Nvidia CUDA Toolkit 4.1 and Parallel Nsight 2.1

Nvidia's CUDA technology has been around for 5 years now, and only 6 months ago blogged about Nvidia's CUDA technology when the CUDA Toolkit 4.0 was released.  Nvidia is keeping up the pace of innovation with a substantive upgrade to both Nvidia CUDA Toolkit (with version 4.1) and Parallel Nsight (now at version 2.1).  

CUDA: What is it?

CUDA is NVIDIA’s parallel computing architecture that enables dramatic increases in computing performance by harnessing the power of the GPU (graphics processing unit) for applications including image and video processing, computational biology and chemistry, fluid dynamics simulation, CT image reconstruction, seismic analysis, ray tracing, and much more.  The current Nvidia (NASDAQ:NVDA) "Fermi" line of GPUs (Graphical Processing Units) provides incredibly powerful parallel computing within reach of most individual users and businesses through rather affordable Nvidia Graphics Cards (and, the upcoming Nvidia "Kepler" GPUs for early 2012 will only be better, faster, and more efficient).

Note: many of these latest CUDA features require a "Fermi"-based GPU (and, using the LLVM-based compiler does).  These cards are worth investing in if you plan to do any CUDA development.  You can get a Fermi-based CUDA-Capable Graphics Card that is quite affordable and power-efficient: I rather like my Quadro 600 card (~$160.00) which uses only 40W for 96 CUDA-processing cores; this card has been very capable for running all my development work on.

New in Nvidia CUDA Toolkit 4.1

LLVM Compiler / Toolchain Support

Nvidia CUDA Toolkit 4.1 now includes a new LLVM-based CUDA compiler along with over 1000 new image processing functions, plus a redesigned Visual Profiler.  Integrating the open source Low Level Virtual Machine (LLVM) toolchain support definitely has my attention (LLVM is a collection of modular and reusable compiler and toolchain technologies).

The first notable benefit of the LLVM compiler is that Nvidia claims this compiler delivers up to 10% faster performance for many applications (compared to their prior in-house developed C/C++ compiler).

But, what strikes me as the most (potentially) important aspect of this move to LLVM is that we could potentially soon see more (programming) language support for using CUDA outside of just C/C++ and/or additional CPU support.  Nvidia has apparently used the Clang C and C++ compilers within the LLVM framework and has hooked in support for the CUDA parallel development environment.

Although Nvidia's (CUDA C and CUDA C++) compiler modifications are not open-sourced, LLVM will provide a foundation for more easily adding language/processor support.  Given Apple's use of LLVM on ARM (platform), I have to wonder if this is going to be a build-target in the not too distant future.  There are also open-source projects for other programming languages to make use of the LLVM toolchain, so the potential does exist for accessing CUDA / GPU-support from other domain-specific languages eventually (perhaps Java, Python, etc) directly.

Other Major New Features in CUDA Toolkit 4.1
(from Nvidia website, with some added comments and details)

New & Improved “Drop-In” Acceleration With GPU-Accelerated Libraries

• Over 1000 new image processing functions in the NPP (Nvidia Performance Primitives) library — this brings to total number of NPP functions to 2200+. These GPU-accelerated functions (building blocks) for image and signal processing include capabilities geared toward arithmetic, logic, conversion, statistics, filters, and more; also, these can execute on the GPU at up to 40x (yes, 40 times!) the speed of Intel IPP (Integrated Performance Primitives).  This is great for media, entertainment, and visual processing applications.
• New Boost style placeholders in Thrust CUDA C++ template library which allow inline functors now.  Thrust includes optimized functions for sort, reduce, scan operations and so on.
• New cuSPARSE tri-diagonal solver up to 10x faster than MKL on a 6 core CPU; this also includes up to 2x faster sparse matrix vector multiplication using ELL hybrid format 
• New support in cuRAND for MRG32k3a and Mersenne Twister (MTGP11213) RNG algorithms 
• Bessel functions now supported in the CUDA standard Math library 
• CuFFT (Fast Fourier Transforms) library has a thread-safe API now (callable from multiple host-threads); also, substantial improvements in speed!
• CuBLAS level 3 performance improvements up to 6X over Intel MKL (Math Kernel Library)
• Batched-GEMM API for more efficient processing of many small matrices (i.e., 4x4 through 128x128 matrices; up to 4X speedup over MKL); up to 1 TFLOPS sustained performance (yes, a teraflop!  Wow)
• Average and rounded-average functions (e.g., hadd / rhadd - signed and unsigned)

Enhanced & Redesigned Developer Tools (On Windows, Mac, & Linux)

