Nvidia GeForce RTX 4060 Ti (8GB) evaluation: Disappointing for $400

At first blush, Nvidia’s $399 GeForce RTX 4060 Ti (8GB) appears prefer it must be a smashing success.

Its predecessor, 2020’s $399 GeForce RTX 3060 Ti, was probably the greatest GPUs in a powerful RTX 30-series technology, providing impeccable 1080p and spectacular 1440p gaming efficiency at an inexpensive value. The brand new RTX 4060 Ti sticks to the identical value whereas weaving in Nvidia’s newest killer RTX 40-series options, like DLSS 3, Reflex, RTX Video Tremendous Decision, best-in-class ray tracing, AV1 encoding, beautiful energy effectivity, and extra.

However peel again the hood and issues aren’t fairly so clear lower.

Nvidia made technical selections—lowering core counts and altering the reminiscence subsystem—that permit the GeForce RTX 4060 Ti to scream at 1080p decision however make it unappealing for 1440p gaming. It additionally makes the 4060 Ti a mediocre generational improve by way of uncooked firepower. The choice to outfit this card with 8GB of reminiscence means it gained’t have the ability to run all video games with eye sweet maxed out even at 1080p, by Nvidia’s personal admission. And whereas the addition of DLSS 3’s AI-generated frames is definitely welcome, picture upsampling applied sciences create noticeable visible artifacts on the much less pixel-packed 1080p decision, rendering DLSS’s performance-boosting benefits much less of an element right here.

So I suppose the underside line is: Are you keen to pay $400 for a 1080p graphics card with 8GB of reminiscence within the yr of our lord 2023?

What are the Nvidia GeForce RTX 4060 Ti’s specs and value?

Nvidia revealed some high-level particulars for the GeForce RTX 4060 Ti 8GB in its announcement (which additionally revealed a $299 RTX 4060 and $499 RTX 4060 Ti 16GB, each coming in July) however didn’t present deeper technical info. Right here that’s, in contrast in opposition to the RTX 3060 Ti:

A couple of key particulars leap out right away. First, the RTX 4060 Ti—constructed upon Nvidia’s new “Ada Lovelace” GPU structure—is tremendously energy environment friendly and affords a lot increased GPU and reminiscence clock speeds than earlier than. Bravo, Nvidia.

Second, this new card really holds considerably fewer CUDA, tensor, and RT cores than the RTX 3060 Ti it replaces. Due to this, the RTX 4060 Ti can really be slower than its predecessor in some conditions, and solely affords a middling efficiency uplift general. Greatest-case eventualities prime out at round 15 to twenty p.c quicker than the RTX 3060 Ti, however you’ll see basically static gen-on-gen efficiency in lots of extra video games and resolutions. It’s a testomony to Lovelace that this GPU reveals any velocity upgrades contemplating what number of fewer CUDA cores it has, particularly given the ultra-low energy draw, nevertheless it’s nonetheless deeply meh as a generational improve.

Brad Chacos/IDG

Third, Nvidia adjusted the GeForce RTX 4060 Ti’s reminiscence subsystem to extra intently mirror AMD’s Radeon Infinity Cache setup, plopping a giant L2 cache on-die whereas nerfing the reminiscence bus width all the way in which right down to 128 bits. (The 3060 Ti has a 256-bit bus.) Nvidia additionally stored reminiscence capability an identical to its predecessor at 8GB of GDDR6, in an period when fashionable video games can now eat greater than that even at 1080p decision. And in terms of VRAM capability, both you could have sufficient otherwise you don’t.

The reminiscence configuration prevents the GeForce RTX 4060 Ti from being a viable Excessive/Extremely 1440p gaming choice within the long-term, and even in lots of conditions now. We defined why in our evaluation through the RTX 4060 Ti’s preliminary reveal:

“The reminiscence bus width is being slimmed down, from 256-bit within the RTX 3060 Ti to simply 128-bit within the 4060 Ti. Consider it like site visitors congestion; if reminiscence requests are vehicles, the bus width is the scale of the highway. 100 vehicles transfer a lot quicker on a four-lane freeway than a single lane dust highway, and that velocity is analogous to the cardboard’s general reminiscence bandwidth. Ostensibly, the GeForce RTX 4060 Ti affords 288GB per second of general reminiscence bandwidth, in comparison with the 3060 Ti’s 448GB/s—an enormous gulf.

It’s not that straightforward although.

