I think of Gao frequently, at least briefly multiple times a month. As I have indicated over the last four years I started my many-years experience with him at MIT in the early 1980's and shared multiple government-funded parallel computing projects with him up until near his sad passing.  The world, our field, and our lives were better for knowing him.

My first memory of Professor Gao dates from the 1980s when he came with his advisor Jack Dennis to visit Lawrence Livermore National Laboratory where I was at the time working in the Sisal functional language project. Our two groups had very lively discussions about dataflow and functional languages, which was great fun. I learned then that Gao was going to teach at McGill University in Montreal, so I decided to go there for doctoral study. It was an excellent decision on my part. Gao was so kind and generous. Very popular with his grad students, who also remain pleasantly etched in my memory. Gao was an enthusiastic teacher. He pushed me to improve my technical writing by encouraging me to explain more clearly the intuitions underlying some rather complicated mathematics. Also, wicked good at  ping-pong. Our friendship and collaborations continued after my graduation. His family, Peggy and Nick, were also very kind to me. Great people,  great memories.

My friendship with Gao spans more than 40 years with casual somewhat distant relationship while at MIT in the early 80's but a growing and substantial close association in the many years following until his passing some 3 years ago. His memorial service was like yesterday and I think of him frequently when memories are triggered by usually minor instances.  We collaborated on multiple projects throughout the years and I can say that I truly miss him as a friend and colleague. 

Happy Birthday Professor Gao! I am sure you are enjoying a wonderful glass of red wine in Heaven - like the many reds we drank together!

Another small story, but one that was important to Gao (and me) at the time. It was the summer and the ICPP meeting at the Pheasant Run resort outside of Chicago. I presume it was in the late 1980s. The meeting had just concluded and he and I decided to go to lunch together a couple of miles down the main road; it was my treat. In the future, he and Peggy almost always picked up the tab for decades. Having just attended the meeting our minds were on the field, its state of the art, and its trends. Our conversation turned to those trends. It was a depressing conversation and one that cemented the likely realities versus our wishful thinking. The 80's had been the golden age of experimental high speed computing; it was a time of originality at creativity, of excitement and future possibilities. Gao and I were part of the national, indeed international, community dreaming about what could become the future and how, if we were lucky, we would contribute. But the dark clouds of VLSI economics were on the horizon and the high speed computing market was about to metamorphize into scalable microprocessor based MPPs and commodity clusters (Linux in the case of Beowulf; but that was 5 years in the future). As anyone reading this knows, Gao was a world expert in the maturing field of Dataflow. I had other vague ideas of custom system architectures. The conversation was more like a memorial for new forms of computers; it was beginning to appear that maybe commercial microprocessor could eventually dominate parallel computing, not because they were best by some aesthetic we never adequately articulated but because of the potential economy of scale advantage with the then workstation market driving up the capabilities and driving down the per unit cost. This was a sobering conversation and it changed my professional perspective for ever. We were sharing a heart-felt moment when we realized our respective professional trajectories were, per force, about to experience a dramatic "singularity" and head off in different directions. The second factor that exposed our naivete was the role of course-grain parallelism that would come to dominate much of the next decade and more. Gao and I had largely focused on near-fine grain parallelism. A door had closed; another door had opened: Gao and I walked through it together. 

I have many many short memories of me and Gao. They are minor but fill in the cracks in the large wall of his life. This was at a hotel in Irvine California; probably in 1995. We had undertaken an NSF sponsored project somewhat under duress to conduct an architecture point design study towards Petaflops performance computing. This was what would become the first phase of the HTMT project. Put simply, we had proposed that multi-threading would be a method to mitigate the intrinsic latencies that would be exposed in a system comprising superconducting logic, optical networking, and holographic storage. Through a series of paper studies we had shown that this was a bad idea and would not work. I was delighted because I was glad to deliver a negative result from a project I did not want to do in the first place. That was the context of this memory. We sat in a nice hotel room overlooking a nice view of southern CA and talked about this. We were ready to give up when we started brain-storming about there being any other possibility. In doing so, we came up with"percolation", a form of message-driven computation: essentially moving work to data when appropriate rather than  always the other way around. Dataflow tokens, and active messages, as well as RPC were earlier primitive forms of this kind of asynchronous operation. But percolation was at a whole new level. We chatted about it for a couple of hours and realized that it would work. The results of this one afternoon's discussion, just the two of us, resulted in our HTMT (terrible name, my bad) project going forward into Phase 2 with multi-million dollar funding. We also engaged in a follow-on NSF project for queuing simulation of percolation to determine its optimization profile. Finally, today, I use "operons" in my current Active Memory Architecture research which is a variant of this original idea that Gao and I first conceived. 

