BRN Discussion Ongoing

Just a thought.

I wish Brainchip had named their product Akaida. Rather than Akida.

That way "AI" is central to their branding.

I keep accidentally typing Akaida anyway, so then I'd be right.:)

Maybe they could think of this for the next generation chips.
In most of the cases that matter our IP will be in someone else's chip so they will brand it anyway they like. Most people won't even know we are involved, but the 1000 eyes will.

SC
 
  • Like
  • Fire
Reactions: 20 users

Iseki

Regular
I'm a bit late here. Is that idea that BRN will produce a new Akida 1000 chip with the Arm component replaced by the risk-v component,
or will it be the akida 500 or 2000
or will it be a completely new chip that risk-v sells with akida IP included in it??
 
  • Like
Reactions: 5 users

IloveLamp

Top 20
I'm a bit late here. Is that idea that BRN will produce a new Akida 1000 chip with the Arm component replaced by the risk-v component,
or will it be the akida 500 or 2000
or will it be a completely new chip that risk-v sells with akida IP included in it??
The last one as far as I can gather
 
  • Like
  • Fire
Reactions: 7 users
April 29, 2021

SiFive and Samsung Foundry Extend Partnership to Accelerate AI SoC Development​

Configurable SiFive RISC-V AI SoC Development Platform is built on 14nm Samsung process technology to accelerate custom Machine Learning solutions
SAN MATEO, Calif., April 29, 2021– SiFive, Inc., the industry leader in RISC-V processors and silicon solutions, today announced the next phase of their partnership with Samsung Foundry, a world leader in advanced semiconductor technology. The extended partnership will enable and accelerate the development of Artificial Intelligence and Machine Learning inference and training SoCs based on SiFive RISC-V processors and built using Samsung Foundry technology infrastructure.
The SiFive RISC-V AI SoC Development Platform combines SiFive RISC-V technology, high-speed peripherals, and multiple AI accelerator blocks. The collaborative effort was further customized with a customer-specific AI inference accelerator IP and an Open Compute Project Microsoft Zipline accelerator, resulting in an AI accelerator SoC tape out on Samsung 2nd generation 14LPP FinFET technology on April 13th, 2021. The ability to rapidly integrate additional IP to the SiFive RISC-V AI SoC Development Platform enables customers to optimize workload processing with custom SoCs.
The AI accelerator SoC tapeout and inference IP integration utilized SEMIFIVE Platform SoC technology. SEMIFIVE Inc. is the pioneer of platform-based SoC design for the Samsung Foundry ecosystem. Featuring verified PCIe Gen. 4 connectivity and quad-channel 32-bit LPDDR4/4X interfaces, SEMIFIVE provides off-the-shelf opportunities to develop custom hardware focused on the AI workload.
"Samsung Foundry is committed to building a rich and resilient ecosystem with our industry partners," said Mijung Noh, VP of Foundry Design Service Team at Samsung Electronics. "We are pleased to work with SiFive to accelerate our customers' AI / ML custom SoC design, illustrating the innovation potential we can achieve together."
“Working in partnership with Samsung Foundry has accelerated SiFive’s ability to deliver our highly-efficient and configurable approach for SoC design and implementation,” said Dr. Yunsup Lee, CTO of SiFive. “We’re excited to continue to co-innovate with Samsung Foundry as we launch our latest SiFive Intelligence products to accelerate the development of next-generation AI SoCs with Samsung’s advanced process technology.”
SiFive’s industry-leading portfolio of RISC-V-based processor Core IP scales from the AI-focused SiFive Intelligence line of Linux-capable 64-bit multi-core processors with enhanced RISC-V Vector capabilities to area-optimized real-time cores. SiFive Core IP is silicon-proven and pre-integrated with SiFive Shield, a whole-SoC security solution, SiFive Insight advanced trace and debug capabilities, and optional SiFive hardware cryptographic accelerator (HCA). Read more information on the latest update for the SiFive 21G1 release on the SiFive Intelligencepage.
Further development of the AI accelerator SoC by integrating SiFive Intelligence products into the chip will create a heterogeneous compute platform for machine learning that can be tuned to meet the needs of edge AI, automotive, 5G/networking, or data center accelerator needs.
About SiFive
SiFive is the leading provider of processor cores, AI accelerators, and SoC IP to enable domain-specific designs based on the open RISC-V instruction set architecture specification. SiFive offers scalable, configurable processor cores pre-integrated with security, trace, and debug features for workload-specific accelerator designs. Founded by the inventors of RISC-V, SiFive has design centers worldwide and backing from Sutter Hill Ventures, SK hynix, Qualcomm Ventures, Western Digital, Intel Capital, Spark Capital, Osage University Partners, and Prosperity7 Ventures. For more information, please visit www.sifive.com.
 
  • Like
  • Fire
Reactions: 35 users
I'm a bit late here. Is that idea that BRN will produce a new Akida 1000 chip with the Arm component replaced by the risk-v component,
or will it be the akida 500 or 2000
or will it be a completely new chip that risk-v sells with akida IP included in it??
IIL is correct.

