Qualcomm Snapdragon 820

By Ashwin Ananthapadmanabhan

Qualcomm Snapdragon 820: More specs revealed along with early benchmark score


QUALCOMM HAS been detailing more about its upcoming Snapdragon 820 SoC, which appears to be about to beat the pants off everything in its wake.
In spite of the Snapdragon 810 having a few teething problems, what with running hotter than a Macdonald's apple pie filling, time waits for no man and its successor is on the table. And it's bonkers.
Qualcomm Snapdragon 820: More specs revealed along with early benchmark scoreIt works on a principle of heterogeneous computing - in other words the CPU, GPU and DSP can share cores. This means that, for example, when streaming a movie, which involves rendering and very little else, all the runtime can go towards the quality of the picture.
This is controlled by Symphony, the chip's new management tool which controls the prioritisation and configuration of the various elements of the CPU.
The Snapdragon 820 will include Kryo, the company's first custom-designed 64-bit quad-core CPU, an Adreno 530 GPU and a Hexagon 680 DSP.
The upshot is that, because of its basis in Samsung's 14nm FinFET fab process, the Snapdragon 820 promises twice the power efficiency and performance of the 810, despite 'only' running at 2.2GHz - not a huge leap over its predecessors 1.8GHz.
The results have been striking. The fastest thing in the office is an Nvidia Shield tablet with the fabled Tegra K1 CPU, running at 2.2GHz. That scored an Antutu benchmark of 40,750. Even the Meizu MX4, the highest benchmarking device in the database, scores under 50,000. But Techgrapple has shown a leaked Antutu score for a device running the 820 of 83,774. That's over double the Nvidia Shield.
There are several pinches of salt to take with this. We know the device is running an HD screen, not a 4K screen, which will make it easier to render. We also know that the device was running Kitkat, not Lollipop.
But whatever way you slice it, that score is just bonkers. And, while we're not expecting to see 80,000+ on retail devices running the Snapdragon 820, you can expect it to be something very, very special if this leak is accurate.

Popular posts from this blog

BitTorrent’s Cool New feature “Altruistic Mode” Makes Torrent Downloads “Never Complete”

By Ashwin Ananthapadmanabhan
Image
Short Bytes:  A new BitTorrent feature has been announced which is designed with a goal of maintaining a sharing ratio of 2:1 in a swarm. The confusing Altruistic Mode feature will make sure that your share more and download less. B itTorrent Inc. has added a new feature that’s strange, confusing, and interesting at the same time. Known as Altruistic Mode, the features promotes the philosophy of giving more taking less on the torrent network. The Altruistic Mode is designed to make sure that your sharing ratio remains more than 2:1 i.e. while downloading a torrent, you upload 2Gb data for every 1GB download. Many of the loyal torrent users seed to benefit the network and the download speed of other users. But BitTorrent says that this does not guarantee a 2:1 ratio. That’s where the Altruistic Mode comes handy and helps a swarm. “Because of this Altruistic Mode downloads will often never complete,” writes Bram Cohen, the creator of the BitTorrent protocol. In a blog p...
Read more

Project Jacquard

By Ashwin Ananthapadmanabhan
Image
Google’s Project Jacquard Turns Clothes Into Touch Screens and Track pad We are currently breathing in a world that is smarter. We have smart phones, smart tabs, smart televisions, smart watches and what not but there’s one thing that Google wishes to add to this list of smart things of our everyday life. Want to know about it – its smart fabric, a fabric inbuilt with sensors, equipped with the ability to sense touch, swipe, send data through the conductive yarns and lot more. This is an effort by Google to generate a technology that would revolutionize the field of textiles and fabrics. The fabric can also facilitate calling, messaging, chatting and lot more alike. The project named as Project Jacquard debuted on Google I/O 2015 as a unit of Google ATAP. The fabric is weaved with a special yarn that is conductive created by Google ATAP team in partnership with American jeans manufacturing company Levis. However, the fabrics are not fully weaved out of the ...
Read more

Run ChatGPT-Like LLM Models offline on your PC

By Ashwin Ananthapadmanabhan
       Ever since ChatGPT hyped the use of AI, everyone wanted to try similar AI models on their devices for daily queries. These are powerful AI models capable of executing high-level tasks. Now with the help of customized tools, users can install various Large Language Models (LLMs) on their computers for offline use. If you want to make your computer an AI chat tool that you can access anytime you want, then the answer is the tool that allows users to install various AI models such as Llama 3, Phi 3, Falcon, Gemma, Replit, and more. Let’s now see the steps to get these tools and setup AI Models. M inimum Requirements: Apple Silicon Mac (M1/M2/M3) with macOS 13.6 or newer Windows / Linux PC with a processor that supports AVX2 (typically newer PCs) 16GB+ of RAM is recommended. For PCs, 6GB+ of VRAM is recommended NVIDIA/AMD GPUs supported Note: You can follow this guide for Windows, Mac, and Linux platforms. AI Models use a high amount of RAM, CPU, and GPU. I...
Read more