According to a Twitter post from @Zed__WangNvidia’s (vanilla) RTX 4060 8GB will use the company’s entry-level Ada Lovelace GPU die known as AD107, rather than the RTX 4060 Ti’s larger AD106 die. While this doesn’t affect the GPU’s official specs, the fact that the RTX 4060 can be used with his AD107 die is Nvidia’s focus on improving power efficiency in this generation’s mid-range and entry-level GPUs. indicates that
Nvidia’s decision to use the AD107 inside the RTX 4060 is neither a leak nor a rumor. In fact, Nvidia has directly told us that in its new mid-range GPUs he will officially use the AD107 die. As a result, this will be his first RTX xx60 class product to work on an entry-level die. Previous generation mid-range RTX GPUs such as the 3060 and 2060 ran on He GA106 or TU106 dies, one tier above their entry-level counterparts ending in the number 7.
The GPU is supposed to use a fully enabled AD107 die. As a result, there is no GPU model better than the RTX 4060 with better specs on the same die such as core count, memory bus configuration, ROP, cache, etc. There could be an ‘RTX 4060 Super’ with higher TDP and clocks, but there is no other GPU with higher physical specs on this die.
| graphics card | RTX4060 | RTX3060 |
|---|---|---|
| architecture | AD107 | GA106 |
| process technology | TSMC 4N | Samsung 8N |
| Transistor (billion) | 18.9 | 12.0 |
| Die size (mm^2) | 158.7 | 276 |
| SMS | twenty four | 28 |
| GPU core (shader) | 3072 | 3584 |
| tensor core | 96 | 112 |
| RT core | twenty four | 28 |
| Boost clock (MHz) | 2460? | 1777 |
| VRAM Speed (Gbps) | 17 | 15 |
| VRAM (GB) | 8 | 12 |
| VRAM bus width | 128 | 192 |
| L2 cache | twenty four | 3 |
| ROP | 32 | 48 |
| TMU | 96 | 112 |
| TFLOPS FP32 (Boost) | 15 | 12.7 |
| TFLOPS FP16 (FP8) | 121 (242) | 102 (sparse) |
| Bandwidth (GBps) | 272 (valid 453) | 360 |
| TGP (Watts) | 115 | 170 |
| release date | July 2023 | February 2021 |
| Release price | $299 | $329 |
For reference, the RTX 4060 specs include 3072 CUDA Cores, 24 SMs, 96 Tensor Cores, 24 RT Cores, 24 MB of L2 Cache, 48 ROPs, 96 TMUs, and a 128-bit bus. Additionally, the boost clock is rated at 2460 MHz and the GPU power target is 115W.
The AD107 gives Nvidia some advantages of the RTX 4060. For one, GPUs are cheaper to manufacture, as they run on the smallest and easiest-to-manufacture GPU die in the Ada Lovelace family. Another is that Nvidia could offload production to another GPU die in the future if it needed to improve yields. For example, Nvidia plans to leverage both AD107 and AD106 in future vanilla RTX 4060s to improve yield and reduce waste of potentially bad AD106 dies that are unusable with all cores turned on. may choose.
For enthusiasts and gamers alike, it would be a shame to see the performance potential the RTX 4060 could have had with Nvidia’s larger AD106 die being regulated to an AD107 die. But at least you get an incredibly efficient GPU that’s easy to power and cool.
