SoylentNews

Quibi Versus the World: Can Jeffrey Katzenberg and Meg Whitman change everything about video on your phone?

Jeffrey Katzenberg insists that his new video-streaming service Quibi isn’t competing against Netflix, Disney+, HBO Max, Peacock, or any of the other streaming services that have launched or are launching soon. You’ve got it all wrong. You’re not even asking the right questions.

“We don’t think we’re in the streaming wars,” Katzenberg, the former boss of Walt Disney Studios and founder of DreamWorks, tells The Verge in a closed-door meeting the day before the company’s grand reveal at a CES keynote. “They’re all battling for this,” he says as he thrusts his arm toward a TV in the room. “We’re going for this,” he says, gesturing toward his phone. “Don’t tell them!”

Katzenberg and Quibi CEO Meg Whitman, who is best known as the CEO of HP and eBay, are publicly announcing Quibi at CES — but not quite unveiling it — after having raised $1 billion on the promise of a roster of Hollywood stars and supposedly revolutionary video-streaming technology that delivers portrait and landscape video at the same time. Everything on Quibi is designed for viewing on a phone, on the go, in 10 minutes or less. These chunks of video are called “quick bites” — hence, “Quibi.”

A streaming service for millennial goldfish. Should be about as successful as Meg Whitman's campaign for Governor of California.

The past is prologue! What has been, will be again. Amor Fati!

Once again, aristarchus submissions are being suppressed! By this I mean, not rejected, but hidden, concealed, obscurated. These three submissions are listed as "accepted", but do not appear in the queue, or anywhere else.
US Army veteran among group of suspected neo-Nazis arrested by FBI on gun charges
White nationalist who spoke at Charlottesville arrested on fugitive warrant charges
3 Alleged Members Of Hate Group 'The Base' Arrested In Georgia

Yes, typical aristarchus submissions. Fake accepted just to mess with me? Or deep-sixed to keep coverage of alt-right terrorism off of SoylentNews? Possibly related to the recent threading bug TMB has discovered? Or another "Athanasius" scenario? QAnon, or Lizard People?

*****
Update: more rejected submissions, as the Hate goes down in Virginia.
Proud Boy "Tiny" Toese Pleads Guilty to 2018 Assault
And,
What is The Base? FBI arrest of alleged white supremacists puts focus on extremist group

And,
‘Emasculated’ Virginia gun nuts mocked for ‘purchasing toys, dressing up in costumes, and fantasizin (one of my personal favorites!)

And,
ALEX JONES CRUISES THROUGH STREETS IN INFOWARS 'BATTLE TANK' AS PROUD BOYS JOIN HIM AT VIRGINIA, the crazy does not get any more crazy than this, unless it is Texans in Congress.
****
New excuse for censorship on SoylentNews! Makes me feel so much better. Here it is:

Thank you for your submission. Unfortunately it does not meet the criteria spelled out in the submission guidelines, which can be found at https://soylentnews.org/faq.pl?op=editorial and has therefore been removed from the submissions queue. Please consider revising it based on the submission guidelines and resubmitting. Alternately you could consider posting it as a journal entry.

Yup, gonna revise this an resubmit! That'll work! Nothing like getting around censorship by doing exactly what the censors want! Hooboy! I feel a spate of submissions about alt-right Neo-nazi really stupid racists and Trump supporters coming on! Pray that the fetuses of Evangelical miscarry! Turn about is fair play. Oh, and TMB is behind on his payments.

CES 2020: Innogrit SSD Controllers Score Multiple Design Wins

Phison At CES 2020: Preparing For QLC To Go Mainstream

NAND flash memory prices are projected to climb in 2020. The manufacturing transitions to 96-layer 3D NAND and beyond are not going to increase bit output as quickly as demand will be growing. This will be a major change from the NAND oversupply that caused price crashes in 2018 and into 2019.

SSD controller vendor Phison is betting that increasing prices will finally push the consumer SSD market to embrace 4 bit per cell QLC NAND flash memory, which thus far has seen only limited success in the retail SSD market and virtually no adoption from PC OEMs. The price premium for SSDs with 3 bit per cell TLC NAND has been small or non-existent across all market segments, so the performance and endurance advantages of sticking with TLC NAND have been worthwhile. Those days may be coming to an end. Phison expects—quite reasonably—that when NAND flash memory supplies are constrained the bulk of the TLC NAND manufactured will be snatched up by the higher-paying enterprise SSD customers, more or less forcing the consumer SSD market to start shifting toward using QLC as the mainstream option.

