BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Learning
Learning to the Top 🕵♂️
That is a very interesting statement from Douglas.
Thanks, Tothemoon24.
If we look through Douglas LinkedIn; all his posts in the last 6 month is promoting BrainChip and recently MorseMicro ( Due to the investment of Megachips to MorseMicro)
What I find intresting is MorseMicro' WiFi Halo is base on the RISC-V architecture. We know BrainChip had partner with SiFive, so in essence, AKIDA should be compatible with RISC-V architecture.
BrainChip + Megachips + MorseMicro ( perhaps a mé·na·ges à trois to dominate WIFI without the internet???)
Edit: WiFi security cameras without internet.
Just my Sunday thought
Learning 🏖
Last edited:
Well you can't use the Akida 1500 FD-SOI IP on vanilla CMOS. The circuit design could be implemented in CMOS, but it would need a different tape-out design adapted for CMOS.
Bravo
If ARM was an arm, BRN would be its biceps💪!
I thought this extract from an interview with Jack Kang (SVP SiFive) was very interesting and confirms why patience is required while everything moves through the development cycle; with automotive having a much longer cycle than mobile or datacentres. Here Jack discusses SiFive's aim to move to 5nm chips and how long it's expected to take.
Speaking of 5nm, in one of @FactFinder's recent posts he said "We also know that AKD1000 was always claimed by Brainchip to be scalable down to at least 5nm. It is a fact that one of the reasons to scale down is that in so doing power usage reduces along with latency. In other words it becomes more efficient as you scale down."
Speaking of 5nm, in one of @FactFinder's recent posts he said "We also know that AKD1000 was always claimed by Brainchip to be scalable down to at least 5nm. It is a fact that one of the reasons to scale down is that in so doing power usage reduces along with latency. In other words it becomes more efficient as you scale down."
Q&A with SiFive SVP Jack Kang: 5 nm chips with SiFive cores likely to be used in cars by 2025-2026
SiFive successfully rolled out a portfolio of automotive CPUs based on the RISC-V standard, and is about to create a new ecosystem of chips, tier-1 auto component providers and OEM partners. DIGITIMES had an exclusive interview with SiFive SVP Jack Kang, who shared what triggered the RISC-V...
www.digitimes.com
You are right - see, eg, @Tothemoon24 's post below.
Fact Finder
Top 20
Just one additional thought.I have been looking at Edge Impulse blogs and noting there are 60,000+ developers running 130,000+ projects the following seems even to this old technophobe significant:
“Deployment Options
Any model that has been developed in the impulse design on Edge Impulse Studio can be deployed to BrainChip Akida target devices. In order to download these models for custom deployments, developers must choose the BrainChip MetaTFTMmodel block (refer Figure 5a) under the deployment stage to obtain a .zip file with the converted model in it. Alternatively, there is also a BrainChip Akida PCIe deployment block (refer Figure 5b) available which will generate a pre-built binary that can be used with Edge Impulse Linux CLI to run on any compatible Linux installation where this board is installed.
The pre-built binary that is provided from the deployment block also has in-built tools that provide performance metrics for the AKD1000 mini PCI reference board (refer Figure 7). This is a very unique integration as developers can not only deploy their favorite projects to BrainChip Akida devices but can also capture information such as efficiency that help build a prototype system with Akida technology”
[IMG alt="Graphical user interface, application, Teams
Description automatically generated"]https://lh3.googleusercontent.com/s...PAD9OP50EUJK1iUijH5qgjrsAroN_zwpRuo_krg[/IMG][IMG alt="Graphical user interface, application
Description automatically generated"]https://lh3.googleusercontent.com/U...XiJMFmxvc2E_2jna6hYv-MQdxORqpE42PxXapCw[/IMG]Figure 6a: Download BrainChip Akida-compatible model for custom deployment on BrainChip Akida devices Figure 6b: Download pre-built binary for AKD1000 mini PCIe development board
BrainChip Akida™ Platform Features in Edge Impulse
This blog highlights the new features with BrainChip technology that are now available in Edge Impulse Studio.www.edgeimpulse.com
Unless I am completely missing the point what Edge Impulse is saying is that any of the existing 130,000 projects can be simply deployed to AKIDA and you will be provided with performance figures so you can see the advantages such as they may be of running on AKIDA.
Later in this blog they talk about things an advanced user could additionally do so again this suggests that checking to see if AKIDA will improve performance is as simple as falling off a technology log.
This suggests at least to me that 1% is a very conservative estimate of what is happening in the Edge Impulse developer world where AKIDA is concerned.