If you remember back to Anil Mankar’s 2021 Ai Field Day presentation he made the point that AKIDA technology is processor and sensor agnostic. All the heavy lifting is done by the AKIDA neural fabric and it is happy to work with anyones processor for the housekeeping matters.

The article I just posted about SiFive and Samsung means that the very 14nm that it refers to could overnight drop off the Ai accelerator and throw on the AKIDA IP neural fabric and rule the world.

My opinion only DYOR
FF

AKIDA BALLISTA
PS: After everything I have read about SiFive once the word gets out I think there is a strong likelihood that the US will understand and value this announcement more highly than the ASX.
 
  • Like
  • Fire
Reactions: 44 users
IIL is correct.

If you remember back to Anil Mankar’s 2021 Ai Field Day presentation he made the point that AKIDA technology is processor and sensor agnostic. All the heavy lifting is done by the AKIDA neural fabric and it is happy to work with anyones processor for the housekeeping matters.

The article I just posted about SiFive and Samsung means that the very 14nm that it refers to could overnight drop off the Ai accelerator and throw on the AKIDA IP neural fabric and rule the world.

My opinion only DYOR
FF

AKIDA BALLISTA
PS: After everything I have read about SiFive once the word gets out I think there is a strong likelihood that the US will understand and value this announcement more highly than the ASX.
Was distracted by my granddaughter and forgot to mention @Diogenese and his many posts explaining this very point about how AKIDA works and what the ARM4 does.

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
Reactions: 19 users

Diogenese

Top 20
In the Greek Alphabet, the letter Kappa, is printed like an English capital K whether it is upper or lower case, so the way ακίδα is printed is correct.

But seriously, leave the Greeks alone - they're still trying to come to grips with Italy's loss to Macedonia in the World Cup Qualifiers :)
... but they are very good at cybersecurity.
 
  • Like
  • Haha
  • Love
Reactions: 21 users
Noice! Very noice!

🎈 🎊🥂🍾🍻🥳




(Extract)

VentureBeat: Does BrainChip use any of Arm’s IP in its chips? Arm is known for low-power and high performance.

Kang:
BrainChip has discussed Arm IP for their product line. Arm processors have built a reputation for low-power based on comparisons to x86-based products. SiFive Intelligence products compare well to Arm products through offering improved performance-per-area of up to 30% combined with a single ISA for simpler programming, and a modular approach that aligns well to working with hardened AI IP such as developed by BrainChip.

VentureBeat: Can you expand upon this statement: “(Brainchip) mimics the human brain to analyze only essential sensor inputs at the point of acquisition.”

Kang:
This statement refers to the ability of humans to focus on what’s important. For example listening to a conversation in a coffee shop while still registering and acknowledging background sounds. The BrainChip solution will mimic this ability to reduce power and increase efficiency by focusing on the important data being processed. This is similar to, but a step beyond, the adoption of mixed and lower precision data types (INT8 vs FP16) to speed up and improve the efficiency of AI/ML processing.


Published 2 hours ago!

🎊
I think Kang has just let a very big cat out of his NASA bag with his description of a ‘HARDENED Ai IP’ DEVELOPED BY BRAINCHIP.

Both Brainchip and SiFive are partnered with NASA.

Vorago was paid to come up with a design to harden AKIDA.

We have not had confirmation that it had succeeded until this Kang statement.

Bravo strikes again.

The Barty Party might have finished but the Brainchip Party is about to step in and fill the void on Earth and in Space.

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Fire
  • Love
Reactions: 52 users

Slade

Top 20
Me trying to explain to my wife why BrainChip is a good investment.
1649222268805.png
 
