[11:58:45 AM] Bryan Downing: there is this http://www.enterprisetech.com/2014/10/02/ibm-accelerates-power8-clusters-gpus-fpgas-flash/
[11:58:52 AM] Bryan Downing: the closest i could find
[11:59:02 AM] Bryan Downing: fpga is complicated but redis is easy to setup and maintain
[11:59:18 AM] Bryan Downing: it is free to so i think the performance could be adequate for 100% redis
Question is: could this FPGA/GPU cluster faster if you choose to be 100% Redis? Or at least is there the potential?
Please let me know by getting in touch via my contact form lower right. Thanks
I thought you might be interested in this journal from Xilinx.
My view on this is simple. I need to develop trading ideas rapidly. I will never have time to hand code the monstrosity of something like OpenCL. Unless there is an intelligent simplified layer built on top of it, I still won’t use it due to it usually being highly constricted. I look at Python as an example on how there are so many options to do something. Same with Java. R unifies everything nice with RStudio but it still has its problems. This why I use my Simulink for rapid trading idea implementation. I will NOT have time to wait it out for some dude to fix a problem in his open source library. Plus we are not even talking about the complex multi threading component within OpenCL.. Maybe I am missing something but I still find Simulink with code generation is the best option I got for potential FPGA deployment.
I could always be wrong based on something I don’t know but this consideration is a long way off!
Someone ask me about my view of how I use the term high-frequency trading. It seems to be a hyped up marketing term that is kind of trendy to people in to learning about one subset of algorithmic-based trading. To be honest, the only path is through the Matlab system ecosystem which includes Simulink. As we all like to see credibility of actual investment banks using it, there is a video out there of a MathWorks document to educate the power and who uses it.
As you know the last week, I’ve posted many models for my Quant elite members. Here’s the latest list of what I posted:
Simulink call put parity options trading model with C or C++ code Matlab Simulink future trading model on basis Initial Equations for Options and Futures trading Raw Equations for Options and Futures trading
Trading Workflow from Research Paper to Matlab to C C++ or FPGA HDL for ultra lowest latency HFT
I have just put up two new postings on how to trade options and futures. There's a definite relationship between these two asset classes but I only focus on options on futures which are non-deliverable.
Here are my two latest postings:
This online course is Free and is a very old-school approach with a fundamental view. It still boggles my mind on how people need to complicate their trading for mediocre results. Do you really want to
focus exclusively on technical analysis based on past pricing behavior? All these techniques learned can easily be applied into a high-frequency trading setting. Guess what my next phase will be?
I will be starting this process sometime next week. Once finished, this will be my biggest milestone yet. Boy am I excited to release it to the world if it comes together. I will also be focusing on all the trading models coming out of this futures and options course which will be implemented using this methodology.
Join my Quant Elite membership as this take shape next week. I will be dedicating an entire Meetup around this. Not only that I'll be focusing a video playlist on my YouTube channel on this as well.
PS. If you are living in Toronto please visit our attendance session coming this Monday night.