https://fuse.wikichip.org/news/5727/mythic-rolls-out-m1000-series-analog-ai-accelerators-raises-70m-along-the-way/ Skip to content Wednesday, September 1, 2021 Latest: * Mythic Rolls Out M1000-Series Analog AI Accelerators; Raises $70M Along The Way * YouTube Accelerates Transcoding * Intel Introduces Thread Director For Heterogeneous Multi-Core Workload Scheduling * Intel Unveils Sapphire Rapids: Next-Generation Server CPUs * Intel's Gracemont Small Core Eclipses Last-Gen Big Core Performance * WikiChip Fuse WikiChip Fuse Your Chips and Semi News [ ] * Home * Account * Main Site * Architectures + x86 + ARM + RISC-V + Power ISA + MIPS * Supercomputers * 14 nm * 12nm * 10nm * 7nm * 5nm Neural Processors Mythic Rolls Out M1000-Series Analog AI Accelerators; Raises $70M Along The Way August 22, 2021August 22, 2021 David Schor 40 nm, analog, Analog Compute Engine (ACE), Analog Matrix Processor (AMP), eFlash, embedded flash, Mythic, neural processors As our readers would know, we've always been a big fan of novel architectures. It should come as no surprise that we had to revisit Mythic as their new M1000-series true analog accelerators go into production. We first wrote about their in-flash memory analog compute architecture in 2019. Lots have happened since. Mythic has grown to over 110 employees with offices in Austin, TX and Redwood City, CA. Late last year, they unveiled their first series of AI analog accelerators - the M1000-series - along with their top chip. More recently they expanded that series with a second, more versatile, chip. In May, Mythic also announced it raised $70 million in Series C funding led by BlackRock and HPE. This puts the total amount raised to date at $165.2 million. Mythic says the new funding intends to accelerate plans for mass production, increase customer support, build out its software portfolio, and help develop its next-generation hardware platform. Analog Compute In-Memory Mythic took a very different approach to accelerate AI workloads. We will only briefly talk about their architecture as we've already covered that more comprehensively before. As you know, the bulk of the computation takes place in the repeated matrix multiply operations. It also happens to be fairly memory intensive, shifting large amounts of data in and out. Addressing this, most accelerators typically rely on a very large amount of SRAM on-die. Alternatively, some choose to rely on external memory and go with a large amount of memory bandwidth either through a high number of memory channels along with conventional DDR4/LPDDR4x or a wider memory bus such as HBM2e. Mythic, on the other hand, went with flash memory. A single flash cell can store multiple values and is therefore considerably denser than an equivalent 8-bit SRAM memory block. But the real advantage in their case comes with how they use that flash memory. By storing stationary weights in flash and using them directly acting as a variable resistor, they can take advantage of Ohm's law to naturally perform matrix multiplication operations by applying a set of voltages across. This allows them to not only store the weights in dense memory but operate on those weights without actually "reading" them. The combination of using flash memory along with analog matrix multiply operations promises to bring a step-function improvement in energy efficiency and performance-per-cost. [mythic-wc-overview] M1000-Series Analog Matrix Processor (AMP) Mythic calls their neural processors the Analog Matrix Processors or AMP. The chip itself features a tiled architecture. Within each tile is the Mythic Analog Compute Engine (ACE), a digital SIMD vector engine, a 32b RISC-V nano-processor, a small amount of local SRAM, and the NoC router. There are a number of additional miscellaneous tiles for control and interfaces. [mythic-wc-m1108-1] [1108m-w-297x300] Mythic's first product which launched last November is the M1108 AMP. This is the flagship model and the chip based on the largest configuration of tiles in this generation. The '108' represents the 108 compute tiles on that die. There are also 4 control and interface tiles including x4 PCIe Gen2 lanes. Within each tile is 1024x1024 flash array which means a whole 1 MiB of memory per tile. For the M1108, this equates to 108 MiB of memory. The entire package is within 19x19 mm2 BGA package and consumes just about 4 W - a fraction of the power consumed by their competitors. [mythic-wc-m1108-2] 108 MiB of storage capacity for weights is a lot. It's a lot of memory even if the chip was on a leading-edge process; but in Mythic's case, they are fabricating the chip on a mature and cost-effective 40-nanometer CMOS. Keep in mind that the target market for those chips is the billions of edge devices such as industrial applications (machine vision, autonomous drones, etc..), video surveillance, vision systems, and other low-power applications. Here, the models can fit entirely on-chip without any external memory. In fact, multiple