  • Haha
  • Like
  • Fire
Reactions: 43 users

Slade

Top 20
My Wife
Confused No Idea GIF by Channel 7
 
  • Haha
  • Like
  • Love
Reactions: 29 users

Diogenese

Top 20
April 29, 2021

SiFive and Samsung Foundry Extend Partnership to Accelerate AI SoC Development​

Configurable SiFive RISC-V AI SoC Development Platform is built on 14nm Samsung process technology to accelerate custom Machine Learning solutions
SAN MATEO, Calif., April 29, 2021– SiFive, Inc., the industry leader in RISC-V processors and silicon solutions, today announced the next phase of their partnership with Samsung Foundry, a world leader in advanced semiconductor technology. The extended partnership will enable and accelerate the development of Artificial Intelligence and Machine Learning inference and training SoCs based on SiFive RISC-V processors and built using Samsung Foundry technology infrastructure.
The SiFive RISC-V AI SoC Development Platform combines SiFive RISC-V technology, high-speed peripherals, and multiple AI accelerator blocks. The collaborative effort was further customized with a customer-specific AI inference accelerator IP and an Open Compute Project Microsoft Zipline accelerator, resulting in an AI accelerator SoC tape out on Samsung 2nd generation 14LPP FinFET technology on April 13th, 2021. The ability to rapidly integrate additional IP to the SiFive RISC-V AI SoC Development Platform enables customers to optimize workload processing with custom SoCs.
The AI accelerator SoC tapeout and inference IP integration utilized SEMIFIVE Platform SoC technology. SEMIFIVE Inc. is the pioneer of platform-based SoC design for the Samsung Foundry ecosystem. Featuring verified PCIe Gen. 4 connectivity and quad-channel 32-bit LPDDR4/4X interfaces, SEMIFIVE provides off-the-shelf opportunities to develop custom hardware focused on the AI workload.
"Samsung Foundry is committed to building a rich and resilient ecosystem with our industry partners," said Mijung Noh, VP of Foundry Design Service Team at Samsung Electronics. "We are pleased to work with SiFive to accelerate our customers' AI / ML custom SoC design, illustrating the innovation potential we can achieve together."
“Working in partnership with Samsung Foundry has accelerated SiFive’s ability to deliver our highly-efficient and configurable approach for SoC design and implementation,” said Dr. Yunsup Lee, CTO of SiFive. “We’re excited to continue to co-innovate with Samsung Foundry as we launch our latest SiFive Intelligence products to accelerate the development of next-generation AI SoCs with Samsung’s advanced process technology.”
SiFive’s industry-leading portfolio of RISC-V-based processor Core IP scales from the AI-focused SiFive Intelligence line of Linux-capable 64-bit multi-core processors with enhanced RISC-V Vector capabilities to area-optimized real-time cores. SiFive Core IP is silicon-proven and pre-integrated with SiFive Shield, a whole-SoC security solution, SiFive Insight advanced trace and debug capabilities, and optional SiFive hardware cryptographic accelerator (HCA). Read more information on the latest update for the SiFive 21G1 release on the SiFive Intelligencepage.
Further development of the AI accelerator SoC by integrating SiFive Intelligence products into the chip will create a heterogeneous compute platform for machine learning that can be tuned to meet the needs of edge AI, automotive, 5G/networking, or data center accelerator needs.
About SiFive
SiFive is the leading provider of processor cores, AI accelerators, and SoC IP to enable domain-specific designs based on the open RISC-V instruction set architecture specification. SiFive offers scalable, configurable processor cores pre-integrated with security, trace, and debug features for workload-specific accelerator designs. Founded by the inventors of RISC-V, SiFive has design centers worldwide and backing from Sutter Hill Ventures, SK hynix, Qualcomm Ventures, Western Digital, Intel Capital, Spark Capital, Osage University Partners, and Prosperity7 Ventures. For more information, please visit www.sifive.com.
Hi FF,

A year ago SiFive was crowing about a "a customer-specific AI inference accelerator".

That is, the SoC silicon had to be designed for each specific customer, or at least for each customer application.

Akida's versatility makes the customer-specific design of SoC's obsolete. Now they can use the one SoC architecture for any customer-specific AI inference application. The "programming" of Akida comes down to a firmware reprogramming of the the model library used in the configuration of the Akida neuron weights.
 
  • Like
  • Fire
  • Love
Reactions: 39 users
Punching above your weight much Slade?

SC
 
  • Haha
  • Like
Reactions: 11 users

Diogenese

Top 20
  • Haha
Reactions: 6 users

Slade

Top 20
  • Haha
  • Like
Reactions: 8 users

TechGirl

Founding Member
Hi FF,

A year ago SiFive was crowing about a "a customer-specific AI inference accelerator".

That is, the SoC silicon had to be designed for each specific customer, or at least for each customer application.

Akida's versatility makes the customer-specific design of SoC's obsolete. Now they can use the one SoC architecture for any customer-specific AI inference application. The "programming" of Akida comes down to a firmware reprogramming of the the model library used in the configuration of the Akida neuron weights.

No Way Wow GIF
 
  • Like
  • Haha
  • Love
Reactions: 37 users
  • Like
  • Fire
  • Haha
Reactions: 10 users

VictorG

Member
Hi FF,

A year ago SiFive was crowing about a "a customer-specific AI inference accelerator".

That is, the SoC silicon had to be designed for each specific customer, or at least for each customer application.

Akida's versatility makes the customer-specific design of SoC's obsolete. Now they can use the one SoC architecture for any customer-specific AI inference application. The "programming" of Akida comes down to a firmware reprogramming of the the model library used in the configuration of the Akida neuron weights.
yes of course, I knew that.
 
  • Haha
  • Like
Reactions: 6 users

Diogenese

Top 20
  • Haha
  • Like
  • Wow
Reactions: 18 users

TechGirl

Founding Member
  • Like
  • Haha
  • Fire
Reactions: 8 users

TechGirl

Founding Member
Owls have more neck vertebrae than giraffes so they can look over their shoulders ...

Sup Reaction GIF
 
  • Haha
  • Like
Reactions: 13 users
Top Bottom