• Redesigned Visual Profiler with automated performance analysis and expert guidance (guided workflow and drill-down expert guidance); during an online presentation, this was described as "almost like having an Nvidia engineer in a box", which sure sounds handy!  You should benefit from the experience of those engineers, and be helped along through attaining best-practice outcomes with these built-in automated analyses/experts.
• Assert() in device code - helpful for debugging!
• CUDA_GDB support for multi-context debugging and assert() in device code
• CUDA-MEMCHECK now detects out of bounds access for memory allocated in device code
• Parallel Nsight 2.1 CUDA warp watch visualizes variables and expressions across an entire CUDA warp
• Parallel Nsight 2.1 CUDA profiler now analyzes kernel memory activities, execution stalls and instruction throughput
• Learn more about debugging and performance analysis tools for GPU developers on our CUDA Tools and Ecosystem Summary Page

Advanced Programming Features

• Access to 3D surfaces and cube maps from device code
• Enhanced no-copy pinning of system memory, cudaHostRegister() alignment and size restrictions removed
• Peer-to-peer communication between processes
• Support for resetting a GPU without rebooting the system in nvidia-smi

New & Improved SDK Code Samples

• simpleP2P sample now supports peer-to-peer communication with any Fermi GPU
• New grabcutNPP sample demonstrates interactive foreground extraction using iterated graph cuts (this is really neat!)
• New samples showing how to implement the Horn-Schunck Method for optical flow, perform volume filtering, and read cube map texture

New in Nvidia Parallel Nsight 2.1 for Visual Studio
Parallel Nsight is a powerful IDE-integration and development tool that allows you to perform the following types of procedures from within Microsoft Visual Studio:

• Debug CUDA Kernels directly on the GPU hardware
• Examine (potentially thousands of) threads that are executing in parallel
• Use on-target conditional breakpoints to locate errors
• Use the CUDA memory-checker
• Perform System-Trace activities to review CUDA activities that span your CPU(s) and GPU(s)
• Perform deep kernel analysis to find performance bottlenecks so you can optimize the code speedup that is possible with CUDA and massively parallel-processed code.
• Profiling capabilities including advanced experiments to measure memory utilization, instruction throughput, and stall conditions

Some of the new capabilities include:

• a "warp watch" ability to watch variables and expressions across an entire CUDA warp (a particular level of granularity that is very useful to watch)
• analyzing kernel memory (alloc/dealloc events, execution stalls, etc)

Summary: CUDA 4.1 Continues Nvidia's Great GPU-Accelerated Application Development Tools Improvements

This latest release of the CUDA Toolkit from Nvidia continues to make life easier for any of us that are into parallel-programming with modern GPUs.  Although GPU-computing can be a bit overwhelming and a requires a different mindset than programming desktop applications or designing a website, if you have an application that can benefit from the power of simultaneous operations, this is a technology worth diving into: it is nothing short of a transformational technology.

VariCAD 2012 Sale : 3D CAD Software that is Affordable

VariCAD 2012 CAD Software Sale ends December 28th, 2011

If you are in the market for an affordable 2D and 3D CAD software package for personal, hobby, or professional use, give VariCAD 2012 a look now while VariCAD corporate is offering a 20% off sale through December 28, 2012, and MV4T.com (the USA VariCAD reseller) discounts it by another nearly 20%!  That puts the current price at $525.00 USD for either the Linux or Windows version including one-year of upgrades and support.

This super-affordable software (especially compared to the "big name" 3D CAD packages like AutoCad and SolidWorks), is very robust and still simple enough to use that a novice can pick it up in little time (like I did). It can produce some fantastic 3D renderings too (from their samples gallery):


 

As you can see, it can be quite useful for prototyping mechanical objects and personal inventions visually. I tend to focus on the 3D-aspects of the software (because that is how I think about the objects I am creating), but the 2D CAD features are also plenty impressive also.

My Own VariCAD Use / Experience

I have been much more productive putting my product-ideas down "on the computer" instead of "on paper" (where they tended to be illegible a few weeks later).


My visual-brainstorming has led to some interesting creations (in wood, metal, and carbon-fiber) that I would otherwise not have attempted building -- especially the carbon-fiber inventions, since the material is expensive and I do not want to waste any! The 3D modeling in VariCAD lets me see how parts interact BEFORE I commit to building something (and, I can tweak my design without wasting materials).

In a prior blog about VariCAD, I posted a rendering of some 3D HVAC-ducting layout that I created with VariCAD on Windows -- this design visualization helped me a great deal with what was a complex geothermal heating-system upgrade where new trunk-lines had to be installed in an existing home while working around various obstacles.

Note: you can always try the software (before buying)... just download the trial from VariCAD.com; if you like it, save some money and buy a license while it is on sale!

Installing Redis Database as Windows Service (Redis.io DB) : Issues and Workarounds

Run Redis.io DB as Windows Service

I recently started experimenting with Redis database (a NoSQL DB) as an alternative to SQL-Server for certain development requirements.