models can usually fit in that space. For this reason, it's actually a single-chip platform without any external memory used for typical use cases which can further reduce the BoM. [mythic-m1108-overview] Mythic offers the M1108 AMP in a couple of ways. Customers can buy the chip itself and integrate it directly into their system however they see fit. Mythic also offers a PCIe evaluation kit as well as an M.2 card for customers that want to take a much simpler route. [1108-m] When comparing their M1108 to Nvidia Xavier AGX, Mythic claims superior performance at a 10th of the power - while keeping the platform size to a fraction (no need for the expensive external memory or complex power delivery). But perhaps the biggest difference is in the price which at the current prices is around 30x lower. Now, we do have to be careful with that comparison as it's far from even. The Xavier SoC is loaded with many other features such as a programmable vision accelerator (PVA), stereo and optical flow engine, and of course, the CPU cores and GPU compute. And while you are paying a premium for those features, for the edge and embedded space where Mythic is targeting, much of this will go unused by many customers. [mythic-wc-comp] M1076 - Smaller form-factor In early June Mythic introduced a second product in the series - the M1076 AMP. As the model number implies, this is a further cut-down version of the M1108 AMP with just 76 tiles. This product will likely be cheaper than the M1108 and Mythic thinks it will also be more popular. The rationale behind this version is the form factor. Whereas the original M1108 was sized for M.2 M key card, with physical dimensions of 22 mm x 80 mm. Mythic says that in the embedded space, there is strong interest in the even smaller A+E key variant which has the physical dimensions of just 22 mm by 30 mm. The original M1108 AMP was simply too big for that. [m1076-m] The M1076 supports up to 25 TOPS and because it's got 32 fewer tiles, it operates at lower power - about 1 W less than the M1108 or 3 W power envelope. One interesting aspect of the new M1076 AMP is the new multi-chip configuration they are offering and letting customers experiment with. Here Mythic is offering a PCIe card solution with up to 16 AMPs. At the largest configuration, a full card with 16 AMPs is maxed at around 75 W and can provide 400 TOPs of compute power along with a whopping 1.1875 GiB of on-die flash storage for close to 1.3 billion weights. Both the M1108 and the M1076 along with the M.2 and PCIe card versions are available for customer evaluation. - Spotted an error? Help us fix it! Simply select the problematic text and press Ctrl+Enter to notify us. - * - YouTube Accelerates Transcoding * Share This Post: Related Articles Fujitsu Semi Sells 300mm Mie Fabs A Look at Cerebras Wafer-Scale Engine: Half Square Foot Silicon Chip Intel Announces Keem Bay: 3rd Generation Movidius VPU UMC Rolls Out 22-Nanometer Hot Chips 30: Nvidia Xavier SoC Cambricon Reaches for the Cloud With a Custom AI Accelerator, Talks 7nm IPs Top Six Articles * Mythic Rolls Out M1000-Series Analog AI Accelerators; Raises $70M Along The Way * Analog AI Startup Mythic To Compute And Scale In Flash * The x86 Advanced Matrix Extension (AMX) Brings Matrix Operations; To Debut with Sapphire Rapids * Intel's Gracemont Small Core Eclipses Last-Gen Big Core Performance * Intel Unveils Sapphire Rapids: Next-Generation Server CPUs * IEDM 2018: Intel's 10nm Standard Cell Library and Power Delivery Ezoicreport this ad Recent * Mythic Rolls Out M1000-Series Analog AI Accelerators; Raises $70M Along The Way Mythic Rolls Out M1000-Series Analog AI Accelerators; Raises $70M Along The Way August 22, 2021August 22, 2021 David Schor * YouTube Accelerates Transcoding YouTube Accelerates Transcoding August 21, 2021August 21, 2021 David Schor * Intel Introduces Thread Director For Heterogeneous Multi-Core Workload Scheduling Intel Introduces Thread Director For Heterogeneous Multi-Core Workload Scheduling August 19, 2021August 19, 2021 David Schor * Intel Unveils Sapphire Rapids: Next-Generation Server CPUs Intel Unveils Sapphire Rapids: Next-Generation Server CPUs August 19, 2021August 19, 2021 David Schor * Intel's Gracemont Small Core Eclipses Last-Gen Big Core Performance Intel's Gracemont Small Core Eclipses Last-Gen Big Core Performance August 19, 2021August 21, 2021 David Schor * Intel Unveils Alder Lake: Next-Generation Mainstream Heterogeneous Multi-Core SoC Intel Unveils Alder Lake: Next-Generation Mainstream Heterogeneous Multi-Core SoC August 19, 2021August 19, 2021 David Schor Ezoicreport this ad Random Picks Qualcomm launches the Centriq 2400 server family Qualcomm launches the Centriq 2400 server family November 8, 2017May 25, 2021 David Schor Intel to launch Skylake-D in Q1 2018, followed by Xeons with integrated FPGA Intel to launch Skylake-D in Q1 2018, followed by Xeons with integrated FPGA November 26, 2017May 25, 2021 David Schor Inside Baidu's In-House Neural