F I F O
Freshly Instigated Frequent Occurrences
Dt. July 26th, 2010 : Beautiful day!! Best birthday present ever...paper got accepted in HiPC. Wait -> shall put up link to the paper soon.
Dt. May 27th, 2010 : Maybe I shouldn't even bother putting in dates.
Dt. April 29th, 2010 : Loads of incoherent news, ruminations and stories. Hope they see the light of canopus some day.
Dt. April 4th, 2010 : I think I might finally realise my dream to work in the overlap of Astrophysics and Computer Science. Thank you ESA and GSoC for at least offering the opportunity. Lost valuable time in a ceremony I had to attend back in Bhubaneshwar (which was fun though - shouldn't insult the little that comes my way :P), but now, I am back and all geared up to start new stuff and end old stuff. "Ahoy! On Board, Man!!" One more thing : Ha Ha at Abhinav for participating in Astro quiz '10 without me! You deserve to not come first :P.
Dt. March 26th, 2010 : One complete month, I see. Not that I haven't anything new going on in my life. Just that, I haven't had the time to think if it in fact was new. Have been thoroughly depressed with the M.S. Application results. As has been oft-said, "The ways of the Admission Committees are mysterious". I have so much in my mind that I have to implement, so much I have to discover, so much I have to invent. But that is not possible without proper guidance, as I have not gotten for the last 3 years. CRL has been good to me. I have found mentors who are really encouraging as they realise my talent and passion. But they won't be around for too long. Just the next 3 months. That is probably why, I am on a grey cell overdrive, trying to think of new, different things, and honestly, I am thoroughly enjoying it. "Enjoy till it lasts" This may as well be my last chance at true research, if anything is to be gathered from the trends shown in the M.S. admissions.
Dt. Feb 26th, 2010 : Brilliant play, "Massage" written by the famous Vijay Tendulkar and performed by Rakesh Bedi. Versatile indeed. It was really hilarious though could have been slightly more profound, though one dialogue was really profound. The translation goes like : "It is truth that we see in pitch darkness, the absolute truth. Light simply reveals all that is false.". End on this note.
Dt. Feb 20th, 2010 : Awesome trip to Malshej ghats and Shivneri fort. Breathtaking beauty and brain boiling heat. Lovely taste of local pop food centre : kadi vada and special milk and lemon juice with ginger. Later in the evening, hangout at shisha cafe at ABC farms - sophistication galore. Paid passers-by tribute to those who died in the blast at the nearby german bakery.
Dt. Feb 19th, 2010 : Visit to shreevatsa. Played and had fun with the orphaned kids. We must indeed consider ourselves lucky. :)
Dt. Feb 17th, 2010 : Aah! Code runs perfectly for preliminary tests. Now putting in really huge matrices and running it on Eka for accuracy testing. Yet to do proper speed testing as well.
Dt. Feb 15th, 2010 : My implementation has improved code performance, but still showing bugs for really large inputs. Need to investigate.
Dt. Feb 14th, 2010 : Terrorists blast right behind my home. Was enjoying in Hyderabad. :(
Dt. Feb 13th, 2010 : Visit to the grand city of Hyderabad. Airport : Impressed. Roads, traffic : Impressed. Food : Delightful. Places to visit : Limited, but worth at least 2 days. Hotel stay : Exquisite. Experience : Maybe shall stay in memory for a while at least.
Dt. Jan 30th, 2010 : First trekking adventure today. Sinhagad fort, Pune. Too tired to type. Took the tougher route. Last stretch was really dangerous, for me at least, as I took a path that wasn't. Stamina, endurance tested. Totally worth it!
Dt. Jan 29th, 2010 : Had our first quiz yesterday which I pretty much aced, for there was nothing even as much as 3*10^42 light years close to being challenging. Ran most of our codes on Eka and got a real feel of the bandwidth, latency values and computational power of the cluster. Then came the deciding moment, that is to test the performance of my implementation of the scatter algorithm. The results were indeed encouraging and are available on the Project Updates page.
Dt. Jan 27th, 2010: Its finally here!! Google Summer of Code 2010!. Gotta have myself in it this time around. My brain is nearly going to fission into two or sprout buds off every peripheral point if I don't get all those ideas in it out there. Any suggestions as to how to go about it? (n00bish, but truly so)
Dt.Jan 25th, 2010: Finally handed over CRL-tailor-knitted SPMV libraries that I can extract thread by thread and understand its underlying fabric structure(really, literally!). I then think of a better way to put these threads together, so that the fabric feels softer (though it is pretty soft already) and I get commended by the knitting(read computer science) community. Returning back from the alternative-hyper-reality, I have already thought of a scheme in which i really do not bother about the partitioning scheme of the sparse matrix itself. Instead, I split the vector it is multiplied with into a pair of, what I would hence call mirrored child vectors, so that the communication step is eliminated altogether. Yet to figure out the mathematics completely. What do you think? If there were a stupid'er' idea, I would not have lived through it. Heck, even death would stay away from such moron-<collective phenomenon suffix>. Current task is to search and fix some redundant data transfer.
Dt.Jan 20th, 2010: Eka was relatively cold in its welcome and this surprisingly did make me feel more of a fanboy. Yes, she is flexible and can be filled in by a lot more(shady interpretations were definitely intended). Yes, she is bloody damn efficient and is in the world's top 30. She needs a lot of air coolers because she gets too hot too easily and that is the only thing that cannot be scaled (well, the money cant scale either). Any suitors?
Dt.Jan 19th, 2010 : Going to visit eka suercomputer today. Only the second supercomputer I have seen after PARAM 10000, though far more superior. Taking up learning scala programming language. Looks interesting and probably is bleeding edge in the field of cloud computing and distributed systems.
Dt. Jan 12th, 2010 : Presently everything related to message passing and benchmarking codes, that is instructions per second, bandwidth, latency, effect of memory alignment, false sharing, scatter algorithms etc. Yes, these have been around for quite awhile, but since I am becoming aware/ coding them right now, they are hot topics to me. "What you see in the present is what exists for your existence. As and when they are removed from your observation field in their entirety, they most probably cease to exist." Inspired by SEP field by Douglas Adams - A concept used extensively and effectively in computer science and has evolved our brain to involve a polling check for SEP every cycle it runs.
