Nvidia has released the RTX 4060 Ti 8G graphics card, and its memory interface width is only 128bit. In contrast, Nvidia’s old RTX 3060 Ti is 256bit wide, and the RTX 3060 is also 192bit wide.
In this regard, Nvidia responded in a blog post, saying that compared with the previous generation GPU with 128-bit memory interface width, the storage subsystem of the new NVIDIA Ada Lovelace architecture increased the size of the L2 cache by 16 times, which greatly increased the size of the L2 cache. Improved cache hit ratio.
As shown in the figure above, the L2 cache bandwidth in Ada GPUs has increased significantly. This makes it possible to transfer more data between the processing core and the L2 cache.
Additionally, Nvidia engineers tested the RTX 4060 Ti with 32 MB of L2 cache versus a special test build of the RTX 4060 Ti that only used 2 MB of L2.
In various games and synthetic benchmarks, a 32 MB L2 cache reduces video memory bus traffic by an average of over 50% compared to the performance of 2 MB L2 cache. This 50% reduction in traffic allows the GPU to use its memory bandwidth more efficiently, up to a nearly 2x increase in efficiency. So, in this case, isolating memory performance, an Ada GPU with 288 GB/s peak memory bandwidth performs similarly to an Ampere GPU with 554 GB/s peak memory bandwidth. Across a range of games and synthetic tests, the greatly improved cache hit rate increased game frame rates by up to 34%.
According to Nvidia, memory bandwidth has historically been used as an important indicator for determining the speed and performance level of new GPUs. However, memory bandwidth alone is not an adequate indicator of memory subsystem performance. Instead, a more complete understanding of storage subsystem design and its overall impact on game performance is helpful.
GeForce RTX 4060 Ti takes advantage of the 256-bit GeForce RTX 3060 Ti and RTX 2060 SUPER from the previous generation thanks to the benefits of the Ada architecture, including new RT and Tensor cores, higher clock speeds, new OFA engine and Ada’s DLSS 3 capabilities Graphics cards are faster while consuming less power.
The RTX 4060 Ti and RTX 4060 graphics card parameter information as follows:
RTX 4060 Ti has 4352 CUDA cores, is equipped with 8GB / 16GB 128bit GDDR6 video memory, TGP power consumption 160W / 165W, using PCIe 4.0*8 connection, the price starts at 3199 CNY and it will go on sale on May 24.
The RTX 4060 graphics card has 3072 CUDA cores, is equipped with 8GB GDDR6 128bit video memory, consumes 115W, and uses PCIe 4.0*8 connection. The price starts at 2399 CNY and will be available in July.
