Do you want to pay less for a better RTX4070?

[dirtyferret]

you may but Nvidia believes you would actually prefer to pay the same for a less capable RTX4070

Nvidia is set to introduce a new version of its RTX 4070 that uses slower GDDR6, rather than GDDR6X memory, and the first of those cards has been spotted revealing that despite having slightly slower memory, the new card won’t be any cheaper than existing models. The new Nvidia RTX 4070 V2 GDDR6 cards are listed as having the same $549 MSRP as other RTX 4070 cards.

Nvidia’s new slower RTX 4070 arriving without a price drop, says leak

The first of a new wave of RTX 4070 V2 graphics cards has been spotted and despite having slower memory, they haven't had a discount.

www.pcgamesn.com

 

[Dr. Dro]

This has been known for some time, and I do not believe this is a significant change, not in this segment. They're using 20 Gbps G6, which has a lower access latency and should be able to close that bandwidth gap easily enough should the user feel like it. It turns out to be 480 GB/s rather than 504 GB/s, it's not a major drop in bandwidth and if anything, the G6 model should enjoy lower memory temps and since G6 tends to use less power, more room for the core to boost within the given TGP. It's a complete nothingburger, hopefully W1zzard will be able to score a review sample.

There is a shortage of 21 Gbps G6X chips due to extreme demand for the RTX 4090, which sells as fast as Nvidia can make them. Other popular SKUs like the 4070/Ti Super are also using the same chip, so it makes sense to shave at the lowest segment. The 4080 Super uses the newer 24 Gbps G6X module that debut with the original 4080 (which is no longer in production), so they aren't affected and you shouldn't expect any revision for this particular SKU.

 

[Assimilator]

If @Dr. Dro is correct then it's only a 5% drop in memory bandwidth, so maybe the right thing for NVIDIA to do would be to drop the overall price by 2.5%, to ~$535... which is also so small as to not really be meaningful, so yeah, I agree that there's no reason for NVIDIA to bother.

 

[chrcoluk]

Not long ago was a news story Nvidia had to reject a batch of GDDR6X as it failed quality testing.

 

[wNotyarD]

This has been known for some time, and I do not believe this is a significant change, not in this segment. They're using 20 Gbps G6, which has a lower access latency and should be able to close that bandwidth gap easily enough should the user feel like it. It turns out to be 480 GB/s rather than 504 GB/s, it's not a major drop in bandwidth and if anything, the G6 model should enjoy lower memory temps and since G6 tends to use less power, more room for the core to boost within the given TGP. It's a complete nothingburger, hopefully W1zzard will be able to score a review sample.

There is a shortage of 21 Gbps G6X chips due to extreme demand for the RTX 4090, which sells as fast as Nvidia can make them. Other popular SKUs like the 4070/Ti Super are also using the same chip, so it makes sense to shave at the lowest segment. The 4080 Super uses the newer 24 Gbps G6X module that debut with the original 4080 (which is no longer in production), so they aren't affected and you shouldn't expect any revision for this particular SKU.

While I agree that the performance drop shouldn't be major in this GPU tier, I'll agree with @Assimilator.
If NVIDIA wants to cut corners so their costs drop, we want their prices to drop as well to match.

 

[the54thvoid]

If @Dr. Dro is correct then it's only a 5% drop in memory bandwidth, so maybe the right thing for NVIDIA to do would be to drop the overall price by 2.5%, to ~$535... which is also so small as to not really be meaningful, so yeah, I agree that there's no reason for NVIDIA to bother.

A %perf delta isn't reflected in any %shift in pricing. The 4090 is proof of that as a halo product. The product stacks are given arbitrary pricing based on what the company expects people are willing to pay. Let's not pretend it's okay to sell an inferior version (no matter the perf drop) at the same price. Even a token reduction would be a symbolic acknowledgement. If this is an accuarte news piece, then NV are flipping the finger to the consumer for a few extra dollars.

Edit: and they'll sell enough to rake in a large profit from the change.

 

[Assimilator]

If this is an accuarte news piece, then NV are flipping the finger to the consumer for a few extra dollars.

I'm not arguing that; but compared to the way they've been flipping the finger to consumers for the past 3 generations, this is a pretty minor one.

 

[Dr. Dro]

There's always the chance this one will also be targeted towards the Chinese market too (a la RTX 4090 D), as they are much more receptive of alternate SKUs overall there.

Until there's official word from Nvidia it's all speculation

 

[dirtyferret]

so maybe the right thing for NVIDIA to do would be to drop the overall price by 2.5%, to ~$535

I could see this if the card came out in April of 2024 but it came out in April of 2023. Make the MSRP $499 just so you don't have the headlines of web sites complaining.

