AT2k Design BBS Message Area
Casually read the BBS message area using an easy to use interface. Messages are categorized exactly like they are on the BBS. You may post new messages or reply to existing messages!

You are not logged in. Login here for full access privileges.

Previous Message | Next Message | Back to AnandTech  <--  <--- Return to Home Page
   Local Database  AnandTech   [31 / 100] RSS
 From   To   Subject   Date/Time 
Message   VRSS    All   Troubled AI Processor Developer Graphcore Finds a Buyer: SoftBan   July 12, 2024
 3:30 PM  

Feed: AnandTech
Feed Link: https://www.anandtech.com
---

Title: Troubled AI Processor Developer Graphcore Finds a Buyer: SoftBank

Date: Fri, 12 Jul 2024 16:30:00 EDT
Link: https://www.anandtech.com/show/21468/troubled...

After months of searching for a buyer, troubled U.K.-based AI processor
designer Graphcore said on Friday that it has been acquired by SoftBank. The
company will operate as a wholly owned subsidiary of SoftBank and will
possibly collaborate with Arm, but what remains to be seen what happens to
the unique architecture of Graphcore's intelligence processing units (IPUs).

Graphcore will retain its name as it will become a wholly owned subsidiary of
SoftBank, which paid either $400 million (according to EE Times) or $500
million (according to BBC) for the company. Over its lifetime, Graphcore has
received a total of $700 million of investments from Microsoft and Sequoia
Capital, and at its peak in late 2020, was valued at $2.8 billion. Nigel Toon
will remain at the helm of Graphcore, which will hire new staff in its UK
offices and continue to be headquartered in Bristol, with additional offices
in Cambridge, London, Gdansk (Poland), and Hsinchu (China).

"This is a tremendous endorsement of our team and their ability to build
truly transformative AI technologies at scale, as well as a great outcome for
our company," said Nigel Toon. "Demand for AI compute is vast and continues
to grow. There remains much to do to improve efficiency, resilience, and
computational power to unlock the full potential of AI. In SoftBank, we have
a partner that can enable the Graphcore team to redefine the landscape for AI
technology."

Although Graphcore says that it had won contracts with major high-tech
companies and deployed its IPUs, it could not compete against NVIDIA and
other pr�t-�-porter AI processor vendors due to insufficient funding. In the
recent years the company's problems were so severe that it had to lay off 20%
of its staff, bringing its headcount to around 500. Those cuts also saw
office closures in Norway, Japan, and South Korea, which made it even harder
to compete against big players.

Graphcore certainly hopes that with SoftBank's deep pockets and willingness
to invest in AI technologies in general and AI processors in particular, it
will finally be able to compete head-to-head with established players like
NVIDIA.

When asked whether Graphcore will work with SoftBank's Arm, Nigel Toon said
that he was looking forward to work with all companies controlled by its
parent, including Arm. Meanwhile, SoftBank itself is reportedly looking
forward to build its own AI processor venture called Project Izanagi to
compete against NVIDIA, whereas Arm is reportedly developing AI processors
that will work in datacenters owned by SoftBank. Therefore, it remains to be
seen where does Graphcore fit in.

For now, the best processor that Graphcore has is its Colossus MK2 IPU, which
is built using 59.4 billion transistors and packs in 1,472 independent cores
with simultaneous multithreading (SMT) capable of handling 8,832 parallel
threads. Instead of using HBM or other types of external memory, the chip
integrates 900 MB of SRAM, providing an aggregated bandwidth of 47.5 TB/s per
chip. Additionally, it features 10 IPU links to scale with other MK2
processors. When it comes to performance, the MK2 C600 delivers 560 TFLOPS
FP8, 280 TFLOPS FP16, and 70 TFLOPS of FP32 performance at 185W. To put the
numbers into context, NVIDIA's A100 delivers 312 FP16 TFLOPS without sparsity
as well as 19.5 FP32 TFLOPS, whereas NVIDIA's H100 card offers 3,341 FP8
TFLOPS.

Sources: Graphcore, EE Times, BBC, Reuters

---
VRSS v2.1.180528
  Show ANSI Codes | Hide BBCodes | Show Color Codes | Hide Encoding | Hide HTML Tags | Show Routing
Previous Message | Next Message | Back to AnandTech  <--  <--- Return to Home Page

VADV-PHP
Execution Time: 0.0177 seconds

If you experience any problems with this website or need help, contact the webmaster.
VADV-PHP Copyright © 2002-2024 Steve Winn, Aspect Technologies. All Rights Reserved.
Virtual Advanced Copyright © 1995-1997 Roland De Graaf.
v2.1.241108