Nvidia GeForce RTX 4070 Ti review: Hobbled and wildly overpriced

Price When Reviewed

$799

This evaluate can be quick and candy. Nvidia’s new $799 GeForce RTX 4070 Ti is a technically hobbled, drastically overpriced graphics card that prices $100 greater than the higher-tier RTX 3080 did final technology. In alternate for that 14 p.c greater price ticket, you normally get 14 p.c greater efficiency at 4K decision, and round 20 p.c at 1440p.

Depending on the way you take a look at it, the GeForce RTX 4070 Ti provides stagnant-at-best worth at comparable costs to final technology, or a painfully steep mark-up over last-gen’s already overpriced RTX 3070 Ti. It’s a beastly card that may topple the earlier RTX 3090 Ti flagship in lots of video games, particularly at 1440p decision or when DLSS 3 swings into motion, however at this chest-clutching value, we merely can’t suggest it. In truth, we suggest avoiding it outright except there are mammoth value reductions.

GeForce RTX 4070 Ti: Specs, options, and design

The specs for the GeForce RTX 4070 Ti shouldn’t shock anybody, because it’s the “unlaunched” $899 RTX 4080 12GB with a brand new identify. Gamers (and us) referred to as shenanigans on Nvidia initially giving this scaled-down GPU the identical 4080 branding because the far more highly effective 16GB RTX 4080, and shortly earlier than AMD revealed its Radeon RX 7900-series graphics cards—which might’ve made this GPU look foolish on the identical $899 value because the 7900 XT—Nvidia canceled its unique launch plans. It’s again with a cheaper price tag that’s nonetheless unreasonably, uncomfortably excessive for what’s being provided.

Here’s a recap of the RTX 4070 Ti’s core specs:

• Graphics Processing Clusters: 5
• Texture Processing Clusters: 30
• Streaming Multiprocessors: 60
• CUDA Cores: 7680
• Tensor Cores: 240 (4th Gen)
• RT Cores: 60 (third Gen)
• Texture Units: 240
• ROPs: 80
• Boost Clock: 2610 MHz
• Memory Clock: 10500 MHz
• Memory Data Rate: 21 Gbps
• L2 Cache Size: 49152 KB
• Total Video Memory: 12GB GDDR6X
• Memory Interface: 192-bit
• Total Memory Bandwidth: 504 GB/sec
• Texture Rate: 626 Gigatexels/sec
• Connectors: 3 x DisplayPort, 1 x HDMI
• Max Display Resolution: 4K at 240Hz or 8K at 60Hz with DSC
• Power Connectors: 2x PCIe 8-pin cables OR 300w or larger PCIe Gen 5 cable (adapter included in field)
• Minimum Power Supply: 700 Watts
• Power: 12W Idle, 20W AV1 Video Playback, 226W Average Gaming Power (AGP), 285 W Total Graphics Power (TGP)
• Maximum GPU Temperature: 90° C
• PCI Express Interface: Gen 4

Yes, the 4070 Ti’s core rely and reminiscence capability are drastically lowered versus the 4080—Nvidia made the precise name (finally) with its identify change. But now that the identify gained’t sow confusion amongst would-be patrons, the extra regarding tweak revolves across the 4070 Ti’s reminiscence bus.

Wider bus lanes allow you to transfer information extra successfully. Think of it like rush hour; a four-lane freeway can hold visitors shifting a lot quicker than a single-lane dust street. Bus width is a vital issue for total reminiscence bandwidth, particularly at greater resolutions the place textures and ray tracing can gobble up reminiscence quick.

The GeForce RTX 4070 Ti packs a pipsqueak 192-bit reminiscence bus. By distinction, the RTX 3080—which, once more, was a higher-tier card that price $100 lower than the 4070 Ti—wielded a 320-bit bus, which is typical for a high-end GPU. Even the already-pandemic-overpriced $600 RTX 3070 Ti had a 256-bit bus. You needed to go all the best way right down to the $330 RTX 3060 to search out an Nvidia GPU with a 192-bit bus final technology.

