As a result, this could my first claim in potentially building an ultra high frequency trading platform with ultra low latency. Learn more on how I do it!!
I really hope you are able to join me on this exciting development,
P.S. Don’t be afraid to rach out if got any questions on this.
Here is a comment posted at one my Youtube videos:
You have few good points. I don’t know why Security is not an issue for you. I’m a Linux guy and I don’t agree on that. As you said building from Linux kernel is a complicated and we have limited number of linux engineers. Also, we do not need all system calls of Linux kernel either. Tweak the kernel too. Just to add, we could even tweak the network packet, cut of few headers in the packet and improve the performance by a fraction of millisecond.
My respons at this current time:
Thanks for your excellent points. I never thought of these for a Linux kernal but it makes total sense. As for security, Windows is always a concern for security but it seems at this point I need to focus on strategy side vs the platform or or OS right now.
Tata Communications has now launched global ultra low fast latency network for potential HFT
Tata Communications has just announced the launch of its low latency network service. The announcement has recieved a significant amount of press coverage.I would be very interested to get views from this group on the announcement. I’ve attached a link below for further details but by all means feel free to contact me directly should you have any questions.
PDF IBM looks solid with Websphere, MQ, Java, Linux, and hardware for ultra <1 ms low latency solutions
This PDF makes IBM look like a solid choice for low latency through a tweaked Java and Websphere solution. This is exactly what a lot of the big banks maybe using for some of their solutions. Just saying.
Nvidia GPU Geforce 8+ is next generation for ultra high performance trading
I have heard many people within the programming community talk about the performance boosts these graphic cards can give. Nvidia GPU (graphic processing units) is leading the way for algo based trading developers and programmers to get extra performance for their applications.
As I looked up Nvidia’s developer documents, I was surprised to learn about application boosts of anywhere from ten percent to well over two hundred percent. All of these application need to be written in C or C++ take full advantage of Nvidia’s application programming interface (API).
The other thing about this Nvidia API is that is can be supported on all major operating system platforms including Linux, Windows, and Mac OS/X. If you ask me, these graphic cards are for extreme gamers but the exciting thing is they are architected for other uses. This of course includes developers who do high performance trading with their custom applications. As said, most of these applications are developed in C or C++. As a result, you can take full advantage of what the Nvidia core API offers.
I am reading a decent set of reviews on this class of graphic cards so they are worth pursuing on affordable level. Other servers like FPGAs could cost multi thousands where a few of these graphic cards may cost anywhere from two hundred dollars to six hundred dollars each. This is a very exciting option indeed.
I learned about this technology in two areas. Some of the really nerdy guys in a Ruby on Rails group I belong to as well as a few press releases from our friends at Marketcetera. Once again, their architecture and design enable their own platform to take full advantage of the Nvidia graphic processing units. You might want to check them out to for your next generation of ultra high performance trading system.