BRN Discussion Ongoing

Steve10

Regular
Prophesee tech is also in Imago Technologies products.


HIGH-SPEED COUNTING


With Prophesee’s patented event-based vision technology, objects are counted as they pass through the field of view, triggering each pixel independently as the object goes by.

By only recording the pixels independently triggered by changes, Prophesee’s Event-Based Vision captures the essential information the system requires and no more.

This new approach allows for unprecedented counting speed and accuracy

>1000 Obj/s.
Throughput

>99.5%
Accuracy


1675810084295.png


KINEMATIC MONITORING


Event-Based Vision unlocks new applications such as Kinematic Monitoring, designed to monitor in real-time any deviation in your machine production cycles.

A machine with parts moving following a regular cycle can be monitored with precision by assessing if each pixel is triggered at the exact time it should be. The system will flag any deviation, spatially or temporarily, before they can cause a production stop.

In addition, it can log KPIs, such as average cycle time and shutdown time.

1%
Motion Period Irregularity Detection

1pixel
Motion Amplitude Defficiency Detection

1675810137548.png


VIBRATION MEASUREMENT


Drastically improve your predictive maintenance strategy by measuring equipment vibrations from 1Hz to 1000Hz continuously, in real time and in a non-intrusive way.

This information now allows your maintenance team to observe and understand any process deviation long before machines malfunction or break down.

In this way, Event-Based Vision will ensure the longevity of your machinery, making your capital investments go further.

1 pixel
Minimal Amplitude Detection

>1000 Hz
Maximum Frequency Detection

1675810182092.png



 
  • Like
  • Fire
  • Wow
Reactions: 40 users

Bravo

If ARM was an arm, BRN would be its biceps💪!
Here's a report just published that lists leading companies in neural networks for radar/lidar for the aerospace and defence industry. It says there have been over 174,000 patents filed and granted in the aerospace and defence industry in the last three years alone. Ford Motor leading the pack with 29 patents.

Screen Shot 2023-02-08 at 9.50.13 am.png


Screen Shot 2023-02-08 at 9.47.53 am.png



Screen Shot 2023-02-08 at 9.48.41 am.png




Screen Shot 2023-02-08 at 9.48.55 am.png


Screen Shot 2023-02-08 at 9.49.03 am.png


 
  • Like
  • Fire
  • Love
Reactions: 32 users

Boab

I wish I could paint like Vincent
Prophesee tech is also in Imago Technologies products.


HIGH-SPEED COUNTING


With Prophesee’s patented event-based vision technology, objects are counted as they pass through the field of view, triggering each pixel independently as the object goes by.

By only recording the pixels independently triggered by changes, Prophesee’s Event-Based Vision captures the essential information the system requires and no more.

This new approach allows for unprecedented counting speed and accuracy

>1000 Obj/s.
Throughput

>99.5%
Accuracy


View attachment 28973

KINEMATIC MONITORING


Event-Based Vision unlocks new applications such as Kinematic Monitoring, designed to monitor in real-time any deviation in your machine production cycles.

A machine with parts moving following a regular cycle can be monitored with precision by assessing if each pixel is triggered at the exact time it should be. The system will flag any deviation, spatially or temporarily, before they can cause a production stop.

In addition, it can log KPIs, such as average cycle time and shutdown time.

1%
Motion Period Irregularity Detection

1pixel
Motion Amplitude Defficiency Detection

View attachment 28974


VIBRATION MEASUREMENT


Drastically improve your predictive maintenance strategy by measuring equipment vibrations from 1Hz to 1000Hz continuously, in real time and in a non-intrusive way.

This information now allows your maintenance team to observe and understand any process deviation long before machines malfunction or break down.

In this way, Event-Based Vision will ensure the longevity of your machinery, making your capital investments go further.


1 pixel
Minimal Amplitude Detection

>1000 Hz
Maximum Frequency Detection

View attachment 28975
Whether we are involved or not (and I hope we are) this technology really is gob smacking.
Love it.
 
  • Like
  • Fire
  • Love
Reactions: 19 users

Kyba

Emerged
Apologies if already posted, still catching up on the thread for the last day.
Interesting article with a number of mentions for Brainchip.

Neuromorphic Chip Market - Growth, Trends, Covid-19 Impact, And Forecasts (2022 - 2027)​

 
  • Like
  • Fire
  • Love
Reactions: 24 users

Steve10

Regular
WBT now $1B MC & BRN now $1.13B MC.

BRN is a few years ahead of WBT who recently taped out their chip.

BRN also has many big companies as partners.

BRN heavily shorted whereas WBT is not.

Is WBT expensive or is BRN inexpensive?
 
  • Like
  • Thinking
  • Fire
Reactions: 39 users
WBT now $1B MC & BRN now $1.13B MC.

BRN is a few years ahead of WBT who recently taped out their chip.

BRN also has many big companies as partners.

BRN heavily shorted whereas WBT is not.

Is WBT expensive or is BRN inexpensive?
I think the shorts are going to get a hold of WBT, which might give us some breathing room
 
Also a roughly a year ago we migrated to this platform
 


At 32min in video Prophesee event based vision sensor is mentioned & will be in smartphones in 2023.