QLC data retention, write endurance, and speed drop-offs: manageable?

CES 2020: Samsung 980 PRO PCIe 4.0 SSD Makes An Appearance

OWC Releases Accelsior 4M2 SSD: Quad-M.2 For Over 6000 MB/s

CES 2020: ADATA Preparing Three PCIe 4.0 Consumer SSDs

The next-gen consoles due for release in late 2020 have been rumored to use faster-than-typical SSDs.

I have become convinced.

There is no bottom.

There is no low that is too low. No ethical boundary that cannot be crossed. No crime that cannot be ignored. No profanity or taking the Lord's name in vain that cannot be hand waved away.

Trump really could parade naked in the streets, with hookers and drugs, shoot some or several random persons in the middle of the street in broad daylight . . .

and his band of FoxNews fed mouseketeers would just wave it away. Fake News! It didn't happen. We won't hear any witnesses. No evidence. Witch Hunt! Anyone who says it happened is a liar! I cannot recall! I would have to check my records. The trial judge is working closely with Trump's lawyers to fix this, so no worry.

The evangelicals cover their eyes or look the other direction. Justify and rationalize it by inches at a time.

Good is called evil and evil is called good.

We are all numbed to the bizarre and irrational. Reporters are afraid to report on the worst of Trump's behavior because it is so bad that they fear people will think they are biased against Trump merely for telling the simple facts. Because it sounds so bad that in normal times you would think the press is biased. Comedians make jokes about it, but it's not even funny any more. Just boring every day events.

Hear no evil, See no evil, Speak no evil.

People who don't like Java shouldn't use it. Use something more to your liking.

Java is not perfect (shocker!)

Java has warts from being designed in the early 90's and released later in the 90's. Computers and their limitations were different back then.

The Java runtime is a big disk footprint. At least if you use other languages, and don't realize what all you're getting in this fat package.

Java likes lots of memory.

Java programs start up slowly.

Java is not ideal for all programming tasks. (shocker! I can't write my bootloader or micro controller code in Java!)

Java (rather the JVM) does not have tail call optimization.

Java has Garbage Collection (GC). (Some people see this as a problem. If GC is a problem, you probably don't want Java, nor the JVM runtime system.)

Terminology clarification

From here on I'll try to use Java to mean a programming language that you write source code in, and compile it to JVM bytecode. I'll use JVM bytecode to mean the object code from compiled source languages, including the Java language. I'll use JVM to mean the Java Virtual Machine, that is the runtime system which executes JVM bytecode -- no matter what source language it was compiled from.

Virtues of Java / JVM -- Garbage Collection

Java and the JVM have Garbage Collection (GC) ! Free clue: all modern languages now have GC. Visual Basic. Visual FoxPro. JavaScript. Python. Arguably: Perl. C#. Erlang. Go. Lisp like languages. Prolog, Haskell and other higher order languages. And many others.

If GC is so horrible, why do so many languages have it? Some will say because programmers do not know how to manage memory. But that is not true. Speaking for myself, I wrote untold amounts of Pascal in the 80's, with complex data structures and understand quite well how to manage memory, avoid, detect and deal with memory leaks.

I would point out: Greenspun's tenth rule

Any sufficiently complicated C or Fortran program contains an ad-hoc, informally-specified, bug-ridden, slow implementation of half of Common Lisp.

(and that includes garbage collection along with other higher order programming idioms)

JVM GC is not comparable to other systems like Python or Node.js. Modern JVMs can run with Terabytes of memory and hundreds of cpu cores and have only 1 ms GC pause times. (Please call me when your Python or Node.js or Go can do that!)

Java offers multiple GCs to choose from. Each one has various tunable knobs and dials. Instrumentation (like VisualVM) can give you insight into the GC behavior of your large program that has many third party libraries so you can tune accordintly. (again, call me when your system can do that)

Red Hat's Shenandoah, and Oracle's ZGC are the current new state of the art GCs on Java. Both of these are open source in the latest Open JDK builds.

Multiple languages

Multiple languages compile to JVM bytecode. Not just Java. Kotlin. Scala. Clojure. And other languages run on the JVM (Groovy, Jython, even C).

Cross Platform

Java object code (JVM bytecode) is cross platform. I can take a desktop GUI program written in Java using Swing in 2004 on Windows, and run that object program on a Raspberry Pi (different OS) which did not even exist when the program was written. It runs perfectly.

Java programs, especially servers, which have no GUI are extremely cross platform.

