In a normal GPU marketplace, Nvidia’s new GPU—the RTX 3070 Ti—would land either as a welcome jump or a power-per-watt disappointment. In the chip-shortage squeeze of 2021, however, both its biggest successes and shortcomings may slip by without much fanfare.
The company’s RTX 3070 launched eight months ago at an MSRP of $499, and it did so at an incredibly efficient power-to-performance ratio. There’s simply no better 220 W GPU on the market, as the RTX 3070 noticeably pulled ahead of the 200 W RTX 3060 Ti and AMD’s 230 W RX 6700XT. That efficiency, unsurprisingly, isn’t repeated with the new model released this week: the RTX 3070 Ti. This device’s MSRP jumps 20 percent (to “$599,” but mind the scare quotes), and its TDP screams ahead at 32 percent. We’ve been here before, of course. “Ti”-branded Nvidia cards aren’t usually as power-efficient as their namesakes, and that’s fine, especially if a mild $100 price jump yields a solid increase in performance.
But the RTX 3070 Ti spec sheet doesn’t see Nvidia charge ahead in ways that might match the jump in wattage. And while the 3070 Ti’s performance mostly increases across the board, the gains aren’t in any way a revolution. That may be less about Nvidia’s design prowess and more about squeezing this thing between the impressive duo of the RTX 3070 and RTX 3080 ($699) on an MSRP basis.
This new card’s performance, unfortunately, doesn’t land as evenly between those two GPUs.
Commonly around seven percent faster
The new GPU’s spec-sheet increases include an upgrade in VRAM technology from the 3070’s GDDR6 to GDDR6X, and there are memory-spec upgrades to match—particularly a 26 percent jump in memory bandwidth. That’s otherwise the same amount of VRAM as the 3070 (8GB). Meanwhile, the rest of the spec sheet is mostly a mild increase compared to the 3070 Ti’s predecessor card. CUDA cores, texture units, tensor cores, and RT cores all increase an identical 4.16 percent over the RTX 3070. Boost clocks increase slightly less than that, and ROPs remain static.
Depending on the test, my RTX 3070 Ti review hardware can jump ahead as much as 14 percent over its older namesake, which is great (and likely the headline that Nvidia would prefer). But those gains are outliers. Just as often, test results land almost tied. The most common result is the 3070 Ti winning at closer to a seven percent increase over the 3070. The good news for the Ti, at least, is its more stable frame rates, as evidenced by the below benchmarks’ solid “one percent low” stats.
The 3070 Ti is decidedly different from last week’s RTX 3080 Ti, which had no problem proving itself to be an across-the-board upgrade over the cheaper RTX 3080. I measured the gains in those tests at 10 percent and up—though, again, the MSRP gap between those two cards was much higher ($500). And speaking of which: the RTX 3080 stomps far ahead of the RTX 3070 Ti, as the below charts make abundantly clear.
The 3070 Ti’s average gain count over the 3070 applies whether ray tracing and DLSS are in the picture or not, with Cyberpunk 2077’s version 1.2 standing out as an exception. In my custom benchmark of that CD Projekt Red title (which sees me walking forward for 60 seconds through the game’s first bustling, open-air zone), I have recorded gains close to 40 percent over the RTX 3070, whether with DLSS on or off. Thanks to Cyberpunk 2077‘s notoriously buggy state, I do not recommend that GPU shoppers pin their ray-tracing performance hopes on that benchmark result. However, it does line up with my ability to run the game at a DLSS-boosted 1440p resolution, with ray tracing set to near-max, while enjoying surprisingly crisp imagery at 60 fps. Using the same “automatic” DLSS upscale to 1440p with the base RTX 3070 is slightly less stable and noticeably blurrier.
During my review process, I ran into bizarre RTX 3070 Ti behavior at first: its performance fell well behind the RTX 3070. This came without any apparent hardware noises, visual flickering, or other signs of losing the GPU lottery. During my initial tests, 3DMark’s synthetic benchmarks did not suffer, suggesting an issue with Windows or drivers. Unhooking everything, reseating parts, and trying again didn’t fix the problem.
As it turns out, my only solution was to enable, and then disable, the Windows 10 setting for “hardware-accelerated GPU scheduling.” At first, this had me convinced that I needed to leave that setting enabled for the 3070 Ti to function properly. Eventually, after more trial and error than I cared to deal with, I learned that the HWAGS setting was ultimately not necessary (and it alternated between boosting and reducing GPU performance, depending on the game). In the end, I disabled it. There’s a chance something else contributed to the 3070 Ti’s initial weird performance, but I can no longer replicate the issue.
But in this GPU universe…
The RTX 3070 and 3070 Ti don’t live in isolation, and the newer card’s mostly meager increases may look fantastic to AMD loyalists who have purchased an RDNA 2-series GPU in the past year. As you can see in the above spec tables, the RX 6800, which debuted last November at an MSRP of $579, absolutely contends with Nvidia’s similarly priced RTX 3070 Ti. Meanwhile, $70 more in MSRP gets you some serious gains in the form of AMD’s RX 6800XT. And depending on what universe you live in, your choice may boil down to whether you believe in the promise of ray tracing and Nvidia’s proprietary DLSS. AMD simply can’t compete with that yet.
But I mention “universes” because, as I have typed a lot as of late, this GPU universe is largely defined by chip shortages and auction prices. Until GPU prices and supply stabilize, Nvidia can admittedly call a mulligan on this GPU’s disappointments—and that’s a pretty loud indication that Nvidia has entered the not-so-fast lane for the time being. What’s the point of aggressively competing with updated SKUs when a boilerplate release can turn a Best Buy parking lot into a BTS-like crowd frenzy?
Still, if you’re eager to upgrade your GPU, are somehow able to snap up an RTX 3070 Ti at a normal retail price, and can stomach a VRAM maximum of 8GB, you can expect a moderately future-proofed 1440p performer, right around the same performance level as the 3070. Let’s be clear—whether it exceeds or meets the original 3070, the 3070 Ti is still doing significant work, whether by leveraging Nvidia’s ray tracing-specific cores or by supporting DLSS as an upscaling champion. (AMD’s new rival standard doesn’t look like it’s going to come close any time soon.)
The same cannot be said for 4K performance. That 8GB VRAM cap will dash your dreams of the highest-res textures. And the rest of the spec sheet doesn’t necessarily scale to raw 4K rasterization, unless you play with settings or luck out with your favorite game(s) supporting the largely successful DLSS standard.
If Ars’ comment sections are any indication, readers had been holding out hope that an eventual “3070 Ti” would mostly copy the original RTX 3070’s spec sheet, only with more VRAM. The 3080 Ti kinda-sorta delivered in that respect by jumping from its predecessor’s 10GB of VRAM to 12GB. But the 3070 Ti instead insists that its upgrade potential makes more sense with faster VRAM, not more. In some test results, that bears out. But that decision arguably does not pan out in enough results—and certainly not when compared to how much an additional $100 nets you from an RTX 3080.
Listing image by Sam Machkovech