Nvidia

Nvidia

Nvidia swiped a web page from AMD’s killer Infinity Cache know-how for RTX 40-series choices, dropping a big on-die cache onto the GPU itself, and the RTX 4060 household is not any totally different. Not solely does the nearer bodily proximity imply this cache is quicker, however deploying it additionally reduces the necessity to ship many requests out to the reminiscence chips arrayed across the GPU itself, as illustrated within the slides above. Whereas AMD’s Infinity Cache makes use of an L3 cache, Nvidia opted for a 32MB L2 cache as a substitute (versus simply 4MB on the 3060 Ti). The corporate says the bigger L2 cache notably helps enhance ray tracing and DLSS efficiency, and leads to the “effective” reminiscence bandwidth proven within the technical specs slide.

Right here’s the factor although: As evidenced in our evaluation of the last-gen Radeon RX 6600 XT, which had 32MB of Infinity Cache and a 128-bit bus, pairing a giant on-die cache with a neutered bus width helps the GPU carry out exceptionally—each in efficiency and energy—on the show decision it’s focused for, however normally leads to lower-than-expected outcomes at increased resolutions. Increased resolutions have increased reminiscence wants, which leads to fewer profitable on-die cache hits. When that occurs, the reminiscence requests exit to conventional reminiscence—and the RTX 4060 Ti’s a lot smaller 128-bit bus would possibly falter when that occurs.”

Now that we’ve examined the RTX 4060 Ti, we will verify that it certainly slows down on the increased 1440p decision. In truth, it will probably’t even grasp robust at 1440p with the RTX 3070, which price $500 in 2020. You’d anticipate a brand new card to at the very least rival an older GPU that price simply $100 greater than it 2.5 lengthy years in the past, however the RTX 4060 Ti fails to clear that low bar.

Nvidia appears to be hitching the 4060 Ti’s star to Lovelace’s energy effectivity and the corporate’s GeForce function set, which is admittedly killer: DLSS 3, RTX Video Tremendous Decision, Nvidia Broadcast, cutting-edge AV1 encoding, class-leading ray tracing, Nvidia Reflex, and extra are on faucet, and so they’re all really glorious.

Brad Chacos/IDG

We’re testing Nvidia’s GeForce RTX 4060 Ti Founders Version, a manageable two-slot iteration of the corporate’s tried-and-tried and attractive in-house design. It’s extra silver than upper-tier fashions, coming with a single HDMI slot and three DisplayPorts.

Brad Chacos/IDG

One quibble: Regardless of solely requiring a single 8-pin connector for energy, Nvidia nonetheless outfit the cardboard with a 12VHPWR cable, necessitating using a brief adapter included in field. It appears awkward and fugly in the event you’re utilizing a case with a windowed facet. The fan blades are additionally a lot much less densely packed than earlier FE fashions, although it doesn’t have an effect on cooling efficiency, as you’ll see afterward.

Brad Chacos/IDG

Sufficient chatter. Let’s get to the video games!

Our check system

We check graphics playing cards on an AMD Ryzen 5900X PC used solely for benchmarking GPUs. We now check with PCIe Resizable BAR (also called Sensible Entry Reminiscence on Ryzen methods) energetic, as most fashionable gaming PCs launched within the final 4 years help the performance-boosting function, both natively or by way of a motherboard firmware replace. Nvidia additionally recommends turning on the optionally available “Hardware-accelerated GPU scheduling” in Home windows to let the RTX 40-series stretch its legs to the fullest, so we’ve made that tweak as properly. A lot of the {hardware} was offered by the producers, however we bought the storage ourselves.

We check 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. Every recreation is examined utilizing its in-game benchmark, sanity checked by Nvidia’s FrameView software, on the highest potential graphics presets except in any other case famous, with VSync, body price 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 very least 3 times and record the typical end result for every check. Outdoors of esports, we’ve restricted our benchmarks to 1440p and 1080p resolutions, because the 4060 Ti really isn’t constructed for 4K gaming. We check options like ray tracing and DLSS in a separate part, as upscaling options like DLSS and AMD’s rival FSR tweak the look of video games to assist them run quicker. 

Normally we current our knowledge in a collection of per-game bar charts, however given some extreme time constraints, we’re operating our knowledge in uncooked charts at this time and supplementing it with further efficiency commentary as a substitute. (Hit me up on Twitter to let me know in the event you like it or hate it!) These time restrictions additionally prevented us from operating new benchmark knowledge for the $330 16GB Intel Arc A770, alas, which is one other sturdy contender on this value phase.

