Knoxville Researcher Wins A.M. Turing Award

schwit1 writes: It's a few weeks old, but ...

A local computer scientist and professor at the University of Tennessee at Knoxville has been named an A.M. Turing Award winner by the Association for Computing Machinery. The Turing Award is often referred to as the "Nobel Prize of computer science." It carries a million dollar prize.

"Oh, it was a complete shock. I'm still recovering from it," Jack Dongarra told Knox News with a warm laugh. "It's nice to see the work being recognized in this way but it couldn't have happened without the support and contribution of many people over time." Chances are Dongarra's work has touched your life, even if you don't know it. If you've ever used a speech recognition program or looked at a weather forecast, you're using technology that relies on Dongarra's software libraries. Dongarra has held a joint appointment at the University of Tennessee and Oak Ridge National Laboratory since 1989. While he doesn't have a household name, his foundational work in computer science has undergirded the development of high-performance computers over the course of his 40-year career...

Dongarra developed software to allow computers to use multiple processors simultaneously, and this is basically how all computer systems work today. Your laptop has multiple processing cores and might have an additional graphics processing core. Many phones have multiple processing cores. "He's continually rethought how to exploit today's computer architectures and done so very effectively," said Nicholas Higham a Royal Society research professor of applied mathematics at the University of Manchester. "He's come up with ideas so that we can get the very best out of these machines." Dongarra also developed software that allowed computers with different hardware and operating systems to run in parallel, networking distant machines as a single computation device. This lets people make more powerful computers out of many smaller devices which helped develop cloud computing, running high-end applications over the internet. Most of Dongarra's work was published open-source through a project called Netlib.

Congratulations!

If only we had a cluster

[DryLand404]

Imagine a Beowulf cluster of Jack Dongarra's

Re:

[kaatochacha]

Man, beat me to it...

Dongarra and MPI

[Leading Edge Boomer]

True story: The late Ken Kennedy and his team at Rice University were working on compilers that automatically parallelized code, and needed a runtime library to target for managing parallel computations. This was for really difficult languages like Fortran and C with all kinds of obscure side effects, etc. He asked his Jack Dongarra for help by defining the library API and building a library of code. This became MPI. It was never intended to be visible to programmers, but it took off independently to become a major tool in manually programming parallel applications. Its explicit use by humans has been, IMHO, a setback in building really good parallel software that non-heroic programmers can understand.

Re:

[John_The_Geek]

First, not true. The initial version of MPI was designed by a very knowledgeable committee (effectively led by Jack) and building closely upon PVM, the previous "standard".

Second, you repeat the canard that because MPI requires skill and attention, that there must be some simpler alternative. Turns out large scale parallel computing is hard.

All these years (and many well-funded government programs) later, MPI is still the only method used at large scale. I do not exaggerate, well in excess of 99% of compute

Re:

[Leading Edge Boomer]

Believe what you will about history, but I was in the room when the story was told by Ken himself.

Re:

[John_The_Geek]

For better and worse, to me it isn't history. I was a PVM programmer and heavily involved in the transition to MPI. So this story simply doesn't make sense. As anyone from that era can readily attest.

Undergirded?

[Badoit]

Is that even a word?

Re:

[fibonacci8]

Yes [m-w.com]

Re:

[drainbramage]

Yes, it is a perfectly cromulent word.

Re:

[GoTeam]

Yes, it is a perfectly cromulent word.

Shazbot!

Re:

[Badoit]

I think the figurative sense is mainly American, which probably explains why I haven't seen it before.

Headline: "Uninteresting person wins award"

[Anonymous Coward]

If only the editors could, you know, edit, and put the important bits up front.

Instead of burying them somewhere invisible and try and make up for it with clickbaiting the headline.

In THIS house

[pr0t0]

While he doesn't have a household name...

• If you've ever used a speech recognition program or looked at a weather forecast, you're using technology that relies on Dongarra's software libraries
• Dongarra developed software to allow computers to use multiple processors simultaneously
• Dongarra also developed software that allowed computers with different hardware and operating systems to run in parallel, networking distant machines as a single computation device.
• Most of Dongarra's work was published open-source through a project called Netlib.

Mr. Dongarra, you're now a name in this household at least. I offer a heartfelt "thank you" for your contributions.

Re:In THIS house

[GoTeam]

So he's done nothing for Bitcoin or NFTs? Why is this story on /.?

Re:

[K. S. Kyosuke]

Dongarra developed software to allow computers to use multiple processors simultaneously

Ironically, if you're currently "[using] multiple processors simultaneously" in scientific computation, that's quite likely not thanks to Dongarra but rather thanks to whoever works on MKL or OpenBLAS. LINPACK was *not* designed for the kind of machine that most people have access to. Dongarra *did* however show the world how to design numerical software properly, it has to be said.

Oh, come on!

[OneHundredAndTen]

Jack Dongarra is not just 'a local computer scientist and professor at University of Tennessee' - the report makes it sound as though they are talking about an all but unknown figure who has all of sudden be propelled to the forefront. The truth is that he has been a very relevant figure in the computer science world for decades now.