There are two commerce reveals yearly that I like. Computex in June is nice, as a result of the dimensions of the business it covers, and Taipei is a superb location. Hot Chips in August is the opposite present, which is wonderful for the extent of depth it offers on the most recent expertise, in addition to upcoming releases. This yr the record of displays for Hot Chips is sort of overwhelming, and we’re wanting ahead to attending.
Hot Chips is now in its 31st model, and is the positioning of a variety of technical disclosures on the most recent chip logic improvements. Almost all the foremost chip distributors and IP licensees concerned in semiconductor logic designs participate: Intel, AMD, NVIDIA, Arm, Xilinx, and IBM are on the most recent record, and builders of huge accelerators like Fujitsu and NEC final yr. Even the foundry companies participate, in addition to among the huge cloud suppliers: Google final yr confirmed its safety chips, that no-one else will ever see. There are notable absences from the likes of Apple, who regardless of being on the Committee, final offered on the convention in 1994.
We’ve come to count on Hot Chips to be the place for microarchitecture overviews in addition to a wider ecosystem examination over the finer factors of {hardware} that may hit a lot of us behind the scenes. There have additionally been parts of the convention devoted to new designs, open-source {hardware}, and safety vulnerabilities. This yr sees a big and big selection of spectacular headlining talks deliberate.
Hot Chips 31 will run on 18-20th of August 2019, and be held at Stanford University. The first day, Sunday 18th, is usually a tutorial day. The detailed talks begin on the 19th.
| Hot Chips 31 (2019) Time Table | |||||
| Day 1 | |||||
| 08:15 | Registration | ||||
| 08:45 | Opening Remarks | ||||
| 09:00 | General Purpose Compute | AMD | Zen 2 | Matisse | |
| Arm | A Next-Gen Cloud-to-Edge Infrastructure SoC Using the Arm Neoverse N1 CPU and System Products | Neoverse N1 | |||
| IBM | IBM’s Next Generation POWER Processor | POWER9 with IO | |||
| 10:30 | Break | ||||
| 11:00 | Memory | Upmem | True Processing In Memory with DRAM Accelerator | ||
| Princeton | A Programmable Embedded Microprocessor for Bit-Scalable In-Memory Computing | ||||
| Intel | Intel Optane | Optane DCPMM | |||
| 12:30 | Lunch | ||||
| 13:45 | Keynote | AMD | Dr. Lisa Su | Delivering the Future of High-Performance Computing with System, Software and Silicon Co-Optimization | |
| 14:45 | Methodology and ML Systems | Stanford | Creating an Agile Hardware Flow | ||
| MLPerf | MLPerf: A Benchmark Suite for Machine Learning From an Academic-Industry Co-operative | ||||
| Zion: Facebook Next-Generation Large-Memory Unified Training Platform | |||||
| 16:15 | Break | ||||
| 16:45 | ML Training | Huawei | A Scalable Unified Architecture for Neural Netowrk Computing from Nano-Level to High Performance Computing | Huawei Da Vinci | |
| Intel | Deep Learning Training at Scale – Spring Crest Deep Learning Accelerator | Spring Crest | |||
| Cerabras | Wafer Scale Deep Learning | ||||
| Habana | Habana Labs Approach to Scaling AI Training | ||||
| 18:45 | Reception | ||||
The first day kicks off with a normal function compute session with AMD, Arm, and IBM. AMD’s speak appears into its latest Zen 2 microarchitecture, which can be powering Matisse based mostly ‘Ryzen 3000’ line of desktop processors in addition to ‘Rome’ server processors. We’re not anticipating a lot new from this presentation, provided that it’s anticipated for AMD to have launched the merchandise to market earlier than mid-August. Following AMD is Arm, with its Neoverse N1 platform we reported that was introduced a number of weeks in the past. IBM’s speak can be attention-grabbing, discussing the most recent POWER processor, which is prone to be its optimized model of POWER9 specializing in IO help.
The second session is about…