Nvidia GeForce RTX 4060 Ti 1080p gaming efficiency

Nvidia itself calls the RTX 4060 Ti “a 1080p performance champ,” so let’s begin there.

Brad Chacos/IDG

All of those graphics playing cards ship nice 1080p graphics efficiency with eye sweet maxed out, even the last-gen Radeon RX 6600, which could be discovered hovering round $200 nowadays. We additionally included outcomes for the Radeon RX 6700 XT. This AMD GPU options 12GB of VRAM and a 256-bit bus, sidestepping issues with the RTX 4060 Ti’s reminiscence configuration. It may be discovered for roughly $350 nowadays.

The brand new RTX 4060 Ti usually leads its RTX 3060 Ti predecessor, albeit by a slim to negligible margin in most video games, as evidenced by the typical FPS proven on the finish of the chart.

The one exception? The legendary Counter-Strike: World Offensive, one of many most-played video games on the planet. GeForce RTX 40-series GPUs have constantly did not beat their predecessors in CSGO, probably as a result of a lot narrower reminiscence bus widths that Nvidia slapped on this technology. When you exclude CSGO and its ultra-high body charges from the averaged outcomes, the RTX 4060 Ti’s lead widens, hitting 130fps on common versus the RTX 3060 Ti’s 119—nonetheless a paltry 9 p.c uplift general.

Nvidia GeForce RTX 4060 Ti 1440p gaming efficiency

We additionally examined these playing cards at 1440p decision, including within the $500 RTX 3070 from final technology and the brand new $599 RTX 4070 for good measure (actually).

Brad Chacos/IDG

As soon as once more, all these graphics playing cards flip in a good exhibiting, and as soon as once more, the brand new RTX 4060 Ti lands uncomfortably near its predecessor. Factoring in CSGO’s excessive body charges and slower RTX 40-series efficiency, the 3060 Ti really holds the lead within the averaged body price right here. Oof. That’s a bit deceptive although, as a result of the RTX 3060 Ti is a lot quicker than the RTX 4060 Ti at 1440p that it skews the general image. Exclude CSGO and the RTX 4060 Ti is roughly 7 p.c quicker than the RTX 3060 Ti. Nonetheless oof.

That oof turns into an OOF when you think about the RTX 3070’s efficiency. It smashes the RTX 4060 Ti to the tune of a 13.7 p.c common with CSGO included, or by 5 p.c with CSGO excluded. Not having the ability to beat a card that price simply $100 extra 2.5 years in the past is deeply embarrassing for Nvidia’s new GPU, and a sign of why the RTX 4060 Ti’s new technical configuration makes it unsuitable for long-term 1440p gaming efficiency.

Talking of, peep that Radeon RX 6700 XT at each resolutions. It performs equally on common to the RTX 3060 Ti, which as we simply established is inside spitting distance of the RTX 4060 Ti, however comes with 12GB of reminiscence and a wider 256-bit bus for $350. When you’re nervous about 8GB not being sufficient VRAM for contemporary video games, or wish to dabble in 1440p gaming and don’t thoughts giving up DLSS and different Nvidia options, it’s price contemplating—particularly at $50 lower than Nvidia’s new GPU. In relation to VRAM capability, both you could have sufficient otherwise you don’t.

Nvidia GeForce RTX 4060 Ti ray tracing, DLSS, and DLSS 3 efficiency

So the RTX 4060 Ti is a mediocre generational improve in terms of conventional gaming horsepower. Nvidia is seemingly hitching this GPU’s hat on energy effectivity (extra on that later), ray tracing, and—given the character of the efficiency metrics Nvidia unveiled within the card’s announcement—the facility of AI-enhanced DLSS efficiency enhancements.

Luckily, most of the video games in our rigorously chosen check suite embody ray tracing and picture upscaling (Nvidia DLSS/AMD FSR) choices. We check the video games at native, native with RT on Extremely, and with RT on Extremely and DLSS/FSR on Balanced to guage every expertise.

BUT!

It’s price noting that I believe DLSS/FSR picture upscaling appears horrible on the 1080p decision this graphics card was constructed for. Picture upscaling works by internally rendering a body at a decrease decision, then utilizing AI and/or software program methods to upscale the ultimate picture to match your chosen decision. It permits you to get quicker efficiency with minimal to no loss in picture high quality—at the very least at increased pixel-packed 1440p and 4K resolutions.

