Nvidia has officially announced the RTX 4060 and RTX 4060 Ti 8GB and 16GB graphics cards. It has also listed the specifications for them.
A day ago, we got to know through leaks about the specifications and the prices for the Nvidia GeForce RTX 4060 and 4060 Ti graphics card series. Looks like those leaks are almost spot on.
Now, Nvidia has officially announced the cards in the RTX 4060 series, which Nvidia claims is made for the 1080p gamers. Let us take a look at them.
RTX 4060 Ti Specs
The Nvidia GeForce RTX 4060 Ti 8GB is the first graphics card which is going to launch in this series.
Both the RTX 4060 Ti 8GB and 16GB versions are going to come with 4352 CUDA Cores. They will run at a base clock of 2.31 GHz and will boost up-to 2.54 GHz.
They will come with 8GB / 16GB GDDR6 VRAM having a 128-bit memory bus. The RTX 4060 Ti will have a bandwidth of 288 GB/s. However, Nvidia claims that as the L2 cache has been increased from 4MB to 32MB, the effective bandwidth is 554 GB/s.
They have a TGP of 160W (8GB) / 165W (16GB) and will consume just 7W at idle.
Nvidia claims that the shaders in the RTX 4060 Ti will perform at 22 TFLOPs, compared to 16 TFLOPs in RTX 3060 Ti and 7 TFLOPs in RTX 2060 Super.
In terms of performance, Nvidia claims that RTX 4060 Ti is just 15% faster than 3060 Ti without DLSS 3 frame generation. However, with DLSS 3 frame generation, that number increases to 70%.
The RTX 4060 Ti 8GB will cost $399 and will release on 24th May. Expect the reviews to be out on 23rd May.
The RTX 4060 Ti 16GB will cost $499 and will release in July along with the RTX 4060.
Nvidia claims there’s no difference in the performance between 8GB and 16GB models in the RTX 4060 Ti. Except in some games, where larger VRAM allows one to play at higher settings.
Tom’s Hardware reports that a lot of card makers have announced their version of the RTX 4060 Ti 8GB graphics cards by now.
RTX 4060 Specs
Nvidia GeForce RTX 4060 (non-Ti) will release along with RTX 4060 Ti 16GB version.
The official specifications for the RTX 4060 seem to be matching exactly with the leaked specs rumored in mid-February.
The RTX 4060 will come with 3072 CUDA cores. It will run at a base clock of 1.83 GHz and will boost up-to 2.46 GHz.
It will come with 8GB of VRAM with a 128-bit memory bus. The RTX 4060 will have a memory bandwidth of 272 GB/s. However, Nvidia claims that, just like RTX 4060 Ti, due to larger 24MB L2 cache, the effective bandwidth is higher at 453 GB/s.
It has a TGP of 115W and uses 7W at idle. Interestingly, Nvidia recommends a 550W PSU for all the graphics cards in the RTX 4060 series.
In terms of raw numbers, RTX 4060’s shaders will perform at 15 TFLOPs, compared to 13 TFLOPs of RTX 3060 and 7 TFLOPs of RTX 2060.
In terms of performance, Nvidia claims that RTX 4060 is just 20% faster than 3060 without DLSS 3 frame generation. However, with DLSS 3 frame generation, that number increases to 70%.
The RTX 4060 will have a MSRP of $299 and will release in July.
Overall Specs
The overall specs for the RTX 4060 series, compared to the RTX 3060 series, is shown in the charts below.
| Model | RTX 4060 Ti | RTX 3060 Ti | RTX 4060 | RTX 3060 |
|---|---|---|---|---|
| Process | 4 nm | 8 nm | 4 nm | 8 nm |
| CUDA Cores | 4352 | 4864 | 3072 | 3584 |
| Base Clock | 2.31 GHz | 1.41 GHz | 1.83 GHz | 1.32 GHz |
| Boost Clock | 2.54 GHz | 1.66 GHz | 2.46 GHz | 1.77 GHz |
| VRAM Size | 8GB / 16GB | 8 GB | 8 GB | 12 GB |
| Memory Bus | 128-bit | 256 bit | 128-bit | 192 bit |
| Memory Bandwidth | 288 GB/s (554 GB/s)* | 448.0 GB/s | 272 GB/s (453 GB/s)* | 360.0 GB/s |
| L2 Cache | 32 MB | 4 MB | 24 MB | 3 MB |
| TGP | 160 / 165W | 200 W | 115 W | 170 W |
| MSRP | $399 / $499 | $399 | $299 | $329 |
* Nvidia claims Memory bandwidth in RTX 40 series is different from Effective Memory Bandwidth due to larger L2 Cache.
As one can look above. Almost all the specs in the RTX 4060 series are a downgrade when compared to the RTX 3060 series. The only exception is the newer cards are clocked higher and have a larger L2 cache to compensate for lesser powerful specs.
Nvidia Counters VRAM Debate
Nvidia has come a under a lot of criticism these days for being ungenerous when it comes to providing enough VRAM in their graphics cards. Games these days are using a huge amount of VRAM, so much so that 8GB of VRAM is not enough for 1080p. Something which Nvidia is marking these xx60 cards for.
So to counter that, Nvidia has published a blog suggesting that graphics cards won’t require that much VRAM with larger L2 cache that RTX 40 provides.
Nvidia says that with larger L2 cache, the cache hits are higher, hence the game is not required to use VRAM that much. Additionally, it claims that larger L2 cache size means that bigger bus width isn’t required, as larger cache reduces the bus usage by almost half.
Nvidia further says that larger VRAM requires more VRAM chips, which can increase the cost of the cards.
Then, Nvidia uses the RAM argument for the VRAM. It says just because a game uses all the VRAM doesn’t mean it needs that. It also goes onto to blame various On Screen Display (OSD) software, which it claims can give wrong information.
Additionally, it is showcasing tools for developers which they can use to prevent excessive VRAM usage.
Conclusion
The specs for the RTX 4060 and RTX 4060 Ti are disappointing. But they aren’t surprising.
We personally think that while what Nvidia is saying about VRAM is true. This is more like a PR job to hide the fact that Nvidia is skimping out on larger VRAM to keep their profits higher.
While it’s true that Nvidia does invest a lot in optimizations of it’s graphics cards, no argument is enough for providing just 8GB VRAM in this era. Not to forget, VRAM chips are just not that expensive as Nvidia is trying to portray.
In-fact, as this tweet by Buildzoid1 explains, it would cost Nvidia only $32 more to make a 16GB card. But that’s not for bulk orders. When bulk orders for VRAM chips are considered, then making a 16GB graphics card is extremely cheap. So Nvidia’s argument on the cost doesn’t stand either.
If that is not enough, this tweet by Hardware Unboxed says that Nvidia has confirmed that extra cache cannot compensate if the VRAM is used up. Means extra cache is an enhancement for the graphics card, not a replacement for larger VRAM.
In all. This series of RTX 40 graphics cards are disappointing. However, before deciding anything, like we always say, wait for the reviews. The reviews for the RTX 4060 Ti 8GB are expected to be out in a few days by 23rd May. Only then will we come to know how good this graphics card actually is.
