Nvidia’s RTX 50-series makes its very first venture listed below the $1,000 mark beginning today, with the$749 RTX 5070 Ti– a minimum of in theory.

The third-fastest card in the Blackwell GPU lineup, the 5070 Ti is still far from “reasonably priced” by historic requirements(the 3070 Ti was$599 at launch ). It’s likewise$ 50 more affordable and a reasonable bit much faster than the outbound 4070 Ti Super and the older 4070 Ti. These are actions in the ideal instructions, if little ones.

We’ll talk more about its efficiency soon, however at a high level, the 5070 Ti’s efficiency falls in the exact same basic variety as the 4080 Super and the initial RTX 4080, a card that released for $1,199 simply over 2 years earlier. And it’s most likely your flooring for regularly playable native 4K video gaming for those of you out there who do not wish to depend on DLSS or 4K upscaling to strike that resolution(it’s likewise most likely all the GPU that the majority of people will require for high-FPS 1440p, if that’s more your speed).

It’s a card I’m ambivalent about! It’s close to 90 percent as quick as a 5080 for 75 percent of the rate, a minimum of if you pass Nvidia’s minimum sticker price, which for the 5090 and 5080 have actually been mainly imaginary up until now. If you can discover it at that rate– which’s a huge “if,” given that every$749 design is currently out of stock throughout the board at Newegg– and you’re desperate to update or are developing a new 4K video gaming PC, you might do even worse. I would not invest more than $749 on it, and it may be worth waiting to see what AMD’s very first 90-series Radeon cards look like in a couple weeks before you leap in.

Fulfill the GeForce RTX 5070 Ti

Nvidia isn’t making a Founders Edition variation of the 5070 Ti, so this time around our evaluation system is an Asus Prime GeForce RTX 5070 Ti supplied by Asus and Nvidia. These third-party cards will deviate a little from the stock specifications noted above, however factory overclocks tend to be extraordinarily moderate, and done primarily so the GPU producer can slap a huge “overclocked” badge someplace on package. We evaluated this Asus card with its BIOS change set to “performance” mode, which raises the increase clock by a whole 30 MHz; you do not require to be a mathematics whiz to think that a 1.2 percent overclock is not going to alter efficiency much.

Compared to the 4070 Ti Super, the 5070 Ti brings 2 things to the table: an approximately 6 percent boost in CUDA cores and a 33 percent boost in memory bandwidth, thanks to the switch from GDDR6X to GDDR7. The initial 4070 Ti had even less CUDA cores, however most significantly for its 4K efficiency consisted of simply 12GB of memory on a 192-bit bus.

The 5070 Ti is based upon the very same GB203 GPU silicon as the 5080 series, however with 1,792 CUDA cores handicapped. There are a lot of resemblances in between the 2, consisting of the 16GB bank of GDDR7 and the 256-bit memory bus. It looks absolutely nothing like the yawning space in between the RTX 5090 and the RTX 5080, and the 2 cards’ similar-ish specifications implied they weren’t too far from each other in our screening. The 5070 Ti’s 300 W power requirement is likewise a bit lower than the 5080’s 360 W, however it’s quite near to the 4080 and 4080 Super’s 320 W; in practice, the 5070 Ti draws about as much as the 4080 cards do under load.

As a Blackwell GPU, the 5070 Ti likewise supports Nvidia’s most-hyped addition to the 50-series: assistance for DLSS 4 and Multi-Frame Generation (MFG ). We’ve currently covered this in our 5090 and 5080 evaluations, however the brief variation is that MFG works precisely like Frame Generation performed in the 40-series, other than that it can now place approximately 3 AI-generated frames in between natively rendered frames rather of simply one.

Particularly if you’re currently performing at a fairly high frame rate, this can make things look a lot smoother on a high-refresh-rate display without presenting distractingly extreme lag or unusual rendering mistakes. The function is generally questionable due to the fact that Nvidia is comparing 50-series efficiency numbers with DLSS MFG made it possible for to older 40-series cards without DLSS MFG to make the 50-series cards appear a great deal faster than they really are.

We’ll release some frame-generation numbers in our evaluation, both utilizing DLSS and (for AMD cards) FSR. Per normal, we’ll continue to focus on natively rendered efficiency– more pertinent for all the video games out there that do not support frame generation or do not benefit much from it, and more pertinent due to the fact that your base efficiency determines how great your produced frames will look and feel anyhow.

