In regards to Tom's article
"Painting a New Picture of Pentium 4 - Tweaked MPEG4 Encoding" an Intel engineer presents a recompiled version of the Flask MPEG encoder that is optimized for the Pentium 4 with admirable results.
While the Pentium 4, with it's updated SSE2 instructions is a really cool new architecture, is it realistic to compile special versions of current or new applications just to make Intel competitive with the Athlon Thunderbird or even the older Pentium 3?
Will these Pentium 4 optimizations adversely affect performance or, heaven forbid, break code on these competing hardware platforms?
As a native code software developer, I am not willing to risk losing performance or introducing errors on the majority of hardware platforms supported by my software.
As a Java software developer, I am dependent on external vendors to fully support the Pentium 4. These external vendors unfortunately have the same fears regarding hardware support as I do when it comes to binary code compatibility. As it stands today my Java software, on a 1.5 Ghz Pentium 4 would run as slow as or slower than on 1.0 Ghz Pentium 3. Interpreted Java is slow enough as it is. I cannot afford to slow it down even further.
My market is made up of a mixture of every Intel-compatible hardware platform available from the lowly i386 to the Pentium 4 and Athlon T-Bird. Today the Pentium 4 makes up such a tiny segment of that audience, it is just not worth the re-engineering effort required to optimize my software for it.
Don't get me wrong. I am not putting down the introduction of new technology in the microprocessor world. The dated Pentium Pro/2/3 architecture is due for an overhaul. I just don't want to re-engineer my code to support it until the majority of my market can benefit from the effort.
You will find that I am not alone in this view.
"Painting a New Picture of Pentium 4 - Tweaked MPEG4 Encoding" an Intel engineer presents a recompiled version of the Flask MPEG encoder that is optimized for the Pentium 4 with admirable results.
While the Pentium 4, with it's updated SSE2 instructions is a really cool new architecture, is it realistic to compile special versions of current or new applications just to make Intel competitive with the Athlon Thunderbird or even the older Pentium 3?
Will these Pentium 4 optimizations adversely affect performance or, heaven forbid, break code on these competing hardware platforms?
As a native code software developer, I am not willing to risk losing performance or introducing errors on the majority of hardware platforms supported by my software.
As a Java software developer, I am dependent on external vendors to fully support the Pentium 4. These external vendors unfortunately have the same fears regarding hardware support as I do when it comes to binary code compatibility. As it stands today my Java software, on a 1.5 Ghz Pentium 4 would run as slow as or slower than on 1.0 Ghz Pentium 3. Interpreted Java is slow enough as it is. I cannot afford to slow it down even further.
My market is made up of a mixture of every Intel-compatible hardware platform available from the lowly i386 to the Pentium 4 and Athlon T-Bird. Today the Pentium 4 makes up such a tiny segment of that audience, it is just not worth the re-engineering effort required to optimize my software for it.
Don't get me wrong. I am not putting down the introduction of new technology in the microprocessor world. The dated Pentium Pro/2/3 architecture is due for an overhaul. I just don't want to re-engineer my code to support it until the majority of my market can benefit from the effort.
You will find that I am not alone in this view.
Facebook
Twitter
Instagram