Every programmer knows (should know) that interpreted languages are the poorest performers in the speed game. Unfortunately, just about all of the web these days is powered by interpreted languages like Ruby and PHP. Their performance hit is a fact hidden to most users because CPU’s interpreting text is still generally faster than the Internet lag between user and host. One of the most dominant sites out there these days has attempted to do something about that, by creating HipHop, a compiler for PHP that converts logic into C++ whereupon a compiler/linker generates binary code. The company claims that performance gains cut server loads by nearly 50%, a factor that can allow smaller server farms to handle the same load (a reduction in energy, hardware and support personnel).
Similarly tagged OmniNerd content:
- Undocumented API Reverse Engineering Accelerates Firefox 3, by VnutZ about 2 years ago
- Google Contributes to WINE, by VnutZ about 2 years ago
- How to Create a REST API in Rails 2.0, by dhinchcliffe about 2 years ago
- A Cloudy Future for Ruby on Rails, by VnutZ about 2 years ago



a current event
article
by
Print Friendly
Write an Article
Compiled PHP Comment from FB Feed by VnutZ :: NR10 :: Show
So my FaceBook friends never click on the RSS feed’s “add comment” and keep leaving their opinions on my FB page.
Isn’t hardware cheap enough to overcome without all this work? IBM has a 100ghz chip in fab on a non-silicon die. Won’t be long before OBE imp … [cut off by the censors I presume]
Well – the true point is in reducing the amount of hardware by allowing those ridiculous fast computers to do stuff more efficiently than less. Is it really such a bad idea?
I see on OmniNerd the effect of interpreted language whenever Mark runs his analysis scripts. The stuff he’s doing should only take about 2 seconds at most but end up running the server at 90% load for nearly 15-20 seconds.
It’s just unnecessary IMO.
Well consider this; at one point programmers were so cheap that they wrote everything in assembler and looked for every last drop of efficiency possible to get the most out of their multimillion dollar mainframe. Advances in hardware and dramatic changes in costs shifting the burden toward labor vice h/w shifted demand from code to labor … See Moreefficiency and we got C and Ada and then interpreted languages because they were cheaper to program than assembly. We are about to embark on a era of Unprecedented h/w cost reduction and performance gain, and its likely to only further reduce focus on code efficiency in favor of labor cost savings and ease of code management.
You said Ada. I hate you.
Seriously though – I agree with the notion that programmers should be able to program to an ideal form rather than having to kludge up shortcuts because hardware is inadequate. And I agree that we’ve approached that nexus point.
But given the ABILITY to simply run code through a pre-processor/compiler … why not?