NVIDIA GeForce RTX 5070 Ti Review – ASUS PRIME Tested

The RTX 5070 Ti arrives amidst MSRP debate and general negativity

NVIDIA’s latest GPU is about to hit shelves, however briefly, and we were able to test out an ASUS PRIME ahead of the launch. And, as I sifted through the benchmark results, I began to get that feeling…the RTX 4060 kind of feeling…where there just isn’t going to be a way to objectively review this product on its own merit without facing scorn. This is not to say that I think it’s a great product, but it is a perfectly serviceable solution that slots in neatly behind the RTX 5080 this generation. Please flame me for saying this in the comments.

Since our review is several hours late as I write this, I have the benefit of seeing other reviewers’ take on the new RTX 5070 Ti – and, as I have come to expect, reactions have been largely negative (GN most notably). Most of this angst seems to stem from pricing, as indicated by these tasteful thumbnails – the top three results from a YouTube search for “rtx 5070 ti review”:

The familiar look YouTube hardware reviews

In case you missed it, YouTuber JayzTwoCents released a video on the 17th about the disparity in list price for “MSRP” units, which seem subsequently to have been adjusted before launch as certain SKUs – including the ASUS PRIME card tested in this review – returned to $749.99 USD in the aftermath of the pushback over retail markup. So, once again, the entire launch will be about pricing (and availability), and not performance.

I won’t waste too much time on preamble here. The GeForce RTX 5070 Ti is a lesser card than the GeForce RTX 5080, and the former is supposed to cost 25% less than the latter. If that was always the case, the card might be a good value relative to its performance. Let us see if that is the case.

The ASUS PRIME Card

“Experience Primal performance with the Prime GeForce RTX 5070 Ti, an SFF-Ready Enthusiast GeForce Card featuring a 2.5-slot design for expansive compatibility, enhanced by a triple-fan setup for supreme airflow design for supreme cooling.”

The Prime card from ASUS is a 2.5-slot design measuring 304 x 126 x 50 mm. It is essentially a stock offering, though a very modest OC mode is available which raises the GPU frequency from the default 2452 GHz to 2482 GHz. You can read the full specs from ASUS by following this link.

Performance Testing

First, a brief note on test methodology. Each of the charts to follow contain scores or FPS representing the average of three identical runs for each card. As usual for all testing, outliers were tossed and testing was resumed after verifying accuracy. It is normal to see a small variance between test runs, but large swings can point to background process interference, or improper resolution scaling settings, inadequate cooldown between tests, or any number of other factors. In the end, the results published here are as accurate as possible, and should be easily reproduced by any test setup identical to the one employed here.

3DMark Tests

Let’s begin with some 3DMark results.

In 3DMark Speed Way (see overview), which is a DirectX 12 Ultimate test (heavy on ray tracing) that is rendered at 2560×1440, the RTX 5070 Ti is faster than the RTX 4080 Super – and far ahead of the Radeon RX 7900 XTX and RTX 4070 Ti Super – though it remains quite a bit behind the RTX 5080.

In 3DMark Steel Nomad (see overview), which UL recommends for high-end gaming PCs and renders at 3840×2160, the GeForce RTX 5070 Ti falls short of both the Radeon RX 7900 XTX and the GeForce RTX 4080 Super.

Finally, we will have a look at the older Time Spy Extreme test:

In 3DMark Time Spy Extreme (overview here), which is another 3840×2160 test but old enough to have just a 4GB VRAM requirement, the GeForce RTX 5070 Ti once again falls short of the GeForce RTX 4080 Super, only out-performing the Radeon RX 7900 XT by a small margin. Ouch.

So to loosely sum up 3DMark results, if you are running something with ray tracing enabled, the relative performance of the GeForce RTX 5070 Ti is a lot more impressive than it would be otherwise, DLSS not withstanding.

Game Benchmarks at 2560×1440

Rather than continue to push 3440×1440 ultrawide testing exclusively like I did all last year, in the past few weeks I have re-built the gaming testbed and started over some more mainstream 2560×1440 results. Sure, that means only testing with ~3.69 million pixels, but that’s still a much more demanding workload than the ~2.07 million pixels of 1920×1080. I can also multiply other resolutions and type in the results here in pixels, but I won’t.

Here are a few charts to look at:

The GeForce RTX 5070 Ti trades blows with the GeForce RTX 4080 Super in three of the charts above – though a range of 1.8% to 4.1% is not very exciting. Of more concern is the final result, with the older Metro Exodus Enhanced Edition showing a 12.3% advantage for the GeForce RTX 4080 Super, which itself was very nearly tied with the newer GeForce RTX 5080 in this test.

So much of the story with the GeForce RTX 50 Series is about software, and to this end here is a look at this generation of cards vs. the previous flagship GeForce RTX 4090 in a DLSS scenario:

The first thing I took from the above chart was just how close the RTX 5070 Ti is to the RTX 5080 without Frame Generation enabled – though the RTX 5080 is enjoying a 10.8% advantage in this case. Looking at the rest of the results, and when considering the hypothetical list prices of these cards, I think it’s fair to say that getting to within about 11% to 13% of the performance of a card that costs 25% more makes the GeForce RTX 5070 Ti a good value. This is all academic if actual pricing differs from MSRP, which we all know is probably going to be the case.

Let is now take a quick look at power draw, and here I will compare the GeForce RTX 5070 Ti to its slightly larger sibling in the current lineup – as well as two more often similarly-performing cards in the GeForce RTX 4080 Super and Radeon RX 7900 XTX:

The GeForce RTX 5070 Ti averaged almost exactly 300 watts during the Speed Way test run, was very close to the power draw of the RTX 5080, and the previous-gen cards from NVIDIA and AMD draw quite a bit more – though actually identifying anything from the confusing spaghetti of the chart above seems like a remote possibility.

Final Thoughts

So where does the GeForce RTX 5070 Ti stand? The way it slots in a little behind the GeForce RTX 5080 on most charts shouldn’t surprise anyone, as NVIDIA is always segmenting their lineup like this. The theoretical MSRPs are the same as last year, relative to the RTX 4070 Ti Super/RTX 4080 Super. But consumers – and YouTube hardware reviewers, apparently – are sick of inflated retail pricing and poor availability, and once again the focus of this GPU launch is ancillary rather than primary. I wish we could judge products based on their relative performance, but the endless price/availability problem with NVIDIA cards, which we have been saddled with since the last mining craze, continues.

AMD is in an enviable position with their upcoming Radeon RX 9070 Series launch, as some real pressure could be applied to the market – or at least to the GPU vendor-agnostic side of it, which is admittedly a much smaller segment of the overall GPU buying public…unless you only look at German retail figures and ignore every JPR market share report (and think Steam survey results are rigged).

The ASUS PRIME version of the GeForce RTX 5070 Ti sits atop the coffee table

Bottom line, NVIDIA is releasing a new GPU that can trade blows with last generation’s $1000 card, but will probably end up selling for a lot more than the planned $749 USD MSRP.