Showing posts with label Intel. Show all posts
Showing posts with label Intel. Show all posts

Thursday, January 14, 2016

Watson on Jeopardy!: The end of the Microsoft/Intel computer era

On this day five years ago, IBM's Watson competed against Brad Rutter and Ken Jennings on the game show Jeopardy!, winning the competition. (The actual competition happened January 14, 2011 with the first broadcast of the two episodes on February 14 and 15, 2011.)

To me, this event marked the Microsoft/Intel computer era that we have been living in for so long and the start of the Watson era, which continues today. To my knowledge, Watson's software and hardware does not contain much, if any, Microsoft code or Intel processors.

The Microsoft/Intel computer era was an age where cheap software running on cheap hardware was king. It was an era where Version 1.0 was garbage and everyone waited until at least Version 1.1 came out. It was an era where Intel microprocessors would take arbitrary text and execute it as a program (allowing over 30 million viruses and other malware to date).

Sadly, other companies fell into this trap of bad software with bad security. Adobe Systems is a leading example of this. In the Watson era, Adobe's Flash is near the end of its life.

I graduated from the University of Illinois with a degree in Computer Science (College of Engineering) in 1976. Some of the best practices of that time were simply ignored by Microsoft and Intel and have only been rediscovered in this decade, some 40 years later. What if the cars of today were designed and built with the technology of the 1970s? Air bags and antilock brakes would not be included. People would be furious! But many business people didn't want to pay for good-quality hardware and software over the last 40 years, so they bought technology that was poorly designed even for the 1970s.

What is different about the Watson era of computing? IBM's Power Systems servers, like the ones running Watson, running the IBM i Operating System have multiple parity bits (including error-correcting memory) and other checks to prevent memory errors and even adder errors from corrupting the operating system, user programs or data. These processors and operating system require that programs be compiled by a compiler and will not run arbitrary text as a program. I know of no viruses or worms that run on IBM Power Systems (with the caveat that there may be malware that affects the two other operating systems that run on IBM Power, AIX and Linux). Ironically, many of IBM Power Systems do run antivirus software, but not as much for themselves: They do it to detect viruses affecting Microsoft/Intel systems which have been uploaded into their file systems by infected computers (how ironic)!

The IBM i Operating System, including its predecessors going back to 1979 (longer than any Windows Server has existed), have always more robust than Windows Server. Today they can manage multiple workloads in multiple virtual machines without conflict or crashing and dependable resource allocation. It's not unusual to have a 10-year-old Power Systems server that has never crashed. There have been cases where one IBM Power Systems server has replaced over 100 Windows and Linux servers. There are companies that have 15,000 active users working on one Power Systems server with sub-second response time.

What did the end of the Microsoft/Intel computer era and the start of the Watson era five years ago mean for the typical computer person? For many, nothing. Companies will still buy cheap: They can buy Windows 2012 R2 running on a server with an Intel processor for less than $1,000. They will still pay lots of money to load this system up with antivirus software, which is unlikely to block zero-day vulnerabilities, and pay technicians to keep this system running and to restart it when it crashes. They will continue taking the server down to install monthly patches from Microsoft. If they need another function, they will buy another server, and another, and another... I know a company that buys a skid of servers with Intel processors (at least 20 servers) whenever they have a planned power outage to replace the servers that will not boot up when the power comes back.

But there's another group of companies out there which understand the false economy of the Microsoft/Intel world. These companies will spend the money for better servers and operating systems, without the need for antivirus software for their operating system, and end up with better results with a lower cost of ownership. And many good computer people will work for these companies, because they don't want to deal with things keeping them from writing dependable programs which can run 24/7 without having to deal with crashes and glitches.


Notes: I have donated $10 in both 2015 and 2014 to Wikipedia for its operations. Have you donated?
This is cross-posted on my company blog at Netburg Services.


Sunday, June 15, 2014

64-bit mobile processors? Only in the Apple iPhone 5S!

John Gruber, writer of Daring Fireball, wrote a four-part post titled "Only Apple." I'm going to look at one statement he made in the second part: "No one else is making 64-bit mobile CPUs and Apple sold tens of millions of them immediately." "Immediately" in this context means beginning with the release date of the iPhone 5S, Sept. 21, 2013.

Here we are, almost nine months after Apple began shipping 64-bit mobile processors in volume. My question: Is anyone other than Apple shipping a mobile device (phone or tablet) with a 64-bit processor? The answer, as you will see below, is "No."