"Redis is an open source, advanced key-value store. It is often referred to as a data structure server since keys can contain strings, hashes, lists, sets and sorted sets."

Since I do most of my development under Microsoft Windows, I was hoping to run my Redis instance on my Windows7 x64 Pro development desktop; more specifically, I wanted to run Redis as a Windows service.

Since Redis natively targets Linux/Unix environments, I went searching to see if there was a Windows port of the Redis database project that included the ability to run it as a service.  I found two open-source projects that, when combined, allow me to run Redis on Windows as a Service.

Running Redis DB on Win-7 x64
Installation Notes

The first thing I did was acquire a Windows port of the Redis server. I ended up using this project's compiled-version: https://github.com/kcherenkov/redis-windows-service. The project is described as:

"Windows 32 and x64 port of Redis server, client and utils. 
It is made to be as close as possible to original unix version."

That "close as possible" statement mainly refers to how Redis commands that would (under Unix) rely on fork() to perform background operations are implemented as foreground operations (in Windows port).  But, for purposes of my software development and testing, this would suffice.  I can run my "production" instance of Redis on one of my Linux virtual machine instances (in particular, I have it running on OpenSuse 12.1 x64).

Get the Redis.io for Windows Build
To begin with, download the actual Redis.io (for windows) builds from here: https://github.com/dmajkic/redis/downloads (In my case, I selected the latest x64.zip build, which was redis-2.4.2-win32-win64-fix.zip).

Within that zip-archive, you will see two sub-directories: one is "32bit" and the other is "64bit".  Those zip file directories include the redis-server.exe and redis-cli.exe files (and redis.conf configuration file, etc).  Simply copy the contents of the archive's "64bit" (or 32bit) directory into the chosen directory where you will run Redis from. For example, I placed the x64 files into c:\Redis\

Theoretically, I simply needed to get the redis-server.exe running as a service now...

How to Install Redis DB as a Windows Service:
Win-7 x64 Installation Notes

OK, I have the Redis for Windows executables in my c:\Redis\ directory. Now it is time to get this database running as a Windows Service.  I found one such project that appeared active enough to merit consideration: (link) Run Redis as Service on Windows project on GitHub.

You need to download the compiled executable (RedisService.exe), which is available as a rather small (7 or 8KB) file on the "downloads" page for the project, and place it in your Redis directory.

Note: you may wish to reference this Microsoft site: using SC to create a service if you wish to understand in more detail what the upcoming commands I discuss are doing.

Although you may experience issues (as I will discuss next), you are now supposedly ready to install and start the RedisService.exe (from the command-line in a Windows console window) with the following command (note: alter the "Redis242" service-name to whatever makes sense for you as a process-label; also, change c:\redis portions to whatever directory location you chose):

sc create Redis242 start= auto DisplayName= Redis242 binpath= "\"C:\Redis\RedisService.exe\" C:\Redis\redis.conf"

IF the above statement *appears* to work, the service may or may not start when you execute the following: 

sc start Redis242

But, if you experience some of what I did, the service may be failing for what I will call "hidden" reasons...

Fix Redis Windows Service Problems
and Potential Issues to Workaround

What I discovered with this RedisService.exe windows-service for Redis is that it is quite typical for open-source code: it makes a lot of assumptions and does little to provide proper dependency-testing and meaningful error-condition notification.

When you create the service (per above code: sc create ...) and/or try to start the service (using sc start) it may appear to just "hang" or otherwise take a very long time to attempt to start prior to failing with timeout errors.

The reason for redis-windows-service failing to start properly will be obfuscated, and here are some reasons why:

• Starting the Service will throw 1053 (timeout) errors without indicating why, but one possible failure reason is that you must have the .NET Framework 4.0.30319 installed for this service to work.
• Next, depending on your security setup (like my Windows-7 Pro security settings), you may need to tell Windows Firewall that it is OK for this process to act on your local network.  The easiest way to do this is run the redis-server.exe from the command-prompt and allow it access (to local network, through Firewall) when prompted.
• Next, if you attempt run the redis-server again, you may see another (otherwise hidden) issue in that the executable is not from a "trusted source" or such: again, this issue can be resolved by choosing to allow this un-trusted process to run when provided the option.

After resolving this list of potential issues, you should be able to execute the sc create command and then perform an sc start redis242 (or whatever name you gave the service), to start the Redis Windows service and no longer experience a 1053 error due to timeouts caused by hidden reasons.

Redis Windows Service-Shutdown Problems

Note: there are problems with shutting down this service!  So far, the only way I have found to truly stop it is to reboot my system.  Also, when attempting to delete the service (sc delete redis242 or such), you will not be able to truly delete it as long as any Windows Service-Manager windows are open.