Along with the following comments from the Prophesee/Brainchip partnership indicate highly likely Akida IP will be in smartphones with Qualcomm chips this year.

“We’ve successfully ported the data from Prophesee’s neuromorphic-based camera sensor to process inference on Akida with impressive performance,” said Anil Mankar, Co-Founder and CDO of BrainChip. “This combination of intelligent vision sensors with Akida’s ability to process data with unparalleled efficiency, precision and economy of energy at the point of acquisition truly advances state-of-the-art AI enablement and offers manufacturers a ready-to-implement solution.”

“By combining our Metavision solution with Akida-based IP, we are better able to deliver a complete high-performance and ultra-low power solution to OEMs looking to leverage edge-based visual technologies as part of their product offerings,” said Luca Verre, CEO and co-founder of Prophesee.


According to Statista 1.2 billion smartphones were sold in 2022.

And Qualcomm states:
In the same online briefing, chief financial officer Akash Palkhiwala predicted smartphone shipments would drop by roughly 5 percent year-on-year in 2022, dropping from prior estimates of 750 million units to between 650 million to 700 million devices.

Thus Qualcomm has 650-700 million / 1.2 billion = 54.16% - 58.33% smartphone market share.

Initially Akida IP will most likely be used in 10% of high end smartphones & later be available in all Qualcomm chip smartphones.

650-700 million chips per year x 10% Akida IP = 65-70 million.

If we allow $15-20 per chip x 3% Akida royalty = 45-60c BRN revenue per Qualcomm chip smartphone.

We'll use 50c BRN revenue per chip x 65-70 million units = $32.5-35M revenue.

When the tech becomes standard in all Qualcomm chip smartphones there will be 650-700 million chips per year x 50c BRN revenue = $325-350M revenue.

Qualcomm have also released Snapdragon Ride Flex SoC for vehicles.


Personally, not expecting Akida in Snapdragon this year but very happy to be wrong.

Some articles delving into the Snapdragon over the past few mths over 2 posts.

Samsung have just announced a few hours ago they will be using Sanpdragon now instead of their own chip.

Haven't seen a hard mention of Prophesee as yet in spec releases but will see whether Qualcomm Ann that (as to how they achieved the deblur) or if leave it to the phone manufacturers themselves?

Suspect that if we are in there somewhere that we would be in the 2nd AI Processor - Sensing Hub?

Believe this targeted use is custom wake words, always on, low power and I think some camera support?



Qualcomm Brings Big SoC Upgrade to Samsung Galaxy S23 Series​

6 hours ago by Jake Hertz

Samsung's new flagship smartphone is leaving behind in-house Exynos processors in favor of Qualcomm’s Snapdragon 8 Gen 2 mobile processor.​


The high-end smartphone market has been mostly dominated by two players: Apple with its iPhone line and Samsung with the Galaxy S series as a formidable competitor.
Now, Samsung is continuing the growth and improvement of its Galaxy S series with its upcoming Galaxy S23 series. Last week, Samsung made headlines with the announcement that all variants of the new phone will be powered by Qualcomm’s Snapdragon 8 Gen 2 mobile processor.

Samsung Galaxy S23 series

Samsung Galaxy S23 series. Image (modified) courtesy of Samsung


In this article, we’ll look at Samsung’s decision to adopt Qualcomm’s processor and what features the processor will bring to the S23.

Samsung Turns to Outsourcing SoCs​

Samsung is choosing to fully outsource the SoCs for its flagship product. Historically, Samsung has launched its Galaxy S line with two different SoCs, which vary based on the region. In places like Europe and India, Samsung Galaxy S phones have traditionally been powered by Samsung’s in-house Exynos processor family, while locations including the U.S., China, and Japan have dominantly been powered by a Snapdragon processor.

Samsung's Exynos 1280 SoC

Samsung's Exynos 1280 SoC. Image courtesy of Samsung


While Samsung has claimed that both variants are functionally equivalent, benchmarking has shown that Snapdragon variants consistently outperform Exynos variants. Hence, to appease consumers, Samsung has chosen to go exclusively with Qualcomm’s Snapdragon SoCs for the Galaxy S23 line.
As major competitors such as Apple and Google are moving toward in-house SoCs for their products, it seems that Samsung is trending in the opposite direction—at least for Galaxy S23.

The Snapdragon 8 Gen 2 Focuses on AI​

To power the new Galaxy S23 family, Samsung has chosen the Qualcomm Snapdragon 8 Gen 2 mobile processor, which is designed with high-performance AI in mind.
At the heart of the SoC is Qualcomm’s AI Engine, which starts off with a 3.36 GHz Kyro CPU and an Adreno GPU. According to Qualcomm, a new microarchitecture in Kyro has enabled a 40% increase in power efficiency, while the Adreno GPU achieves up to 25% improved performance and 45% better power efficiency over previous generations.

The Snapdragon 8 Gen 2 block diagram

The Snapdragon 8 Gen 2 block diagram. Image courtesy of Qualcomm


Alongside the CPU and GPU in the AI Engine is Qualcomm’s Hexagon processor. This subsystem features a number of important features for boosting AI performance, such as a tensor accelerator, a scalar accelerator, a fused AI accelerator, a dedicated power delivery system, and an industry-first implementation of micro-tile inferencing. The AI Engine also includes the Qualcomm Sensing Hub, which features always-on and ultra-low power for audio and sensors.
Qualcomm claims that the Snapdragon 8 Gen 2 achieves 4.35x faster AI performance than the previous generation.

