Nvidia GeForce RTX 50-series product value

Boss

Regular
Newcomer
Nvidia GeForce RTX 5070 Ti review: 4080 territory, or more with an overclock A decent value - if MSRP cards are properly available

Here we go again with Digital Foundry and their interesting takes on GPU reviews. The review starts off like this:
The RTX 5070 Ti is here - and it's not a bad value, actually. In fact, with a small bump to raw performance and the inclusion of multi frame generation for $50 less than the previous generation card, it may be the best performing GPU you can get for $750... assuming that it's actually available at its MSRP for more than a fleeting moment at launch.
It's a bad value actually and objectively even at msrp. We are in an era where the new xx70ti card can't even beat the last gen xx80 card. In the past, the new xx70 card would be beating the last gen xx80 card so, to go from that to this is laughable. The 5070ti is 7% on average faster than a 4070ti super.

I don't think I've read a more biased review today that this one. The funny part is that even before the reviews, the product launch, we knew availability at MSRP would be scarce. Pricing was leaked and even if you didn't see the leaks, it'd be logical to assume that there is a very low chance that availability at MSRP would be great. Shortly after the announcement of these products and pricing, Trump announced tariffs for an additional 10% on products from China. Anyone who has been paying attention for the last 5 years already knew that this would have an impact on the Western pricing of good regardless of nation. Combine that with the historical unfavorable exchange pricing for electronics, plus the predicted lack of availability and it's certainly not a recipe to hit MSRP.

At this point, Eurogamer/DF should just stop sharing their opinions in their GPU "reviews" because it's clear that it doesn't align with reality at all. They should just publish the data and stick to the facts if they're unwilling to provide an unbiased review of the product. The amount of mental gymnastics that goes on in their reviews ever since they began their relationship with Nvidia is just too much.
 
TPU and HUB are my mainstays for GPU benchmarks and I don’t think I’ve seen such vastly different results before. They bench quite a few of the same games and had very different numbers for the 5070 Ti.

HUB removed the 30Mhz factory overclock on the 5070 TI which is basically nothing and TPU’s 4070 Ti super results are on a slightly older driver. Nothing that can explain the huge differences.

5070 Ti vs 4070 Ti Super @ 4K

Counter strike
TPU: 25%
HUB: 10%

Cyberpunk
TPU: 29%
HUB: 18%

Veilgard
TPU: 23%
HUB: 11%

Ragnarok
TPU: 19%
HUB: 17%

Stalker 2
TPU: 20%
HUB: 5%

Starfield
TPU: 14%
HUB: 4%

Outlaws
TPU: 28%
HUB: 6%

Hogwarts:
TPU: 20%
HUB: 14%

Last of Us
TPU: 22%
HUB: 12%
 
TPU and HUB are my mainstays for GPU benchmarks and I don’t think I’ve seen such vastly different results before. They bench quite a few of the same games and had very different numbers for the 5070 Ti.

HUB removed the 30Mhz factory overclock on the 5070 TI which is basically nothing and TPU’s 4070 Ti super results are on a slightly older driver. Nothing that can explain the huge differences.

5070 Ti vs 4070 Ti Super @ 4K

Counter strike
TPU: 25%
HUB: 10%

Cyberpunk
TPU: 29%
HUB: 18%

Veilgard
TPU: 23%
HUB: 11%

Ragnarok
TPU: 19%
HUB: 17%

Stalker 2
TPU: 20%
HUB: 5%

Starfield
TPU: 14%
HUB: 4%

Outlaws
TPU: 28%
HUB: 6%

Hogwarts:
TPU: 20%
HUB: 14%

Last of Us
TPU: 22%
HUB: 12%
Are they running like for like tests? If they're not, you're comparing unlike data.
 
Are they running like for like tests? If they're not, you're comparing unlike data.

All reviews aim to answer the same question - how much faster is the new card. With that much variation in the same games you’re going to get vastly different answers depending on what you read/watch.
 
No, "objectively" it's not. You get $1000 4080/S performance for $750.

Maybe objectively was being used ironically. The 5070 Ti is the best value card today for $750 which is Rich’s (obvious) point. If the argument is that the best value card is still poor value and nobody should buy GPUs anymore well that’s not going to happen.
 
No, "objectively" it's not. You get $1000 4080/S performance for $750.
Yea for sure, lets just ignore the fact that the 7900xtx could be had for $819.99 on prime day a mere 5 months ago.... If I wanted that level of performance at a relatively similar price point, I could have had it a while ago. The 5070ti is bad value for money period.
 
Yea for sure, lets just ignore the fact that the 7900xtx could be had for $819.99 on prime day a mere 5 months ago...
So a card with worse performance could've been had for more than the new card's price which is why it is "objectively" bad value?
I'm at a loss trying to understand whatever you're saying.

Last time you got $2000 RTX 3090 Ti performance for $800.
Times change. And yeah people were saying the same "bad value" thing last time too IIRC.
 
