That's an NVIDIA DGX-2 supercomputer, $399,000, off the shelf.
More after the jump.
From CTRM Center (Commodity Trade Risk Management):
Automated trading in European intraday power and gas markets is just the tip of the iceberg. According to a CFTC report published last year, there have been substantial increases in automated trading across almost all commodities. However, beyond simply automation of trading is the use of machine learning and Artificial Intelligence to actually decide when and what to trade. This too is on the rise as evidenced by the interest in solutions offered on the software market. To catch up with the latest trends, I often reach out to recruiters. I spoke with Carl Vellenoweth of Commoditas Partners and James Richmond of MethodSearch – who better to know what the hot trends are than those tasked with hiring people to perform those activities? To get some feedback on the implications of this trend, I spoke with Tim Rogers of Contigo and Aviv Handler of ETR Advisory as well.
I initially spoke with Carl Vellenoweth, MD Commoditas Partners and an ad hoc contributor to CTRMCenter, to get his views and he was very clear – “It’s all about data – there is more data, it’s cheaper and more accessible…” he told me. “Put another way, ten years ago, we hired traders based on their knowledge of trading, their contacts and record now we ask how are your programming skills!” In his opinion, traders are now quants and quants are data scientists and he fully anticipates that the entire front office will become automated in the future.But what are the implications of this trend? Well, firstly perhaps, the days of the trader as a ‘demigod’ earning big bonus payments might well be numbered. The new breed of trader/programmer is more economical to hire it seems and the playing field has been levelled a bit. Having skills in Python and other languages might be more relevant than knowing a bunch of other traders in the industry? Of course, there must be experts on the underlying commodities, instruments and markets but in the end, these may act more like trader analysts working with programmer/traders to build intelligent robots and stay abreast of the other robots out there. While it may read like a cheap Sci-Fi novel, it really does seem that things are headed in that direction.I also asked James Richmond, MD of MethodSearch his thoughts,“the main commodity futures I see being traded algorithmically are power + gas, certain agri commodities (wheat, cattle, corn, Soybeans) and some metals. In these markets with relatively simple / less long-term trading strategies, I see a rise in automation. Oil I am not so sure, but I guess we need to look at what comes of VAKT and Komgo. Within oil, the data science projects around vessel tracking, refinery optimisation, thermal imaging should enhance the capability of the trader and enable better decision making.With this in mind I see the ‘traditional trader’ declining within these asset classes as a variety of machine learning, AI techniques as well as Bot’s automate some / all of the trading process. The team that drives such projects is combination of right IT and Quant Analysts that develop trading strategies.The role of the ‘traditional trader’ could morph into that of a ‘product owner’ within such projects. Of course, some of the ‘best of the best’ traders that hold dominant positions in their respective markets may remain successful, but will need to embrace technology and adapt in the long term to avoid getting left behind. Many of the better ‘bulge bracket’ banks have successfully replaced traders with programmers…What I am not saying is the role of the trader will vanish from the face of the earth. This is definitely not the case. More that the role of the trader will change, headcount will reduce over time and the combination of traders combined with assisted / automated trading will be good for profitability and also enhance the role of the trader whilst diminishing some of the reliability of the traders’ gut feeling.”The move to automated trading in commodity markets has happened quickly with many exchanges reporting rapid increases in this type of trading. The impact of automated trading has also been blamed for ‘disconnecting’ markets from fundamentals in coffee, sugar and cocoa to name a few commodities impacted, driving volatility and unexpected price movements. A number of high profile hedge fund managers exiting the business have even gone so far as to blame automated trading in part, for the exit...MORE
The DGX-2 is designed to handle massive datasets with the first deliveries to the U.S. National Laboratories:
Oak Ridge National Laboratory, in Oak Ridge, Tennessee;
Pacific Northwest Laboratory, in Richland, Washington;
Sandia National Laboratories, in Albuquerque, New Mexico.
The work of the labs involves data manipulation in complex/complex—chaotic systems; in fusion research, climate simulation and human genomics.
The 'puters were not built for trading but with a few tweaks you've got deep learning/borderline autonomous AI High Performance Computing that teaches itself algos and can feed into low latency systems.
We've had a few posts on the beast:
April 3, 2018
April 4, 2014
June 4, 2018
This is a pretty big deal.
High Performance Computing has been a separate and distinct class of machines since the University of Manchester's Atlas computer went online in 1962.
And now NVIDIA is blurring the line in a few different ways....
High Performance Computing has been a separate and distinct class of machines since the University of Manchester's Atlas computer went online in 1962.
And now NVIDIA is blurring the line in a few different ways....
September 4, 2018
NVIDIA’s DGX-2: Sixteen Tesla V100s, 30 TB of NVMe, only $400KOr, if you are handy with tools there's this approach from May 2015's "Nvidia Wants to Be the Brains Of Your Autonomous Car (NVDA)":
We've mentioned, usually in the context of the Top 500* fastest supercomputers, that:
Long time readers know we have a serious interest in screaming fast computers and try to get to the Top500 list a couple times a year. Here is a computer that was at the top of that list, the fastest computer in the world just four years ago. And it's being shut down.That was from a 2013 post.
Technology changes pretty fast.
Among the fastest processors in the business are the one's originally developed for video games and known as Graphics Processing Units or GPU's. Since Nvidia released their Tesla hardware in 2008 hobbyists (and others) have used GPU's to build personal supercomputers.
Here's Nvidias Build your Own page.
Or have your tech guy build one for you.
