According to a well-known hardware leaker, wxnod, Nvidia and graphics card makers will reportedly release the GeForce RTX 5060 Ti graphics cards on April 16 at 9 p.m. to compete against the best graphics cards. Meanwhile, VideoCardz has allegedly received and published the final specifications of Nvidia’s first mainstream Blackwell-based GPUs today. Keep in mind that the information does not come from official sources, so take it with a grain of salt.
Nvidia’s upcoming GeForce RTX 5060 Ti, purportedly expected around mid-April, may include two versions: 8GB and 16GB of GDDR7 memory—both potentially featuring the GB206-300 graphics processor with 4,608 CUDA cores and a 128-bit memory interface. The memory subsystem featuring 28 GT/s GDDR7 will provide a peak bandwidth of 448 GB/s, a 55% improvement compared to the GeForce RTX 4060 Ti. The boards will be rated to 180W total board power.
The increase in the number of CUDA cores, GPU clock, Blackwell microarchitecture optimizations, and software refinements almost guarantee a tangible increase in GeForce RTX 5060 Ti performance compared to its direct predecessor, the GeForce RTX 4060 Ti. But there is a major catch.
Nvidia GeForce RTX 5060 Ti Specifications
|
GPU Name |
RTX 5070 |
RTX 5060 of** |
Rtx 4060 of the |
RTX 5060* |
RTX 4060 |
RTX 5050* |
|---|---|---|---|---|---|---|
|
Family |
Blackwell |
Blackwell |
Ada Lovelace |
Blackwell |
Ada Lovelace |
Blackwell |
|
Board Name |
? |
PG152-SKU10/15 |
PG190-SKU361 |
PG152-SKU25 |
PG173-SKU371 |
PG152-SKU50 |
|
GPU Core |
GB205-300-A1 |
GB206-300-A1 |
AD106-350-A1 |
GB206-250-A1 |
AD107-400-A1 |
GB207-300-A1 |
|
CUDA COLORS |
6,144 |
4,608 |
4,352 |
3,840 |
3,072 |
2,560 |
|
SMs |
48 |
36 |
34 |
30 |
24 |
20 |
|
Bus Width |
192-bit |
128-bit |
128-bit |
128-bit |
128-bit |
128-bit |
|
Memory |
12GB |
16GB/8GB GDDR7 |
16GB/8GB GDDR6 |
8GB GDDR7 |
8GB GDDR6 |
8GB GDDR6 |
|
TGP |
250W |
180W |
165W/160W |
150W |
115W |
130W |
*Specifications are unconfirmed.
That catch is the GeForce RTX 5070, which is based on Nvidia’s relatively big GB205 GPU, configured to 6,144 CUDA cores for the desktop version. This GPU produces more compute oomph, and there is no chance that the GeForce RTX 5060 Ti will rival this product.
To that end, the appeal of the GeForce RTX 5060 Ti will depend on its price. At a $549 MSRP, the GeForce RTX 5070 looks entirely plausible (by today’s standards), though AMD’s Radeon RX 9070-series AIBs are on the horizon and cannot be owned at their recommended prices. If Nvidia’s GeForce RTX 5060 Ti starts at $499 MSRP, it may not be as appealing as the GeForce RTX 5070 for $549.
Given uncertainties about performance and price, let us briefly discuss compatibility. If unofficial reports about the GeForce RTX 5060 Ti are correct, it is rated for 180W, 150W from an eight-pin PCIe auxiliary connector, 30W from the slot, or 180W from a single 16-pin power plug. The RTX 5070 model is now rated for 250W total graphics board power, so it will require a better power supply.
In any case, remember that we are talking about specifications that do not come from an official source, so take this information with a grain of salt.
