Wednesday, July 31, 2013

Google Needs to buy Nvidia (If they let them).

It's no secret, I'm a Nvidia fan, but disregarding that I say Google needs to buy Nvidia. Nvidia would give Google a strategic advantage in pretty much every part of their company. Here's a list of reasons why I think Google needs to buy Nvidia.

Google uses computers a lot



He's Using a Computer too
The obvious thing is that Google uses computers a lot. All of their employees have laptops, cell phones, some have desktops, and the company has servers that they use to handle request from billions of people each year. If Google is able to produce the computer GPUs for their employees, servers and mobile devices they would all together spend less money. 



Nvidia's servers easily outperform Google's


Incredibly powerful server tech
Just recently I read an article stating that Nvidia in conjunction with Stanford developed a neural network that surpassed Google's with merely 16 servers. Sure you may see that as a whatever thing, but keep in mind that Google used 1000 servers for their neural network. That's a big difference. Nvidia's methods are cheaper than Google's and more power efficient.  Nvidia's GPUs has up to 3000 cores within them. Basically each core can be used for parallel computing, therefore the server is able to add another 1000 processes into a server with ease. 

Google already is producing phones


The Moto X
Google is already producing phones with their consumer hardware company Motorola. They've recently been emphasizing on GPUs for their products, and Nvidia happens to be very good at it. Also, if Google creates phone processors custom tailored to their devices, they would have a greater deal of perfection within their devices.


Nvidia is filled with computer hardware geniuses


Looks fun to work there
Google is made of software geniuses, while Nvidia is filled with hardware geniuses. I can fairly say that Google has been pretty bad at creating low leveled software. Since Motorola became part of the company, Google has been able to speed up Android's kernel development by a lot. It shows off with the performance and battery life increases with Android 4.3. With Nvidia, they would be able to advance low level software further and faster compared to before. When you see things like DirectTouch coming from Nvidia, you really begin to see how far these Nvidia can go. Google needs to get a piece of the action.

Can quickly design chips for anything


If Google has an idea, they can easily get a chip designed between Motorola and Nvidia in no time, prototype a device with the chip they design, make quick modifications, and deploy products much faster than any of their competition (until you look at Samsung). 

Monday, July 29, 2013

Android 'could' be a Killer Laptop/Desktop OS

In the last two days I had a stroke of genius - okay, maybe not genius, but still good - and I have come to the conclusion that he Android operating system could actually turn out pretty well inside of a laptop. It could get to the point in which it competes with the likes of Mac OSX, Ubuntu and Windows, which hasn't happened yet with any other operating system. I'm not just saying it will because I love Android, which I do and I'm sometimes pretty bias, but I considered many different components from all sides of the "Computer Spectrum". Here are my list of reasons why the Android OS could be "the Killer Laptop/Desktop OS":

Software

It's Linux based


Android originally began as a major fork of the Linux kernel 2.6. For those of you on android that don't know what a kernel is, fear not. A kernel is the bridge between the physical hardware and applications. Windows uses the word "driver", which is why most people have not heard the term before. Below is a diagram of the structure of android. Never mind the green, blue and yellow sections, focus on the red. If you look, you'll see a number of drivers that are in the kernel to help the hardware connect to the overall operating system.


If you're like most people, you're probably asking "Where are you getting at?", so allow me to explain. Well, Linux is the fastest moving operating system in the world. According to the video here, Linux has on average 8000 developers developing the operating system collaboratively from all over the world, from multiple companies each year. This means the kernel is changed every few months instead of every few years like with Mac OSX and Windows. This also means there's less issues in the operating system, because usually people will fix a problem and send a patch to the creator and the people in his office for widespread change. With it's support, Linux has brought some features most mainstream OS's haven't implemented like 3D printing and compiler changes that would make the OS snappier and use less memory. 


While Android's kernel originally began as something drastically different due to the quest of keeping the OS small, Android is now going back to the mainline of the Linux kernel as it has proven to keep the performance up while making sure the resources used are low. The "K Release" , also rumored as Android Key Lime Pie, will likely introduce an almost fully merged Linux Kernel, which will create the opportunity for the OS to add a massive amount of features that you'd find from Ubuntu, Debian and Red Hat. This means you'd be able to add plenty of compilers (mostly like more a versatile GCC) on the android laptop/desktop OS. IDE's, and other special development tools  for programmers can be imported to the Android operating system. That would make the tech guys pretty happy. 


Android So Far...


So far, Android has been able to reach a deeper sleep than it ever has with the latest release, Jelly Bean 4.3. Before now, Android has had a problem reaching a deep sleep, which is why it ran hot people's pockets. It's also one of the main reasons manufacturers needed to place massive batteries into their phones compared to the iPhone where the battery is extremely small. The latest android introduced a few other modifications like security, and performance increases. 

