arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2606.25313cs.DCcs.AR
收藏

具有1,000,000个p比特的可编程概率计算机

Programmable Probabilistic Computer with 1,000,000 p-bits

University of California, Santa Barbara(加州大学圣塔芭芭拉分校);King Fahd University of Petroleum & Minerals(法赫德国王石油与矿产大学);Carnegie Mellon University(卡内基梅隆大学);Siemens Digital Industries Software(西门子数字工业软件)
Navid Anjum Aadit, Xiuqi Zhang, Shuvro Chowdhury, Kevin Callahan-Coray, Kyle Lee, Saleh Bunaiyan, Sanjay Seshan, Clayton Thomas, Jason Twigg, Andrew Seawright, … 展开作者

Navid Anjum Aadit, Xiuqi Zhang, Shuvro Chowdhury, Kevin Callahan-Coray, Kyle Lee, Saleh Bunaiyan, Sanjay Seshan, Clayton Thomas, Jason Twigg, Andrew Seawright, Forrest Brewer, Tathagata Srimani, Kerem Y. Camsari

AI总结:

通过联网FPGA构建百万p比特的Ising机,实现超过万亿次/秒的吉布斯采样,并揭示分布式采样中边界刷新频率与局部更新频率的时序比η对精度的影响。

摘要:

基于p比特构建的概率计算机已被提议作为采样和优化Ising模型的硬件加速器,但现有系统局限于单个芯片,受其容量和内存带宽限制。在此,我们通过将FPGA联网成一个远超任何单个设备容量的Ising机,突破了这一限制,实现了一台具有一百万个p比特的可编程概率计算机。该机器以超过万亿次翻转每秒的速度执行吉布斯采样,同时将所有耦合权重保留在本地片上内存中。执行过程中,设备之间仅交换1比特边界状态。这种架构引出了一个对任何分布式采样器都至关重要的问题:对于分区机器而言,边界信息必须以多高的频率刷新才能表现得像未分区机器一样。利用三维Edwards-Anderson自旋玻璃,我们证明答案由单个时序比η = f_comm/f_p-bit决定,即边界交换频率与本地p比特更新频率之比。在高于拓扑相关阈值时,分布式机器与单片GPU参考匹配。低于该阈值时,残余能量仍按幂律衰减,但指数减小,将并行性转化为可量化的吞吐量-精度权衡。理论团簇平均场模型再现了相同行为,表明这种权衡是分区随机动力学的普遍性质。这些结果提供了一个可编程的百万p比特平台,并在自旋玻璃、Max-Cut和布尔可满足性问题中得到验证,同时为将概率计算机扩展到单芯片极限之外提供了定量设计规则。

英文摘要:

Many problems in physics and machine learning require sampling the collective states of large numbers of interacting variables. On modern graphics processors, repeated movement of interaction data between memory and compute units can limit how efficiently these calculations can scale. Here we demonstrate a programmable probabilistic computer with one million interacting probabilistic bits (p-bits) across 18 field-programmable gate arrays, sustaining $10^{12}$ probabilistic updates per second. Coupling weights remain beside the update circuits in on-chip memory, and only single-bit boundary states pass between devices. Experiments on three-dimensional and quasi-two-dimensional spin glasses show that the ratio of boundary-exchange frequency to local update frequency controls the effect of communication delays on optimization. A bound derived from the graph partition and interconnect identifies operating conditions that recover the residual-energy decay of an unpartitioned GPU reference. Infrequent exchange reduces the decay exponent, although faster local updates can shorten the time to reach a target energy. A cluster mean-field model reproduces this dependence, while ferromagnetic sampling experiments reveal a corresponding dependence of magnetization relaxation on the same frequency ratio. These results establish a million-p-bit computing architecture in which processing and local memory scale together, and provide a quantitative design principle for exploiting tolerance to communication delays in larger probabilistic computers.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