News RTX 4060 Ti 8GB 128-bit Bus Config Seemingly Confirmed in New Listings

It reportedly has 36% more TFLOP compute performance, so it could be significantly faster than its predecessor in non-gaming situations.

Based on the rumored 2685 MHz boost clock on some custom models, the card could deliver up to 23.3/24 TFLOPs of compute horsepower which seems roughly 44% higher than the NVIDIA GeForce RTX 3060 Ti.

But the actual performance increase should be 30-40% faster than the RTX 3060 Ti.
 
Last edited by a moderator:
  • Like
Reactions: atomicWAR
Disappointing. Nvidia is lagging behind where game developers and the console vendors are going. Fortunately, there will probably be a bunch of variants of the 4060 just like how there were many variants of the 3060. Hopefully this base model is just the start.

How much longer until Nvidia starts producing Light Artificial Intelligence (LAI) cards? They had Light Hash Rate (LHR) cards. Bring on the LAI variants!
 
I don't see it on this site as of this posting, but Nvidia just announced a new texture compression method,

Yeah, I read about it on Nvidia's blog, the neural texture compression method. The method sounds kind of interesting.

Random-Access Neural Compression of Material Textures/NTC:


They also shared a lot of technical details on this texture compression method in this PDF file.

 
Based on the rumored 2685 MHz boost clock on some custom models, the card could deliver up to 23.3/24 TFLOPs of compute horsepower which seems roughly 44% higher than the NVIDIA GeForce RTX 3060 Ti.

But the actual performance increase should be 30-40% faster than the RTX 3060 Ti.
except the 4060 ti has lower memory bus than the 3060 ti. (if the leak was correct ofc)

so imho it wont be 30%+ better.

not the 1st time a card would be heldback by its memory bus being shafted.
 
except the 4060 ti has lower memory bus than the 3060 ti. (if the leak was correct ofc)

so imho it wont be 30%+ better.

not the 1st time a card would be heldback by its memory bus being shafted.
Ada is not a very memory-constrained architecture. e.g. the 3090Ti vs. the 4080: 2/3 of the bus width, 2/3 of the memory capacity, slightly lower core count, but ~35% (25% - 50%) greater performance. That's close enough to the increase in clock speed (135%) to show that Ada is more tolerant of tighter memory capacities and bandwidths clock-for-clock and core-for-core.
 
It would be interesting to test some VRAM overclocking with it to see how much of an impact the memory bandwidth has on it, especially considering that 288GB/s is getting really close to the lower mid range of cards from 2014. Hopefully there will be some overclocking tests to see how much of a bottleneck the low bandwidth is.

While more cache helps, it doesn't eliminate the need for fast VRAM.