The new 4070 Ti provides 504GB/s of complete reminiscence bandwidth; the 3080 hit 760GB/s and the 3070 Ti provided 608GB/s, whereas the 3060 and its far slower GDDR6 non-X reminiscence hit an applicable 360GB/s.

Brad Chacos/IDG

It issues. Compared to the RTX 3080, the 4070 Ti’s uplift in 4K gaming is measurably slower than its 1440p uplift. You can nonetheless play 4K video games with every thing cranked and nonetheless hit 60 frames per second with out concern more often than not, however deliberately hobbling the 4K efficiency of a freaking $799 graphics card leaves a bitter style within the mouth. The 4070 Ti is tangibly slower than all different new-gen GPUs and the RTX 3080 in Counter-Strike: Global Offensive at 4K and 1440p resolutions, and I wouldn’t be shocked if the neutered bus width components into that. It could be extra acceptable if this GPU prices lots of of {dollars} much less.

Fortunately, the RTX 4070 Ti is far more environment friendly than the 3080 was, rated for 285 watts of complete graphics energy in comparison with the 3080’s 320W. It’s quicker, too. Nvidia’s new Ada Lovelace GPU architecture impresses there.  

The customized Asus Tuf 4070 Ti we’re reviewing at present—Nvidia isn’t launching a Founders Edition model of this GPU—nonetheless makes use of the brand new 16-pin 12VHPWR connector, nonetheless, with an adapter included within the field. (Make sure you plug it in snugly.) It’s additionally an enormous 3.25-slot graphics card regardless of the decrease energy draw. Asus in any other case did a high-quality job with Tuf, which contains a hanging industrial design, a dual-BIOS swap (we take a look at at default), and never one, however two HDMI 2.0b connectors to associate with three DisplayPorts. Finding twin HDMI ports on a GPU is uncommon lately, and Asus solely fees a $50 premium for this model of the Tuf—although, once more, the RTX 4070 Ti is wildly overpriced to start with.

Brad Chacos/IDG

Nvidia maintains its lead by way of providing well-rounded options and capabilities that AMD merely can’t match. The Radeon RX 7900 duo is tangibly quicker than the 4070 Ti by way of uncooked rasterization (learn: conventional gaming) efficiency, however Nvidia’s GPU holds a large ray tracing efficiency benefit, bolstered by AI-powered DLSS and DLSS Frame Generation applied sciences. They’re fiercely spectacular. Generally talking, Nvidia additionally thwomps AMD in CUDA assist for machine studying duties (duh), total content material creation prowess, and with different life-enhancing options like Nvidia Broadcast, Reflex, and Shadowplay. If you do extra along with your GPU than merely play video games, or in case you think about peak ray tracing efficiency a precedence, goal an Nvidia GPU.

Let’s get to the video games.

Our take a look at system

We take a look at graphics playing cards on an AMD Ryzen 5900X PC used completely for benchmarking GPUs. We now take a look at with PCIe Resizable BAR (also referred to as Smart Access Memory on Ryzen techniques) energetic, as most trendy gaming PCs launched within the final 4 years assist the performance-boosting characteristic, both natively or through a motherboard firmware replace. Nvidia additionally recommends turning on the elective “Hardware-accelerated GPU scheduling” choice in Windows to let the RTX 40-series stretch its legs to the fullest, so we’ve made that tweak as effectively. Most of the {hardware} was offered by the producers, however we bought the storage ourselves.

We’re evaluating Nvidia’s $799 GeForce RTX 4070 Ti towards all the fashionable GPUs launched by each Nvidia and AMD to this point, in addition to final technology’s RTX 3080, 3080 Ti, and 3090 and Radeon RX 6900 XT. Time constraints prevented us from benchmarking the $600 GeForce RX 3070 Ti, the 4070 Ti’s ostensible predecessor, however given this new card’s jacked-up value level the RTX 3080 is a extra sincere comparability anyway.

