• 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.

NVIDIA 4000 Series

Confirmed?

That tweet just says "can"

That would also seem like a bizarre release, especially anytime soon.
Yeah that Twitter guy should be ignored imo.
He has leaked a lot of Nvidia releases. Remember everyone doubting the Super releases? He leaked them IIRC.

Or Nvidia emergency rejigging the GTX200 series pricing and releasing the GTX260 216 and GTX275? Done at quite short notice.

If sales are not good, then it's most likely Nvidia will let the current range run down:

Already rumours of an RTX4070 production cut. In a couple of months Nvidia releases the RTX4070 Super 16GB and RTX4060 Super 16GB.

The RTX4000 series launched over six months ago.

Nvidia has then milked the ones who couldn't wait.

Nvidia had no problems screwing over it's earlier customers before.
 
Last edited:
He has leaked a lot of Nvidia releases. Remember everyone doubting the Super releases? He leaked them IIRC.

Or Nvidia emergency rejigging the GTX200 series pricing and releasing the GTX260 216 and GTX275? Done at quite short notice.

If sales are not good, then it's most likely Nvidia will let the current range run down:

Already rumours of an RTX4070 production cut. In a couple of months Nvidia releases the RTX4070 Super 16GB and RTX4060 Super 16GB.

Nvidia had no problems screwing over it's earlier customers before.

I am not saying a refresh with an AD103 16gb 4070 included isn't possible. I just doubt it is imminent (if it happens, probably end of this year/start of next)
 
Last edited:
I am not saying a refresh with an AD103 16gb 4070 isn't possible. I just doubt it is imminent (if it happens, probably end of this year/start of next)
Nvidia within months has made moves like they did with the GTX200 series. We already are half way through the year now.

Do you honestly think they were going to make an RTX4060TI 16GB? All these games using a lot of VRAM happened in the last month or two.

That is launching to retail in July apparently.

It's why the moment I saw the way they launched the series it was obvious they were targetting the ones who jump quickly onto new generations,and also the fact AMD is late with RDNA3.

This is a company who made two Pacal based Titans too and outdated the Titan with the GTX1080TI with only 1GB less VRAM.
 
Last edited:
I don't get it, why wouldn't a 4070 Super variant be based on AD104?

Is it related to the memory bus configuration of AD104?

They'd have to cut down AD103 a lot if the rumour is true.

Are 'Super' cards even a thing still? they didn't bother with them for the last gen.
 
Last edited:
I don't get it, why wouldn't a 4070 Super variant be based on AD104?

Is it related to the memory bus configuration of AD104?

They'd have to cut down AD103 a lot if the rumour is true.

Are 'Super' cards even a thing still? they didn't bother with them for the last gen.
He said AD103. AD103 has a 256 bit bus.
AD104 is a GA106 replacement so 192 bit bus, so would mean 24gb. If they put 24GB on it people might use it as a cheap rendering card and cost them sales elsewhere.

Honestly it's so weird people think Nvidia won't betray launch customers. They did it to a lot of launch Turing and GTX200 series customers. The 8800GT also did the same to 8800GTS 320MB/640MB customers within a year. Even AMD has done it before, cough, Zen4.

Also last generation there was no upselling so you had chips were they should have been.

The TI launch was there to unofficially raise RRP for the next generation because people overpaid for dGPUs. Some of us told you next generation Nvidia would try another Turing.
 
Last edited:
I don't get it, why wouldn't a 4070 Super variant be based on AD104?

Is it related to the memory bus configuration of AD104?

They'd have to cut down AD103 a lot if the rumour is true.

Are 'Super' cards even a thing still? they didn't bother with them for the last gen.

AD104 (4070Ti) with 7680 shaders already is a fully enabled core - there is nothing left to "switch on" for a Super variant.

Equally cutting down an AD103 probably isn't a great option - aside from the cost of using such a big core for a lower product, in order to keep 16GB of RAM, you can't cut the memory bus, which then means that cutting the core configs etc leads to quite an unbalanced product, possible outside of the range that it's designed to work
 
Last edited:
Hmm. with the RTX 2000 series, the 2080 Super card was slower than the respective TI model.

