BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
It looks like we are in the Renesas R-Car Gen 5 system on chip.... planned for 2027. I can't see that Akida 1000/1500 is in R-Car S4 Gen 4?
Is there any evidence that Akida is in anything in the near future? Given Renesas did the 1500 tape out in Dec 2022... I'm hoping that there is a Renesas product planned before 2027.
www.eenewseurope.com
Is there any evidence that Akida is in anything in the near future? Given Renesas did the 1500 tape out in Dec 2022... I'm hoping that there is a Renesas product planned before 2027.
Renesas moves away from uncancellable contracts as inventory rises
Renesas Electronics is moving away from the ‘uncancellable contracts’ as it sees inventory rising and flat revenue.
www.eenewseurope.com
AusEire
Founding Member.
@Diogenese this is one for you mate. Looks and sounds like Akida but is it? 
https://www.nordicsemi.com/News/202...160-in-the-Vodafone-Asset-Solar-asset-tracker
Thanks to Daniel on one of the Facebook group chats for finding this
https://www.nordicsemi.com/News/202...160-in-the-Vodafone-Asset-Solar-asset-tracker
Thanks to Daniel on one of the Facebook group chats for finding this
Rise from the ashes
Regular
@Diogenese this is one for you mate. Looks and sounds like Akida but is it?
https://www.nordicsemi.com/News/202...160-in-the-Vodafone-Asset-Solar-asset-tracker
Thanks to Daniel on one of the Facebook group chats for finding this
Huawei and Nordic Cellular IoT Enter into Licensing Deal
Huawei Technologies has today entered into a patent license agreement with Nordic Semiconductor.
Nordic ports TinyML to cellular IoT chip
Nordic and Edge Impulse’s easy-to-use development tools, designers can now add TinyML embedded machine learning (ML) nRF9160 SiP cellular IoT prototypes
www.eenewseurope.com
Last edited:
Cartagena
Regular
Qualcomm's 'Holy Grail': Generative AI Is Coming to Phones Soon | Edge Impulse (a Qualcomm company)
Qualcomm wants its next-gen Snapdragon chips to use AI for more than just improving camera shots. "Instead of needing an entire server farm, Qualcomm's solution is to have a device's existing silicon brain do all the thinking needed — at no extra cost." cc: CNET, David Lumb, Ziad Asgharwww.linkedin.com
I can smell Akida inside these Qualcomms premium phone coming next year to markets via Generative AI on the device. Great read and interesting article!!
And please watch this awesome on-device AI demo by Qualcomm: https://lnkd.in/gy92AZJZ
Dozzaman1977
Regular
Neuromorphia has put the links after your post . This was spoken of in great detail a few months back .
DYOR GLTAH
Cartagena
Regular
DK6161 said:
Can you point us to where it says mass production containing our IP please? Eager to read and confirm
Here is the answer.. production volumes to be available in Q1 2024, so from January next year, not long to go
Attachments
Last edited:
Schnitzel lover
Regular
In my view, if the product revenue from this was going to be meaningfully market sensitive, we should be seeing some sort of industry buying of Brainchip shares, as they would be the ones who have visibility over the scale and or coming adoption of the product.
Just putting it out there. But just watched the TENNs video out by brainchip I think they highlighted the main reason we are where we are. "Generations ahead of the industry". What we have seems magical, but does the industry need it today? It seems it will come, but is it a case that we are too far ahead and the industry is saying we don't need to invent the wheel again just yet. I think we need Megachips or the like to come out with something so needed that others must follow. Yes the cloud is getting busy, but are we providing a solution not needed yet? Just a few thoughts, but maybe I'm not understanding enough, maybe the need is now?
Hi Cartagena,
I don't believe there is any evidence that Qualcomm uses Akida, although I wish they did.
They have their in-house Hexagon:
https://www.qualcomm.com/news/onq/2...ile-computing-performance-for-windows-laptops
https://www.theregister.com/2022/11/15/qualcomm_snapdragon_8_gen_2/?td=readmore
Qualcomm pushes latest Arm-powered Snapdragon chip amid bitter license fight
The Snapdragon 8 Gen 2 system-on-chip features eight off-the-shelf cores from Arm, which is locked in a bitter legal fight with Qualcomm over licenses and contracts.
...