Other Features of Snapdragon 8 Gen 2​

Beyond AI support, the Snapdragon 8 Gen 2 SoC includes other important features.
The processor is part of Qualcomm’s first-ever mobile platform with a cognitive image signal processor (ISP). The ISP featured on this SoC is a Qualcomm Spectra, which is an 18-bit triple ISP that includes a hardware accelerator for computer vision, an image signal processor, and a cognitive ISP. With this, the Samsung Galaxy S23 will be able to support a 200 MP camera as well as 8K 30fps video capture.

Micro-tile inferencing

Micro-tile inferencing on the Hexagon and Snapdragon 8 Gen 2 processors. Screenshot courtesy of Qualcomm


Additionally, for connectivity, the SoC features the Snapdragon X70 5G Modem-RF System, which achieves 5G with download and upload speeds up to 10 Gbps and 3.5 Gbps, respectively. The device includes Qualcomm FastConnect 7800 for Wi-Fi 7 connectivity up to 5.8 Gbps.


Qualcomm Snapdragon 8 Gen 2 4nm Mobile Platform official: 35% faster CPU, 25% faster GPU​


Snapdragon-8-Gen-2-Mobile-Platform-1024x636.jpg

Qualcomm just introduced the Snapdragon 8 Gen 2, the company’s latest flagship Mobile Platform at the Qualcomm Snapdragon Tech Summit 2022 event, as expected.

CPU​

Snapdragon-8-Gen-2-CPU-features-1024x544.jpg

Compared to the predecessor, the new Kryo CPU based on latest ARM-V9 architecture promises 35% performance improvements and 40% improved power efficiency with new 1-4-3 microarchitecture. It has 1 x Prime core, up to 3.2 GHz, 4 x Performance cores up to 2.8 GHz and 3 Efficiency cores, up to 2.0 GHz.

GPU​

Snapdragon-8-Gen-2-GPU-1024x567.jpg

The Snapdragon 8 Gen 2 has a new Adreno GPU that promises up to 25% faster performance, with up to 45% better power efficiency compared to the predecessor. There are new Snapdragon Elite Gaming features, including real time hardware-accelerated ray tracing for life-like light, reflections, and illuminations to mobile games.
Snapdragon-8-Gen-2-Ray-Tracing-partners-1024x520.jpg

The company said that gamers can experience photorealistic human characters in their games with the world’s first mobile optimized support for Unreal Engine 5 Metahumans Framework on Snapdragon. It is also the first mobile platform to support Vulkan 1.3 API.


AI​

Snapdragon-8-Gen-2-AI-features-1024x567.jpg

The upgraded Qualcomm Hexagon Processor offers faster natural language processing with multi-language translation and advanced AI
camera features.
The Hexagon processor features brand-new upgrades to the architecture,
including micro tile inferencing and a bigger tensor accelerator for up to 4.35X increased AI performance.
This is also the first Snapdragon mobile platform to support INT4—a revolutionary AI precision format—with 60% performance/watt improvement for sustained AI inferencing.
And now, the Qualcomm Sensing Hub features dual-AI processors for the first time, powering exciting new experiences like direct-to-app voice assistance for convenient control of your favorite apps.

ISP​

Snapdragon-8-Gen-2-ISP-1024x481.jpg

The Spectra 18-bit triple Cognitive ISP is the first ever AI-powered camera processor from Qualcomm. This new architecture powers real-time Semantic Segmentation, to recognize and optimize each aspect within a
frame—like faces, hair, clothes, backgrounds, and more.
Snapdragon-8-Gen-2-Quad-exposure-Sony-sensors-1024x484.jpg

Sony Semiconductor Solutions is the first to develop quad digital overlap HDR technology, and which is fine-tuned for Snapdragon. This is available in Sony IMX800 1/1.5″ and the IMX989 1″ sensors.
The Samsung ISOCELL HP3, the first 200-megapixel image sensor optimized for Snapdragon 8 Gen 2, promise professional quality photos and videos.
It offers up to 200 MP photo capture and 8K HDR video capture in 10-bit HDR. This is also the first Snapdragon to include an AV1 codec with support for video playback up to 8K HDR at 60 frames per second.

Connectivity​

Snapdragon-8-Gen-2-connectivity-1024x578.jpg

It has Snapdragon X70 5G Modem RF System, making it the first in the world with a dedicated 5G AI processor that offers 5G+5G/4G Dual-SIM Dual-Active, which harnesses the power and flexibility of two 5G SIMs at once.
It supports Wi-Fi 7 offering up to 5.8 Gbps speed and dual Bluetooth connectivity for sustained immersion, thanks to new Qualcomm FastConnect 7800 Mobile Connectivity System.
This is the world’s first commercial Wi-Fi 7 SoC, with advanced High Band
Simultaneous Multi-Link that enables peak Wi-Fi 7 performance globally, allowing for smooth, blazing connections when streaming, gaming, and more.