So a card with worse performance could've been had for more than the new card's price which is why it is "objectively" bad value?
I'm at a loss trying to understand whatever you're saying.
5070ti.png
According to techpower up, nah... Keep in mind that I referenced $819.99 on prime day, but if you were willing to buy a used card, you could have had this level of performance for as low as $650-$700 on facebook marketplace 1 year ago. Like I said, the 5070ti is just a bad product. 7% performance increase over the 4070ti super at real world prices approaching the 4080super new prices.
 
lets just ignore the fact that the 7900xtx could be had for $819.99 on prime day a mere 5 months ago
The 7900XTX is an objectively worse product, way slower in ray tracing, has no decent upscaler or latency reduction solution, and lacks many visual/utility features that the 4080/5070Ti enjoys (Ray Reconstruction/RTX HDR ... etc).
 
I think the people you are talking to have zero interest in having a good faith discussion, so this effort is meaningless.

They will compare against (a) used products, (b) something that was available 15 moons ago for a super sweet deal when Venus's orbit aligned perfectly with Mercury's, (c) a hypothetical pie-in-the-sky product that exists in their imaginations. All to feed into their sense of righteous indignation.
That's simply not true. Degustator and I both used launch MSRPs for said comparisons. My point was to show why many might not consider it such a good value when last time you got almost twice the value on "same upgrade"
 
I think the people you are talking to have zero interest in having a good faith discussion, so this effort is meaningless.

They will compare against (a) used products, (b) something that was available 15 moons ago for a super sweet deal when Venus's orbit aligned perfectly with Mercury's, (c) a hypothetical pie-in-the-sky product that exists in their imaginations. All to feed into their sense of righteous indignation.
This post is ironic. Firstly you argue that it's bad faith to compare against a used product as if consumers don't make the decision between new vs used products everyday based on their budgetary constraints. Secondly you suggest that the prime day deal was a super sweet deal except it's not as there were many deals like this... I'm not even going to touch the righteous indignation part, I'll leave that to the mods.

The reality of the situation as seen from Microcenter is that 5070ti will realistically cost $900USD before tax with many models costing more. So like you said, lets not speak in hypotheticals but in reality. The reality is that the 5070ti is a bad product. Price aside, it's a bad product because the performance gains when compared to it's predecessors historically is well below average.
 
Last edited:
My point was to show why many might not consider it such a good value when last time you got almost twice the value on "same upgrade"
Isn't it funny how you've used the same way to compare 4070Ti with the outgoing Ampere stack as I've been doing back at 4070Ti launch trying to highlight why it's not that bad of a gain in perf/price?
While people here (not sure if that was you specifically) were pointing out that 3090Ti is just some 20% faster (at best) than a 3080 12GB which was selling at a similar $800 price.
Well here we don't have that "crypto inflated" top end in Ada's lineup so there's no such comparison.
 
So is the 5070 $750 or $900us?

The 5070 MSRP is $550 and will probably go for $650+ at launch. The 5070 Ti is essentially $900-$1000 at launch. Everybody’s falling over themselves with surprise that the first few weeks of any GPU release is a shit show. AMD is yet to play their hand and it’ll take a few weeks/months after that for things to settle after the scalpers and impatient have had their turn.
 
That's simply not true. Degustator and I both used launch MSRPs for said comparisons. My point was to show why many might not consider it such a good value when last time you got almost twice the value on "same upgrade"
Last time doesn't matter. As much as we want to, last time != this time. Times have changed dramatically, due to silicon technology limitations and macro-economic conditions. Some of those conditions are local to the US, but unfortunately these things seem to transplant across the oceans. We can -- and do -- complain about these global issues (cue old man screaming at clouds), but specific product values need to be compared vs. what else is available contemporaneously.

It's totally fair to reserve judgment until AMD shows their hand, which we know will happen in a matter of days/weeks. If AMD comes out with some decent value plays, by all means, declare these offerings as poor value. It's also fair to compare vs. last-gen products that are still available on the market today. But the comparison has to be against something a prospective buyer can purchase today, or a handful of days in the near future. It doesn't make sense to any potential buyer to assess value by comparing against a historical generational uplift -- that's totally an imaginary and emotional benchmark.
 
The 7900XTX is an objectively worse product,
Depends on the buyer?
way slower in ray tracing,
True
has no decent upscaler
I don't like upscalers.. It is well known.
or latency reduction solution,
Anti-lag + exists and is functionally equivalent to reflex
and lacks many visual/utility features that the 4080/5070Ti enjoys (Ray Reconstruction/RTX HDR ... etc).
RTX HDR is completely useless to me and is no way equivalent in importance to ray reconstruction. Ray reconstruction is also quite unfinished with numerous well documented problems. So if the pitch is for unfinished underutilized proprietary technology, I'm not interested.

As someone who owns a 4080 super, in a way, it's reassuring that due to Nvidia's mediocrity, there are still only 3 Nvidia GPU's faster than mine. Usually, it would be 5(5090, 4090, 5080, 5070 ti, 5070) and I'd start to ask myself questions about whether I should upgrade or not. If Nvidia continues to wallow in mediocrity, my wallet will receive some much needed respite.
 
Back
Top