This includes an AI acceleration engine that is, we're told, up to 4.35 times faster than the previous generation, and with a potential 60 percent increase in performance-per-watt, depending on how it's used. This unit can be used to speed up machine-learning tasks on the device without any outside help, such as object recognition, and real-time spoken language translation and transcription. The dual-processor engine can handle as low as INT4 precision for AI models that don't need a lot of precision but do need it done fast on battery power, which the 4-bit integer format can afford developers, according to Qualcomm.
Qualcomm is pushing the INT4 capabilities as a precision ideal for modern mobile apps. It said a cross-platform Qualcomm AI Studio is due to be made available in preview form in the first half of next year that will optimize developers' models for this precision as well as other formats. This studio looks like a typical IDE in which programmers can organize their training workflows.
...
The SoC supports up to 200MP image capture and 8K HDR video capture in 10-bit HDR, according to the specifications. Qualcomm said it worked with Samsung and Sony to develop large sensors that the 8 Gen 2 can handle. There are direct paths in the chip that link the Hexagon AI engines to, say, the image-processing units so that pictures and video can be manipulated more efficiently.
The processor, according to Qualcomm, can also be made to reduce the amount of data read and written during neural network inference – which saves power – by breaking input data into not just tiles as other chipsets do, but micro tiles that apparently do a better job of cutting down information transfer.
This is from a quote in a now-deleted Stable genius post on TSEx:
Qualcomm uses all of the Snapdragon SoC’s processing elements for AI processing and calls the combination of these processing elements the “AI engine.” Among the enhancements incorporated into the AI engine was a dedicated power plane and a doubling of the tensor processing cores within the Hexagon processor. The result is a 4.35x improvement in performance and an equally impressive 60% improvement in performance per watt efficiency. Qualcomm also added support for transformer neural network models which are critical for applications like natural language processing (NLP) for speech-to-text and text-to-speech translation. The Hexagon can splice the neural NLP model into smaller elements to run on micro tiles allowing for more efficient use of the processing cores. Additionally, Qualcomm added support for Int4 data structures. In many cases, this lower precision data structure can be used by neural network models like computational photography image enhancement without a noticeable loss of accuracy while improving speed and power efficiency. The end result is faster and more efficient processing of neural network models.
This is from a 2021 Qualcomm patent application:
WO2023049655A1 TRANSFORMER-BASED ARCHITECTURE FOR TRANSFORM CODING OF MEDIA 2021-09-27
Systems and techniques are described herein for processing media data using a neural network system. For instance, a process can include obtaining a latent representation of a frame of encoded image data and generating, by a plurality of decoder transformer layers of a decoder sub-network using the latent representation of the frame of encoded image data as input, a frame of decoded image data. At least one decoder transformer layer of the plurality of decoder transformer layers includes: one or more transformer blocks for generating one or more patches of features and determine self-attention locally within one or more window partitions and shifted window partitions applied over the one or more patches; and a patch un-merging engine for decreasing a respective size of each patch of the one or more patches.
...
[0075] The SOC 100 may also include additional processing blocks tailored to specific functions, such as a GPU 104, a DSP 106, a connectivity block 110, which may include fifth generation (5G) connectivity, fourth generation long term evolution (4G LTE) connectivity, Wi-Fi connectivity, USB connectivity, Bluetooth connectivity, and the like, and a multimedia processor 112 that may, for example, detect and recognize gestures. In one implementation, the NPU is implemented in the CPU 102, DSP 106, and/or GPU 104. The SOC 100 may also include a sensor processor 114, image signal processors (ISPs) 116, and/or navigation module 120, which may include a global positioning system.
Snapdragon8_2Hexagon
Snapdragon 8 Gen 2 deep dive: Everything you need to know (androidauthority.com)
Snapdragon 8 Gen 2 deep dive: Everything you need to know
All you'll ever need to know about Qualcomm's latest Snapdragon 8 Gen 2 mobile platform for next-gen phones can be found here.
www.androidauthority.com
...
Qualcomm doubled the physical link between the image signal processor (ISP), Hexagon DSP, and Adreno GPU, driving higher bandwidth and lowering latency. This allows the Snapdragon 8 Gen 2 to run much more powerful machine-learning tasks on imaging data right off the camera sensor. RAW data, for instance, can be passed directly to the DSP/AI Engine for imaging workloads, or Qualcomm can use the link to upscale low-res gaming scenarios to assist with GPU load balancing.
Schnitzel lover
Regular
The same issue in the energy space.