Audio​

Snapdragon Sound Technology, now features spatial audio with head-tracking (when using compatible Snapdragon Sound products) for complete
surround-sound immersion.
It supports for 48kHz lossless music streaming and ultra-low latency Bluetooth streaming <48ms for lag-free gaming.
Snapdragon-8-Gen-2-features-1024x576.jpg

Qualcomm Snapdragon 8 Gen 1 (SM8550) specifications

  • 1 x Kryo Prime CPU (Arm Cortex-X3 based) at up to 3.2GHz, 4 x Kryo Performance CPUs (2 x Arm Cortex-A715 and 2 x Arm Cortex-A710 based) at up to 2.8GHz, 3x Kryo Efficiency CPUs (Arm Cortex-A510 based) at up to 2GHz, ARMv9 architecture
  • 4nm Process Technology
  • Built-in Qualcomm Snapdragon X70 5G modem-RF system
  • 5G speeds of up to 10Gpbs down, 5G mmWave and sub-6 GHz, standalone (SA) and non-standalone (NSA) modes, mmWave: 1000 MHz bandwidth, 8 carriers, 2×2 MIMO, Sub-6 GHz: 300 MHz bandwidth, 4×4 MIMO, Qualcomm 5G PowerSave Gen 3, Qualcomm Smart Transmit 3.0 technology, Qualcomm Wideband Envelope Tracking, Qualcomm 5G Ultra-Low Latency Suite. Global 5G multi-SIM, including 5G-5G/4G DualSIM Dual-Active (DSDA)
  • Qualcomm Adreno next-gen GPU, Vulkan 1.3 API support, HDR gaming (10-bit color depth, Rec. 2020 color gamut), API Support: OpenGL ES 3.2, OpenCL 2.0 FP, Vulkan 1.3, Real-time Hardware-Accelerated Ray Tracing
  • Hardware-accelerated H.265 and VP9 decoder, HDR Playback Codec support for HDR10+, HDR10, HLG and Dolby Vision
  • Qualcomm Aqstic audio codec, New Qualcomm Aqstic smart speaker amplifier, Total Harmonic Distortion + Noise (THD+N), Playback: -108dB, Qualcomm Audio and Voice Communication Suite, Spatial Audio with Head Tracking
  • Up to 36 MP triple camera, Up to 64+36 MP dual camera, Up to 108 MP single camera with ZSL, Up to 200MP single camera, Qualcomm Spectra Image Sensor Processor (Cognitive 18-bit triple-ISP)
  • Rec. 2020 color gamut photo and video capture, Up to 10-bit color depth photo and video capture, 8K HDR Video Capture + 64 MP Photo Capture
    10-bit HEIF: HEIC photo capture, HEVC video capture, Video Capture Formats: HDR10+, HDR10, HLG, Dolby Vision, 8K HDR Video Capture @ 30 FPS, 4K Video Capture @ 120 FPS, Slow-mo video capture at 720p @ 960 FPS, Bokeh Engine 2 for Video Capture, Pro Sight Video Capture , Video super resolution, Multi-frame Noise Reduction (MFNR), Locally Motion Compensated Temporal Filtering, Multi-Frame and triple exposure staggered/digital, overlap HDR dual sensor support, AI-based face detection, auto-focus, and auto-exposure
  • On-Device Display Support: 4K at 60 Hz / QHD+ at 144 Hz, Maximum External Display Support: up to 4K @ 60 Hz, 10-bit color depth, Rec. 2020 color gamut, HDR10 and HDR10+, Demura and subpixel rendering for OLED Uniformity
  • Qualcomm Hexagon Processor, Fused AI Accelerator, Hexagon Tensor Accelerator, Hexagon Vector eXtensions, Hexagon Scalar Accelerator, Support for mix precision (INT8+INT16), Support for all precisions (INT4, INT8, INT16, FP16), Micro Tile Inferencing
  • Support for LP-DDR5x memory up to 4200 MHz
  • Wi-Fi 7 Peak speed: 5.8 Gbps, Wi-Fi 6E, 802.11ax (Wi-Fi 6) Multi-gigabit, Integrated 802.11ac 2×2 (2-stream) MU-MIMO, 2.4 GHz, 5 GHz and 6GHz, Bluetooth 5.3, LE Audio, Dual Bluetooth antennas
  • Bluetooth audio: Snapdragon Sound Technology with support for Qualcomm aptX Voice, aptX Lossless, aptX Adaptive, and LE audio, NFC
  • GPS, Glonass, BeiDou, Galileo, QZSS, NavIC capable, Dual Frequency GNSS (L1/L5), Sensor-Assisted Positioning, Urban pedestrian navigation with sidewalk accuracy, Global freeway lane-level vehicle navigation
  • USB 4.0, USB-C
  • Qualcomm Quick Charge 5 technology

Availability

Snapdragon-8-Gen-2-availability-1024x568.jpg

Qualcomm confirmed that Snapdragon 8 Gen 2 platform will be adopted by global OEMs and brands including ASUS ROG, HONOR, iQOO, Motorola, nubia, OnePlus, OPPO, REDMAGIC, Redmi, SHARP, Sony, vivo, Xiaomi, XINGJI/MEIZU, and ZTE, with the first commercial devices expected by the end of 2022
 
  • Like
  • Fire
  • Thinking
Reactions: 28 users

Snapdragon 8 Gen 2: Meet the AI super chip that should power Samsung Galaxy S23​

By Philip Michaels
last updated November 22, 2022
Here's what to expect from the chipset likely to feature in the biggest Android phones next year

Photo of Snapdragon 8 Gen 2 chip

(Image credit: Future)

WAILEA, Hawaii — The chipsets powering smartphones are getting smarter, and the ones made by Qualcomm are no exception.