Android 5.0

Android 5.0 is when Android reaches practical perfection. The operating system is supposed to be able to be used in many different configurations like in Google glass, watch, and even laptop like form. The entire theory that Android will soon come in laptop form came from analyst reports. This OS will have many internal modifications for speed and battery purposes just like 4.3 brought. 5.0 will be ever closer to a fully functional linux laptop OS in no time.

App And Developer Support


Android is known for it's wide selection of applications. As I'm writing this blog, there has been 806676 android apps added to the market. An Android Laptop will have at least 50,000 of those apps within the first year or two. With chrome for android, Android will have a unified experience between laptops, web browsers, and other mobile devices. 

Gaming


Nvidia's Shield
Facing the facts, Android has a robust game developer support. It could get to the point where it can compete with Windows. Seeing they are now working on solid multiplayer API, and Nvidia's Tegra graphics are getting to the same point as the PS3, I think Android laptops and Gaming systems can compete directly with Xbox and Sony's Console.



Floating Apps and Windows


One true advantage mainstream OSs like OSX and Windows have on Android is Window multitasking support. We've seen with Facebook Messenger that Android can have applications that sit on top of each other, and Samsung has proven that Android apps can be viewed side-by-side too. Android has the capacity for full window support as well. I think Google will implement full window support into Android 5.0 to allow the OS to compete with big league Desktop OSs.




Hardware


No matter what form of common hardware you use (ARM or x86), Linux and therefore Android will be able to run on it, the Motorola Razr i is a good example of how x86 architecture works with it. ARM is on 95% of android phones to date, therefore there's no need for further proof that it works with ARM.

Outside of the basic architecture, ARM has a wide array of designers and manufactures that could make the chip's architecture go viral. Here is a list of unique processors that could bring it's technology to Android.

Nvidia's Kepler GPU


Look at the tessellation. So cool 


Nvidia's Kepler GPU technology is truly a very efficient technology. Compared to most other GPU technologies, the ARM Kepler is said to outperform every mobile GPU technology by up to 5 times while using around 1/3 of the power. Now I've been looking at the special details for this, and I couldn't find anything extremely specific. I did, however, find a bit of info about Kepler graphics in general. 



Mobile graphics will make a major leap in the next year


How Kepler Works


Dynamic Parallelism vs Conventional
The main goal of Kepler chips is to reduce the work load of your CPU.  The common CPU/GPU model requires the GPU to send a response every time an action takes place. Sure it works, but it could be a lot more efficient by requiring the GPU to respond less while still getting the sane amount of work completed. To do this Nvidia brought a couple of parallel computing concepts into their Kepler GPU. The first is Dynamic Parallelism, the second Hyper-Q, the third GPUDirect


Perfect Graphic Control
Dynamic Parallelism allows one GPU core to send processes to other GPU cores without having to send a signal back to the CPU. Therefore the CPU is able to reduce the load taken while completing graphic intensive operations. The GPU is also able to do calculations where the based only on what moves much more efficiently.


Hyper-Q allows each CPU core to complete work on the entire GPU at the same time. That allows the GPU to work at full potential all the time and reduce the latency of the CPU. I particularly love this, as this means the processor can finally compute graphical information to the fullest more efficiently. 


GPUDirect enables the GPU to directly make transfers to other 3rd party devices. This means devices such as SSDs or NICs (Network Interface Controllers) can transfer information almost instantaneously, which is why Nvidia technologies can be used with cloud computing. This technology can also allow for other clustering/parallel technologies like insanely powerful mobile peer-2-peer technology or using SSDs to cache large sets graphic information instead of normal mobile disk. The download speeds can greatly be increased too, as downloads are directly connected with the NICs. You'll notice that nvidia's shield has powerful download speeds. Some is contributed to MIMO Wifi, some is also contributed to a slight integration with this technology. Hopefully Nvidia uses this to the full potential. 

Here's a picture of Ira
He's so Cooooool
Nvidia's technology can make the phone in the next 12 months up to 5x more powerful. While many people reading this blog are probably annoyed that I haven't talked much about other people like the creators of PowerVR GPUs, or Qualcomm's Adreno, but understand I've looked at them both and they didn't impress me enough to focus on. That's mostly because while they will provide powerful graphics, they won't be as efficient as Nvidia's Kepler technology. Qualcomm's Adreno technology won't be licensed out to other chip designers either, making them slightly less relevant in 2014 as the main players in the smartphone business now want to design their own chips. 

Motorola's X8 


The X8 Chip
Motorola's new X8 processor impressed some, didn't impress others, but it certainly impressed me. The new chip is meant to have the usual dedicated CPU and GPU within the chip, but they introduced a concept I personally didn't think would come about. They made a special core for each natural language and sensors!