So, I was thinking a cut down 4070 TI could be the Super variant.
 
Last edited:
Nvidia within months has made moves like they did with the GTX200 series. We already are half way through the year now.

Do you honestly think they were going to make an RTX4060TI 16GB? All these games using a lot of VRAM happened in the last month or two.

I have no idea. Probably. They made a 12gb 3060 which had more vram than the 3060ti, 3070 and 3080.
 
I think it would be better to just cut the price of the 4070 TI, and produce a lot more :D

No need for RTX 4070 Super then.
 
Last edited:
I have no idea. Probably. They made a 12gb 3060 which had more vram than the 3060ti, 3070 and 3080.
Nvidia hates bad PR.Now it's clear games are pushing VRAM, they will do something to manage it.
If that RTX4060TI is real it's about optics. So even if it leads to an odd looking stack,they will try and rejig it now.
I think it would be better to just cut the price of the 4070 TI :D

No need for Super model then.
They didn't do that for Turing. Releasing new "better value" models means reviews have to be done again. Nvidia gets praise for listening to their customers. The older negative ones of the "old legacy models" are now history. It's all about optics.

AD104 (4070Ti) with 7680 shaders already is a fully enabled core - there is nothing left to "switch on" for a Super variant.

Equally cutting down an AD103 probably isn't a great option - aside from the cost of using such a big core for a lower product, in order to keep 16GB of RAM, you can't cut the memory bus, which then means that cutting the core configs etc leads to quite an unbalanced product, possible outside of the range that it's designed to work

But TBF,the AD103 is really the GA104 replacement,ie, similar die size and memory bus width. So the upsell is huge.

Everyone was wondering why there wasn't an SKU made from a salvaged AD103? Technically that should have been the RTX4070TI.

There is plenty of scope to release a slightly faster RTX4070TI,for similar money with a wider memory bus. The RTX2080 was only 7% faster, according to TPU, when compared to the RTX2070 Super.

Hmm. with the RTX 2000 series, the 2080 Super card was slower than the respective TI model.

So, I was thinking a cut down 4070 TI could be the Super variant.

The RTX2070 Super essentially replaced the RTX2080:

It was essentially 70% of the price of the RTX2080 and the RTX2080 was only 7% faster.

The RTX4080 is 20% faster than an RTX4070TI at qHD and the latter is 2/3 the price. So if you split the difference,and move to 16GB at the same RRP as the RTX4070TI that is your RTX4070 Super. Nvidia sells it at RTX4080 level performance at 2/3 the price.

Then push down the current RTX4070 series in price. Might also see an RTX4060 Super in the future too. Could be a slightly slower RTX4070 12GB.

This was what the RTX2060 Super was to the RTX2070! So you could see essentially:
1.)RTX4070 Super(AD103 based)
2.)RTX4070TI(AD104 based)
3.)RTX4060 Super(AD104 based)

The RTX4070 Super and RTX4070TI will be close in performance,but Nvidia gets to upsell it still because of more VRAM.

The difference between the RTX4060 Super and RTX4070TI will be greater than 20% so Nvidia gets to upsell more performance.
 
Last edited:
But what if you just invested in a 4070?...

Nvidia will make a slightly slower RTX4060 Super at some point then! Remember Turing V2? They didn't officially drop the RTX2070 pricing(IIRC) just made an RTX2060 Super which was almost an RTX2070. But you could find clearance RTX2070 cards for not much more.

Nvidia will find a way to manage this. Turing had a lot of negativity,but once the Super range came out a lot of that went away and people forgot about. You can see this as how people already have forgotten that Ada Lovelace V1 is basically the same move,after another mining boom.
 
Last edited:
I'll take a guess and say they will market a 16GB 4060ti so much above a 8GB model that one might as well pay the extra for a 4070. You want VRAM... its not going to be free. :p

So not only will these be overpriced these cards will last abit longer than a 8GB card but will suffer esp at higher res due to 128 bit memory bus / bandwidth / grunt compared to cards that could properly use it.
 