Qualcomm is showing off the flagship silicon that will power many of the best Android phones in 2023 — likely including the Samsung Galaxy S23 — and a beefed-up AI engine leads the parade of new features coming to the latest Snapdragon chip.
Specifically, the Snapdragon 8 Gen 2 unveiled at the Snapdragon Summit November 15 has been "purpose built for AI," according to Cisco Cheng, a senior director of product marketing at Qualcomm. That means a new Qualcomm AI Engine powered by an upgraded Hexagon processor that supports features like multi-language translation, customized wake words and camera capabilities that tap into artificial intelligence.


The Snapdragon 8 Gen 2 and its AI improvements arrive as rival chip makers are paying more attention to the neural processing capabilities of their own silicon. Think Google, which now designs its own Tensor chips to power a number of smart features unique to the company's handsets like the Pixel 7. MediaTek also touts the AI capabilities of the new Dimensity 9200 system-on-chip it announced a week ago.


(Image credit: Tom's Guide)
But it's the Snapdragon 8 Gen 2 that figures to power more of the marquee Android phones we're expecting to see in the coming year. And for that reason, it's worth paying attention to the upgrades Qualcomm's made to the successor to the Snapdragon 8 Gen 1 chipset introduced a year ago.

Here's what we know so far about the Snapdragon 8 Gen 2 and what it means for upcoming Android devices.

Snapdragon 8 Gen 2 availability​

Qualcomm says the Snapdragon 8 Gen 2 will begin to appear in phones before the end of this year. We'd imagine that those early devices will be debuting in China initially, though Qualcomm is very particular about leaving specific phone announcements to its hardware-making partners.
To that end, the company did share the name of multiple phone makers who've committed to releasing Snapdragon 8 Gen 2 devices, including Motorola, OnePlus, Oppo, Redi Sony, Xiaomi and ZTE among the companies whose handsets are sold in the U.S. and Europe. That Qualcomm also called out makers of gaming phones like Asus Republic of Gamers, Nubia and RedMagic suggests that mobile gaming ranks high among the Snapdragon 8 Gen 2's feature set.
Conspicuously missing from that list of phone makers is Samsung, though that shouldn't raise any eyebrows. Samsung is usually left out of such Qualcomm announcements, even though the leading phone maker includes Snapdragon silicon in a lot of its handsets, including its Galaxy S flagship line.

Galaxy S23 leaked render (Image credit: OnLeaks/SmartPrix)
In fact, the next version of that phone — the Samsung Galaxy S23 — has been tipped to feature the Snapdragon 8 Gen 2 in all models, not just the ones released in the U.S. and other markets. (Galaxy S phones in Europe have run on Samsung's own Exynos chips in recent years.)
Samsung may not have been mentioned during the Snapdragon 8 Gen 2 unveiling, but we won't have to wait long to see if the chipset makes its way into Samsung's next flagship — rumors have the S23 appearing as soon as late January.

Snapdragon 8 Gen 2: AI Engine​

Qualcomm executives are talking up capabilities like the ability to translate multiple languages at once or blur the background in videos — features that will sound very familiar to Pixel 7 users.
The Qualcomm AI Engine on the Snapdragon 8 Gen 2 is the fastest Qualcomm has ever included on its silicon. The engine's tensor accelerator is twice as large as before and the Hexagon processor now support micro tile inferencing, a processor for speeding up neural network processing to power better AI experiences.
Those kinds of hardware changes translate to a 4.35x boost in AI performance over the Snapdragon 8 Gen 1, according to Qualcomm's math. Support for INT4 precision — the first time that's been included on a Snapdragon mobile platform — figures to bolster performance-per-watt by 60%.

(Image credit: Qualcomm)
A faster, more power efficient AI Engine should enable new features on Snapdragon 8 Gen 2-powered phones that sound a lot like what Google's Tensor G2 chipset is bringing to the table for the latest Pixel 7 models. Qualcomm executives are talking up capabilities like the ability to translate multiple languages at once or blur the background in videos — features that will sound very familiar to Pixel 7 users.
In another change, Qualcomm is adding a second AI processor to the Qualcomm Sensing Hub, a feature that collects and analyzes data from all around you. In the past, that's fueled features like triggering AI-based noise cancellation in noisy settings or making your ringtone louder so that you can hear it over background noise. But dual AI processors could enable custom wake words to control specific apps, Qualcomm says.