The only pain points I can think of is (1) serial port communication, (2) doing extremely platform specific things (uh, say, accessing /proc or /sys). There are some nice cross platform solutions for serial communication; I had to use one a few years ago. I had to deal with whether to use names like /dev/ttyS1 or COM3:.

Compiled JVM bytecode runs on amazingly diverse platforms. From Blu Ray players to giant IBM Mainframes with seemingly strange architectures. Smart cards, car infotainment, single board Linux (eg, "Pi" type boards), and who knows what else.

Speed, yet dynamic access

The JVM runtime interprets JVM bytecode. It dynamically profiles every function to see how much CPU time it is getting. (How "hot" it is) Hot functions are immediately compiled (C1 compiler). C1 rapidly compiles the JVM bytecode into unoptimized native code. The function is added to a list to be compiled later by the C2 compiler.

Later, when the C2 compiler comes along, it spends significant time and effort recompiling that function into highly optimized native code.

C2 is one of, if not the most sophisticated compiler on the planet. The product of a couple decades of much research. It has only one source language to compile: JVM bytecode. It has multiple target instruction sets to compile to.

C2 compiles to the instruction set of the actual hardware it is running on. Something that an ahead of time C compiler, for example, cannot know in advance. C2 knows which instruction set extensions your actual processor has. Does your processor have SSE, MMX, ISSE instructions?

C2 also has global knowledge of the entire linked runtime program. Also something that a C compiler does not have advance knowledge about. C2 could know that a certain function could be called efficiently in two different ways from two other parts of the program, and compile two separate versions accordingly.

All method references in Java (JVM) are "virtual" (to use C++ terminology). But in practice many or most methods do not actually need to be virtual. C2 can (and does) prove that a method is never called in a way that it needs a vtable entry, and can compile efficiently accordingly. The programmer never needs to make decisions about whether functions should be virtual or not.

Research has shown that in many cases even when a virtual polymorphic function is called, from a specific call site, it often always calls the same concrete function at that call site. The new GraalVM takes advantage of this and cache which virtual function to call for next time. If at this particular call site (where function is called from) this time a different virtual method will be invoked, then a runtime error occurrs, is handled, the right virtual function computed and it is now cached.

C2 aggressively inlines code for performance. It is after speed not small code size. Memory is cheap. You can never get back time.

Now suppose that YOUR function A calls MY function B. When C2 compiles your function A, it may inline my function B inside the native code of your function to avoid function call overhead. Now suppose that the class which has my function B is dynamically reloaded. Oh no! Your function A now has a stale version of my function B inside of it! Not to worry, the JVM de-optimizes your function A back to being bytecode interpreted. If your function is still "hot" it will very soon get recompiled by C1 and then later by C2.

This "hotspot" and C1/C2 behavior is one reason why Java programs seem to "warm up". They start up slowly and then within a few minutes become very fast. If you have a program which needs to be restarted very infrequently, runs for a very long time between restarts, then this is for you! (example: long running servers) If you have a program that is rapidly and frequently started from the command line and needs to do something very quickly and exit, then this is definintely NOT for you!

When running a 64bit JVM with less than 32 GB of memory, JVM can do a pointer optimization trick. It can use 32-bit pointers instead of 64-bit pointers. All objects start on 8 byte boundaries (on x86/x64). So the low three bits of a 32-bit pointer are always zero. So why have them. Thus a 32-bit poiner can reference objects in 32 GB of memory. Next time you start the program with more than 32 GB of memory, it will have to use normal 64-bit pointers everywhere.

GC is the lubricant between different code libraries

In C or C++ there may be different memory management disciplines. Different calling conventions. Different ideas of responsibility about who "owns" something and is responsible to dispose of it. Even different allocators. Are there instances where glue code is needed to adapt the conventions of one code base with another that are both used in the same program?

An overlooked and maybe little known benefit of GC is that these problems go away. Everything uses one memory management discipline. Same calling conventions. A library written two decades ago can be passed data structures from another library that was only recently written.

GraalVM

This is so new I am not very familiar with it.

The JVM runtime is written in C++. Such a complex runtime, with GC, hotspot, C2/C2, native code interfaces, dynamic code reloading, etc is getting hard to maintain. Especially with multiple GCs to choose from. One of the things GraalVM does is let much more of the JVM be written in Java. Graal VM can also run code from LLVM, and other languages like R, Python, etc. So your C, Python and Java code could call functions within each other in the same runtime, and pass parameters transparently.

