
Excessive Efficiency at Minimal Leakage
At its 2023 Know-how Symposium TSMC revealed some extra particulars about its upcoming N4X technology that’s designed particularly for high-performance computing (HPC) functions. This node guarantees to allow ultra-high efficiency and enhance effectivity whereas sustaining IP compatibility with N4P (4 nm-class) course of know-how.
“N4X really units a brand new benchmark for the way we are able to push excessive efficiency whereas minimizing the leakage energy penalty,” mentioned Yujun Li, TSMC’s director of enterprise growth who’s accountable for the foundry’s Excessive Efficiency Computing Enterprise Division.
TSMC’s N4X know-how belongs to the corporate’s N5 (5 nm-class) household, however it’s enhanced in a number of methods and is optimized for voltages of 1.2V and better in overdrive mode.
To attain larger efficiency and effectivity, TSMC’s N4X improves transistor design in three three key areas. Firstly, they refined their transistors to spice up each processing velocity and drive currents. Secondly, the foundry integrated its new high-density metal-insulator-metal (MiM) capacitors, to offer dependable energy underneath excessive workloads. Lastly, they modified the the back-end-of-line steel stack to offer extra energy to the transistors.
Particularly, N4X provides 4 new gadgets on prime of the N4P machine choices, together with ultra-low-voltage transistors (uLVT) for functions that should be very environment friendly, and extremely-low threshold voltage transistors (eLVT) for functions that must work at excessive clocks. For instance, N4X uLVT with overdrive provides 21% decrease energy on the identical velocity when in comparison with N4P eLVT, whereas N4X eLVT in OD provides 6% larger velocity for important paths when in comparison with N4P eLVT.
Marketed PPA Enhancements of New Course of Applied sciences Knowledge introduced throughout convention calls, occasions, press briefings and press releases |
||||||||||
TSMC | ||||||||||
N5 vs N7 |
N5P vs N5 |
N5HPC vs N5 |
N4 vs N5 |
N4P vs N5 |
N4P vs N4 |
N4X vs N5 |
N4X vs N4P |
N3 vs N5 |
||
Energy | -30% | -10% | ? | decrease | -22% | – | ? | ? | -25-30% | |
Efficiency | +15% | +5% | +7% | larger | +11% | +6% | +15% or extra |
+4% or extra |
+10-15% | |
Logic Space
Discount % (Density) |
0.55x
-45% (1.8x) |
– |
– |
0.94x
-6% 1.06x |
0.94x
-6% 1.06x |
– |
? |
? |
0.58x
-42% (1.7x) |
|
Quantity Manufacturing |
Q2 2020 | 2021 | Q2 2022 | 2022 | 2023 | H2 2022 | H1 2024? |
H1 2024? | H2 2022 |
Whereas N4X provides vital efficiency enhancements in comparison with N4 and N4P, it continues to make use of the identical SRAM, commonplace I/O, and different IPs as N4P, which permits chip designers emigrate their designs to N4X simply and cheaply. In the meantime, retaining in thoughts N4X’s IP compatibility with N4P, it’s logical to count on transistor density of N4X to be kind of consistent with that of N4P. Although given the main target of this know-how, count on chip designers to make use of this know-how to get excessive efficiency moderately than most transistor density and small chip dimensions.
TSMC claims that N4X has achieved its SPICE mannequin efficiency targets, so clients can begin utilizing the know-how right now for his or her HPC designs that can enter manufacturing generally subsequent 12 months.
For TSMC, N4X is a vital know-how as HPC designs are anticipated to be the corporate’s important income development driver within the coming years. The contract maker of chips anticipates HPC to account for 40% of its income in 2030 adopted by smartphones (30%) and automotive (15%) functions.