You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: WHITE_PAPER.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -43,7 +43,7 @@ While ARM NEON provides high absolute throughput for dense binary math, the Tern
43
43
44
44
## 4. Related Work
45
45
46
-
The Ternary Fabric sits at the intersection of Ternary Neural Networks (TNNs) and domain-specific accelerators.
46
+
Ternary Fabric sits at the intersection of Ternary Neural Networks (TNNs) and domain-specific accelerators.
47
47
48
48
***Google TPU:** Utilizes large systolic arrays for 8-bit integer math. While highly efficient for dense GEMM, it lacks native support for ternary logic and doesn't exploit fine-grained sparsity to the same degree as the Fabric's Zero-Skip logic.
49
49
***MIT Eyeriss:** A research-focused spatial accelerator that prioritizes data movement efficiency. Like the Ternary Fabric, it values energy efficiency but remains centered on binary-encoded weights.
0 commit comments