Snapdragon 8 Gen 2: Performance improvements​

Any new chipset is going to generate interest in how well it performs. Like its predecessor, the Snapdragon 8 Gen 2 is built on a 4-nanometer process, putting it in the same league as another 4nm chipset, the A16 Bionic.
The A16 Bionic delivers the best performance of any mobile chip we've tested, so to combat that, Qualcomm is going with a new Kryo CPU built around a prime core, four performance cores and three efficiency cores. That moves one of the efficiency cores from previous Snapdragon 8 chipsets into the performance core realm.
Qualcomm says its new CPU should be 35% faster than the Snapdragon 8 Gen 1. That kind of gain in, say, Geekbench 5 results would fall short of what the A16-powered iPhone 14 Pro can do on that benchmark, which measures overall performance. But it would be very close to the results we've seen from A15 Bionic-powered iPhones, which is particularly relevant as Apple still uses that chipset in its entry-level iPhone 14.

The A16 Bionic chip powering the iPhone 14 Pro Max is the current pace-setter in our tests. (Image credit: Tom's Guide)
We had the chance to run some Snapdragon 8 Gen 2 benchmarks, and the new chip certainly delivers the promised performance improvements, with our Snapdragon 8 Gen 2-powered test device posting much better numbers than Snapdragon 8 Gen 1 phones like the Galaxy 22 Ultra in both general performance and graphics tests. Even more impressive, the Snapdragon 8 Gen 2 beat the A15 Bionic-powered iPhone 14 in a lot of the tests we ran, though the A16 Bionic still seems like the better performer.
The more intriguing number cited by Qualcomm involves the power efficiency of the Snapdragon 8 Gen 2's Kryo CPU. It's supposed to improve power consumption by 40%. A common complaint when we reviewed Snapdragon 8 Gen 1-powered phones this past year was unimpressive battery life, especially when we ran our custom battery test in which phones are made to surf the web over cellular until they run out of power.
We noticed an improvement to battery test results with the Snapdragon 8 Plus Gen 1 chipset that came out later in the year. Hopefully, the Snapdragon 8 Gen 2 leads to even bigger gains in helping phones to last longer on a charge.

Snapdragon 8 Gen 2: Snapdragon Elite Gaming​

Like the Kryo CPU, the Adreno GPU on board the Snapdragon 8 Gen 2 also gets a boost over its predecessor. Qualcomm expects a 25% bump in graphics rendering and a 45% boost to power efficiency, each when compared to the Snapdragon 8 Gen 1.
Qualcomm is also adding to the array of Snapdragon Elite Gaming features, the chip maker's initiative for bringing console-quality gaming to smartphones. The Snapdragon 8 Gen 2 will be Qualcomm's first chipset to offer real-time hardware-accelerated ray tracing. That should mean more realistic light and reflections when you're playing graphically demanding mobile games.
Also new is Snapdragon's support for the Unreal Engine 5 Metahumans framework that's been optimized for mobile. The added support should allow games to feature more photorealistic human characters.
Qualcomm says the Snapdragon 8 Gen 2 is the first mobile platform with Vulkan 1.3 support, boosting Vulkan performance by 30%.

Snapdragon 8 Gen 2: Image signal processor and camera capabilities​

The AI neural engine can recognize faces, facial features, hair, clothes and sky among other things, optimizing and image tuning each part of a photo.
Cameras support remains a critical part of Qualcomm's system-on-chip, so the Snapdragon 8 Gen comes equipped with the Spectra 18-bit triple cognitive image signal processor (ISP). For the first time, the Spectra ISP is powered by artificial intelligence with the help of an AI neural engine.
The AI neural engine can recognize faces, facial features, hair, clothes and sky among other things, optimizing and image tuning each part of a photo. It's known as real-time semantic segmentation, allowing the ISP to recognize and optimize each aspect within a frame. Snapdragon chipsets have supported semantic segregation since the Snapdragon 865, but now it's part of the hardware itself. That means the AI engine is talking to the ISP, telling it whether each pixel makes up a face or sky or some other part of the picture; the ISP then processes that pixel accordingly.

(Image credit: Tom's Guide)
Last year's Snapdragon 8 Gen 1 chipset put bokeh blurs in the hands of the hardware. That feature's being improved with the Snapdragon 8 Gen 2 so that the size and shape of the blur can change on the fly.
The Snapdragon 8 Gen 2 can support photo capture of up to 200MP — good news for all those rumors of a Samsung Galaxy S23 Ultra with a 200MP main camera. You'll also get support for 8K HDR video capture in 10-bit HDR.
We got a chance to see some camera demos during the November Snapdragon Summit, and there are 5 killer camera features supported by the Snapdragon 8 Gen 2 that you should get excited about.

Snapdragon 8 Gen 2: Connectivity​

On the connectivity front, the Snapdragon 8 Gen 2 features the Snapdragon x70 modem Qualcomm announced earlier this year. That modem features a dedicated AI processor for optimizing 5G performance.

(Image credit: Qualcomm)
Just as significantly, the Snapdragon 8 Gen 2 will support 5G+5G/4G Dual-SIM Dual-Active (DSDA), which allows for the use of two 5G+5G or 5G+4G SIM cards at once for greater flexibility.
Chip makers are adding Wi-Fi 7 support to their silicon these days, even if Wi-Fi 7-ready devices aren't widely available yet. Nevertheless, the Snapdragon 8 Gen 2 joins MediaTek's Dimensity 9200 in supporting the new wireless standard.
According to Qualcomm, the Snapdragon 8 Gen 2's FastConnect 7800 system with Wi-Fi 7 taps High-Band Simultaneous Multilink to deliver massive throughput. Wi-fi speeds can reach up to 5.6 Gbps, doubling the performance of Wi-Fi 6.

