New owner/maintainer for CSVChart Drupal module

Back in ~2008, I had created a small Drupal module that used CustomFilter and Google Chart Tools: Image Charts, called CSV Chart. It was a simple module that took embedded CSV data and presented them as Google charts. It was very handy for what I wanted -and still want- to do, present benchmarks for my work on Altivec (and now in general SIMD).

However, I don't really code in PHP anymore -haven't coded PHP since 2010- so the module was left to bitrot. I did some minor adjustment to run with the current D7-based site, but that was it.

Thankfully, others found the module useful to use and adopt and the result has been that in the last days I transfered ownership of the CSV Chart to Pierre Vriens, so that he can continue development and maintenance of this tiny module. I would like to publicly thank Pierre for his work!

And that's the beauty of Free Software!

SIMD on javascript, MHO

Just read the Mozilla announcement on SIMD.js and I can say I got mixed feelings about this.

I don't really comment other news/blogs/announcements, but this is an exception.

On one hand, I definitely welcome more SIMD use everywhere, being a SIMD advocate and enthusiast for many years (since 2004 actually). So seeing more of it and by someone such as Mozilla, that's even better! On the other hand, wait, that's SIMD in Javascript?!? Really? Why? As if we already covered native coverage of SIMD in every other part, including the browser itself (No browser uses SIMD extensively in its core, though that would prove to be of actual benefit, the only SIMD code I know of is in the media playback code, which is usually some external library like ffmpeg/x264/etc that already has SIMD optimized parts anyway.

So, instead of using resources to optimize the core browser with SIMD -I'm sure there will be plenty of opportunities in the codebase for such optimizations- so that every web application, including Javascript ones will be optimized, or even the JS JITs themselves, yet Mozilla wants to push the effort to the web app developer to use their SIMD.js to do the equivalent of what SIMD coders have been doing to native apps for a long time now, only for JS apps.

Ok, so what's the gain? I read the PDF presentation that shows mandebrot.js to go from 9 FPS to 37FPS using SIMD.js. Admittedly that's impressive. But it also proves that the whole buzz about lower energy footprint computers, power efficiency is just useless. Why is that? For comparison running Xaos (fractal/mandelbrot program) on my very low end PC (2-core Athlon X2, AM2 socket, so DDR2) gives me ~250FPS, and I'm not even sure it's using SSE at all (from a simple check it doesn't). Zooming is realtime and at full detail. In the same talk, there was a benchmark of LLVM Javascript being as fast as C++ or 1.5x the native running time. I admit haven't tried the tests listed, but the mandelbrot test was using asm.js and 9 is definetely not 250/1.5. But I guess I'm just picky.

So, the latest trend of moving everything to the browser and JS,means that instead of optimizing my apps to run great on native, instead of making stuff running faster on my 5Wt big.Little 8-core ARM SoC, I have to get a much more power-hungry CPU to see the same performance. I'm totally against that. I want my newer CPUs, who are more energy-efficient and faster to actually feel like THAT. I don't want to upgrade just to experience the performance of a 486 20 years ago!

The talk mentioned HTML5 (and hence javascript) overtaking all other platforms for application development everywhere, including the smartphones. I certainly hope that's not the case, and I know of many people who also don't feel that way. We're not buying the "Everything on the web/cloud" paradigm, but I guess we're just a minority.

I could go on for a long time, but I have an actual SIMD-related bug to fix, cheers.

Note: I used to have comments enabled on my blog, but moderating spam was too time consuming, even with CAPTCHAs, so I disabled them entirely, if anyone could suggest of a better method, I'd gladly take advice -have been thinking about disqus, not sure if it's actually a good solution).

Eigen NEON port extended to ARMv8!

Soon after the VSX port, and as promised I have completed the ARMv8 NEON (a.k.a. Advanced SIMD) port. Basically this extends support to 64-bit doubles and also provides faster alternatives to division as ARMv8 has builtin instructions for division both for 32-bit floats and 64-bit doubles. Preliminary benchmarks (bench_gemm):

SIMD optimizations, cont.

A friend of mine told me that I should advertise my passion and know-how about SIMD more, and I decided to follow his advice. Though I am terrible at marketing and even more at personal marketing, I've made an attempt to do just that, advertise the fact that I'm offering SIMD Optimization Services (with emphasis on PowerPC AltiVec/VMX/VSX, and ARM NEON, but I'm ok with SSE as well, the logic is pretty much the same, though the difference(s) are in the details). For this reason I'm offering a free evaluation of your performance critical code (open/closed, able to sign NDAs if needed) to let you know if it's worth optimizing it, what kind of a performance gain you would get and how much it would cost you to get that result.
You can read more here.