Processor Inside Baidu's In-House Neural Processor July 11, 2021July 12, 2021 David Schor ISSCC 2018: Intel's Self-Powered Intelligent IoT Edge Mote ISSCC 2018: Intel's Self-Powered Intelligent IoT Edge Mote April 2, 2018May 25, 2021 David Schor Arm Launches the Cortex-M55 and Its MicroNPU Companion, the Ethos-U55 Arm Launches the Cortex-M55 and Its MicroNPU Companion, the Ethos-U55 February 10, 2020May 25, 2021 David Schor Random Tags 2.5D packaging 3D packaging 5 nm 5nm 7 nm 7nm 10 nm 10nm 12nm 14 nm 16nm AI AMD ARM ARMv8 ARMv9 chiplet Coffee Lake Core i5 Core i7 edge computing EMIB EUV FinFET Foveros GlobalFoundries Hot Chips IBM Ice Lake IEDM inference Intel ISSCC multi-chip package neural processors process technology RISC-V Samsung subscriber only (general) Sunny Cove Supercomputers TSMC VLSI Symposium x86 Zen x86 WorldView All Intel Introduces Thread Director For Heterogeneous Multi-Core Workload Scheduling Desktop Processors Mobile Processors Intel Introduces Thread Director For Heterogeneous Multi-Core Workload Scheduling August 19, 2021August 19, 2021 David Schor Intel introduces the Intel Thread Director for heterogeneous multi-core workload scheduling Intel Unveils Sapphire Rapids: Next-Generation Server CPUs Architectures Server Processors Intel Unveils Sapphire Rapids: Next-Generation Server CPUs August 19, 2021August 19, 2021 David Schor Intel's Gracemont Small Core Eclipses Last-Gen Big Core Performance Architectures Data Processing Unit Desktop Processors Mobile Processors Intel's Gracemont Small Core Eclipses Last-Gen Big Core Performance August 19, 2021August 21, 2021 David Schor Intel Unveils Alder Lake: Next-Generation Mainstream Heterogeneous Multi-Core SoC Architectures Desktop Processors Mobile Processors Intel Unveils Alder Lake: Next-Generation Mainstream Heterogeneous Multi-Core SoC August 19, 2021August 19, 2021 David Schor Intel Details Golden Cove: Next-Generation Big Core For Client and Server SoCs Architectures Desktop Processors Mobile Processors Server Processors Intel Details Golden Cove: Next-Generation Big Core For Client and Server SoCs August 19, 2021August 19, 2021 David Schor Intel Launches 3rd Gen Ice Lake Xeon Scalable Architectures Server Processors Intel Launches 3rd Gen Ice Lake Xeon Scalable April 6, 2021May 23, 2021 David Schor Random ISSCC 2018: MIT's low-power hardware crypto RISC-V IoT processor ISSCC 2018: MIT's low-power hardware crypto RISC-V IoT processor March 17, 2018May 25, 2021 David Schor Intel Announces 20A Node: RibbonFET Devices, PowerVia, 2024 Ramp Intel Announces 20A Node: RibbonFET Devices, PowerVia, 2024 Ramp July 26, 2021July 26, 2021 David Schor A look at Nvidia's NVLink interconnect and the NVSwitch A look at Nvidia's NVLink interconnect and the NVSwitch May 6, 2018May 25, 2021 David Schor Hot Chips 30: Intel Kaby Lake G Hot Chips 30: Intel Kaby Lake G September 9, 2018May 25, 2021 David Schor AMD Discloses Initial Zen 2 Details AMD Discloses Initial Zen 2 Details November 18, 2018May 25, 2021 David Schor ISSCC 2018: Intel's Skylake-SP Mesh and Floorplan ISSCC 2018: Intel's Skylake-SP Mesh and Floorplan March 9, 2018May 25, 2021 David Schor Fujitsu Completes Post-K ARM CPU Prototype Fujitsu Completes Post-K ARM CPU Prototype June 24, 2018May 25, 2021 David Schor ARM WorldView All Marvell Launches 5nm Octeon 10 DPUs with Neoverse N2 cores, AI Acceleration Data Processing Unit Marvell Launches 5nm Octeon 10 DPUs with Neoverse N2 cores, AI Acceleration June 28, 2021June 28, 2021 David Schor Arm Introduces Its Confidential Compute Architecture Architectures Arm Introduces Its Confidential Compute Architecture June 23, 2021June 23, 2021 David Schor A Look At Trishul: Arm's First High-Density 3D Logic Stacked Test-Chip IEDM 2020 Interconnects Packaging Subscriber Only Content A Look At Trishul: Arm's First High-Density 3D Logic Stacked Test-Chip June 11, 2021June 11, 2021 David Schor Arm Launches New Coherent And SoC Interconnects: CI-700 & NI-700 Architectures Interconnects Network-on-Chip Arm Launches New Coherent And SoC Interconnects: CI-700 & NI-700 May 25, 2021May 25, 2021 David Schor Arm Launches The DSU-110 For New Armv9 CPU Clusters Architectures Interconnects Mobile Processors Arm Launches The DSU-110 For New Armv9 CPU Clusters May 25, 2021May 25, 2021 David Schor Arm Launches Its New Flagship Performance Armv9 Core: Cortex-X2 Architectures Embedded Processors Mobile Processors Arm Launches Its New Flagship Performance Armv9 Core: Cortex-X2 May 25, 2021May 25, 2021 David Schor About WikiChip WikiChip is an independent publisher based in New York. The WikiChip Fuse section publishes chips and semiconductor related news with our main site offering in-depth semiconductor resources and analysis. WikiChip Links * Main Site * WikiChip Fuse * Newsletter * * Main Site * WikiChip Fuse Copyright (c) 2021 WikiChip LLC. All rights reserved. Spelling error report The following text will be sent to our editors: Your comment (optional): [ ] [ ] [ ] Send Cancel