In addition to turning on PCIe Resizable BAR by default, we’ve additionally moved to a brand new set of video games in our testing suite for this new technology. We take a look at a wide range of video games spanning varied engines, genres, vendor sponsorships (Nvidia, AMD, and Intel), and graphics APIs (DirectX 9, 11, DX12, and Vulkan), to attempt to signify a full vary of efficiency potential. Each recreation is examined utilizing its in-game benchmark, sanity checked by Nvidia’s FrameView tool, on the highest attainable graphics presets except in any other case famous, with VSync, body charge caps, real-time ray tracing or DLSS results, and FreeSync/G-Sync disabled, together with every other vendor-specific applied sciences like FidelityFX instruments or Nvidia Reflex. We’ve additionally enabled temporal anti-aliasing (TAA) to push these playing cards to their limits.

We run every benchmark at the least 3 times and checklist the common outcome for every take a look at. Outside of esports, we’ve restricted our benchmarks to 4K and 1440p resolutions, as these are the pure matches for these juggernaut GPUs (regardless of the 4070 Ti’s paltry bus width).

We’re going to indicate these benchmarks with out particular person recreation commentary. In basic, take note of how a lot quicker the 4070 Ti is in comparison with the $100 cheaper 3080, and the way the 4070 Ti positive factors extra pace on the 3080 at 1440p than 4K. Also be aware how the equally overpriced $899 Radeon RX 7900 XT outperforms the $799 4070 Ti by a major quantity throughout the board. We didn’t have time to totally benchmark the 4070 Ti’s ray tracing prowess, however included a chart displaying Cyberpunk 2077 operating with RT enabled (with DLSS and FSR each on, and off) to substantiate that, yep, Nvidia’s GPU structure continues to blow AMD’s ray tracing efficiency out of the water. It isn’t even shut.

Gaming efficiency benchmarks

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

Brad Chacos/IDG

GeForce RTX 4070 Ti: Power and thermals

We take a look at energy draw by looping the F1 22 benchmark at 4K for about 20 minutes after we’ve benchmarked every thing else (to heat up the GPU) and noting the very best studying on our Watts Up Pro meter, which measures the ability consumption of our total take a look at system. The preliminary a part of the race, the place all competing automobiles are onscreen concurrently, tends to be essentially the most demanding portion. 

This isn’t a worst-case take a look at; this can be a GPU-bound recreation operating at a GPU-bound decision to gauge efficiency when the graphics card is sweating exhausting. If you’re enjoying a recreation that additionally hammers the CPU, you possibly can see greater total system energy attracts. Consider your self warned.

Brad Chacos/IDG

Despite being quicker than the RTX 3080, the 4070 Ti consumes far much less energy each when idle and beneath load. Even if Nvidia isn’t shifting the needle by way of worth with the 4070 Ti, its effectivity is little doubt enhancing.

We take a look at thermals by leaving GPU-Z open through the F1 22 energy draw take a look at, noting the very best most temperature on the finish.

Brad Chacos/IDG

The customized Asus Tuf cooling design delivers completely frigid temperatures beneath full load. Overclockers might have a area day with this explicit mannequin. You hear some fan noise to attain these spectacular temperatures, nevertheless it isn’t overly loud or annoying. I’ve no complaints in regards to the Asus Tuf’s total thermal design.

Should you purchase the GeForce RTX 4070 Ti?

Please don’t. It’s a beast of a card that exhibits what Nvidia’s stellar {hardware} and software program engineers are able to—however not that this eye-watering value.

In a vacuum, the $799 GeForce RTX 4070 Ti’s rasterization efficiency underwhelms in comparison with the $899 Radeon RX 7900 XT besides in ray tracing efficiency, however Nvidia’s plentiful software program and have choices—Studio drivers, DLSS and Frame Generation, Broadcast, Reflex, CUDA, et cetera—could make that hole simpler to swallow in case you use your GPU for extra than simply gaming. In actuality, the RTX 4070 Ti—just like the RTX 4080 and Radeon 7900 XT earlier than it—carry value tags that make their worth stagnant at greatest and borderline insulting at worst.