Once the RedisService.exe is installed and actually working, even "sc delete" requires a system reboot to take effect, since you can not otherwise truly stop the service.

The good news...
Although this service is problematic (as of when I wrote this tech blog entry), the program will run as a service and the client (redis-cli) can now be executed against the service-induced redis-server to test SET/GET of keys, etc.

If you are interested in accessing Redis.io from JavaScript, you may want to read my blog about NPM (Node Package Manager) where my example for installing a Nodejs module used a node-redis module.  I am able to access my Redis (NoSQL) DB from both Linux and Windows versions of Nodejs via the node-redis module's functionality from within Javascript.

NPM (Node-Package-Manager) Package Install Error

NPM (Node-Package-Manager) Package Install Errors:
Misleading Error Messages

I recently found myself encountering some nearly useless NPM errors when I was trying to install the redis-node package (i.e., Redis.io database connectivity layer for Nodejs), and it did not instantly occur to me why I was seeing error messages during NPM's archive-unpacking operation, especially as the Node-Package-Manager was dumping out a list of errors that made little sense.

Under Linux, NPM install was producing errors that implied I did not have permission to the tmp / temporary directory or directories the package was being unzipped/unpacked into for installation from (i.e., the temporary location the tarball / tar.gz file would be unpacked in).  Under Windows, the same NPM install (under git-bash window) produced different errors that made it appear the downloaded .git package could not be unpacked for god knows what reason, all hidden in a massive error dump that had nothing to do with the REAL reason for the error.

Well, I figured out why the errors occurred (as detailed in this blog entry), and I also learned that this NPM software is a perfect example of what I consider a non-user-friendly interface in that it presents the user with all sorts of completely meaningless error messages in the event of very simple-to-detect issues.  Bottom line: the node package manager is written by geeks, for geeks.  This is not "enterprise grade" software by any means, as it does not have robust error-detection/traps/messages.  In fact, the error-traps it does employ seem to mislead more than assist: this is simply poor design.  (note: I give credit to ANY open-source effort like this though, and I understand why people focus on other functionality vs. "usability").

Node Package Manager Windows Behavior and NPM Error Messages due to using wrong .git file URL

I have been using Node (Nodejs) for "server-side javascript" development on a Windows 7 Pro x64 machine, with Node "installed" as simply the Node.exe download placed in a local directory of c:\node

From within that directory (using the git-bash terminal window on Windows that was installed with the Git SCM tool version 1.7.7.x Windows .exe installer), I executed the following command:

c:\node>node ./npm install -g https://github.com/mranney/node_redis.git

...but, that produces the following error dump. Can you see why? It was all too clear to me a bit later...

c:\node>node ./npm install -g https://github.com/mranney/node_redis.git


npm ERR! couldn't unpack C:\TEMP\npm-1322598499268\1322598499268-0.7676450146827847\tmp.tgz to C:\TEMP\npm-1322598499268\1322598499268-0.7676450146827847\contents
npm ERR! Error: ENOENT, no such file or directory 'C:\TEMP\npm-1322598499268\1322598499268-0.7676450146827847\contents\package\package.json'
npm ERR! Report this *entire* log at:
npm ERR!     <http://github.com/isaacs/npm/issues>
npm ERR! or email it to:
npm ERR!     <npm-@googlegroups.com>
npm ERR!
npm ERR! System Windows_NT 6.1.7601
npm ERR! command "node" "c:\\node\\npm" "install" "-g" "https://github.com/mranney/node_redis.git"
npm ERR! cwd c:\node
npm ERR! node -v v0.6.3
npm ERR! npm -v 1.0.105
npm ERR! path C:\TEMP\npm-1322598499268\1322598499268-0.7676450146827847\contents\package\package.json
npm ERR! code ENOENT
npm ERR!
npm ERR! Additional logging details can be found in:
npm ERR!     c:\node\npm-debug.log
npm not ok

Notice the URL I mistakenly specified for the install package... THAT is what causes the error.  Yes, something as simple as copying and pasting the incorrect URL from the project's github page will lead to this tragic error-dump.

This is where I went wrong:

...the default URL shown in that textbox on Github (image above) is for the HTTP version of the URL.  I had accidentally copied that and pasted to my command prompt for my "npm install" command, where instead I needed to do this:

Now I could copy and paste the proper git:// prefixed URL address of the node_redis.git package that I wanted to install with NPM.

c:\node>node ./npm install -g git://github.com/mranney/node_redis.git

...which produces the simple one-line output as a result of a "successful" package install:

redis@0.7.1 c:\node\node_modules\redis

Ah, that is better!

Now, on to what this same issue presents like under Linux...

