Compiling

We put a particular emphasis on compiling because it stresses the entire system (hard drive, processor and memory) but also because any *nix user knows, compiling is no fun on a slow machine.

GNU Make 3.79.1 / GCC 3.4.2

While GCC isn't multithreaded, we can run multiple jobs using the -j command in make. Below, you can see the significant improvement on performance going from 1 to 3 to 5 jobs. We used the commands as below to compile the Linux 2.6.4 kernel from kernel.org. The kernel is set for a cross-compilation of a default x86_64 machine:

# yes "" | make config
# time make -jX

There were minor advantages of the Solaris compilation over the SLES one. Like most of our other tests, as our compilation moved closer to nine jobs, the reduction in processing time was not quite linear. We do see considerable improvements in speed, but we really aren't getting all of the bang for our buck.

We also threw in some compile tests of entire GCC base, which take significantly longer than the Linux kernel to compile.

You'll notice pretty much the same problems from the Linux kernel compilation test; scaling becomes an issue as we increase the number of jobs. We see a 43% performance increase over the quad Opteron 250 V40z; certainly impressive but we would like to see more.

Rendering Benchmarks Final Thoughts
POST A COMMENT

47 Comments

View All Comments

  • nottlv - Tuesday, July 05, 2005 - link

    #46; the 8 way record claim was for a 64-bit JVM. The PowerPC result you reference was for a 32-bit JVM. specjbb stresses the memory architecture heavily (there's no I/O); the submitted result from IBM include a machine with 64GB of RAM, while the Sun Opteron box had 16GB, and they are running different JVM versions. If you look at the lower numbered runs you'll notice they're pretty close (the Opteron being slightly ahead), but that it hits it's wall much earlier due to significantly less RAM. Reply
  • jjames5 - Tuesday, July 05, 2005 - link

    8 way jbb2000 world record - right!

    This result is a year old and still bests the sun with over 50%:

    http://www.spec.org/jbb2000/results/res2004q3/jbb2...
    Reply
  • nserra - Friday, July 01, 2005 - link

    #34 That is plain stupid, it isnt from been taken out that it will protect him (if thats the true), or the microsoft guys cant read forums where the article have been already posted.

    Microsoft only have to look for who usually talk with anand.

    Reply
  • Opteron - Friday, July 01, 2005 - link

    Forget my last comment...


    ps. i missed 4 way Xeon and Itanium systems :D
    Reply
  • Opteron - Friday, July 01, 2005 - link

    There is miscalculations about percentages, those are calculated wrong.

    In http://images.anandtech.com/graphs/sun%20fire%20v4...

    comment is:
    " We see a 43% performance increase over the quad Opteron 250 V40z; certainly impressive but we would like to see more."

    But actually it's almost 64% since there is no point in comparing 5 threads vs 5 in a 8core system..
    Reply
  • Googer - Friday, July 01, 2005 - link

    Where did the article go? Reply
  • KristopherKubicki - Thursday, June 30, 2005 - link

    Den: Very true but recall that our previous test was done on Opteron 850s instead of Opteron 852s. The 852 performs a bit better than th 850.

    opus13i: We have been stuckin MI redtape land for some time. It wouldn't make sense to change the benchmark at this point either because our previous tests used the 32-bit single core solution. Since they don't seem to have much desire to provide us with the correct license we will probably drop that benchmark in favor of something a little more versatile.

    Kristopher

    Kristopher
    Reply
  • jkostans - Thursday, June 30, 2005 - link

    I dont care who made the hardware, its a hell of a machine. I wish Intel had to rely on inovation and good products to survive like AMD does. Intel really doesn't have many products capable of out performing AMDs equivalent anymore. Reply
  • opus13i - Thursday, June 30, 2005 - link

    ugh.

    Did you even bother to check with mental images? With a simple phone call you could have had the proper licensing in place for 8 cores, as well as every possible variation of 64bit possible.

    "We include Mental Ray and Shake as a point of reference, although both applications are strictly 32-bit at this time. Mental Ray is further hindered by the fact that the version we have is not SMP-aware."

    way to go detective, i dont suppose you actually looked at teh specifications did you?

    http://www.mentalimages.com/2_1_2_configurations/i...

    no 64 bit indeed.



    Reply
  • Xunilla - Thursday, June 30, 2005 - link


    You folks are giving Sun a bit too much credit on the hardware portion of this review. In reality, the system OEM is a company called Newisys, a subsidiary of Sanmina-SCI.
    Reply

Log in

Don't have an account? Sign up now