Brad Chacos/IDG

At $799, the GeForce RTX 4070 Ti prices $100 greater than final technology’s beautiful RTX 3080—and it’s a lower-tier GPU. It’s $200 greater than the already-overpriced RTX 3070 Ti was. In alternate for that 14 p.c value hike over the 3080, you normally get a stagnant 14 p.c extra efficiency than the 3080 at 4K, or roughly 20 p.c quicker speeds at 1440p. Nvidia minimize the 4070 Ti’s reminiscence bus again to RTX 3060 ranges which slows it down at 4K. That’s flat-out unacceptable in a $799 graphics card.

Gamers are in a nasty place. Nvidia and AMD’s flagships, the GeForce RTX 4090 and Radeon RX 7900 XTX, supply beautiful efficiency positive factors over their predecessors for comparable costs. But every thing beneath these carry downright painful MSRPs in comparison with prior generations. Nvidia and AMD should nonetheless assume we’re within the midst of a pandemic with roaring cryptocurrency worth. We’re not. Miners have been those shopping for graphics playing cards for painful, outrageous costs, not avid gamers. These graphics card prices are ludicrous.

Nvidia’s RTX 4080 prices $500 extra than the 3080 did. Gross. Rather than seizing the second, AMD priced its feature-lacking Radeon RX 7900 XT at $899, when spec-wise it most likely ought to’ve been referred to as the Radeon RX 7800 XT and priced a pair hundred {dollars} decrease. But AMD’s thirst for 7900 XT revenue margins additionally allowed Nvidia to reintroduce the “unlaunched” disastrous RTX 4080 12GB with a brand new identify and a price ticket simply $100 lower than earlier than. It’s nonetheless extraordinarily unappealing. You shouldn’t should pay $800 {dollars} for what’s being actively marketed as a 1440p graphics card in 2023, particularly with its 4K efficiency being hobbled by the subpar reminiscence bus.

Bah. It’s like WOPR mentioned in WarGames: “A strange game. The only winning move is not to play.”

Part of the explanation why Nvidia and AMD are conserving costs so excessive is to maneuver their stockpiled last-gen GPUs now that cryptocurrency has crashed, at rigorously managed excessive costs. Nvidia specifically retains attempting to indicate the RTX 40-series as costly choices on high of the 30-series lineup, moderately than the generational substitute it actually is—a worryingly anti-consumer enterprise resolution. Buy a used graphics card instead in case you aren’t going all-in on a 4090 or a 7900 XTX. You can discover equally performing GPUs just like the Radeon RX 6800 XT and 6900 XT or the RX 3080 for lots of of {dollars} lower than Nvidia and AMD are charging for these wildly overpriced new choices. We have tips to help you avoid buying a used lemon.

And I’m saying that regardless of being a man who typically recommends that you avoid used graphics cards.

Because paying $800 for a hobbled, drastically overpriced 1440p-focused graphics card in 2023 ain’t it. The launch of this new technology largely mirrors the RTX 20-series launch, the place stagnant gen-on-gen worth led to horrible gross sales and the introduction of an RTX 20 Super collection shortly after to assist kickstart issues…although that launch coincided with the discharge of AMD’s glorious, sanely priced Radeon RX 5000-series. This technology, AMD appears content material to experience the coattails of Nvidia’s extreme pricing. Flagship GPUs shouldn’t supply the very best price-per-dollar worth, and the remainder of your stack shouldn’t ship efficiency that scales virtually linearly with the pricing of last-gen fashions.

Discrete graphics playing cards gross sales simply hit a 20-year low. Unless one thing drastic adjustments, like a contemporary introduction of Super collection equivalents, releases just like the GeForce RTX 4070 Ti and 4080 will seemingly solely exacerbate the ache.

Not simply go, however exhausting go. If you should purchase a next-gen GPU round this value level, the Radeon RX 7900 XT can be atrociously priced nevertheless it’s noticeably quicker and comes with 20GB of reminiscence, 8GB greater than the 4070 Ti. If you want Nvidia options and should purchase a 4070 Ti, I extremely suggest the Asus Tuf mannequin.

But we’re within the midst of an odd recreation. Please don’t purchase both. Not at these costs.