Snapdragon 8 Gen 2: Sound and security​

Other Snadpragon 8 Gen 2 features include support for Qualcomm's Snapdragon Sound audio standard. That includes support for spatial audio with dynamic head-tracking and as well as 48kHz lossless music streaming support.

(Image credit: Qualcomm)
Security comes through Snapdragon Secure, which promises the latest support for isolation, cryptography, key management and attestation among other security features. Qualcomm says its new chipset gets an updated face unlocking system, too.

Snapdragon 8 Gen 2: Outlook​


(Image credit: Tom's Guide)
Because of Qualcomm's prominence as a chipmaker, there's no doubt we'll see the Snapdragon 8 Gen 2 in some very significant phones. And with the release timeframe Qualcomm is promising, devices powered by the Snapdragon 8 Gen 2 are going to hit retail shelves very soon.
When they do arrive, we'll be looking forward to seeing how performance compares to Apple's A16 Bionic and A15 Bionic as always. And the improvements to the Qualcomm AI engine makes us curious to see how phone makers take advantage of that: will Snapdragon 8 Gen 2 devices give Google's Tensor-powered Pixels a run for their money when it comes to AI-fueled experiences?
We can't answer these questions until Snapdragon 8 Gen 2 phones arrive. But 2023 is already shaping up to be a very interesting year on the smartphone front.
 
  • Like
  • Fire
  • Thinking
Reactions: 25 users

Diogenese

Top 20


At 32min in video Prophesee event based vision sensor is mentioned & will be in smartphones in 2023.

Along with the following comments from the Prophesee/Brainchip partnership indicate highly likely Akida IP will be in smartphones with Qualcomm chips this year.

“We’ve successfully ported the data from Prophesee’s neuromorphic-based camera sensor to process inference on Akida with impressive performance,” said Anil Mankar, Co-Founder and CDO of BrainChip. “This combination of intelligent vision sensors with Akida’s ability to process data with unparalleled efficiency, precision and economy of energy at the point of acquisition truly advances state-of-the-art AI enablement and offers manufacturers a ready-to-implement solution.”

“By combining our Metavision solution with Akida-based IP, we are better able to deliver a complete high-performance and ultra-low power solution to OEMs looking to leverage edge-based visual technologies as part of their product offerings,” said Luca Verre, CEO and co-founder of Prophesee.


According to Statista 1.2 billion smartphones were sold in 2022.

And Qualcomm states:
In the same online briefing, chief financial officer Akash Palkhiwala predicted smartphone shipments would drop by roughly 5 percent year-on-year in 2022, dropping from prior estimates of 750 million units to between 650 million to 700 million devices.

Thus Qualcomm has 650-700 million / 1.2 billion = 54.16% - 58.33% smartphone market share.

Initially Akida IP will most likely be used in 10% of high end smartphones & later be available in all Qualcomm chip smartphones.

650-700 million chips per year x 10% Akida IP = 65-70 million.

If we allow $15-20 per chip x 3% Akida royalty = 45-60c BRN revenue per Qualcomm chip smartphone.

We'll use 50c BRN revenue per chip x 65-70 million units = $32.5-35M revenue.

When the tech becomes standard in all Qualcomm chip smartphones there will be 650-700 million chips per year x 50c BRN revenue = $325-350M revenue.

Qualcomm have also released Snapdragon Ride Flex SoC for vehicles.


Hi Steve,

There is a separate Qualcomm thread.

Qualcomm have their in-house AI:

https://www.qualcomm.com/content/da...ocuments/Snapdragon-8-Gen-2-Product-Brief.pdf

https://www.qualcomm.com/content/da...sets/documents/prod_brief_qcom_sd888_5g_0.pdf
 
  • Like
  • Sad
Reactions: 20 users

Bravo

If ARM was an arm, BRN would be its biceps💪!
WBT now $1B MC & BRN now $1.13B MC.

BRN is a few years ahead of WBT who recently taped out their chip.

BRN also has many big companies as partners.

BRN heavily shorted whereas WBT is not.

Is WBT expensive or is BRN inexpensive?


WBT Shares on Issue: 173,640,855
BRN Shares on Issue: 1,767,058,145
 
  • Like
  • Sad
Reactions: 14 users

TechGirl

Founding Member
President Biden is talking right now at the State of the Union & he specifically mentions the semiconductor industry & how they created the chips act & how USA used to supply 40% of the worlds chips & now they only supply 10% of the worlds chips. He talked about how during covid there was a shortage of overseas chips & how it shut down car manufacturing at home & how people lost their jobs, he said new cars have 3000 chips in each car & he also mentioned how intel IFS is building its own foundries to address these problems. (y)
 
  • Like
  • Fire
  • Love
Reactions: 74 users

Dozzaman1977

Regular
Last edited:
  • Like
  • Thinking
Reactions: 16 users
Hmmm....didn't see posted (probs has been?)