Guy and Gao sharing a glass of Port wine — with Guy Tremblay and Prof. Gao

I never had Dr. Gao as a professor or as an advisor. However, in the many years that I spent going for my B.S., M.S., and PhD, Dr. Gao always asked how my progress was going even when I was having a very rough time at Delaware. He was one of the very few that genuinely cared, and I never forgot that. While I don't associate myself with Delaware anymore and have since changed careers, Dr. Gao was a bright spot indeed, and truly "old school" to the point that he cared about the research and the students, even those outside of his group like me. When I heard this news I was shocked since he was one of the legends not only in Evans Hall but also in his field. Condolences to the family on your loss - he will certainly be missed. 

We are so sorry for the loss of dear Professor Gao. I came to know him in 1992 when I arrived at McGill University as a Master student from Hunan, China. I was so eager to learn under his guidance. He was such a kind person who not only gave me academic guidance, but also helped me find financial aid. He had high expectations for his students, which pushed me to excel academically and lay the foundation for my future career. I will miss him dearly. Rest in peace.

Renhua Wen

听到高老师去世的消息的时候，非常震惊，因为就在1、2周以前，高老师还在频繁的和我们讨论书稿，那情景仿佛就在眼前。因此很长一段时间，都不能接受高老师离开我们的事实，直到刚才，在Zoom里看到追悼会现场，高老师安详的躺在棺木中，才觉得，高老师真的离我们而去了...

记得和高老师第一次见面，是在华中科技大学的学术活动中心高老师下榻的房间，和高老师聊了半小时，高老师就决定邀请我到CAPSL访问。那只是高老师无数次回国讲学中的一次，我也只是高老师提携的众多学生中的一个，但于我而言，却非常幸运，能认识高老师，并因此开始一段不同的人生。

高老师学术上取得了极高的成就，但更令人折服的，是高老师的人格魅力。高老师始终怀着一颗赤子之心，对人如是，对研究亦然。和高老师相处，如沐春风，常常让人觉得高老师不仅是老师，更像是和蔼可亲的长辈和朋友。还记得在Delaware的时候，有时去高老师家里蹭饭，有次从冰箱拿出Peggy包好的饺子去煮，高老师和我兴致勃勃的讨论下冰冻的饺子，用不用等水烧开，Peggy觉得要等水开，高老师和我一致觉得冰柜的饺子不用等水开（顺便说，高老师家的自己做的酱配饺子非常好吃）。高老师平时的生活饮食就是如此简单，但美食是高老师喜欢的，闲暇时兴致勃勃的和我谈论回国时吃的各种美食，其中不乏红烧肉、土豆丝这样的家常菜，但高老师甘之如饴；正如高老师对数据流的研究，无论高潮低谷，外界如何评价，高老师始终如一的坚持数十年，并在最后的日子，还在全力推动出版数据流的教材。

讨论数据流书稿的时候，高老师谈起本来准备应我们邀请回国，但种种原因没有成行，感觉有点遗憾，我说等疫情过去，再接高老师过来，吃点好吃的，没想到，竟再没有机会了...

愿高老师安息！

Dr Gao was my mentor throughout undergrad and graduate school and his guidance gave me opportunity to succeed and I still think of his teachings when I work with students.

He will be missed.

Following keynote@HPCS 2004

What a great person. Prof. Gao not only had vision for his students and staff's career path but also took care of them opening his door for frequent get together with lots of good food. He and Mrs. Gao remembered each and every name including children.  Prof. Gao was excellent and quick in analysing people’s talent and inspire & involve them in research projects appropriately. He was kind enough to accept our (Parimala and I) invitation for a keynote at HPCS Canada held in Winnipeg in 2004. Our students got lot of inspiration from him in just one afternoon. He introduced me into computational finance ( emerging in late 90s), which became my career in computer science.  While at UDel we travelled many places together and many interesting incidents flash through my mind.  

• In each moment
• There was sunlight
• Passing
• Amongst the leaves

• And when night
• Had fallen
• Only emptiness
• Could be seen

• But tomorrow repeating
• The luminescence
• Had shown
• In ourselves

• Forever
• As one is many
• The tree
• Branches

CAPSL Group Photo. Published on Multiple SuperComputing Conference Posters