There’s technology available to supercede oil, gas, coal, and wind generated power, and has been since the mid 1900s.
Unfortunately it’s not whether the technology is good. It’s whether it gets adopted.
BRN clearly have the lead in the AI Edge space, but they are busting every valve to try to get it used and adopted, which is the ultimate challenge in my view..
They probably need to get the right powerful people invested in BRN to then get it the adoption that would make it ubiquitous.. IMO DYOR
Rise from the ashes
Regular
You think part of the lack of adoption is to do with that some companies have already made deep investments in developing other tech and want to see a return on that to save face before turning a new leaf and looking at other options?
Unrelated but interesting.
Mineralogy meets zero-shot computer vision
Mineralogy meets zero-shot computer vision
Identifying minerals is a complex and time-consuming problem for geologists, often taking anywhere from 30 minutes to several days per sample. Further complicating the situation is the fact that a sufficient portion of minerals remain inadequately researched, leaving us with just a few hundred...
robsmark
Regular
I agree that one cow needs to be milked before they move onto another.You think part of the lack of adoption is to do with that some companies have already made deep investments in developing other tech and want to see a return on that to save face before turning a new leaf and looking at other options?
Unrelated but interesting.
Mineralogy meets zero-shot computer vision
Mineralogy meets zero-shot computer vision
Identifying minerals is a complex and time-consuming problem for geologists, often taking anywhere from 30 minutes to several days per sample. Further complicating the situation is the fact that a sufficient portion of minerals remain inadequately researched, leaving us with just a few hundred...techxplore.com
Bravo
If ARM was an arm, BRN would be its biceps💪!
There are a few things that really boost my confidence when thinking about Qualcomm and they are, in no particular order:
“Although neuromorphic computing is still in its infancy, systems like these will be available on the market in just a few years,” Mercedes said. “When applied at scale throughout a vehicle, they have the potential to radically reduce the energy needed to run the latest AI technologies.”
There's probably a few more things I could have added but that's hopefully enough to demonstrate my point for now.
- Teksun had an existing ecosystem partnership with Qualcomm and yet they still chose to partner with us, presumably because we can offer technological capabilities that Qualcomm isn't in a position to provide. This is what Brijesh Kamani (Founder & CEO of Teksun) said at the time.
- In a similar vein, we know that Mercedes also works with Qualcomm but decided to announce publicly the employment of our technology within the Vision EQXX concept car with its emphasis on intuition, perception and real-time prediction and with confirmation that this technology will be employed in production versions. Interstingly Google will be deeply integrated within the Mercedes user interface. (UI/UX), the same place where the "Hey Mercedes" voice assistant resides, with hot word detection made 5-10 times more efficient than conventional voice control thanks to AKIDA. Here's some of what Mercedes had to say in January 2022.
“Although neuromorphic computing is still in its infancy, systems like these will be available on the market in just a few years,” Mercedes said. “When applied at scale throughout a vehicle, they have the potential to radically reduce the energy needed to run the latest AI technologies.”
- Then there's Prophesee and I doubt you could get a better endorsement of our technology that those given in the past by Luca Verre and they have agreed to a multi-year collaboration with Qualcomm. So, hopefully, we'll go wherever Prophesee goes.
There's probably a few more things I could have added but that's hopefully enough to demonstrate my point for now.
Last edited:
Getupthere
Regular
I agree… I think we desperately need a corner stone investor like SoftBank is to arm.
That would make people more accountable and drive results.
Schnitzel lover
Regular
Yeah I think it’s an age old problem that is confronting BRN.You think part of the lack of adoption is to do with that some companies have already made deep investments in developing other tech and want to see a return on that to save face before turning a new leaf and looking at other options?
Unrelated but interesting.
Mineralogy meets zero-shot computer vision
Mineralogy meets zero-shot computer vision
Identifying minerals is a complex and time-consuming problem for geologists, often taking anywhere from 30 minutes to several days per sample. Further complicating the situation is the fact that a sufficient portion of minerals remain inadequately researched, leaving us with just a few hundred...
Rob, Antonio and Sean are obviously feverishly working at trying to fast track their contacts and connections to work this challenge and build the Ecosystem and make it the ecosystems benefit to adopt..
Bravo
If ARM was an arm, BRN would be its biceps💪!
Bravo
If ARM was an arm, BRN would be its biceps💪!
There's hearsay and there's heresy, and Qualcomm ain't hearsay.
Similar threads