I feel some of my readers probably don't care all that much about this feature, and call it bluff, but there's a reason it's the exact opposite. This setup is very similar to the way our brain works. Our brain has something called the Limbic System. While we have the frontal lobe which does allows us to use logic and reasoning skills, the parts in the Limbic System and other lobes control other important functions: memory, speech, emotional responses, cognition, body regulation, visual perception, color recognition, spacial orientation, touch and information processing. The mere fact that they now have specialized cores for specialized task means that they can grow ever closer to a human like robot. 


So, the CPU is like the frontal cortex and the GPU is like the Occipital lobe, which is responsible for visual perception and some processing . The Natural Language Chip would be a lot like the speech processing part of our brain, the auditory cortex, and the new sensor chip is much like the Thylamus, which is responsible for registering and transmitting senses within our own brain. 


You can probably see where the Motorola X8 is going now. It's heading to the point of having a chip/core for each dedicated task. I had no idea Linux had this type of capability within it's kernel, but now I see it does, I can't wait to see what this will bring about. 

Adapteva's Parallella Cores


It's coming!!!
Parallella uses the same concept of Nvidia's GPU parallel processing, only they increase the number of CPU cores instead of GPU. The CPU cores run in parallel in order to increase the speed of certain compilations. As of now, Linux barely supports parallelism within small processors. Usually parallel processing is only found within GPUs and within multi-million dollar super-computing systems. The video below shows the computing difference when parallel CPU processing is utilized. 



Skip to 1:05


Android will be a killer Laptop system because it will be able to run smoothly and implement the best parts of hardware and software well before other mainstream operating systems do, much like how normal Linux does. With all of the above Android would seriously be one Killer Laptop OS

Thursday, July 18, 2013

Google Might Not Reach Q2 Expectations

This is a 5 minute microblog article I'm writing prior to the Q2 earnings. It's more of a checklist, but I don't think that matters. Google might not reach Q2 expectations. Here's why:
Google's results will certainly be interesting. Especially when you consider all of the variables to consider. Check the Q2 earnings here.

Sunday, July 14, 2013

What are the next biggest things?

So this week has been glorious. I've had to do massive amounts of work therefore I didn't have a lot of time to make my post. The Title of the blog post implies that this is to be the be a Nexus/Android/Google blog, but you'd notice that my last post was more about the web itself and it only included the topic of Android and Google inside of it. This post will specifically speak about technology as a whole. Sure it will include a few Android devices, but the focus will be on the consumer devices as a whole. These are a few trends I've noticed in the last few weeks in the technology world. The trends suggest what the next big thing is.

The cell phone world will use technology from other parts of the tech industry


I have no idea why I didn't think of this before, but since I read the news article, In One Big Nudge, Google To Spend $500 Million To Market Moto X Phone - ReportI became consumed with the thought that Google was nudging the practice of bringing already established technologies from other parts of the tech world such as car like monitors, Kodak "clearpixel" cameras for low light photography and dual channel LTE for up to an average of +50 MB/s into the mobile tech industry. This could mean devices wont be entirely concerned with beefy processors, screen sizes(and pixel density), but more concerned with introducing new technologies to change the experience of phones. Nokia and Samsung have done similar things until now, each bringing something like NFC with Samsung (and all android), and new camera technologies(like 41 pixels for extreme digital zoom and image stabilizers) in Nokia devices. The Moto X might be the main drive to take the integration of foreign technology from other industries into the mobile industry.

Getting to the last 5 Billion while making the world better

This presentation gave me a good idea of how the time around 2020 will be for the 3rd world country with mobile technology. This failed to even mention any advancements to cooling, agriculture, metal extraction, medicine, 3-D Printing technology, or other revolutionary technologies that I haven't thought of. But generally speaking, the message is all pretty clear. Companies are investing into the final 5 billion people. Investing in emerging markets cost relatively little but usually have high gains after a while. Just think of how the banker J.P Morgan originally invested in the US's Electrical, Steel, Oil and Railroad systems and you'll see exactly what I'm talking about. Nokia's introduction of the Asha 501 devices, and how Google is expanding 3g like wireless service to almost the entire world via Project Loon are key indicators that they are indeed taking emerging markets seriously.

