Show HN: NHE – Eliminating Frame Drops in 4K 144Hz via Direct-to-Silicon Logic

eggpine84 — Sun, 01 Mar 2026 13:28:38 +0000

Traditional high-end GPUs fail to maintain 4K 144Hz not because of raw power, but due to the inherent bottlenecks of the Von Neumann architecture. NHE-588 solves this by replacing the rendering pipeline's "road," not the driver.

*Theoretical Foundation:*

- *Decoding-Free (*≤1 *Cycle):* We bypass the 5~12 cycle decoder lag. Using our Addr=(Sidx ×588)+(sidx ×28)+Oidx formula, rendering instructions are mapped directly to physical coordinates on the silicon. - *No-Scheduler Architecture:* We removed the GPU scheduler (30-40% of chip area) and replaced it with raw computation cores. Deterministic routing ensures 0% collision without traffic management. - *Physical Thermal Barrier:* The "588" constant creates a 10.58μm isolation barrier in 2nm processes, serving as a dedicated heat dissipation path to prevent thermal throttling.

*Practical Verification:*

- *Hardware V-Sync (Pin 16 SCR):* NHE implements a *hard-wired 6.9ms physical timeout (for 144Hz)*. If a frame isn't finished within the deadline, the chip bypasses OS synchronization and *'Direct-Fires'* the current buffer to the monitor via Pin 16, physically eliminating stuttering at the hardware level. - *Verilog IP Verified:* Logic Design is completed, and 1-cycle response has been verified through EDA simulation waveforms. - *Energy Discharge Model:* We use a discharge model that dumps energy immediately after computation to GND (0V), preventing the heat accumulation that typically halves frame rates in 4K sessions.

https://github.com/eggpine84-del/NHE-CODING

Comments URL: https://news.ycombinator.com/item?id=47206522

Points: 1

# Comments: 0