From: Ilias Lazaridis Date: 2011-08-31T18:09:38+09:00 Subject: [ANN] SAMSUNG to produce "Ruby on Rails in Silicon" System on a Chip (public draft) Samsung Electronics, Co., Ltd., one of the worlds greatest advanced semiconductor producer, announced the latest of it's popular ARM11 based System on a Chip (SoC) solutions, the RORIS6440 "Rails in Silicon" web-application processor. Based on the Samsung’s advanced 45nm low power CMOS process technology, the "Rails in Silicon" chip offers a high performance, low power and cost effective solution for next generation web applications. The System on a Chip consists of 4 subsystems occupying each 1/4 of the SoC : a) The ruby language interpreter in silicon b) The ruby gems subsystem (Xilinx FPGA) c) The Rails Framework subsystem d) The normalization FPGA (normalizing inconsistencies of the other 3 subsystems) The RORIS6440 web-application processor is available in samples for selected customers. It is scheduled for volume shipment in the fourth quarter of this year. The chip is housed in a 13×13 FBGA package with a ball pitch of 0.65mm. - Yukuhiro Matsumoto, the ruby language designer commented: "Samsung engineers gave me some requirements for the necessary code- refactoring, in order to simplify the integration of the core interpreter into silicon. I can say that they were really professional till the latest cell of their body. And they listen, too! I said to them "I'm passionate about ruby, and many people love it". They said: "Don't worry, we will place some hearts at the side of each wafer with an inscription "With love, Ruby". - Well, they did it. I'm sitting here at my electron microscope (a present from Samsung's CTO), looking at the wafer's inscription. Just Lovely!" Asked what happened with the code-refactoring, Mr. Matsumoto replied: "I don't know, I got problems with my stomach after one day doing the refactoring. It was finally outsourced to undisclosed contractor, I think somewhere in Europe, but am not sure." - David Heinemeier Hansson, the designer of the initial Rails framework commented: This is the natural flow of things. Several people have contributed to new versions of the Rails framework (which were build based on much more specifications than the initial one. They increased the speed and stability of the Ruby interpreter. Even a "Computer Science Company" got involved, increasing the speed of web-server execution to 30%. Twitter has twitted like crazy in order to make things work - but then they moved to J...! Still, we we're not able to catch up with J... implemented systems - and in no way with systems implemented in the so called "King of Languages" (C++). After one year of reworking ruby and Ruby on Rails, we went nuts. Even a whole book "REWORK" didn't help - things become even worser, and people started to want 3 working days in summer. One step before we were forced to move to J [Mr. Hansson always got a hick-up when trying to say "Java"], Samsung contacted us with the offer to "go silicon". The chip increases the execution speed of rails applications to a factor of 5 to 10 (compared to mainstream intel/amd chips), and reduces the daily restarts to just 1 to 3 (in a typical Rails application). And communication with demanding Rails developers and users has become really easy: "Get used to it, it's silicon.". - Rails developer and ruby contributor Tenderlove commented: "OMG! I feel so happy. I went to the FAB an hugged all the 40 workers in the high-sterility environment, giving each and every of them a few of my favorite flowers. Ok, I've ruined their clean-room, knocking-out the production for 2 weeks. But who cares. OMG!!! A CHIP!!!" http://www.sciencephoto.com/media/347991/view - Ryan Davis (aka "The Release Berserker") said: "I am very proud that a complete RubyGems subprocessor was integrated. Eric Hodel and I insisted that the subsystem will be field- programmable, thus we can still release code fast (and depracate api's). Samsung engineers understood perfectly. They integrated the rubygems subsystem into the 6th generation Xilinx's Spartan-6 FPGA Family, right into the the chip. I have no idea what this chip is about. All that I know is, that I can require 'roris_fpga_upload' and then push a new release by ... (forgot the new API call, I've refactored it already 3 times). Anyway, I like my title. I'm Davis, the Release Berserker - and this will stay even with silicon, thank's to the excellent team at Samsung." http://www.xilinx.com/products/silicon-devices/fpga/spartan-6/ - James Edward Gray II commented: "I've written books about ruby, and half of the stuff I've documented and explained, was refactored and normalized away, in order to make ruby integratable in silicon." he told to the reporter in a slight sad tone. "How does it look now, people ask me now, why did I wrote books about those 'features', instead of normalizing them away."