EfikaMX updated wheezy and jessie images available

A while ago, I promised to some people in powerdeveloper.org forum that I would provide bootable armhf images for wheezy but most importantly for jessie with an updated kernel. After a delay -I did have the images ready and working, but had to clean them up a bit- I decided to publish them here first.

So, here are the images:

http://freevec.org/files/efikamx-wheezy-armhf-20140921.img.xz (559MB)
http://freevec.org/files/efikamx-jessie-armhf-20140921.img.xz (635MB)

VSX port added to Eigen!

Being the SIMD fanatic that I am, a few years ago I did the PowerPC Altivec and ARM NEON port for the Eigen linear algebra library, one of the best and most popular libraries -and most ported.

Recently I thought it would be a good idea to extend both ports to 64-bit, and it would also help me with the SIMD book, using VSX in one case and ARMv8 NEON (or Advanced SIMD as ARM likes to call it) in the latter. ARMv8 hardware is a bit scarce at the moment, so I thought I'd start with VSX. Being in Debian, I have access to a number of porterboxes in several architectures, and luckily one of those was a Power7 (with VSX) running ppc64. So I started the porting -or rather extending the code- to use VSX in the 64-bit doubles case. Unluckily, I could not test anything because Debian kernels do not have VSX enabled in wheezy -which is what the porterbox is running and enabling it is a non-option(#758620). So, running VSX code would turn out to be quite hard.

SIMD book, "Sponsored by ARM"!

Ok, took a while but I got the final word about this and can announce that the sponsor who donated 500 EUR to the Indiegogo campaign was ARM itself! I have to thank my friends at ARM@Cambridge and especially Dr Monika Biddulph, General Manager, Partner Enablement Group at ARM. When the book goes to print you can be sure it will include "Sponsored by ARM" somewhere! :)

Also a friendly reminder that even if the campaign is over, I still welcome the support in the form of preorders/sponsorships.

SIMD book, update 3! Addition/Subtraction mostly finished

Finally. Apologies for the delay, but it's been a busy month. This time I will hold true to my word and upload a PDF for people to see (attached to this page).

So, what's new? Here is a list of things done:

* Finished ALL addition/subtraction related instructions for all engines and major derivatives (SSE*/AVX, VMX/VSX, NEON/armv8 NEON). With diagrams (these were the reasons it has taken so long).
* Reorganized the structure (split the book into Parts I/II, the instruction index will be in Part II, Part I will carry the design analysis of each SIMD engine.
* Added an TOC/index.
* So far, with just Addition/Subtraction Chapter and the rest empty sections, it has reached 175 pages (B5, again I'm not fixed on the size, it might actually change)! My estimate is that the whole book will surpass 800 pages with everything included.

TODO:

SIMD book, second update!

From the Indiegogo page:

https://www.indiegogo.com/projects/simd-engines-comparative-reference-guide/x/4966960#activity

  • Added titlepage (simple, but will do the job)
  • Reorganized ALL instructions to include both unsigned/signed in the same entity
  • Added Saturated Addition, Subtraction and Saturated Subtraction
  • Added ARMv8 NEON instructions taken from ARM infocenter draft
  • Fixed some instructions (added 64-bit arithmetic for NEON)
  • Added some special addition/subtraction, like add/sub with carry(vmx/vsx), addsub(SSE3/AVX)
  • Added some in-vector sum additions, sum reductions but no descriptions yet
  • Added diagram for 8-bit addition/subtraction (still need lots more).
  • Removed VMX128, couldn't find enough information, an email to some IBM toolchain developers was left unanswered, so I guess noone really will really care if that engine is left out, if enough people insist on it, please also be kind enough to provide some documentation on it.

SIMD book, first draft published!

Check activity here:

https://www.indiegogo.com/projects/simd-engines-comparative-reference-guide/x/4966960#activity

From the update:

Ok, I've been busy the past days, I started writing the book (using LaTeX :), and I'd like to say that progress has been good. I fixed the current list of SIMD engines that I'm going to include and it's a long one:

Pages

Subscribe to Front page feed