GeForce RTX 4060 spotted on Geekbench with the same performance uplift as RTX 3060/2060
NVIDIA RTX 4060 is set to launch next week.
The first AD107 based desktop graphics card has been pushed forward by NVIDIA to launch this month. The company and its partners should now be shipping review samples to media and influencers around the world. This means that first benchmark results may now start to randomly show up, on websites such as Geekbench.
According to BenchLeaks, the first RTX 4060 tests are already there, including compute tests with OpenCL and Vulkan APIs. The graphics card has allegedly scored 105K points in OpenCL test and 99K in Vulkan. We can quickly put those results against popular mid-range GPUs, including the recently launched RTX 4060 Ti and Radeon RX 7600 GPUs:
| Geekbench 6 Compute Benchmarks | ||
|---|---|---|
| VideoCardz.com | Vulkan API ↓ | OpenCL API |
| GeForce RTX 4060 Ti | ||
| GeForce RTX 3060 Ti | ||
| GeForce RTX 4060 | ||
| Radeon RX 7600 | ||
| GeForce RTX 3060 | ||
| GeForce RTX 2060 | ||
Please note that this data is based on official Geekbench ranking. The ranking averages all results, so graphics cards which were available for longer time will have more accurate data.
Based on this data, one can see that RTX 4060 is 17 to 18% faster than RTX 3060. Although this may appear like a low result, the same performance uplift was observed with generation upgrade from Turing RTX 2060 to Ampere RTX 3080 (18% on average). Furthermore, the GPU is 4% (Vulkan) to 32% (OpenCL) faster than Radeon RX 7600 featuring Navi 33 GPU. Of course, this goes without saying, but Geekbench data will not always correspond to gaming workloads. What we are looking at are simply raw compute benchmarks with relatively short testing duration.
GeForce RTX 4060 tested, Source: Geekbench
The RTX 4060 features an AD107 GPU with 3072 CUDA cores and 8GB of GDDR6 memory attached to a 128-bit memory. This graphics card will use the same interface as the RTX 4060 Ti: PCI Express 4.0 limited to 8 lanes, which might cause a performance bottleneck on older Gen3 platforms. This should not affect the results above, as the platform used for those tests features a high-end ASUS Z790 ROG APEX motherboard combined with a Core i5-13600K CPU and DDR5-6000 memory.
NVIDIA has put a $299/€329 price tag on RTX 4060, which is just about €30 more than RTX 3060 with 12GB memory and 192-bit interface.
| NVIDIA GeForce RTX 40 Series Specs | |||
|---|---|---|---|
| VideoCardz.com | GeForce RTX 4060 Ti 16GB | GeForce RTX 4060 Ti 8GB | GeForce RTX 4060 |
| Picture |  |  |  |
| Board/SKU | PG190 SKU 363 | PG190 SKU 361 | PG173 SKU 371 |
| Architecture | Ada (TSMC 4N) | Ada (TSMC 4N) | Ada (TSMC 4N) |
| GPU | AD106-351 | AD106-350 | AD107-400 |
| CUDA Cores | |||
| Boost Clock | |||
| Memory | |||
| Memory Clock | |||
| Memory Bus | |||
| Memory Bandwidth | |||
| TDP | |||
| Interface | PCIe Gen4 x8 | PCIe Gen4 x8 | PCIe Gen4 x8 |
| MSRP | |||
| Release Date | July 2023 | May 24, 2023 | June 29, 2023 |
Source: Geekbench (Vulkan) (OpenCL) via @BenchLeaks
