We have posted a total of four GeForce RTX 3070 Ti reviews today: MSI RTX 3070 Ti Suprim X, NVIDIA RTX 3080 Ti Founders Edition, Palit RTX 3070 Ti GameRock OC, and Zotac RTX 3070 Ti AMP Holo.

Just last week, NVIDIA launched their GeForce RTX 3080 Ti, and today, we have the RTX 3070 Ti reviews. Both cards were announced earlier this month at Computex. With the GeForce RTX 3070 Ti, NVIDIA is pushing their Ampere x70 lineup forward to better compete with AMD's Radeon RX 6700 XT and RX 6800 non-XT. To achieve their goal, NVIDIA is using the same GA104 GPU as on the RTX 3070, but with all its 6144 cores enabled. The GeForce RTX 3070 non-Ti has 5,888 active cores, a 4% difference, which by itself isn't big enough to justify a new SKU. That's why NVIDIA switched the memory chips from GDDR6 to GDDR6X, which improves memory bandwidth by 35%.

Memory size on the GeForce RTX 3070 Ti has remained at 8 GB, probably because the only other feasible option is 16 GB, which would have increased cost significantly without major performance improvements. The underlying reason is that VRAM capacity is tied to the memory bus width on the card. In theory, a 12 GB 192-bit design like the RTX 3060 is possible, but the performance loss due to the narrower memory bus would more than negate any gains from the larger memory buffer. On the other hand, AMD is offering 16 GB VRAM on the Radeon RX 6800, so NVIDIA achieving parity would have certainly had a psychological effect. Personally, I'm not a fan of going all out on VRAM size, none of our benchmarks show any noteworthy performance issues arising from 8 GB VRAM capacity. Actually, it seems likely DirectStorage, a technology that was first pioneered on the new consoles, will reduce VRAM pressure by optimizing the disk to GPU memory path.

Averaged over our 22-game-strong test suite at 1440p resolution, the NVIDIA GeForce RTX 3070 Ti Founders Edition achieves a 5% performance lead over the RTX 3070 non-Ti, 7% at 4K. To be honest, that's less than what I expected from this new model, especially considering the investment by NVIDIA: full GA104 GPU, GDDR6X, new PCBs, and a completely new cooler design. Compared to the Radeon RX 6800, the gap shrinks to only 4%, and the RX 6700 XT is clearly beat with a 13% performance difference. GeForce RTX 3080 is 15% faster than the RTX 3070 Ti, and the newly released RTX 3080 Ti is 24% faster. Last generation's GeForce RTX 2080 Ti flagship is 10% behind the RTX 3070 Ti, and the difference to the RTX 2070 Super is 30%.

With those performance numbers, the RTX 3070 Ti is the perfect choice for the huge 1440p gamer crowd, but the card also has enough muscle to drive many titles at 4K 60 FPS, especially if you are willing to dial down settings a little bit. The RTX 3070 Ti is also a great choice for 1080p Full HD if you want to drive a high-refresh-rate monitor with 120 or 144 Hz. For just 1080p at 60 Hz, it's overkill unless next-generation titles go overboard with their hardware requirements, which is highly unlikely. Raytracing performance of the RTX 3070 Ti is better than the Radeon RX 6800 because NVIDIA executes more raytracing functions in hardware and is on their second-generation of the technology. Differences vary between titles, though. The new consoles are built using AMD RDNA 2 technology, so going forward, game developers may invest more resources into optimizing RT for AMD's architecture, or they simply dial down the RT effects to reduce the performance hit, which is what happened recently with Resident Evil 7.