?' he then continues with tears dropping from his eyes "They even ask me, why I didn't saw those inconsistencies all those years, why I documented them like features. Didn't I know? Or did I just want to publish books, thus I'm called an expert?". After a few seconds of silence he stood up and shouted "the worst thing is, that Samsung selected this Zombie named Lazaridis in order to normalize and refactor the source-code base, thus it becomes able to be integrated. This guy knows nothing about ruby, even not "puts", how can he normalize the language?" Gladly, Tenderlove was present, and gave him five rations of hugs, and some flowers that Samsung engineers had trowed after him (those from the clean room). So "little James" (as Tenderlove calls him tenderly) calmed down soon and added with a great smile: "Well, I guess I'll write a new book now: "Ruby on Rails in Silicon" Reference for Beginners. Thank you, Samsung!" - The Ruby Core Developers announced simply "We love our spaghetti-code. It's a good code. And only we can grasp it. That's good, this should stay this way. Chips are not necessary." - Mr. Lazaridis, the first ever seen "troll" which solves C-core-level language-design issues, commented: "The truth is, that the hype around Rails had opened the doors to the headquarter of the Korean Chip-Giant. A java hating CTO (his wife left him for the highest-paid Korean Java-CTO) introduced Rails to some departments. The departments started to implement their applications themselves (as they hated their IT guys), and had a productivity boost of a factor around 5 to 10. It spread quickly within the company, even a dedicated (but unofficial) IT department was introduced, referred to as "Section 31"." Lazaridis continues in his typical criticizing tone "Then the problems and the productivity loss started. The ROM methodology (Relational to Object Mapping, the reverse of ORM - Object Relational Mapping) negates most benefits of the Object Orientation, making the new "Rails- Nija's" collapse and then pay $400 and more for a simple advertisement to find "Rails Experts". Rails Experts which know, that Rails has it's natural limits, dictated by laws of physics and mathematics. But it's a job, and they do it." Lazaridis continued straight and openly, not afraid to loose even the last tiny opportunity for a contract within the ruby domain. He stated the most relevant fact, which everyone knows, but no one want's to speak out: "The technology-lock-in machinery of Rails worked nice. Everything is reinvented, reimplemented, renamed, presented with fun, love and things opposite to what people hate. But how deep got Samsung trapped?. The answer is: VERY deep. The financial departments estimated the "lock-out" costs, and finally found out that producing a new chip would be much cheaper than getting Rails specialists from overseas to solve the problems or to migrate away from Rails. This solution had the additional benefit from preventing the CTO from committing suicide. Who want's to admit publicly that he has been trapped by a marketing machine? This simply does not happen after the lessons that Windows 3.x teached." With a smile in his face, Lazaridis told the reporter: "Many people would get a stroke, if they'd know that I was the undisclosed subcontractor, who normalized (with a self-selected distributed team of 3 people) the ruby-core, and created the basic SystemVerilog2009 simulations. Please don't write this, thus everyone stay's happy!" [Editors note: Mr. Gray released this information already.] - - - Wake up. Could it become really that worse? Possibly not, but anyone who assesses the code-quality of the ruby- source-codes know: The Ruby Language System needs a rework, immediately. - Stricter processing rules, especially for issue-tracking and coding - Clean, decoupled, strict modular and self-documenting code - Open project which invites for code-level contributions It's up to the professional companies around ruby to do something, to act. To assess the code-quality, to admit that there's a major problem, and to ensure that someone dives into the sources and refactors them, reworks them. Who can do such a rework? See a suggestion here, which contains a work-example based on the vm_method.c unit: http://dev.lazaridis.com/base/wiki/RubyRework . -- http://lazaridis.com