Ankhdjet:基于开放PDK的掩码编程三元存算一体的开源编译器
Ankhdjet: An Open-Source Compiler for Mask-Programmed Ternary Compute-in-ROM on an Open PDK
浏览论文内容
中文总结 AI 辅助
本文提出开源编译器Ankhdjet,可将HuggingFace的三元模型检查点转换为开放SKY130 PDK的存算一体宏掩码程序,完成开源工具链验证,为开放PDK的存算一体芯片提供了可复现的开源方案。
中文摘要 AI 辅助
大语言模型推理受限于权重移动:每生成一个token都要重新读取所有权重。三元量化(BitNet b1.58)将每个权重压缩至1.58比特,据报道在20亿参数规模下仍保持性能,小到足以将权重硬连线到只读掩码中,且商用芯片Taalas HC1已在封闭工具链的先进节点上验证了硬连线权重。本文探讨能否用完全开放的基础设施实现可复现的特定模型硅芯片。我们提出Ankhdjet,一款开源编译器,可将HuggingFace的三元检查点(BitNet b1.58及其同类模型)转换为开放SKY130 PDK上固定存算一体宏的通孔掩码程序,通过开源工具完成端到端验证。我们论证两个主张:(1)首个基于可制造开放PDK的开源权重转掩码存算一体编译器,经完整开源工具链签核(KLayout设计规则检查零错误、netgen版图与电路图一致性检查零错误、时序干净),两次使用不同权重矩阵通过相同流程(仅掩码不同)完成;(2)首个基于开放PDK提交制造的存算一体宏(TinyTapeout/ChipFoundry SKY130 shuttle ttsky26c,预计2027年流片)。该编译器的读取为全数字(全摆幅位线采样到标准单元,因此输出端口可用于综合);同机器的模拟比较器读取作为测量变体保留,替换为数字读取后面积缩减至八分之一。我们还报告了每层级对抗性验证方法,其发现了3处设计规则检查不可见的隐性短路,在130nm工艺下每晶体管存储2个三元权重的结果为负,以及哪些模拟传感技术在哪些节点适用。读取能量来自提取寄生参数的仿真(每个传感权重0.98-1.73皮焦);未声称实际能量测量,制造的载体为演示器。
英文摘要
Large-language-model inference is dominated by weight movement: every generated token re-reads every weight. Ternary quantization (BitNet b1.58) shrinks each weight to 1.58 bits with reported parity at the 2B-parameter scale, small enough that hardwiring the weights into a read-only mask becomes plausible, and a commercial chip (Taalas HC1) has validated hardwired weights on an advanced node behind closed tooling. This paper asks whether model-specific silicon can be made reproducible with entirely open infrastructure. We present Ankhdjet, an open-source compiler that lowers a HuggingFace ternary checkpoint (BitNet b1.58 and its kin) to a via-mask program of a fixed compute-in-ROM macro on the open SKY130 PDK, verified end to end with open tools. We defend two claims: (1) the first open-source weights-to-mask compute-in-ROM compiler on a fabricable open PDK, taken through full open-toolchain signoff (KLayout DRC zero, netgen LVS zero, clean timing) twice with two different weight matrices through an identical flow in which only the mask differs; and (2) the first compute-in-ROM macro submitted for fabrication on an open PDK (TinyTapeout/ChipFoundry SKY130 shuttle ttsky26c, silicon expected 2027). The compiler's read is fully digital (full-swing bitline sampling into standard cells, so the emission ports to any node by synthesis); an analog comparator readout of the same machine is retained as a measured variant, replaced by the digital readout at one eighth the area. We additionally report the per-level adversarial verification methodology that caught three DRC-invisible silent shorts, a negative result on storing two ternary weights per transistor at 130 nm, and which analog sense techniques survive at which nodes. Read energy is from extracted-parasitic simulation (0.98-1.73 pJ per sensed weight); no energy measurements are claimed, and the fabricated vehicles are demonstrators.