For the GeForce RTX 3070 Ti Founders Edition, NVIDIA designed a completely new FE cooler that brings the through-flow technology we've seen on the RTX 3080 and 3090 to the RTX 3070 Series. Visually, the design is easily recognizable as an Ampere Founders Edition card—this card is beautiful. The engineering behind the thermal solution is also impressive; everything has been crafted with fantastic precision and lots of thought for little details. NVIDIA also went the extra mile to make their card easy to disassemble/reassemble, no glue is used. While the new cooler is definitely an improvement over the RTX 3070 FE—it lands roughly in the middle between the RTX 3070 FE and RTX 3080 FE, as shown by our heatsink apples-to-apples comparison test—I still feel like it's a little bit on the weak side. Temperatures are "just OK," reaching 82°C GPU under full load. The better approach would have been to simply use the RTX 3080 FE cooler, a design that already exists, so a lot of development cost could have been saved. Noise levels are slightly high, too. With 40 dBA, the card is well audible, but not "loud." However, the RTX 3070 non-Ti Founders Edition runs much quieter at only 34 dBA, which is very quiet. AMD's RDNA 2 reference designs are all quieter than the RTX 3070 Ti, too. With Ampere, NVIDIA introduced idle fan stop on their Founders Edition, which previously was one of the biggest reasons for gamers to look at other brands. In idle, during desktop work, internet browsing, and light gaming, the RTX 3070 Ti will turn off its fans completely for the perfect noise-free experience.

The underlying reason why the cooler seems a bit on the weak side is that the RTX 3070 Ti uses a lot of power. We measured the RTX 3070 non-Ti to consume only 220 W—the most energy-efficient Ampere design so far. The RTX 3070 Ti consumes 300 W on the other hand, which is an 80 W increase, or 36%. If you compare that with the 7% performance difference you'll have to wonder where NVIDIA lost all this efficiency. One part of that is the GDDR6X memory, which is known to use more power than GDDR6. This isn't the only reason, though. NVIDIA has increased the GPU voltage of the RTX 3070 Ti vs. what we saw on RTX 3070—the reason seems to be that NVIDIA is clocking their chip higher than on any other GA104 card before. I also suspect that these chips didn't make the cut for the energy efficient Ampere laptop design.

While comparing voltage-frequency curves of RTX 3070 Ti with RTX 3070, I noticed that the RTX 3070 actually runs at higher clock frequency on average than the RTX 3070 Ti: 1861 MHz vs. 1882 MHz even though the rated boosts suggest the opposite: 1770 MHz vs. 1725 MHz. In NVIDIA's defense, actual clocks are still higher than the boost rating, which is more of a "rough indication." I still find it surprising that a +45 MHz promise turns into a -21 MHz difference when measured. This looks like the missing 4% performance needed to catch up to the RX 6800.

NVIDIA has announced a $600 price point for the RTX 3070 Ti, which is $100 higher than the RTX 3070. If we only look at MSRPs, even that $100 increase is hard to justify. The 7% performance increase can account for $40.Technically, the cooler is better, but these improvements get cannibalized by the higher power consumption, and the net result is a louder card that runs higher temperatures. In the current market, MSRPs are irrelevant, and any responsible reviewer must take into account the craziness happening out there: the RTX 3060 sells for $900, RTX 3070 for $1300, RX 6700 XT for $1000, and RX 6800 for $1400. That's why I think a more realistic price for RTX 3070 Ti is the same $1300 the RTX 3070 goes for. Being faster than the latter would justify a price premium, but I feel the low-hash-rate (LHR) mining limiter on the RTX 3070 Ti will reduce its perceived (resale) value slightly. Don't get me wrong, I support the mining limiter 100%. We need graphics cards in the hands of gamers to ensure the PC ecosystem can thrive and people don't just buy consoles. It also puts a new kind of value proposition on the RTX 3070 Ti: If neither RTX 3070 nor RX 6800 are in stock and the RTX 3070 Ti is available, I'm sure many gamers will be happy to swallow the higher power consumption pill of the RTX 3070 Ti.

I can recommend the card if power, heat, and noise are only secondary to you, and you can find the RTX 3070 Ti at a reasonable price, whatever that may be in your estimation, but only barely. Strong alternatives are the RTX 3070 non-Ti, Radeon RX 6800, and RX 6700 XT.