Activating picture upscaling isn’t so easy at 1080p. At this decrease finish of the size, the picture upscaling know-how is pulling from so many fewer pixels that the visuals that seem onscreen can look blurry and jagged, even at higher-quality DLSS/FSR presets. The inner picture getting used to render video games at 1080p decision with DLSS Balanced on is below 720p—and it appears prefer it, in my view, although DLSS appears superior to AMD’s FSR at this decision.

I wouldn’t purchase a 1080p graphics card with upscaling in thoughts, full cease. Sadly, the decrease upscaled picture constancy additionally makes operating ray tracing harder at 1080p, because it takes away from the improved visuals that ray tracing can present. One of the best lighting on this planet doesn’t look nice when it’s paired with jagged, blurry fashions, alas. It’s fantastic sufficient if want be, although.

Picture upscaling is simply a part of the DLSS worth proposition with the RTX 40-series nonetheless. This technology additionally marks the introduction of DLSS 3 Body Technology, which leverages the tensor cores and enhanced optical circulation accelerator inside Nvidia’s Lovelace GPUs to insert absolutely AI-generated frames between each historically rendered body.

This is a real potential feather within the cap of the RTX 4060 Ti. You don’t must activate DLSS 2 picture upsampling and downgrade picture constancy to run DLSS 3—you’ll be able to simply flip on Body Technology and watch the body price fly, particularly in CPU-bound video games like Microsoft Flight Simulator. It’s implausible.

That mentioned, DLSS 3 isn’t fairly set-and-forget both. Inserting AI frames can typically end in visible artifacts in choose video games, particularly round UI parts. And since you’ll be able to’t really management inputs throughout these AI-generated frames, latency is added. DLSS 3 requires builders to additionally combine Nvidia Reflex to fight these issues, which regularly helps video games be barely extra responsive than when rendering natively, nevertheless it doesn’t end result within the soar in responsiveness you’d usually anticipate when supercharging your body charges. Which means the really feel of the sport won’t match what your mind and palms anticipate—a recreation may be operating 50 percent-plus quicker with body technology on, for instance, however solely really feel 10 to twenty p.c quicker in responsiveness.

DLSS 3 makes video games look superb to make sure, however that element is type of bizarre, and value noting. Our DLSS 3 deep dive goes into every thing in regards to the know-how in a lot larger element.

Brad Chacos/IDG

We examined ray tracing and DLSS at each 1440p and 1080p decision, however are solely revealing the 1080p outcomes above, given a mixture of time constraints and the truth that you must solely actually take into account the RTX 4060 Ti for 1080p gaming.

As you’ll be able to see from these numbers, the RTX 4060 Ti maintains Nvidia’s ray tracing lead on this phase, although it doesn’t functionally prolong it past what the 3060 Ti was already able to years in the past for a similar value. We check at Extremely, however in the event you dial some graphics settings again to Excessive or Medium, you’ll be able to come near 60fps efficiency at 1080p in lots of video games even with ray tracing enabled and DLSS off. DLSS 3 is a recreation changer when it’s accessible, however notice that it’s solely at the moment accessible in about 50 titles.

How are the RTX 4060 Ti’s energy effectivity and thermals?

Glorious.

We check 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 best studying on our Watts Up Professional meter, which measures the facility consumption of our total check system. The preliminary a part of the race, the place all competing vehicles are onscreen concurrently, tends to be essentially the most demanding portion. This isn’t a worst-case check; this can be a GPU-bound recreation operating at a GPU-bound decision to gauge efficiency when the graphics card is sweating laborious. When you’re enjoying a recreation that additionally hammers the CPU, you may see increased general system energy attracts. Think about your self warned.

The GeForce RTX 4060 Ti completely sips energy regardless of flinging out quick body charges, drawing 83 fewer watts than the RTX 3060 Ti regardless of being 7 to 9 p.c quicker in efficiency. And that idle energy is totally bonkers.

Brad Chacos/IDG

We check thermals by leaving GPU-Z open through the F1 22 energy draw check, noting the best most temperature on the finish.

No shock right here. The tiny two-slot Founders Version design does a fantastic job preserving this ultra-efficient GPU operating chilly. Customized 4060 Ti fashions with mammoth coolers would possibly run quieter—the FE delivers fantastic noise ranges, albeit not silent—however there’s no want to enhance on temperatures alone.  

Do you have to purchase the GeForce RTX 4060 Ti?

All of it relies on your reply to the query posed proper up prime: Are you keen to pay $400 for a 1080p graphics card with 8GB of reminiscence within the yr of our lord 2023?

