Главная
Study mode:
on
1
Intro
2
maybe even the fastest in the world?
3
Who wants these machines?
4
OAK IBM POWER
5
Intel x86
6
Summit: Science research Astrophysics Materials Cancer Research Systems Biology
7
Titan 2012 27
8
Metric household
9
Summit: 13 MegaWatts
10
Summit: USD $200 Million
11
550 households
12
1 Sydney house
13
Summit: 300 km of cables
14
Sierra: National Nuclear Security Administration's Stockpile Stewardship Mission
15
How do you build this thing?
16
IBM 2 computers: • Infrastructure • Compute
17
POWER8 based?
18
100Gbps Networking
19
Mellanox CX-5
20
Hybrid approach CPUS + GPUS
21
Compute: Witherspoon AC922
22
How do we build them?
23
Timelines?
24
Sierra release: December 2017
25
Infrastructure nodes are first
26
Linux • Firmware • Systems • GPU interfaces
27
24 Core SMT4
28
8 Billion transistors
29
POWER9 is major refresh POWER
30
Major Architectural changes: • Radix/Linux Based MMU • New interrupt controller • Direct attach DDR4 DIMMs
31
New Slice Microarchitecture
32
First through 14nm fab
33
POWER9 chip development
34
Minor releases too
35
DD1: January 2017
36
Planning for Linux and Firmware
37
Design: Radix MMU
38
Radix MMU: • Simpler • Better performance • KVM allocations
39
Simulation: • Functional • Cycle Accurate
40
Teach Linux basic feature
41
Bringup: Everything is broken
42
Get Linux up
43
Bringup: • Identify issues • Work around • Get out of the way • Find real fix
44
Develop items that need real hardware
45
Testing • More systems • Systems getting more sophisticated • Devs - Machines futher separated
46
Release: Yay!
47
Staged release
48
POWER9 not backwards compatible with POWERS
49
IBM - RedHat strong relationship
50
IBM & RedHat partnered on RHEL7 for POWERS
51
Deliver Linux to customers
52
End of Moore's Law
53
Drive accelerators
54
Binary Linux kernel driver
55
Helped prove out: • Link training • Firmware
56
Coherent memory
57
CUDA Unified memory
58
Design • IOMMU looks like PCle ATS • IOMMU directly uses Radix MMU
59
Simulation with P9
60
Bringup: March 2017
61
Testing: Data integrity
62
Baseboard Management Controller
63
Little computer that turns on your big computer
64
Firmware?
65
Infrastructure nodes: Supermicro based BMC
66
Compute node OpenBMC
67
Compute nodes first OpenBMC release
68
Like a distro
69
Features: • On/Off • Monitor
70
Solutions
71
Pervasive
72
So how did it end up?
73
Fastest computer in the world?
Description:
Explore the journey of creating the world's fastest supercomputer, Summit, powered by POWER9 technology in this 43-minute conference talk from linux.conf.au. Dive into the technical challenges, architectural innovations, and organizational strategies that led to the development of this groundbreaking system. Learn about the major shifts in POWER9 design, including new interrupt controllers, memory management architectures, and advanced IO capabilities like NVLink and OpenCAPI. Discover how open source software played a crucial role in bringing POWER9 from concept to reality, and gain insights into the complex process of building and managing a supercomputer of this scale. Understand the applications of Summit in various scientific fields, including astrophysics, materials science, cancer research, and systems biology. Get a behind-the-scenes look at the timeline, hardware specifications, and software development process that made Summit possible, and explore the future of supercomputing in the post-Moore's Law era. Read more

Climbing the Summit with Open Source and POWER9

linux.conf.au
Add to list
0:00 / 0:00