From WikiChip
Editing bitmain/sophon/bm1680
Warning: You are not logged in. Your IP address will be publicly visible if you make any edits. If you log in or create an account, your edits will be attributed to your username, along with other benefits.
The edit can be undone.
Please check the comparison below to verify that this is what you want to do, and then save the changes below to finish undoing the edit.
This page supports semantic in-text annotations (e.g. "[[Is specified as::World Heritage Site]]") to build structured and queryable content provided by Semantic MediaWiki. For a comprehensive description on how to use annotations or the #ask parser function, please have a look at the getting started, in-text annotation, or inline queries help pages.
Latest revision | Your text | ||
Line 1: | Line 1: | ||
{{bitmain title|Sophon BM1680}} | {{bitmain title|Sophon BM1680}} | ||
− | {{ | + | {{mpu |
|name=Sophon BM1680 | |name=Sophon BM1680 | ||
|image=bitmain sophon bm1680.png | |image=bitmain sophon bm1680.png | ||
Line 12: | Line 12: | ||
|process=28 nm | |process=28 nm | ||
|technology=CMOS | |technology=CMOS | ||
− | |||
|v core=0.9 V | |v core=0.9 V | ||
|v core tolerance=5% | |v core tolerance=5% | ||
Line 23: | Line 22: | ||
|package module 1={{packages/bitmain/fcbga-1599}} | |package module 1={{packages/bitmain/fcbga-1599}} | ||
}} | }} | ||
− | '''Sophon BM1680''' is a [[neural processor]] designed by [[Bitmain]] and introduced in 2017. The BM1680 is capable of performing both network inference and network training. Capable of delivering up to | + | '''Sophon BM1680''' is a [[neural processor]] designed by [[Bitmain]] and introduced in 2017. The BM1680 is capable of performing both network inference and network training. Capable of delivering up to 2 TFLOPS (SP), the BM1680 has a typical power dissipation of 25 W with a peak power of 41 W. The chip was [[taped-out]] in April 2017 and began sampling in June. |
Manufactured on [[TSMC]]'s [[28 nm|28HPC+]] process, the BM1680 is capable of 80 billion algorithmic operations per second. Bitmain claims the chip is designed not only for inference, but also for training of neural networks, suitable for working with the common [[artificial neural network|ANNs]] such as [[convolutional neural network|CNN]], [[recurrent neural network|RNN]], and [[deep neural network|DNN]]. | Manufactured on [[TSMC]]'s [[28 nm|28HPC+]] process, the BM1680 is capable of 80 billion algorithmic operations per second. Bitmain claims the chip is designed not only for inference, but also for training of neural networks, suitable for working with the common [[artificial neural network|ANNs]] such as [[convolutional neural network|CNN]], [[recurrent neural network|RNN]], and [[deep neural network|DNN]]. | ||
== Overview == | == Overview == | ||
− | The chip consists of five subsystems: [[NPU]], [[MCU]], Chip Link, Memory, and Peripherals. | + | The chip consists of five subsystems: [[NPU]], [[MCU]], Chip Link, Memory, and Peripherals. The MCU Subsystem is a low-power embedded [[ARM]] microcontroller. The NPU Subsystem consists of 64 [[NPUs]], the hub, and an NPU Schedule Engine. The scheduling engine is in charge of controlling the data flow to the individual NPUs. Bitmain has not many intimate details of the NPU cores but each core is known to have 512 KiB of program-visible [[SRAM]] and supports 64 single-precision operations. With a total of 64 NPUs, the chip has a total of 32 [[MiB]] of [[cache]] and a peak performance of 2 [[TFLOPS]] (single-precision). |
− | |||
− | The MCU Subsystem is a low-power | ||
− | |||
− | The NPU Subsystem consists of 64 [[ | ||
:[[File:BM1680 block diagram.png|650px]] | :[[File:BM1680 block diagram.png|650px]] | ||
Line 45: | Line 40: | ||
Current offering by [[Bitmain]] is limited to two BM1680s chips per [[accelerator card]], however in theory the chip can be chained to form a much larger network. The primary limitations of the network are power consumption and thermal dissipation as well as PCIe bandwidth which Bitmain requires to be x4 lanes per node (or 3.9 GB/s, 8 GT/s). For example, current offerings by Bitmain features two nodes on a PCIe Gen3 x8 card. | Current offering by [[Bitmain]] is limited to two BM1680s chips per [[accelerator card]], however in theory the chip can be chained to form a much larger network. The primary limitations of the network are power consumption and thermal dissipation as well as PCIe bandwidth which Bitmain requires to be x4 lanes per node (or 3.9 GB/s, 8 GT/s). For example, current offerings by Bitmain features two nodes on a PCIe Gen3 x8 card. | ||
− | |||
− | |||
− | |||
− | |||
− | |||
== Memory controller == | == Memory controller == | ||
Line 63: | Line 53: | ||
|bandwidth qchan=79.47 GiB/s | |bandwidth qchan=79.47 GiB/s | ||
}} | }} | ||
− | |||
− | |||
− | |||
− | |||
− |
Facts about "Sophon BM1680 - Bitmain"
Has subobject "Has subobject" is a predefined property representing a container construct and is provided by Semantic MediaWiki. | Sophon BM1680 - Bitmain#package + |
core voltage | 0.9 V (9 dV, 90 cV, 900 mV) + |
core voltage tolerance | 5% + |
designer | Bitmain + |
family | Sophon + |
first announced | October 25, 2017 + |
first launched | November 8, 2017 + |
full page name | bitmain/sophon/bm1680 + |
has ecc memory support | true + |
instance of | microprocessor + |
io voltage | 1.8 V (18 dV, 180 cV, 1,800 mV) + |
io voltage tolerance | 5% + |
ldate | November 8, 2017 + |
main image | + |
manufacturer | TSMC + |
market segment | Artificial Intelligence + |
max memory | 16,384 MiB (16,777,216 KiB, 17,179,869,184 B, 16 GiB, 0.0156 TiB) + |
max memory bandwidth | 79.47 GiB/s (81,377.28 MiB/s, 85.33 GB/s, 85,330.263 MB/s, 0.0776 TiB/s, 0.0853 TB/s) + |
max memory channels | 4 + |
max operating temperature | 125 °C + |
min operating temperature | 0 °C + |
model number | BM1680 + |
name | Sophon BM1680 + |
package | FCBGA-1599 + |
peak flops (single-precision) | 2,000,000,000,000 FLOPS (2,000,000,000 KFLOPS, 2,000,000 MFLOPS, 2,000 GFLOPS, 2 TFLOPS, 0.002 PFLOPS, 2.0e-6 EFLOPS, 2.0e-9 ZFLOPS) + |
process | 28 nm (0.028 μm, 2.8e-5 mm) + |
supported memory type | DDR4-2666 + |
tdp | 41 W (41,000 mW, 0.055 hp, 0.041 kW) + |
tdp (typical) | 25 W (25,000 mW, 0.0335 hp, 0.025 kW) + |
technology | CMOS + |