: Either they design specific board revisions for every single possible combination of disabled components (so that the resulting boards will always be fully populated), or they could design a single board revision that's compatible with every GPU die (so that the boards will always have some areas that are unpopulated).

Nvidia and their board partners went the latter route, because it's significantly cheaper and more efficient. That means that every 3080 has 1/6 of it's VRAM attachement points left empty, because the parts of the GPU die that would talk to those memory chips have been permanently disabled.

Dabadger84<snip>& Quality DLSS @ 3440 x 1440</snip>

If you're using DLSS, you're not actually rendering at the display resolution. If you run DLSS in all games that support it, that's likely a big part of why the RAM usage you're seeing is as low as it is.

The PCB thing is accurate, it's the same PCB in the case of the cards we're talking about.  NVidia's FE designs are quite different, the 3090 is a larger PCB etc, but on most AIB maker cards, the PCBs for both SKUs are the same base layout.  The GPU and other things are different, and yes, the 3080 is a 3090 die with some parts disabled... but the PCB of the FTW3 Ultra for both the 3080 & 3090 are identical, just some traces are absent because the extra vRAM is not present, etc.
This is confirmed in PCB/VRM/etc reviews that have been done on the cards, in particular the eVGA Models. 
 
As for the DLSS Quality bit, the Quality setting specifically results in better image clarity with 2.0, especially in Cyberpunk 2077, look it up.  You can actually read/see things with DLSS Quality on that you CANNOT read/see at native resolution.  vRAM usage is higher with it off, still, but it's not a gigantic difference:
 
DLSS Quality on: (note the warning signs in the middle left are readable)
 

 
DLSS Off: 
 

 
And of course there's the whole... 28 vs 48 FPS.  I didn't pick the best "scene" to do a comparison of because there's not a lot of distant detail that DLSS makes "better", but there's consistent examples of it in both Hardware Unboxed & GamersNexus' review of DLSS in Cyberpunk 2077.
 
Yes, it's still rendering at a lower resolution then using the technology to antialias and upscale it, but the Quality setting is really the best of both worlds, visually clear & nice AA, and better performance at the same time.
 
Also I wouldn't call 7.4-9.4GB of vRAM usage "low" usage. lol
 
 

gvarv

So Brin_ in order to have a 3080 class card with more RAM we would need either a redesign with higher density RAM pieces or a GPU that hasn't been neutered?

The 3080 has 10 VRAM chips sharing a 320-bit bus (all 10 chips on the "front" of the board with the GPU). The 3090 has 24 VRAM chips sharing a 384-bit bus (12 on each side of the board). The simplest way to think of that is basically that the 3080 has 5 memory channels with 2GB each, while the 3090 has 6 memory channels with 4GB each. The rumored specs for the rumored 20GB 3080 called for it to have 5 memory channels with 4GB each.

Dabadger84The PCB thing is accurate, it's the same PCB in the case of the cards we're talking about.  NVidia's FE designs are quite different, the 3090 is a larger PCB etc, but on most AIB maker cards, the PCBs for both SKUs are the same base layout.  The GPU and other things are different, and yes, the 3080 is a 3090 die with some parts disabled... but the PCB of the FTW3 Ultra for both the 3080 & 3090 are identical, just some traces are absent because the extra vRAM is not present, etc.

This is confirmed in PCB/VRM/etc reviews that have been done on the cards, in particular the eVGA Models.

It's a bit pedantic, but the traces aren't absent from the board itself, they're just unused. The traces run from each VRAM chip BGA to the GPU BGA, regardless of whether there is going to be a VRAM chip on that spot. I strongly suspect that there is an electrical connection all the way back to the GPU die itself for every VRAM position on the board, even those that go unused--doing anything else would add significantly to the complexity of manufacturing, while offering zero net benefit.

Dabadger84As for the DLSS Quality bit, the Quality setting specifically results in better image clarity with 2.0, especially in Cyberpunk 2077, look it up.

I think we're communicating poorly. The discussion was about whether or not there are circumstances where modern games will realistically use more than 10GB of VRAM.

I wasn't saying "DLSS = bad" I was just pointing out that using it likely has an impact on the VRAM usage that you're seeing.

Don't get me wrong: I'm entirely open to the notion that 10GB of VRAM might not end up being a bottleneck for these cards long-term, particularly as the graphical fidelity of games continues to increase--We're really at the dawn of GPUs truly powerful enough to run 4k at playable FPS, and as time goes on 30 series cards will likley need to leverage DLSS in order to tread water at 4k vs whatever GPUs come next.

Good morning, fellow hunters!

....

Whomever is in charge needs to put an end to this so they can catch up and take care of there current customers and meet there other obligations this is ridiculous. There comes a point where this program which started out as a good thing is now just digging EVGA a hole they cannot dig out of as More and more people realize they will not be getting a card anytime soon like next 6 months ..sigh

Im pretty sure it does reset you unfortunately

skleidnbfoies
Does anyone know the original pricing for the 08G-P5-3663-KR and 08G-P5-3667-KR before the price hike? I am still in the queue so it should be the original pricing but I can't seem to find it anywhere...

IIRC they were ~400 and 450 respectively.

xkaosx

Jflo38
Whomever is in charge needs to put an end to this so they can catch up and take care of there current customers and meet there other obligations this is ridiculous. There comes a point where this program which started out as a good thing is now just digging EVGA a hole they cannot dig out of as More and more people realize they will not be getting a card anytime soon like next 6 months ..sigh

Are you allowed to change your step up card? Currently I had it at a 3080 ftw ultra but now want the 3080 ftw ultra hybrid- would that reset me in "line"?
 

First, that card isn't even available through step-up. For a list of what you can currently step up to, see .

Second, If you're within 90 days of your purchase date, you can step out of one step-up queue, and then step into another. For example, if you bought a 3060ti on launch day, and immediately got in line for the 3070, you could still switch to a 3080 or 3090. That said, you'd be starting at the back of the line again. If you're PAST 90 days, you can cancel your current step-up, but you can't start a new one.

 


First, that card isn't even available through step-up. For a list of what you can currently step up to, see .

Second, If you're within 90 days of your purchase date, you can step out of one step-up queue, and then step into another. For example, if you bought a 3060ti on launch day, and immediately got in line for the 3070, you could still switch to a 3080 or 3090. That said, you'd be starting at the back of the line again. If you're PAST 90 days, you can cancel your current step-up, but you can't start a new one.




Well I am like 30 seconds away for the queue line for the 3080 hybrid so that will prob come first too
 

Im saying they are scamming people by continuing to take step ups and adding people to ques when they are so far behind that they really have no ability to step people up in a timely manner. When you go through the step up process they state we can not give people a time of when they will get there cards....ok but almost every single person I have talked to has been like "ok well ill have to wait a month or 2" more like 6 months lol they have not even been able to fulfill orders from the first 10 days of the launch i mean they are still working on people who qued up right at the beginning. They have take thousands of step ups and ques at this point. Realistically even if they increase production (in reality it has slowed if anything) people will be waiting for cards for 6 months if not longer maybe even much longer.
That's misleading people wether they have a disclaimer saying they cannot give any timeliness or not. Misleading people giving you money is scamming people in my opinion...shrug...

Pardon me if im venting and forgive my grammar and mechanics im typing on a broken cell phone