I found this article in PCWorld: "Qualcomm overtakes Apple with eight-core, 64-bit mobile processor." The money quote: "The Snapdragon 615 will go into high-end smartphones and tablets, which should become available in the fourth quarter." We're not even in the third quarter yet.

MediaTek has announced the MT6732 64-bit mobile processor. According to their news release: "The MT6732 platform will be commercially available by Q3 of 2014, with devices expected by the end of the year." So no devices shipping yet.

Samsung, of course, will also have a 64-bit mobile processor. The best information I can find says that these processors will be under the Exynos brand. However, Samsung appears to be waiting for 64-bit Android to be released later this year. Again, no shipping devices yet.

Intel issued a news release touting the "2.13GHz Intel Atom processor Z3480 ('Merrifield')" that is a "64-bit ready SOC" (SOC means "System on a chip"). Intel explains that "64-bit ready" means that the processor will run at 32 bits until 64-bit software is ready. It doesn't actually claim to be a 64-bit processor. It is expected to launch the second quarter of this year. I looked for devices with this processor and found this article from the Motley Fool: "Was Intel Corporation’s Atom Z3480 a Failure?" It claims that Dell is going to begin shipping Venue 7 and 8 tablets with this processor by July 1. I did find the Venue 8 Pro shown on Dell's web site but it is running an Atom Z3740D which is a 32-bit processor although it supports 64-bit instructions. It has the Windows 8.1 32-bit operating system installed. Dell may be shipping a device in the third quarter but it may only have a 32-bit processor, not a 64-bit processor.

Finally, Nvidia has announced the Tegra K1 processor in two versions, one 32 bit and one 64 bit. The 64-bit processor is expected to ship in devices in the second half of the year.

I have to conclude that Apple is the only company shipping a device with a 64-bit mobile processor even though it's now nine months after the iPhone 5S began shipping.

Saturday, July 27, 2013

Seven months since leaving Slashdot

At the end of December 2012, I left Slashdot. Cold turkey, no more reading, no more story submittals, no more comment moderation. I had good karma and I have no idea what they did with it. Like others, I posted under a handle so you won't find my contributions without working at it.

In case you are not familiar with Slashdot, it "is a technology-related news website owned by the US-based company Dice Holdings, Inc. The site, which bills itself as 'News for Nerds. Stuff that Matters,' features user-submitted and evaluated news stories about science and technology related topics." (Source: Wikipedia)

I left for four reasons: I have better sources for "News for Nerds," they refused to track a story back to its original source, the editing of stories was atrocious and I got tired of the stupid comments. Let me expound on these.

Better sources: I have several blogs and other news sources for my tech news. As a whole, they are faster and have better reporting and editing than Slashdot. One of the last straws in December was when Slashdot posted a story that I had read elsewhere two days earlier.

Original source: Slashdot relied on reader-submitted stories. While there is some original reporting on the Internet, many news sites simply scrape and rewrite content from the originators. How many times have you read a story in a major media source and realize that it was written by the Associated Press? The readers who submitted content rarely took the effort to track the story back to its roots, something I always try to do on this blog. This resulted in incorrect information and incomplete stories.

Sometimes, they had ridiculous situations where an Australian news publication was used a source for a United States story. It also resulted in a couple of cases where someone was able to sneak through a story that was a couple of years old.

Atrocious editing: Readers are nerds and their submissions need editing. This didn't happen as much as it should have. Ranging from misspelled words and wrong word usage to summaries that totally misrepresented the stories, this got really obnoxious after a while. Also, the problems listed in "Original source" were passed right on through by the editors.

Stupid comments: Every blog or other site that allows comments has to have a way to deal with stupid ones. Slashdot had an excellent moderation system that used people like me with good karma to keep the stupid comments in line. It was just too much work and there was just too much stupidity to deal with.

I really have a problem with tech people who work exclusively in a Wintel world (Windows + Intel) world and have no clue that there's other worlds out there that have nothing to do with Microsoft or Intel. They would be amazed to know that there are servers out there that don't run any antivirus software and yet are not vulnerable to malware!

Last year, I was wasting too much time cussing at stories and comments on Slashdot. Now I have that time back to do other things, like write this blog.

Updated Aug. 9: Replaced "stupid commenters" with "stupid comments." I really shouldn't call people stupid. People aren't stupid, they just do stupid things, like post stupid comments. As Forrest Gump's mother used to say, "Stupid is as stupid does."