Node Package Manager Linux Behavior and NPM Error Messages due to using wrong .git file URL

I have been using Node (Nodejs) under OpenSuse 12.1 x64 KDE.  When I first encountered this issue under Windows, I quickly jumped over to my Linux VMware virtual-machine to see if for some reason it was a Windows-implementation-only issue (since, Node and NPM have been more mainstream on Linux, and the NPM under Windows is considered "experimental" yet).

This quick test under Linux helped me see the error in my ways quickly, since I encountered a similar mess of misleading error-messages spewing forth from the node-package-manager when I copied and pasted my npm install command to Linux and executed it:

~/node> node ./npm install -g https://github.com/mranney/node_redis.git

Yes, it failed similarly to the Windows NPM version, but with an error message (pasted here) that sure made it appear like the the "tar" (unpack) command failed for some inability to work with the temp .tgz file created as part of the install process...

npm ERR! Failed unpacking /tmp/npm-1322603639405/1322603639405-0.840085425414145/tmp.tgz
npm ERR! couldn't unpack /tmp/npm-1322603639405/1322603639405-0.840085425414145/tmp.tgz to /tmp/npm-1322603639405/1322603639405-0.840085425414145/contents
npm ERR! Error: `tar "-zmvxpf" "/tmp/npm-1322603639405/1322603639405-0.840085425414145/tmp.tgz" "-o"`
npm ERR! failed with 2
npm ERR!     at ChildProcess.<anonymous> (/home/mike/node/npm/lib/utils/tar.js:217:20)
npm ERR!     at ChildProcess.emit (events.js:70:17)
npm ERR!     at maybeExit (child_process.js:359:16)
npm ERR!     at Process.onexit (child_process.js:395:5)
npm ERR! Report this *entire* log at:
npm ERR!     <http://github.com/isaacs/npm/issues>
npm ERR! or email it to:
npm ERR!     <npm-@googlegroups.com>
npm ERR! 
npm ERR! System Linux 3.1.0-1.2-desktop
npm ERR! command "node" "/home/mike/node/npm" "install" "-g" "https://github.com/mranney/node_redis.git"
npm ERR! cwd /home/mike/node
npm ERR! node -v v0.6.2
npm ERR! npm -v 1.0.106
npm ERR! 
npm ERR! Additional logging details can be found in:
npm ERR!     /home/mike/node/npm-debug.log
npm not ok

Having performed successful node module installs with NPM under my Linux host already, I knew that I must be missing something obvious.  And, without any help at all from those meaningless error messages, I finally saw that I pasted the "https://" address of the github node module address instead of the git:// version.  Fixing this mistake, the module installed just fine under both Linux and Windows.

This revelation (about my mistaken "https" vs. "git" prefix), coupled with my frustration at misleading NPM error messages, led to the writing of this blog in case anyone else runs into this issue and gets misled by utterly meaningless error messages.

I hope someone working on the NPM (package manager for node) will eventually make time to implement proper condition-testing (and meaningful error-reporting) to test the format of command-line parameters as important as the actual package/module URL if the format makes such a tremendous difference in results.  Had this software simply tested the URL for a valid format, and told me "sorry, you must use git:// prefix" or suggested another equally effective alternative, that would be ideal and would save myself, and surely others, from wasting time with misleading error messages.

Redis (Redis.io) Database via Node

As an aside, I can report that I am able to access my Redis (NOSQL) DB from both Linux and Windows versions of Node via the node_redis module's functionality from within Javascript.  I hope to add more blog material about my experiences with installing and using Redis with Node at a later date.

Saving Money in Information Technology and the Data-Center

Green Technology : Energy-Savings Techniques

Saving Money by Reducing Energy Consumption

I was just reading the Computerworld top 12 green-IT organizations list for 2011, and after many pages of reading, it was clear that certain energy-savings strategies were quite common across the range of companies featured in the article. I am all for saving energy, as this not only reduces the carbon-footprint of an organization, but it also saves money!

Some of the techniques noted should spur ideas about how you and your technology organization can save money by focusing on being "green".  Here is a summary of how the top-12 companies made substantial strides in reducing their energy consumption through energy-savings initiatives that led to real and measurable financial cost savings too:

• Virtualization: this is a significant savings opportunity for many of the companies featured in the roundup of green IT organizations.  Virtualization is moving beyond server and network virtualization and into the virtualization of desktops also, for some of these companies.  The ultimate objective of this push into virtualization is to consolidate (computing) workloads onto as few machines as possible and achieve very high physical-machine utilization rates.
  
  Allstate Insurance was featured for its ability to reduce its total computing server and device count by 3,000 units while realizing a cumulative energy-reduction of nearly 40% in the past several years. NBC Universal virtualized 60% of their physical servers and shut down 2,000 physical machines. Northrop Grumman posted substantial energy savings through widespread virtualization also, and they are considering thin-client desktops for further savings. Citigroup has gone as far as to require all new servers be virtual, unless physical servers are justified; this has reduced their power and cooling requirements by 73%.