I'm not arguing that; but compared to the way they've been flipping the finger to consumers for the past 3 generations, this is a pretty minor one.

doesn't mean we can't or should not call them out for it

 

[N/A]

Although worded confusingly. I'd totally take a better 4070, for example, the one and only Super 7168 Cuda 80 Rops for less money. Personally, I wouldn't mind a 4070 GDDR6 for the same money if I were in the market for a new card, but this is late in the cycle. No, I expect 5070 for the same, but that's still 9 months from now.

 

[ShrimpBrime]

Either way, it's fast as Fk compared to a 4060 ti, by like 50% faster, faster. Around or at the same price with 6X Vram, meh nothing to haggle over really. It should be priced nearly the same. It's well within it's class.

 

[wheresmycar]

I wouldn't be surprised if V2's slower memory will have much impact on diminishing game performance and there is the added benefit of less power/heat. For Nvidia thats a "hope-they-shut-up-and suck-it-up" "UPGRADE" as long as people are willing to fork out the pocket lettuce.

 

[eidairaman1]

Go on Social Media and Youtube and call them out on a video

 

[Knight47]

Yes, I want to pay less for a better 4070. But 4070 Super prices started to go up(Gddr6x shortage?) and TiS costs almost 50% more.

 

[redeye]

YES, bring back the 250us or 330can Graphics card. or the 4080 for the same price as a console (ps5pro) or 770can or 550us.

 

[Dr. Dro]

YES, bring back the 250us or 330can Graphics card. or the 4080 for the same price as a console (ps5pro) or 770can or 550us.

I think we can confidently say that $550 flagships are never coming back. Not even $650 flagships are coming back. They'll be hanging around $1-1.5K for the foreseeable future, if a hike to $1999 isn't necessary by the next node due to market conditions and the economy's health. The bill of materials, enormous research and development as well as software development costs make that impossible. Values in USD, of course.

 

[dgianstefani]

I think we can confidently say that $550 flagships are never coming back. Not even $650 flagships are coming back. They'll be hanging around $1-1.5K for the foreseeable future, if a hike to $1999 isn't necessary by the next node due to market conditions and the economy's health. The bill of materials, enormous research and development as well as software development costs make that impossible. Values in USD, of course.

Ended when consumer GPUs became viable for making money. Whether through mining or AI, it doesn't matter.

 

[the54thvoid]

Ended when consumer GPUs became viable for making money. Whether through mining or AI, it doesn't matter.

I hate this but it's true.

 

[Dr. Dro]

Yeah, that too. It became less of a component and more of a tool, which serves as justification for a lot of people to bear that extra cost

 

[dgianstefani]

I hate this but it's true.

Man you could probably say this about a lot of my posts

 

[Ruru]

That should be called 4060 Ultra or 4070 SE or something...

I think we can confidently say that $550 flagships are never coming back. Not even $650 flagships are coming back. They'll be hanging around $1-1.5K for the foreseeable future, if a hike to $1999 isn't necessary by the next node due to market conditions and the economy's health. The bill of materials, enormous research and development as well as software development costs make that impossible. Values in USD, of course.

1080 Ti was really last of its kind, dammit.

 

[Dr. Dro]

1080 Ti was really last of its kind, dammit.

Technically the 3080 ($649 MSRP), although we all know what happened that gen...

 

[Ruru]

Technically the 3080 ($649 MSRP), although we all know what happened that gen...

I bought mine used 4 months ago for 450EUR (with a fullcover block), came with the original receipt, it was bought about the time when it was released, cost 729EUR for the first buyer.

But I agree, 3080 for its MSRP had hella good bang for the buck.

 

[wheresmycar]

1080 Ti was really last of its kind, dammit.

Yep, that was the last somewhat commendable buy in that upper performance segment. I say "somewhat" because back then it did seem expensive but overtime the 1080-TI lived up to the game... a bad-ass road-worthy warrior built to last. Post 1080-TI i've had a tough time parting with my hard earned cash to justify the splurge. Its not a matter of affordability but value-driven circumstantial sensibility. Both of my current cards, namely 2080 TI and RTX 3080, long after release were heavily discounted and purchased "used" from trusted/familiar faces. I don't like buying used but it feels way better than feeding the ravenous beast our hard earned cash for the new.

 

[Ruru]

Yep, that was the last somewhat commendable buy in that upper performance segment. I say "somewhat" because back then it did seem expensive but overtime the 1080-TI lived up to the game... a bad-ass road-worthy warrior built to last. Post 1080-TI i've had a tough time parting with my hard earned cash to justify the splurge. Its not a matter of affordability but value-driven circumstantial sensibility. Both of my current cards, namely 2080 TI and RTX 3080, long after release were heavily discounted and purchased "used" from trusted/familiar faces. I don't like buying used but it feels way better than feeding the ravenous beast our hard earned cash for the new.

Practically 1080 Ti was the best card back then, a good custom one overclocked better than a Titan X/Xp so it didn't matter even when it had some shaders and mem/membus cut.

even after 7 years, it still holds up.