~/Brainstorm Board
Capstone 20264th Year

Brainstorm Board

Capture and vote on research ideas, hypotheses, and design directions.

Sort:
6 total ideas·2 pinned·28 total votes
Optimization

Use INT4 mixed-precision quantization

Apply INT4 for weights and INT8 for activations to reduce model size by 2x while maintaining accuracy within 2% of FP32 baseline.

Stephanie (Yixin) · 2026-03-10
Architecture

Tile-based attention computation

Partition the attention matrix into tiles that fit in on-chip BRAM to avoid expensive DRAM accesses during the softmax computation.

Jerry (Chenjia) · 2026-03-10
Research Direction

Explore Swin Transformer for local attention

Swin's window-based attention has O(n) complexity vs O(n²) for standard ViT. Could significantly reduce hardware resource requirements.

Stephanie (Yixin) · 2026-03-10
Optimization

Pipelined HLS design for FFN layers

Use HLS PIPELINE pragma with II=1 to fully pipeline the feed-forward network layers, maximizing throughput.

Tiffany (Yiling) · 2026-03-10
Systems

Double-buffering for weight loading

Pre-fetch the next layer's weights while computing the current layer to hide DRAM latency.

Winnie · 2026-03-10
Research Direction

Compare ZCU104 vs Alveo U250 targets

Evaluate whether the embedded ZCU104 or the datacenter Alveo U250 better fits our latency/power budget.

Jerry (Chenjia) · 2026-03-10