• Cooling system efficiency: modern data centers pack incredible amounts of computing power into confined areas, and as such they require substantial cooling systems (i.e., air-condition).  Kaiser Permanente was a focus company in this area, as they achieved energy-consumption savings across their three data centers that are nothing short of incredible, "cutting an eye-popping 7.2 million kilowatt-hours of power from overall data center operations [this year] -- and over $770,000 from power budgets."
  
  How did they do it? They focused on everything from sealing up air leaks (which provided the biggest win) and a sophisticated real-time monitoring system for measuring cooling system efficiency.  They were able to reduce cold-spots in the data center (which are signs of inefficiency) and avoid overprovisioning power distribution and cooling infrastructure in general.  NBC Universal was also mentioned for having implemented similar smart power distribution units and rack-level environmental and power metering sensors, allowing them to increase rack densities by as much as 200%.


• Raising ambient temperature (in the data center): the IT group at KPMG was featured for its efforts, which included "raising the ambient temperature in the data center to improve efficiency by more than 5%, raising the temperature of the water in the cooling tower to improve efficiency by 5%".  Raytheon was also featured for how IT cut energy use in each telecommunications closet by 30% when it raised the temperature by 10 degrees Fahrenheit to 75 degrees
  
  One related news bit I did not see mentioned in this Computerworld article was how Dell has recently announced that some of their servers (currently the R610 1U and R710 2U varieties) are capable of running at what is essentially typical outdoor-ambient-air-temperatures (in nearly all of the USA, and nearly all year round).
  
  These run-hot-capable Dell products (and Dell, the company and the stock: NASDAQ:DELL) are worth watching, as the energy-savings implications are huge for these high-operating-temperature Dell servers. There are also Dell storage arrays, switches, and power-distribution units that are certified to run hot.  I expect more computers (especially servers) will end up being certified to run hot like this, which could vastly reduce cooling costs for IT facilities as ambient air temperatures could approach outdoor levels (as such, lower cost air-movement by way of fans vs. chillers could perhaps become the norm for much of any year). 


• Blade-Server technology: KPMG was again featured for its migration to blade server technology, with "the average blade server consuming about 50% less power than a comparably configured rack-mounted server".  Other firms were mentioned for their push to increase rack computing densities through such means.


• Data-Compression and storage de-duplication to reduce physical hardware required to house company data.  The finance firm State Street was noted for its efforts in this area where they reduced storage use by between 40% and 50%.  This should lead to about a similar percentage savings on both power and equipment costs related to storage.


• Private Cloud Computing and High-Performance-Computing (HPC) Clusters were mentioned by some of the top-12 energy-saving firms. "Baker Hughes created a high-performance computing (HPC) cluster that incorporates wake-on-LAN technology. In this environment, machines are turned on -- or "woken up" -- via a network message, and the company can wake up machines for use in the HPC cluster as needed. This setup uses 40% less energy than a dedicated HPC pool."


• Solar Panels have been installed by some of these firms (KPMG), and surely other type of alternative-energy will become more common in data-center planning in the future.  I didn't see any mention of things like backup power based on modern fuel-cell technology or micro-turbines and such, but I would expect that some of this is in play for various firms.


• Telepresence to reduce travel. Various firms, including KPMG and Allstate, were mentioned for taking steps to reduce employee travel, not just locally, but nationwide inter-office travel as well.  Technology for streaming video and web-conferencing are employed to make this possible.
  
  To me, it seems incredibly obvious that one of the largest "green" moves any firm can make is to enable workers to work from home whenever possible.  Though, simply put, I do believe that many workers are just utterly incapable of managing their time and remaining focused while working at home as they let all the distractions of home interfere; this explains why, with all this modern technology that would make a vast number of jobs possible from home, there still is not widespread work-from-home opportunity.  It will probably require $10/gallon gasoline to make a substantial change in this area.

SVG onload event not firing : Firefox bug / feature with Shortcuts

FireFox not firing SVG onload event

(Windows) Shortcut handling to blame...

I do a fair amount of work with SVG (Scalable Vector Graphics) images / files that contain embedded JavaScript for various event-driven interactive-SVG components. The onload() event, within SVG files, is something I regularly use too. Today I ran into a strange "feature" or "bug" that shows up in Firefox but not in the Google Chrome / Chromium browser — related to this onload event in an SVG document.

I use Chrome as my default browser, especially because I like the included Developer tools a lot, so most initial testing of my web-page HTML, SVG, and JScript code takes place in Chrome before I move onto testing in other browsers (like FireFox). I have been working on my custom SVG RAD Components (tis' what I currently call them), and because I use one particular .SVG file for the main "test rig", I kept a Windows Shortcut on my Windows 7 desktop for a quick reference to that .SVG file.  I just click the shortcut to launch my SVG "application" (in Chrome) or drag the shortcut into a Chrome tab, and that works just fine.  Ah, but not in FF!