My opinion only DYOR
FF
AKIDA BALLISTA
Edge Impulse by providing benchmarking of AKIDA against a GPU running at 900 megahertz has encouraged an extremely large portion of its developer community to consider the advantages that AKIDA might offer EVEN well back from the far Edge.
In so doing this could lead the developers who were reliant on a GPU running at 900 megahertz to rethink how they could actually sit at the Edge or as Dell prefers the End Point unconnected and running at extremely low power without the need for cooling.
My opinion only DYOR
FF
AKIDA BALLISTA
Fact Finder
Top 20
Can’t help myself.
Well anyone who has ever said “It’s all a blur to me.” would want one.
Sorry.
toasty
Regular
Of recent times I have begun to feel some impatience with the lack of price sensitive information flow. Holding for 7 years was proving to be a little wearying. However, the recent discoveries by the 1K eyes of the involvement of Megachips in the design of AKIDA1500 has boosted my enthusiasm for BRN again. Its caused me to do a little dot joining of my own.
There now seems to be a strong link between Brainchip - Megachips - RISC V, as per the post by @Learning. We also know that the GF process is a preferred platform for US agencies, including NASA and DoD and that GF is a trusted supplier to those agencies. We also know that RISC V is being used, or intending to be used, by NASA for space craft. We also know that NASA innovations are often picked up for military applications and, eventually, find their way into commercial products.
Here's my hypothesis. Megachips is working with BRN to implement smart sensors using AKIDA 1500 which, after initial processing (eg inference, SNN to CNN conversion), send the results to a "main" RISC V processor which acts on the smart sensor inputs. Potential applications for such a configuration include the NASA/Dod type alluded to above but could also include a VERY wide range of use cases involving RISC V. Lets think automotive, robotics, wearables, healthcare and comms just to start with. Could this (AKIDA 1500) be Megachips, Morse Micro and Brianchip proving up multiple use cases for AKIDA/RISC V. Is this why PVdM said that the market has not yet come to understand the significance of the Brainchip/Megachips relationship?
Once agin, these are the ramblings of an unknown internet poster so please DYOR and make up your own mind?
There now seems to be a strong link between Brainchip - Megachips - RISC V, as per the post by @Learning. We also know that the GF process is a preferred platform for US agencies, including NASA and DoD and that GF is a trusted supplier to those agencies. We also know that RISC V is being used, or intending to be used, by NASA for space craft. We also know that NASA innovations are often picked up for military applications and, eventually, find their way into commercial products.
Here's my hypothesis. Megachips is working with BRN to implement smart sensors using AKIDA 1500 which, after initial processing (eg inference, SNN to CNN conversion), send the results to a "main" RISC V processor which acts on the smart sensor inputs. Potential applications for such a configuration include the NASA/Dod type alluded to above but could also include a VERY wide range of use cases involving RISC V. Lets think automotive, robotics, wearables, healthcare and comms just to start with. Could this (AKIDA 1500) be Megachips, Morse Micro and Brianchip proving up multiple use cases for AKIDA/RISC V. Is this why PVdM said that the market has not yet come to understand the significance of the Brainchip/Megachips relationship?
Once agin, these are the ramblings of an unknown internet poster so please DYOR and make up your own mind?
Food4 thought
Regular
Can i the taste of APPLES
But you can be sure that the BrainChip patents will be under the microscope in China.
Bravo
If ARM was an arm, BRN would be its biceps💪!
Qualcomm and Mercedes-AMG PETRONAS F1Team collaborate to create innovative and disruptive experiences | Automotive World
Qualcomm Technologies, Inc. and the Mercedes-AMG PETRONAS F1 Team announced a multi-year agreement featuring the Snapdragon brand
www.automotiveworld.com
Dhm
Regular
Apologies for yet again bringing this up, but I still can't get my head around who is Prophesee is partnering with to get their -blur free- camera ready for inclusion within Snapdragon at Qualcomm. Who ever it is has a multi year contract with Qualcomm. I know we can do this with one hand tied behind my back, but the general feeling here is no, not us.Can’t help myself.
Well anyone who has ever said “It’s all a blur to me.” would want one.
Sorry.
Please please put me out of my misery once and for all.
Attachments
Fact Finder
Top 20
Yes but they cleverly wrote them in English so the secret sauce will be lost in translation. By the time they try it out using Fish Sauce Peter van der Made and Anil Mankar with assistance from Carnegie Mellon will be sitting on top a cortical column.