3-D Printing Is Not a Fab, it's One of the Main Biggest Things


I want to make everything clear. 3D printing is far from a Fab, it's gonna go mainstream. You can really tell because Organizations like GE and NASA are investing into it, and NASA especially doesn't waste money. I have a story of my own to prove how inefficient our current process is. About a year ago I entered the racing car development club in my college for the hell of it. No, I'm not some dude who loves cars, I don't fantasize about have a Mercadez when I grow-up, I want a car that will run on something like 10000000 miles a gallon -I have my dreams. As I said, I joined, but only because I was bored, and I quit after the first week. It was all cool designing the car, but the process for making all of it's parts was unbearable, and the people there were extremely rude and antisocial, which I thought was annoying. People's personalities aside, let me tell you how the parts were made in a step-by-step format:
  1. Find out what piece we have to make.
  2. Look at all the measurements
  3. Figure out what machines to use
  4. Figure out the specifications for creating the piece
  5. Shave the piece to make it look as smooth as required for the car(By hand)
  6. Drink a quarter gallon of water to prepare for one long ride
  7. Get a spot at the machine you need
  8. Get the pieces necessary for the machine
  9. Begin taking an extreme number of measurements so you can make a precise first cut/drill/saw
    1. If You mess up, start from step one again and skip step 6 if an hour hasn't already gone by
  10. Use measurements from the sheet and check for accuracy continuously and keep going to until you can go no more.
    1. Done? Probably took 3 hours of your life for that one piece, if you have special machinery for this the process took around 1.5 hours.
      1. If you messed up, start from square one again. Make sure you enjoy step 6 or add an extra step for flirting, because if you don't you're gonna have one hellish night.
The process for mass production is not too different either. It goes a lot like this:
  1. Engineers look at the design
  2. They determine the dimensions of the parts
  3. Create a machine for each hole, curve from a specific block (Takes forever to design)
  4. Create special machines for special processes, like for moles and special sprays (Also taking forever)
  5. Generate massive amounts of waist from excessive drilling and shiming of metals and heat to cool things. (Not very effective)
I feel like I'm missing little details too, which is a shame. Basically I could call the club "Build one part for 2 or more hours" club, because that's what we were doing. Designing the car took only two weeks, but building it like this took about 29 weeks. Even though the Mechie kids building the car had massive amounts of pride in the car, I thought that with Future Technology, 3-D Printing, the club could have had 6-7 cars up and moving in the time it took for 1. My point is, 3-D printing technology will make this type of situation almost disappear. Just imagine having one machine do all the work for you  in any material by creating the part in a CAD program then just sending it in. Car and plane manufacturing could be speed up by 10 to 25 times the norm, the manufacturing price would be reduced for most items including some micro-electronic parts, and people could make anything mechanical in their garage in their backyards.

They say in the NASA article that 3-D printing cut it's production cost by 70% and shaved off months from the time used. Just imagine what could happen in cars, boats, planes, electronic (and microelectronic) parts, computing device shells(laptops, cell phones etc) and more. This technology can go widespread fast, and will mean everything will be cheaper for everybody in the world. Because everything will be so much cheaper for the general population, printing technologies go hand and hand with targeting the Final 5 Billion, which is a section from earlier in the article.

Wearable technology


Wearable technology goes completely with my previous article, which I hope explains well on why wearable technologies are so important, even if Google Glass is not the most successful thing out there. They will connect with the beginning of the Web 4.0, which should let nerds rejoice.


Quantum Computing, That's right it's coming



Quantum Computing is insane. I'm not just gonna let you take my word for it, take this evidence. Quantum computing in the realm of research will allow humans to discover things that we never would have in no time flat. Generally speaking Quantum computing devices will replace Supercomputers by allowing you to do insanely complicated analysis that would take current supercomputers thousands of years to complete in seconds. Imagine the results: People could become Superheroes by figuring out
which genes will give people powers, we could design warp drives in a short period of time to get us from one Galaxy/Nebula to another in mere seconds, learn many of the properties of the universe faster than before, control gravity to prevent(or create) black holes, find medicines that could cure cancer, create AI that would be more human and intelligent than us, the list is infinite.

What These Technologies Mean


These technologies are just the tip of the iceberg, but they do mean a whole lot. Here's what these technologies mean.

To The Consumer


Prepare to have insane products from all over. Cars may only cost $100 if they're not flying by then, plane rides will only cost $10 with the 600 person plane only costing up to $1m for production, smartphones could be $3 dollars, and food prices can be reduced significantly with just 3-D printing alone. Quantum technologies could increase out scientific discovery by up to 1,000,000 times the normal and bring about new AI systems. Meaning we could live into the thousands (would be boring after a while) and increase our overall intelligence beyond what we've ever had. The standard of living will be like our own or better in emerging nations. Companies that invest in the last 5 billion will become ever wealthier, and that wouldn't hurt you one bit. You could also see 3-4 processor cellphones(like the 3 processor Moto X: Sensor processor, main CPU, GPU) sooner than you think.

To the Investor


The average investor will be very busy for the next 5-10 years. They'll need to find the next big companies that will represent the next big things. Forget app developer fabs, people representing these technologies are the ways to go. I plan on joining the investment gang in about 3-4 years after I get out of college, these are the first things I'm going for. If you are an investor, you can make tons of money.