FireFox apparently does not resolve shortcut properly if dragged into browser

Being a creature of habit, when I was ready to test my latest SVG file and Javascript code within Firefox (using version 7 currently), I dragged my Windows shortcut (to my SVG file) onto the FF browser and poof... it seemed to load the SVG file, but my onload() event code simply failed to run.  I would have sworn I did this exact same drag-to-load (my SVG) with prior versions of Firefox successfully, but either way, it is not working now.

So, I loaded the page again via drag-and-drop of my Windows-shortcut to my SVG file, but this time with Firebug running (debugger / developer tool).  I quickly see that an error is being generated whereby the event-code referenced in the onload() event was shown as "myOnloadFx is not defined" within the onSVGLoad() code in FireFox (evt=SVGLoad). Clearly something strange was going on here, as this code "works" and has worked in Firefox before.

I played around with the code inside the SVG file a bit, and moved the onload() code from the opening SVG tag's onload="myOnloadFx()" to an inline-script (using <script> tags) just before the SVG's closing tag... and, the problem persisted.  So, what the heck?  After wasting more time on this than I ever should have, I then decided to go to the directory in which the SVG file really existed (vs. using the Windows shortcut to open it), and I dragged the .SVG file onto the Firefox window where it opened fine and ran the onload() event code just as Chrome did.  So, the shortcut-dereferencing/resolution is apparently to blame.

FireFox : want your SVG Javascript onload to fire? Do not open the SVG by dragging a shortcut onto FF

Now I know.  Note: the code executed in my onload() event was in an external Javascript file that is "included" in the SVG by way of code like this: <script type="text/javascript" xlink:href="myExternalSVGcode.js"/>  

Firefox is not properly converting, storing, and subsequently referencing the proper file-locations for included-code like this, but is instead looking in the directory where the shortcut appears (in my case, the desktop).  I confirmed this to be the problem simply by moving a copy of the Javascript (referenced) file onto the desktop along with the Shortcut (.lnk) and voila!  It "fixed" the issue.  UNREAL. 

I have not tested other types of scenarios where this could be a problem, and it is unlikely most people will ever encounter this unless they do software development (and perhaps even just SVG/Javascript with included external javascript files).  But, just in case, I figured I would post my notes here for anyone else that may encounter this weird onload behavior in Firefox.

Embarcadero Delphi XE2 / FireMonkey Review Findings : Not Ready, Incomplete

Delphi XE2 : Reviewers / Users Not Pleased

I am a long-term user of the Embarcadero Delphi (formerly Borland, Inprise, and Codegear branded Delphi) RAD (Rapid Application Development) IDE, programming language (which is object-pascal) and VCL (Visual Component Library).  Recently I wrote a blog about the Exciting New Features in Embarcadero Delphi XE2 and how much I was looking forward to putting the new FireMonkey components to use in particular.  Also, while reading this, keep in mind the fact that (for me) FireMonkey was the one thing that was going to keep me interested in this product for my Windows development platform of choice.

Well, here is the simple fact of the matter: Delphi XE2 / FireMonkey is not ready for prime-time.  I have discovered that I am not alone in this opinion, and it appears that Embarcadero has done itself no favor by releasing what many consider an unfinished product.  Perhaps I am premature in predicting the final demise of this platform and product, but that is essentially what I am doing: it is over... at least for me.  I have used Delphi since version 2.0, going all the way back to 1995, and have been a loyal and avid fan of the development language, IDE, and components.  But, I am 99% sure that I am finally calling it quits. Here's why...

Poor and Missing Documentation

I have totally had it with the total CRAP documentation that has come with Delphi since ever since Delphi 7.  I have heard promises of improvements for years, and with every small improvement comes a host of missing documentation, bad or missing hyperlinks within the documentation, a poor (and god-awful slow and inefficient) documentation-browsing interface, etc.  I long for the old days where I could quickly find all the help I needed and quickly navigate links to related information; those days are long gone.

I just can not take it anymore.  It is absolutely inexcusable how pathetic the documentation efforts have been, regardless of which company has owned the rights to Delphi (Borland, Codegear, Embarcadero).  Clearly documentation is barely as step above an afterthought for these companies lately as they probably consider it nothing but an "expense"; never mind that this expense is what leads to a product I actually may want to purchase!  With all sorts of new features simply nowhere to be found in the documentation, my level of frustration is epic.  I have seen as good, or perhaps better, documentation efforts for open-source projects (you know, all those projects that are notoriously lacking documentation). 