Last edited:
Realinfo
Regular
What would he know ??
I’ve just read the latest famous annual letter penned by Warren Buffet. Always interesting, I thought people here might like to hear a couple of his current thoughts.
After almost sixty years, Berkshire Hathaway has returned an average compounded growth of 19.8%, but in explaining this, Warren says…’Our satisfactory results have been the product of about a dozen truly good decisions, that would be about one every five years. Over time, it takes just a few winners to work wonders’.
Another of his remarks that pricked my ears, because it’s also been a topic of discussion here was…’let your winners run’.
Relating to the markets as we often do here, he said…’One advantage of our publicly traded segment is that episodically, it becomes easy to buy pieces of wonderful businesses at wonderful prices. It’s crucial to understand that stocks often trade at truly foolish prices, both high and low. Efficient markets exist only in textbooks’.
He concluded by saying…’At Berkshire, there will be no finish line’.
If it’s good enough for Warren, it’s good enough for me…but what would I know ??
I’ve just read the latest famous annual letter penned by Warren Buffet. Always interesting, I thought people here might like to hear a couple of his current thoughts.
After almost sixty years, Berkshire Hathaway has returned an average compounded growth of 19.8%, but in explaining this, Warren says…’Our satisfactory results have been the product of about a dozen truly good decisions, that would be about one every five years. Over time, it takes just a few winners to work wonders’.
Another of his remarks that pricked my ears, because it’s also been a topic of discussion here was…’let your winners run’.
Relating to the markets as we often do here, he said…’One advantage of our publicly traded segment is that episodically, it becomes easy to buy pieces of wonderful businesses at wonderful prices. It’s crucial to understand that stocks often trade at truly foolish prices, both high and low. Efficient markets exist only in textbooks’.
He concluded by saying…’At Berkshire, there will be no finish line’.
If it’s good enough for Warren, it’s good enough for me…but what would I know ??
Bravo
If ARM was an arm, BRN would be its biceps💪!
What about this?
Canon using what it describes as “cutting-edge technology," has created a new deep-learning image processing technology that aims to achieve 'true-to-life' photography, and focuses on three areas: noise reduction, color interpolation, and aberration diffraction correction (i.e lens blur correction).
New Canon AI processor fixes camera 'issues' - Australian Photography
Canon has shared information about a new deep learning-powered AI image processing technology that the company says can dramatically improve image quality.
Earlyrelease
Regular
Can’t help myself.
Well anyone who has ever said “It’s all a blur to me.” would want one.
Sorry.
FF
Are you saying what I think you’re saying. Akida can cure a hangover
Bravo
If ARM was an arm, BRN would be its biceps💪!
FF
Are you saying what I think you’re saying. Akida can cure a hangover
If AKIDA could do that, we'd be rich beyond our wildest dreams!!!!!
Fact Finder
Top 20
It is really quite simple.
AKIDA is SNN. Prophesee produces spikes so they just plug in to each other and work.
When you travel to the UK from Australia and you want to recharge your laptop you need a converter which you plug into the wall socket then plug in your laptop. This allows you to then charge your laptop. The extra wires and bits and pieces in this converter adds time (latency)to how long the power takes to get from the wall socket to your lap top and also adds resistance which uses up some of the charge travelling from the socket to your laptop. Your laptop will still charge up and work.
Should add if you leave it on charging for a while you will notice the converter becomes warm so you are losing energy through heat.
So Prophesee if it is not connected to a natively spiking processor like AKIDA sends out its spikes just the same but to be used by Qualcomm it needs to convert the spikes into something else so it’s processor can understand and process the information in the original spikes.
In the process of conversion there is some loss of energy/information so the converted spikes produce suboptimal information for Qualcomm to process.
It still works but not as well as it might if it did not have to convert the spikes first before they were processed.
In the camera on your mobile for taking happy snaps this is probably not a particular issue however the extra power draw needed to convert the spikes so they can be processed and the extra power used by a neuromorphic processor that does not use SNN could be significant and reduce battery life to such an extent that the selfie generation might get annoyed an choose an iPhone.
There is also the question of extra heat and the actual cost of the electronics involved.
Where suboptimal performance of Prophesee’s vision sensor would not be acceptable would be in tracking hypersonic missiles and planes where you do not want to waste a single spike or milli second. Even counting pills if complete accuracy is required would necessitate every spike being accounted for.
Hope this helps.
My opinion only DYOR
FF
AKIDA BALLISTA
Last edited:
Similar threads