The GeForce RTX 4060 Ti delivers completely excellent energy effectivity, main ray tracing efficiency, fashionable AV1 encoding, and quick 1080p gaming for prime refresh price displays, backed by Nvidia’s knockout software program suite: DLSS 3 Body Technology, Nvidia Reflex, RTX Video Tremendous Decision, and Nvidia Broadcast are simply some of the killer options accessible to the RTX 4060 Ti, with DLSS 3 solely being accessible on Nvidia’s latest GPU on this value phase. When you’re nonetheless on a GTX 1060 or RTX 2060, the RTX 4060 Ti will likely be a implausible improve (albeit costly).

It’ll be a tough promote for a lot of people, although—similar to the RTX 4080, 4070 Ti, and 4070 have been thus far this seemingly misplaced technology.

Brad Chacos/IDG

Nvidia pitches the GeForce RTX 4060 Ti as “a 1080p performance champ,” and in the event you’ve adopted together with our technical dissection and benchmarks, that’s precisely what it’s. Reminiscence and GPU configuration selections made by Nvidia preserve this GPU from flexing its muscle groups at 1440p (although you’ll be able to definitely recreation at 1440p in the event you don’t thoughts tweaking graphics settings). That’s exacerbated by its 8GB of onboard reminiscence.

As {Hardware} Unboxed not too long ago chronicled, 8GB of reminiscence isn’t sufficient to run all fashionable video games maxed-out now that builders are benefiting from the 16GB shared reminiscence swimming pools within the Xbox Collection X and PlayStation 5. In Nvidia’s personal early RTX 4060 Ti efficiency teasers, it needed to scale back graphics settings in A Plague Story: Requiem and Resident Evil Remake even at 1080p, whereas latest video games like The Final of Us and Hogwarts Legacy additionally wrestle with 8GB on the decrease decision except you drop settings.

Body Technology is fairly damned cool, however Nvidia’s vaunted DLSS upsampling loses fairly a little bit of its luster at 1080p (together with AMD’s rival FSR) due to the restricted quantity of pixels accessible at this decision. I wouldn’t run 1080p video games with DLSS or FSR energetic apart from the particular DLSS 3 Body Technology function. Even then, whereas DLSS 3 is off to a scorching begin, it’s solely at the moment accessible in round 50 video games.

The RTX 4060 Ti can be a deeply uninspiring improve gen-on-gen in terms of uncooked GPU horsepower, solely besting the RTX 3060 Ti by 9 p.c at 1080p decision and seven p.c at 1440p. It has fewer CUDA, RT, and tensor cores than its predecessor, which is disappointing. It flat-out loses to the RTX 3070 at 1440p, which is much more disappointing.

So: Are you keen to pay $400 for a 1080p graphics card with 8GB of reminiscence within the yr of our lord 2023? I’m not, particularly with DLSS/FSR benefits minimized on this phase. (Given the RTX 4060 Ti’s general efficiency, I don’t suppose the $500 16GB model will likely be very interesting when it launches in July both.)

When you have to purchase proper now, I’d quite spend my cash on a Radeon RX 6600 for round $200 if I used to be gaming on a 1080p/60Hz monitor, or go for the Radeon RX 6700 XT and its 12GB of reminiscence on a 256-bit bus for $350 in the event you’re desirous about dabbling in 1440p, nervous about 8GB of VRAM being sufficient at 1080p, and don’t thoughts giving up DLSS 3. (Intel’s $330 Arc A770 affords 16GB of VRAM, on that notice.) Because the RTX 4060 Ti doesn’t present a tangible efficiency improve over the RTX 3060 Ti and is slower than the RTX 3070, you may additionally have the ability to snap up both of these on sale to reap the benefits of Nvidia’s superior options, although notice that these each lack DLSS 3 help and are additionally restricted to 8GB of reminiscence, so search for steep gross sales.

That mentioned, I’d maintain my horses if I might. Nvidia already teased a $299 RTX 4060 with DLSS 3, AV1, and excessive energy effectivity for July. Plus, the rumor mill is screaming that AMD might launch a $300 Radeon RX 7600 any minute now. That value level is much more palatable for 1080p gaming on 8GB in the event you don’t want Nvidia’s deep function set.

The GeForce RTX 4060 Ti would have been extra interesting if it provided 16GB of reminiscence for $399 and ditched the 8GB choice, or provided 8GB of reminiscence with the identical degree of efficiency for $300 to $325. Because it stands, Nvidia’s RTX 40-series upgrades stay uninspiring at finest and this GPU sadly falls right into a no-man’s land of kinds. Look elsewhere.