Nvidia's Next Generation Lovelace Architecture May Not Differ All That Much From Ampere
Nvidia’s next generation Lovelace architecture may not differ all that much.
The RX 6000 series and Nvidia RTX 30 series have been with us for over a year now.
In his latest tweet they indicate the fundamental architecture of the upcoming Lovelace.
The Nvidia RTX 40 series GPUs are not all that different from current Ampere RTX 30 GPUs.
30 series cards are manufactured on Samsung’s 8nm process, derivative of its 10nm node.
AMD uses TSMC’s 7nm node.
Nvidia is unable to achieve clock speed as high as AMD because of this reason.
They are expected to use TSMC’s 5nm node.
Nvidia can claw back some of its current disadvantages if all other things are considered equal.
A smaller node will allow them to add in many shader clusters.
A higher clocked card with 50 percent or more shaders would still be huge.
We can expect to see higher clocks, more CUDA cores, ROPs, Tenser cores and RT cores with more bandwidth.
There are still many more months out from the launch of RTX 40 cards.
RTX 40 Lovelace cards and AMD’s RDNA3 cards will start a great battle later in 2022.