Testbed notes

We checked the 5070 Ti in the very same upgraded testbed and with the exact same upgraded suite of video games that we began utilizing in our RTX 5090 evaluation. The heart of the construct is an AMD Ryzen 9800X3D, guaranteeing that our numbers are restricted as low as possible by the CPU speed.

Per normal, we focus on screening GPUs at resolutions that we believe the majority of people will utilize them for. For the 5070 Ti, that implies both 4K and 1440p– this card is perhaps still overkill for 1440p, however if you’re attempting to strike 144 or 240 Hz (and even more) on a display, there’s a great case to be produced it. We likewise utilize a mix of ray-traced and non-ray-traced video games. For the video games we evaluate with upscaling made it possible for, we utilize DLSS on Nvidia cards and the latest supported variation of FSR (generally 2.x or 3. x) for AMD cards.

We’ve checked and re-tested several cards with current motorists in our upgraded testbed, we do not have a 4070 Ti Super, 4070 Ti, or 3070 Ti readily available to check with. We’ve offered some numbers for those GPUs from previous evaluations; these are from a PC running older motorists and a Ryzen 7 7800X3D rather of a 9800X3D, and we’ve put asterisks beside them in our charts. They must still paint a fairly precise image of the older GPUs’ relative efficiency, however take them with that little grain of salt.

Efficiency and power

Regardless of consisting of less CUDA cores than either variation of the 4080, some mix of architectural enhancements and memory bandwidth increases assist the card equal both 4080 cards practically completely. In the majority of our tests, it landed in the narrow strip right in between the 4080 and the 4080 Super, and its power intake under load was likewise practically similar.

In every manner in which matters, the 5070 Ti is basically an RTX 4080 that likewise supports DLSS Multi-Frame Generation. You can see why we ‘d be slightly passionate about it at$749 however less and less satisfied the better the cost sneaks to $1,000.

Being close to a 4080 likewise indicates that the efficiency space in between the 5070 Ti and the 5080 is normally quite little. In the majority of the video games we checked, the 5070 Ti hovers right around 90 percent of the 5080’s efficiency.

The 5070 Ti is likewise around 60 percent as quick as an RTX 5090. The efficiency is a lot lower, however the price-to-performance ratio is a lot greater, perhaps showing the reality that the 5070 Ti really has other GPUs it needs to take on( in non-ray-traced video games, the Radeon RX 7900 XTX usually equals the 5070 Ti, however at this late date it is primarily out of stock unless you’re ready to pay method more than you should for one).

Compared to the old 4070 Ti, the 5070 Ti can be in between 20 and 50 percent quicker at 4K, depending upon how restricted the video game is by the 4070 Ti’s narrower memory bus and 12GB bank of RAM. The efficiency enhancement over the 4070 Ti Super is more soft, varying from as low as 8 percent to as much as 20 percent in our 4K tests. This is much better than the RTX 5080 did relative to the RTX 4080 Super, however as a generational leap, it’s still quite modest– it’s clear why Nvidia desires everybody to take a look at the Multi-Frame Generation numbers when making contrasts.

Waiting to put theory into practice

Having the ability to get RTX 4080-level efficiency for numerous hundred dollars less simply a number of years after the 4080 released is sort of amazing, though that enjoyment is leavened by the still high-ish $749 price (once again, presuming it’s really offered at or anywhere near that rate ). That definitely makes it feel more like a next-generation GPU than the RTX 5080 did– and whatever else you can state about it, the 5070 Ti definitely seems like a much better buy than the 5080.

The 5070 Ti is a quick and 4K-capable graphics card, quick enough that you must have the ability to get some excellent arise from all of Blackwell’s brand-new frame-generation hoax if that’s something you wish to have fun with. Its price-to-performance ratio does not delight me, however if you do the mathematics, it’s still a better worth than the 4070 Ti series was– especially the initial 4070 Ti, with the 12GB allocation of RAM that restricted its effectiveness and future-proofness at 4K.

2 factors to hold back on purchasing a 5070 Ti, if you’re thinking of it: We’re waiting to see how AMD’s 9070 series GPUs clean, and Nvidia’s 50-series launch up until now has actually been type of a mess, with low schedule and rate gouging both on retail websites and in the pre-owned market. Pay much more than$749 for a 5070 Ti, and its fragile worth proposal fades rapidly. We ought to understand more about the AMD cards in a number of weeks. The supply circumstance, a minimum of up until now, appears like an issue that Nvidia can’t (or will not) determine how to resolve.

103 Comments