FireMonkey Bugs, 64-bit bugs, Missing Features

I am not going to re-hash everything that people are experiencing with FireMonkey, as it just furthers my frustrations with this incomplete and buggy product.  But, here are some of the things that are worth noting, as pointed out on blogs like Delphi Haven and AnalogMachine:

• the omission of Actions and action lists
• aside from the very basics, Keyboard handling is crippled
• Property/method reference documentation (as I more than hinted at earlier)
• TMemo has serious bugs.
• Substantial differences (and missing capabilities) in FM form-designer behavior vs. VCL-forms designer behavior.
• the 64-bit Compiler has some serious bugs.
• And many more...

Delphi is now looking more and more like a total dead-end for me and certainly will not help my career in the least bit; I am no longer willing to wait for bug-fixes to back their technological improvements that ship without proper testing.  And, I do not want to continually pay for the right to deal with such poor quality.

Delphi XE2 Update-1 and Update-2

Embarcadero has released the first couple "updates" (Update 1) for Delphi XE2 and (Update 2) for Delphi XE2 that will supposedly begin the process of correcting some of the deficiencies in this product. But, I am at the point I do not care. Why? Because we all know (based on past behavior) how this will go... they will issue approximately 4 "updates" (i.e., BUG FIXES) to the product during its life-cycle and before Delphi XE3 is released, at which point you then will be forced to PAY AGAIN for what should be free and continued bug-fixes!


And, like darn near every single "release notes" published in the past 5 years by these folks, this XE2 Update 2 release-notes document also has screwed up hyperlinks to the supposed "list of fixes".  Currently, in the "General" section of this document, see where it says (I cut/pasted this from their site):

"For a complete list of the specific problems fixed in this update, see the "List of Bug Fixes in Delphi and C++Builder XE2 Update 2" at: <a href="http://edn.embarcadero.com/de/article/40984/">http://edn.embarcadero.com/de/article/40984/</a>"

Well, guess what folks... the above quoted HTML really links to some OLD Delphi XE Update 1 list of fixes and even points to the German location (/de).  THIS IS SO TYPICAL!  AGAIN, WHERE IS THE QUALITY-CONTROL!?  I really would have liked to see what is truly "fixed" in this XE2 update.  Ughghgh!  Do we now have to post a QC (Quality Central ... heh, "quality", yeah... ok) entry to report this messed up URL?  Again, this is TYPICAL and has been the case for years.  I think it is done ON PURPOSE since they really have no list of updates prepared!  But, who is dumb enough to keep doing this and keep peeving their customer base?!

This practice of shipping bug-ridden, unfinished software products is common among MANY software firms these days... they put out an unfinished/buggy product, patch it a few times for "free", then force users to purchase yet another "upgrade" to get any future fixes/improvements regardless of how strong the case is that there are still substantial bugs in the product you PAID for!  It gets old, and it gets very darn expensive! And, I have seen long-standing substantial bugs go unfixed through multiple major-product-versions/iterations (Delphi and their help-system clearly have had long-standing substantial issues).

Delphi XE2 : the end of the road for my Delphi use

For the reasons laid out already, and for the added fact that there are VERY FEW Delphi Jobs / Contracts available anywhere these days, I am leaning toward totally ditching closed-source development tools like Delphi.  There are some great free and open-source products and technologies out that I am going to be evaluating in depth and preparing to move to.

I am likely go to just HTML5 / SVG / JavaScript wherever possible, and I may totally ditch "native Windows applications" except for those situations where I absolutely require the performance and sophistication I can only achieve with compiled code.  I have prior versions of Delphi that I will continue to use to support some of the systems I have written for my own use of course, but I also plan to begin the inevitable migration to other technologies.  I am not looking forward to migrating my systems include lots of SQL-Server interaction (using VCL dbExpress or ADO components), but such is.

Although I believe Delphi has provided a super-productive development environment and language for developing Windows applications over the years, I also see the writing on the wall: Delphi is a niche tool whose niche continues to shrink as people move to web-based everything.  And, Embarcadero is not helping Delphi's cause when they release buggy unfinished software with equally pathetic documentation.  Perhaps I will again assess Delphi when the inevitable Delphi XE3 appears on the scene, but for now, I am avoiding any further investment of my time and money into this product.

UPDATE (January 2012) for anyone still interested in Delphi XE2: Embarcadero has released Delphi XE2 Update 3 and has put a new Delphi XE2 ISO Image with Update 3 online.  I have not taken time to download and install it, though I did read through the release notes of fixes in Delphi XE2 Update 3, and it appears they are slowly working through the massive pile of major errors and issues.  Surely it must be "better" than the initial premature release of XE2. But, I am still a major skeptic and will only consider a hands-on Delphi re-evaluation when Delphi XE3 is released -- and, Embarcadero had better take time to ensure a much higher quality product if they are to convince me to ever use their development tools again.