Current Events > The 12 GB 4080 isn't even xx70 quality, let alone xx80

(+) Yeah! (+)

I've been reading the specs, and the 12 GB 4080 has a 192-bit bus, which is a spec NVIDIA has traditionally used for xx60 cards (1060, 2060, 3060). xx70 and xx80 cards typically get a 256-bit bus.

Fucking embarassing, NVIDIA

The bus speed isnt the end all of specs... how does it actually proform?

lol

The 4080 12GB is fucking sad. Imagine having only the same bus width as a low tier card. They're really banking on PCIe 5.0, but none of the cards support it yet despite motherboards having it.

ZangsBeard posted...

The bus speed isnt the end all of specs... how does it actually proform?

They're pushing base clock speeds and boost clock speeds a lot. It's why I want to see what the performance is in rasterization. DLSS is good, but it's a smokescreen right now.

https://gamefaqs.gamespot.com/a/user_image/3/9/8/AARLwzAADtKG.jpg

When my laptop GPU has a wider bus than the 4080

Also this is the first generation where the xx90 isn't the full die. Which means it's probably more like:

4090 = 4080
4080 16 GB = 4070
4080 12 GB = 4060

Leftover 3080s are gonna be rebranded as 4060s watch. Nvidia has been caught doing this before. The only difference between a 1080 and a Quadro was what the card thought it was so if you had enough tech skill you could flash the Quadro signature over the GTX and save like $3000 per card.

They fixed that, though.

Tyranthraxus posted...

Leftover 3080s are gonna be rebranded as 4060s watch. Nvidia has been caught doing this before. The only difference between a 1080 and a Quadro was what the card thought it was so if you had enough tech skill you could flash the Quadro signature over the GTX and save like $3000 per card.

They fixed that, though.

I doubt they could do that by virtue of the fact that the 3080 doesn't have the architecture to support dlss 3, although honestly I wouldn't put it past Nvidia at this point

The 4080 12GB is the Engar-dug of graphics cards

DarkRoast posted...

I doubt they could do that by virtue of the fact that the 3080 doesn't have the architecture to support dlss 3, although honestly I wouldn't put it past Nvidia at this point

Nvidia is the type of company to put it in their driver to automatically downscale you to DLSS2 without telling you if you select DLSS3 on a card that doesn't support it.

I watched a hardware impressions of the 4000 series and one of the things they mentioned is that dlss3 actually is not as advanced as dlss 2, because it's actually a more traditional previous frame upscaling technique. So they suspect DLS S3 will be less about improved visuals and more about significantly improved frame rates.

Does it have more TFLOPS than Titan X?

World: Whew thank goodness the chip shortage is over. Back to norm-

Jensen: Can't let you do that, Star Fox

Tell you what, these cards are going to be a bitch to make mobile variants of. With power requirements of 400 watts, good luck cutting that down.

DarkRoast posted...

When my laptop GPU has a wider bus than the 4080

It's called, Jensen wants to be a "Tightwad" and charge you a ridiculous amount so he can enjoy his "Phat Profit Marginz".

Why else would he be charging you the price of a old ##80 for a Hybrid Bastard child of a ##70 & ##60 model.

Imagine what Jensen Huang is going to charge you for the REAL 4070, 4060, etc.

DarkRoast posted...

World: Whew thank goodness the chip shortage is over. Back to norm-

Jensen: Can't let you do that, Star Fox

Tell you what, these cards are going to be a bitch to make mobile variants of. With power requirements of 400 watts, good luck cutting that down.

Imagine how "Stingy" he's going to get with the Mobile variants.

DarkRoast posted...

Also this is the first generation where the xx90 isn't the full die. Which means it's probably more like:

4090 = 4080
4080 16 GB = 4070
4080 12 GB = 4060

You forgot that he's saving those pristine Dies for the 4090 Ti/Super & 4080 Ti/Super variants.

Here's the original Rumored Specs that got confirmed.
https://gamefaqs.gamespot.com/a/user_image/9/8/8/AAAFvAAADskc.jpg

Straight from nVIDIA's website:
https://gamefaqs.gamespot.com/a/user_image/5/4/4/AAAFvAAADtMY.png
The original rumors for the SKU specs was confirmed and 99.9% accurate.

But nVIDIA is holding out on the "Full Dies" for AD102 & AD103
https://gamefaqs.gamespot.com/a/user_image/5/4/5/AAAFvAAADtMZ.jpg
That means the

• 4090 Ti/Super = 144 SM's.
• 4090 _Vanilla = 128 SM's
• 4080 Ti/Super = _80 SM's.
• 4080 _16 GB = _76 SM's

Here's the rest of the lineup:

https://gamefaqs.gamespot.com/a/user_image/5/4/6/AAAFvAAADtMa.jpg
AD106 = 36 SM's for a "Perfect Die"
AD107 = 24 SM's for a "Perfect Die"

We'll see how much nVIDIA has to cut down for binning before releasing the "Perfect Die's" as the Ti/Super variants

The 40 series is for 4K enthusiasts anyways. No one that games at 1440p should even consider a 40 series card if they're smart. Now that the 30 series is affordable and in stock there's no good reason to not upgrade to a 30 series for anyone that's been holding out for stock to be more frequent.

Kamen_Rider_Blade posted...

It's called, Jensen wants to be a "Tightwad" and charge you a ridiculous amount so he can enjoy his "Phat Profit Marginz".

Why else would he be charging you the price of a old ##80 for a Hybrid Bastard child of a ##70 & ##60 model.

Imagine what Jensen Huang is going to charge you for the REAL 4070, 4060, etc.

Imagine how "Stingy" he's going to get with the Mobile variants.

The dude is a grade A douchebag. It's kind of shameful that you can get more vram on a laptop 3080 than you can on the entry 4080. Considering laptops tend to max out at 1440p.

The 3080 10 GB model got slack for having a relatively low amount of vram for a 4K gpu. To only go up two gigabytes is pretty insulting.

DarkRoast posted...

The dude is a grade A douchebag. It's kind of shameful that you can get more vram on a laptop 3080 than you can on the entry 4080. Considering laptops tend to max out at 1440p.

The 3080 10 GB model got slack for having a relatively low amount of vram for a 4K gpu. To only go up two gigabytes is pretty insulting.

This is before we even get to Ti / Super Variants.