In a new tweet, Kopite7kimi says that NVIDIA’s GeForce RTX 4060 graphics cards may not be very fast, possibly due to the utilization of the x8 PCI interface. The leaker says that the NVIDIA AD106 GPU, which should be at the heart of the RTX 4060 series of graphics cards, is not very powerful. We don’t know which specific configuration is being talked about, but it’s likely he’s referring to the full-blooded model.
A supposed 3DMark Time Spy Extreme benchmark has been shared and the GPU is said to score under 7,000 points. Such scores would make the AD106 GPU slower than the RTX 3070 Ti and only comparable to the RTX 3070, which doesn’t seem like the huge upgrade that the high-end AD102 and AD103 GPUs should get.
With Time Spy Extreme scores between 6500-7000, this should make the AD106 GPU (GeForce RTX 4060) 18% faster than the RTX 3060 Ti and 69% faster than the RTX 3060. Now, a performance increase of over 50% is pretty huge, but it’s not as big as the 2x increase we see with high-end cards. If a 50-60% boost is seen in synthetic benchmarks, then the corresponding games will only see a 30-40% performance boost at best. So it looks like we’ll be getting RTX 3070 levels of performance in the next **60 card.
In addition, the leaker mentions that the AD106 and AD107 “Ada Lovelace” GPUs will use the PCIe x8 interface. This makes sense considering that these GPUs are mainly aimed at the mobile segment (laptops/notebooks). For desktops, this means that the GeForce RTX 4060 and 4050 will get an x8 interface, which should not be a problem for most PCs considering that AMD and Intel’s next generation systems have 5th generation PCIe slot support while running on older 3rd generation platforms will result in reduced bandwidth and transfer speeds.
NVIDIA GeForce RTX 4060 “expected” specifications
Based on NVIDIA’s decision to mix AD104/AD106 or use AD106 in its RTX 4060 series lineup, we could see a neutered RTX 4060 or a full-blooded version, while the Ti could use a neutered AD104 configuration. the AD106 GPU will come with 32MB of L2 cache and up to 48 ROPs.
Previously leaked 3DMark Time Spy benchmarks for the NVIDIA GeForce RTX 4090 graphics card were reported to clock up to 3GHz, which is two times faster than the RTX 3090.
The clock speed of the new card has not been confirmed, but given that the TSMC 4N process is being used, we expect the clock to be between 2.0-3.0 GHz. The higher than usual clock speeds come from the fact that NVIDIA is making a two-node jump, considering that the Ampere GPU using Samsung’s 8nm node is actually a 10nm process node with some optimizations. NVIDIA is skipping 7nm and going straight to a 5nm node, not even a normal variant but an optimized version of it. NVIDIA made a huge frequency leap on TSMC’s 16nm node Pascal, and we can expect a similar leap this time.
As for memory specifications, the GeForce RTX 4060 is expected to use 8GB of GDDR6X capacity, possibly at a faster 20+ Gbps with a 128-bit bus interface to achieve over 320GB/s of bandwidth. If the “Ti” variant ends up with an AD104 GPU, it could provide 12GB of memory via a 192-bit bus interface. The GeForce RTX 4060 card is said to have a TGP of 220W, a 30% increase over the RTX 3060 and a 10% increase over the RTX 3060 Ti. This is an increase that counts for a lot, and the performance needs to be very good for NVIDIA to maintain its efficiency ratio.