From a few days ago....Renesas using Edge Impulse...Raspberry Pi board....not saying we there but like that we are compatible and used with those as well :unsure:


Avnet, Edge Impulse, Processors, Renesas, Software, Tools / February 2, 2023
qI_1w6-aTx0-1024x576.jpg


Dirk Seidel, Strategic Business Development Manager at Renesas Electronics, demonstrates the company’s latest edge AI and vision technologies and products at the December 2022 Edge AI and Vision Innovation Forum. Specifically, Seidel demonstrates the company’s MPU RZ/V series solutions, along with partner (and fellow Alliance Member) Avnet’s RZ/V2L single-board computer (SBC).

The RZ/V series includes a power-efficient AI accelerator, the DRP-AI. The entry-class RZ/V2L device delivers low power and high performance when running fellow Alliance Member company Edge Impulse’s FOMO object detection model on the Renesas EVK, as well as when running a multiple-person pose estimation model on the Avnet RZ/V2L (a SBC in the Raspberry Pi form factor), all without need of a fan or heat sink. Seidel also demonstrates the newly-released TVM development tool for the DRP-AI, running the DeepPose facial landmark model.

 
  • Like
Reactions: 24 users

Tothemoon24

Top 20
  • Haha
  • Like
  • Fire
Reactions: 12 users

JK200SX

Regular
A couple of comments.

Tapeout of the AKIDA 1500 has occurred.

In electronics and photonics design, tape-out or tapeout is the final result of the design process for integrated circuits or printed circuit boards before they are sent for manufacturing. The tapeout is specifically the point at which the graphic for the photomask of the circuit is sent to the fabrication facility. (Taken from Wikipedia)

I've stated in a previous message that I believe that our customers are waiting to see the results of the AKIDA 1500 before they commit to signing contracts/deals (ie they know that a better performing chip is coming out and they want that extra functionality).

Now I'm going to ask a question that I don't want to ask. It is a hard question, that will have an equally hard answer. I say this because I've been a holder of BRN shares for the last 6 or so years, have not sold a single share over that time, and truly believe they have a tech that is unique and can change the face of compute forever (and I want them to succeed for all the reasons we talk about on these forums).

But, here is the question:
What happens if at the conclusion of the testing phase, the result is that the AKIDA1500 does not performed to the expected specification?
 
  • Like
  • Thinking
Reactions: 9 users

VictorG

Member
WBT now $1B MC & BRN now $1.13B MC.

BRN is a few years ahead of WBT who recently taped out their chip.

BRN also has many big companies as partners.

BRN heavily shorted whereas WBT is not.

Is WBT expensive or is BRN inexpensive?
WBT is trading at its all time high and has a mc of $1b
BRN at its all time high had a mc of $4b

If WBT is fair value then I think BRN is grossly under priced.
 
  • Like
  • Fire
  • Wow
Reactions: 18 users

VictorG

Member
A couple of comments.

Tapeout of the AKIDA 1500 has occurred.

In electronics and photonics design, tape-out or tapeout is the final result of the design process for integrated circuits or printed circuit boards before they are sent for manufacturing. The tapeout is specifically the point at which the graphic for the photomask of the circuit is sent to the fabrication facility. (Taken from Wikipedia)

I've stated in a previous message that I believe that our customers are waiting to see the results of the AKIDA 1500 before they commit to signing contracts/deals (ie they know that a better performing chip is coming out and they want that extra functionality).

Now I'm going to ask a question that I don't want to ask. It is a hard question, that will have an equally hard answer. I say this because I've been a holder of BRN shares for the last 6 or so years, have not sold a single share over that time, and truly believe they have a tech that is unique and can change the face of compute forever (and I want them to succeed for all the reasons we talk about on these forums).

But, here is the question:
What happens if at the conclusion of the testing phase, the result is that the AKIDA1500 does not performed to the expected specification?
Wouldn't BRN have run AKIDA 1500 through tests before making it public?
 
  • Like
  • Love
  • Thinking
Reactions: 18 users

alwaysgreen

Top 20
A couple of comments.

Tapeout of the AKIDA 1500 has occurred.

In electronics and photonics design, tape-out or tapeout is the final result of the design process for integrated circuits or printed circuit boards before they are sent for manufacturing. The tapeout is specifically the point at which the graphic for the photomask of the circuit is sent to the fabrication facility. (Taken from Wikipedia)

I've stated in a previous message that I believe that our customers are waiting to see the results of the AKIDA 1500 before they commit to signing contracts/deals (ie they know that a better performing chip is coming out and they want that extra functionality).

Now I'm going to ask a question that I don't want to ask. It is a hard question, that will have an equally hard answer. I say this because I've been a holder of BRN shares for the last 6 or so years, have not sold a single share over that time, and truly believe they have a tech that is unique and can change the face of compute forever (and I want them to succeed for all the reasons we talk about on these forums).

But, here is the question:
What happens if at the conclusion of the testing phase, the result is that the AKIDA1500 does not performed to the expected specification?
Tim Allen Kill GIF by PeacockTV
 
  • Haha
  • Thinking
Reactions: 10 users

JK200SX

Regular
WHAT IS WRONG WITH YOU?
Do you think the SP will rocket with that result?
You know the answer. I don't know how old you are or how well educated but that shouldn't be a question from a "6 years holder". Very disappointed at you.
Umm, testing needs to be done after the chip is made.
 
  • Like
  • Wow
  • Haha
Reactions: 6 users
Top Bottom