I'll take a guess and say they will market a 16GB 4060ti so much above a 8GB model that one might as well pay the extra for a 4070. You want VRAM... its not going to be free. :p

So not only will these be overpriced these cards will last abit longer than a 8GB card but will suffer esp at higher res due to 128 bit memory bus / bandwidth / grunt compared to cards that could properly use it.

The along comes Hogwarts Legacy:The Second Coming which needs 12.1GB of VRAM at 1080p,and suddenly the RTX4060TI beats the RTX4070! :P
 
  • Haha
Reactions: G J
Nvidia hates bad PR.Now it's clear games are pushing VRAM, they will do something to manage it.
If that RTX4060TI is real it's about optics. So even if it leads to an odd looking stack,they will try and rejig it now.

They didn't do that for Turing. Releasing new "better value" models means reviews have to be done again. Nvidia gets praise for listening to their customers. The older negative ones of the "old legacy models" are now history. It's all about optics.



But TBF,the AD103 is really the GA104 replacement,ie, similar die size and memory bus width. So the upsell is huge.

Everyone was wondering why there wasn't an SKU made from a salvaged AD103? Technically that should have been the RTX4070TI.

There is plenty of scope to release a slightly faster RTX4070TI,for similar money with a wider memory bus. The RTX2080 was only 7% faster, according to TPU, when compared to the RTX2070 Super.



The RTX2070 Super essentially replaced the RTX2080:

It was essentially 70% of the price of the RTX2080 and the RTX2080 was only 7% faster.

The RTX4080 is 20% faster than an RTX4070TI at qHD and the latter is 2/3 the price. So if you split the difference,and move to 16GB at the same RRP as the RTX4070TI that is your RTX4070 Super. Nvidia sells it at RTX4080 level performance at 2/3 the price.

Then push down the current RTX4070 series in price. Might also see an RTX4060 Super in the future too. Could be a slightly slower RTX4070 12GB.

This was what the RTX2060 Super was to the RTX2070! So you could see essentially:
1.)RTX4070 Super(AD103 based)
2.)RTX4070TI(AD104 based)
3.)RTX4060 Super(AD104 based)

The RTX4070 Super and RTX4070TI will be close in performance,but Nvidia gets to upsell it still because of more VRAM.

The difference between the RTX4060 Super and RTX4070TI will be greater than 20% so Nvidia gets to upsell more performance.

The 2070 super came out 9 months after the 2070/10 after the 2080.

Again, i'm not saying there will not be a refresh or a 4070 super because there is always something new around the corner.

I just doubt it will be in the next couple of months....The 4070 came out less than a month ago!
 
Last edited:
AD104 (4070Ti) with 7680 shaders already is a fully enabled core - there is nothing left to "switch on" for a Super variant.

Equally cutting down an AD103 probably isn't a great option - aside from the cost of using such a big core for a lower product, in order to keep 16GB of RAM, you can't cut the memory bus, which then means that cutting the core configs etc leads to quite an unbalanced product, possible outside of the range that it's designed to work
They'll have to do something because most people simply won't pay the higher asking prices for cards which have a poorer generational uplift compared to the past, £1200 for a 4080 which is only 17% faster than last gen while for the same price previous you got a 3080ti which was 40% faster than previous gen. The 4070ti is 7% slower than last gen for £800 yet the 3080 was 25% faster for £650, so your basically paying £150 more for 30% less performance than you got previously, thats a big swing and its the same for the 4070, +£100 more for 30% less performance.
 
The 2070 super came out 9 months after the 2070/10 after the 2080.

Again, i'm not saying there will not be a refresh or a 4070 super because there is always something new around the corner.

I just doubt it will be in the next couple of months....The 4070 came out less than a month ago!

The report already said RTX4070 production is being reduced. I said it might take a few months to run through what is already made. So three to four months,ie,over summer. That makes it four to five months after launch,and nearly a year after the RTX4000 series was launched. Normally you could argue just reducing prices by a large amount might work,but remember Nvidia literally released the whole Super range just not to cut the legacy pricing. The one time they did it with the GTX260,they had to send rebates to older customers and it was bad PR.
 
Last edited:
Back
Top Bottom