The bulk of PC gamers buy and use the sub-$500 graphics cards that Nvidia is starting to introduce with its new Ada Lovelace GPU architecture. Both good and terrible news are available.
The GeForce RTX 4060 Ti will be available on May 24 for $399, which is the same price as the RTX 3060 Ti’s launch price in December 2020. This is wonderful news because it is the first GPU introduction in a long time without an automatic price increase. It supports DLSS 3’s AI-assisted frame rate-boosting technology and AV1 video encoding compatibility, like other RTX 4000-series GPUs, and is far more power-efficient than its predecessors. The bad news is that, while having the same 8GB bank of RAM as the 3060 Ti and (according to Nvidia) just slightly better performance than the 3060 Ti in games without the DLSS Frame Generation function activated, the 4060 Ti is an extraordinarily modest boost over the 3060 Ti.
The Founders Edition 4060 Ti from Nvidia uses the same 12VHPWR connector as the Founders Edition RTX 4070, while cards from Nvidia’s partners sometimes choose to use a conventional 8-pin power socket instead.
Later in July, Nvidia will complete the 4060 portfolio with a $499 16GB 4060 Ti that will sit between the 8GB model and the $599 RTX 4070. It’s not a bad offer for the first generation-over-generation price drop we’ve seen from Nvidia in years; the lower-end non-Ti RTX 4060 will also go on sale for $299, a slight reduction over the 3060’s $329 launch price for what Nvidia claims should be 20% better performance. Due in part to the fact that 8GB of graphics RAM is starting to become a bottleneck in newer games at higher resolutions, the 8GB 4060 Ti and 4060 are both marketed as 1080p cards.
Meet AD106
RTX 4090 | RTX 4080 | RTX 4070 Ti | RTX 4070 | RTX 4060 Ti | RTX 4060 | RTX 3060 Ti | RTX 3060 | |
---|---|---|---|---|---|---|---|---|
CUDA Cores | 16,384 | 9,728 | 7,680 | 5,888 | 4,352 | 3,840? | 4,864 | 3,584 |
Boost Clock | 2,520 MHz | 2,505 MHz | 2,610 MHz | 2,475 MHz | 2,535 MHz | ? | 1,665 MHz | 1,777 MHz |
Memory Bus Width | 384-bit | 256-bit | 192-bit | 192-bit | 128-bit | 128-bit | 256-bit | 192-bit |
Memory Clock | 1,313 MHz | 1,400 MHz | 1,313 MHz | 1,313 MHz | 2,250 MHz | ? | 1,750 MHz | 1,875 MHz |
Memory size | 24GB GDDR6X | 16GB GDDR6X | 12GB GDDR6X | 12GB GDDR6X | 8GB or 16GB GDDR6 | 8GB GDDR6 | 8GB GDDR6 | 12GB GDDR6 |
TGP | 450 W | 320 W | 285 W | 200 W | 140 W | 110 W | 200 W | 170 W |
The AD106 GPU die, the fourth and smallest (so far) of the chips Nvidia has released in the RTX 4000 series after the AD102 (4090), AD103 (4080), and AD104, serves as the foundation for the whole 4060 series (4070 series). The RTX 4070 laptop variants also employ the AD106 die.
The AD106 chip has 4,608 Nvidia CUDA cores on a 128-bit memory interface when fully enabled, but both 4060-series GPUs use a partially disabled version of this architecture (see the table above for details). When directly compared to the 3060 Ti, which featured more CUDA cores and a 256-bit bus, the 4060 Ti’s specifications are a little lacking, as was expected, which adds to its less-than-average improvement over the previous-generation card. The remaining RTX 4000-series GPUs have typically had as many CUDA cores as their RTX 3000-series counterparts, while every GPU aside from the 4090 has historically had shorter memory bus widths.
The 4060 Ti gains from the architectural advancements of Ada as well as noticeably faster clock rates to bridge that gap. In order to lessen the impact of the smaller memory bus, Nvidia has also introduced additional L2 cache. This will reduce the amount of times the GPU must access main memory (the 4060 Ti has 32MB of L2 cache, and the 4060 has 24MB, compared to 4MB and 3MB for the 3060 Ti and 3060).
The majority of these specifications were not directly compared in Nvidia’s press conference or PowerPoint deck; instead, generation-over-generation specifications were compared using TFLOP figures. People who follow GPU news online are unable to remain calm when they believe that a product has been named incorrectly, positioned incorrectly, or priced incorrectly, so it is obvious that the intention is to drive home the point that the 4060 Ti is a faster card than the 3060 Ti despite these specs (expect a fair amount of outrage about this to be expressed via YouTube thumbnails).
The RTX 4070’s power efficiency was one of the features that most pleased us, and both RTX 4060 GPUs appear to perform admirably in that regard as well. The TGP of the 4060 Ti is 140 W as opposed to the 3060 Ti’s 200 W, whereas the 4060 only requires 110 W as opposed to 170 W. That’s dangerously close to the 75 W of power that a PCI Express slot can deliver without an external power connector; a hypothetical RTX 4050 could reach that level, which may be good news for those trying to install a GPU in a low-power budget desktop tower from Dell or HP that doesn’t have any spare 8-pin power connectors.