My Next Pentium Is A DEC Alpha

Is a DEC Alpha a solution if you want a really fast Linux system? Here is one person's experience that may help you decide.
After the Storm the Calm

Of course, I caused my own disaster, but that makes the repairing much more rewarding (you have that relief factor). After the first weekend of fun, what could I do for amusement? Porting over some of my favorite utilities would be fun. Both Red Hat and Craftworks take some of the fun out by already porting over the Apache Web Server and browsers, but I could still have some fun by moving other stuff over. So, the next question is, “What do you have to know to port code to Alpha Linux?”

Porting Hints

X-Windows applications that use IMakefiles basically configure themselves out of the box. GNU software that uses autoconf/configure to figure out what system it is running on tends to get confused. The machine configuration string that it synthesizes looks like alpha-unknown-linuxaout. This is confusing because it is an alpha that is not running DEC Unix, nor is it Linux running on an Intel system. What to do? Well, I usually put in the following code segment into configure:

alpha-dec-osf3* )
    machine=alpha opsys=decosf3-1
  ;;
  ## We're Alpha Linux
  alpha-*-linux* )
    machine=alpha_linux opsys=linux_axp
 ;;

  ## Altos 3068
  m68*-altos-sysv* )
    machine=altos opsys=usg5-2
  ;;

But that means I have to write a ./src/m/alpha_linux.h (which I would make by blending alpha.h, and removing anything cd DEC Unix specific), and ./src/s/linux_axp.h (which would be made from linux.h, minus instructions on how to make sharable libraries). None of that is too difficult. Later releases of most software will come with pre-built configuration files as autoconf gets updated, and developers begin to use the new version.

The other issue you get involved with is the fact that several programs publicly available assume 32-bit addresses and 32-bit ints. Linux for the Alpha is a 64-bit operating system, with 64-bit addresses. Frequently, this provides harmless warnings about adding 32-bit offsets to 64-bit pointers.

Then there are the programs that attempt to override the definition of operating system calls. System calls that have been standardized to take parameters like size_t (but is being redefined for unsigned int) will cause complaints from the compiler.

The really insidious things, though, are those programs that do bitwise manipulations without regard to portability. Generally, I've learned to become suspicious of any program that isn't packaged with autoconf/IMakefile, which runs only on one platform (e.g., it runs on Linux; you tell us if it runs on Solaris, BSD, HPUX, etc).

The Future Looks Bright

Many packages compile out of the box. With ELF support comes the ability to port the Java JDK over. Sun, HP, and other notables are releasing their 64-bit processors to the marketplace. And while everyone argues what the 64-bit standard for Unix will be, we will already have been there, and have moved on to more interesting projects.

Bryan W. Headley (bheadley@interaccess.com) has been working with Unix since 1978 except for an interruption by that interloper, MS-DOS. A Unix applications developer by day, he becomes a Linux hacker by night. There isn't a compiler or kernel that he doesn't find worth playing with.

______________________

Webcast
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers

Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.

Learn More

Sponsored by AMD

White Paper
Red Hat White Paper: Using an Open Source Framework to Catch the Bad Guy

Built-in forensics, incident response, and security with Red Hat Enterprise Linux 6

Every security policy provides guidance and requirements for ensuring adequate protection of information and data, as well as high-level technical and administrative security requirements for a system in a given environment. Traditionally, providing security for a system focuses on the confidentiality of the information on it. However, protecting the data integrity and system and data availability is just as important. For example, when processing United States intelligence information, there are three attributes that require protection: confidentiality, integrity, and availability.

Learn more about catching the bad guy in this free white paper.

Learn More

Sponsored by DLT Solutions