Concluding remarks

I know it is fashionable to hate Java. Yet Java is consistently year after year the number one language on various job sites and programming language surveys. Java is used by many major corporations. Red Hat, Oracle, SAP, IBM, Azul Systems and others see enough major players needing commercial support that they all seem to have a very tidy business with Java. Amazon and Microsoft Azure both bend over backwards to provide optimized Java runtime systems for these major customers with bucketloads of money. Microsoft recently partnered with Azul to provide their optimized runtine for free to Azure Java users. They did't do that for no reason.

Java must be doing something right. It is the best at what it does. There is nothing else that comes close to doing what I have described above.

Even if it is not right for you.

It's here to stay for a long time. Even if you hate it. But why hate it? Just don't use it.

If there were one perfect programming system for every use, we would all be using it already.

Black Shark 3 could be the first phone with 16GB of RAM

The most RAM you can get in a smartphone right now is 12GB, but that could soon change, as the Black Shark 3 gaming phone from Xiaomi is rumored to boost that spec to 16GB.

That’s according to Sudhanshu (a leaker on Twitter with a reasonable track record). Of course, whether the Black Shark 3 will be the first phone to arrive with that much RAM depends on when it launches, but with the phone having been rumored for a while – and given that the Black Shark 2 launched in March 2019 – it’s likely that it will arrive soon, possibly at MWC 2020.

Whether or not the Black Shark 3 is first to 16GB, it should make good use of it, as gaming phones need all the power they can get – and with that much RAM, coupled with a top-end chipset (the Snapdragon 865 is likely), it could take us another step closer to console-type power on a phone.

All Galaxy S20 Models Rumored to Feature 12GB LPDDR5 RAM as Part of Their Base Configuration

The tip comes from Ice Universe who claims that the base Galaxy S20 memory configuration will start from 12GB RAM, and it’s not just any other RAM; it’s likely the 12GB LPDDR5 DRAM Samsung announced in July last year. Thanks to a higher data rate, the new mobile memory is 1.5 times faster than LPDDR4x which was featured in previous phones. A new circuit design also makes the new chip more power-efficient as it needs 30 percent less power than its predecessor. With these features, 12GB LPDDR5 RAM will be able to make the best use of the speeds offered by 5G.

Samsung had already announced that the production of 16GB LPDDR5 RAM will begin in 2020, and thus it’s entirely possible that Galaxy S20 memory configuration will go as high as the aforementioned capacity.

Leak Suggests Top Galaxy S20 Ultra Model Will Have 16GB of RAM, Only Variant to Have microSD Card Slot

This is DOA. I need 1,280 GB of RAM in my next smartphone.

Previously: Samsung Begins Mass Producing 12 GB DRAM Packages for Smartphones
Samsung Mass Producing LPDDR5 DRAM (12 Gb x 8 for 12 GB Packages)
Get Ready for Smartphones with 16 GB of RAM

In recent days I have been seeing many posts with pictures of "homemade labs." This fucking pisses me off. Your lab could be endangering the lives of millions, if not billions of others - and I'm not even counting your own life, and you won't even learn until it's too late. I propose that amateur chemistry be universally banned. Only professionals who have at least a bachelor's degree should be allowed to set foot in a lab.

You don't even know all the weird shit that happens when people even get near chemicals. One of the guys I work with once touched isopropanol to his skin, and then the next day he just disappeared off the face of the earth. He was a well-trained chemist and respected at his institution, and then he was just...gone. This can happen to you, kids. Think twice before touching a chemical.

Exoneration for everybody! All transactions and phone calls were perfectly innocent!

SNAFU!

It’s Bernie’s moment. But it’s Bloomberg’s race. (archive)

If it can be summed up, then, the Democratic “mood” is basically this: “We like Bernie. He’s a warrior. But we’re afraid if we nominate him, he’ll lose in the fall. We need someone to get the job done.”

If the two men who might be that someone — former vice president Joe Biden and former South Bend, Ind., mayor Pete Buttigieg — lose to Sanders in Iowa and New Hampshire, that would make their “electability” somewhat less convincing. Defeat can be contagious. There are not many voters who say: “I like him — he loses a lot.”

So it’s Bernie’s moment, which has sent a wave of panic through the Democratic ecosystem. It’s like waking up from a nightmare, only to realize that you’re waking up in a nightmare.

Which helps explain why Democrats across the country will soon find themselves with a newfound appreciation for the virtues of one Mike Bloomberg, former Republican mayor of New York and billionaire founder of a financial data services empire. He might not have been exactly what they had in mind, but by Super Tuesday he’ll look like Brad Pitt.

Two b(m)illionaires competing for the Presidency is the kaiju battle we need right now. So let's repeat the mistake success of 2016 and finally send Bernie to a retirement home.