AI for Drug Discovery
Good approach. This is the way forward for future.
12 publicly visible posts • joined 6 May 2013
Even over a decade after the availability, GPU clusters and in the last 4 years, Xeon -Phi clusters are still to take off on mass usage. Very few applications are available. Also even those ones that are running have very low real application performance. The cost advantage of a Accelerator provided cluster Vs the large CPU cluster is very minimal considering the real time application performances.
Very interesting product. Can be of great value to Banking,, Pahrama and Research industry. This will enable the companies to achieve Geographical De-risking of their IT setup . It will also substantially lower the IT costs as companies with large data requirement can now opt for mid range / entry level storage boxes that generally have low scalability.
The market seems to be and will need to move towards applications availability and scaling rather than topping TF/ PF. in the real world except for some one off applications, there are not many practical research benefits that are accruing with the accelerator based systems. Also if we discount the benefits of reduction in foot print , the accelerator based systems are not really cost saving. Wider Application porting, scaling and availability is the crucial step that needs to be crossed before the market really makes the next big jump.
IBM Bluegene has not been a commercial success. Primarily due to the fact that they are peiced higher. Also the number of applciations that have been made availble on Blue gene is very very low. Except for some niche and IBM committed clients, there are hardly any takers for that. HP Moonshot is also likelyt o face a similar fate. While it sounds good on the space utlization front, many of the X86 ported applications may not perfoorm well on the Atom of ARM processors. HP shoudl explore the design option for putting multiple pure X86 processor blade cards in a chasis.
While in terms of the overall bussines volumes, it may not make a difference to IBM, it will have an long term impact. There is an increasing trend towards open systems, particularly in the Non-US markets. Non US markets are going to be global drive engines in the HPC segment. Not having X-86 portfolio will sure hurt IBM. Also retaining a part of HPC bussiness may not work in the long term as IBM's clout over Intel to extract best prices on X-86 configured systems wlll drastically fall and hence they will turn out to be loosers in the competetive sapce. IBM has to pick one boat to sail