NVIDIA Claims TSMC 20nm will not Scale?
Interesting article from Joel Hruska on ExtremeTech: Nvidia deeply unhappy with TSMC, claims 22nm essentially worthless . The title is a bit dramatic (poetic license) but the charts are accurate to the degree that 20nm costs will be significantly higher from the start and will continue to be higher throughout production and maturity.
One reason is double patterning, which is required at 20nm (double patterning splits a design into separate masks when devices are too close together). If 28nm masks cost $3M, 20nm masks will cost $5M or more. Another reason is design complexity: layout dependent effects (LDE) and process variation will get worse at 20nm, simulation and verification requirements will explode, and the list goes on…
To read the full article, click here
Related Semiconductor IP
- MIPI D-PHY and FPD-Link (LVDS) Combinational Transmitter for TSMC 22nm ULP
- 12-bit, 400 MSPS SAR ADC - TSMC 12nm FFC
- General use, integer-N 4GHz Hybrid Phase Locked Loop on TSMC 28HPC
- Process/Voltage/Temperature Sensor with Self-calibration (Supply voltage 1.2V) - TSMC 3nm N3P
- 25MHz to 4.0GHz Fractional-N RC PLL Synthesizer on TSMC 3nm N3P
Related Blogs
- Samsung 20nm test chip includes ARM Cortex-M0 processor core. How many will fit on the head of a pin?
- Intel will NOT build ARM chips!
- TSMC versus Intel at 20nm!
- Apple Will NOT Manufacture SoCs at Intel
Latest Blogs
- ReRAM in Automotive SoCs: When Every Nanosecond Counts
- AndeSentry – Andes’ Security Platform
- Formally verifying AVX2 rejection sampling for ML-KEM
- Integrating PQC into StrongSwan: ML-KEM integration for IPsec/IKEv2
- Breaking the Bandwidth Barrier: Enabling Celestial AI’s Photonic Fabric™ with Custom ESD IP on TSMC’s 5nm Platform