It'll be a shame if they don't dump a chunk of that money on the Mill. However there may very well be some string attached unsuitable to Ivan Godard et al.
U.S. Invests $258M in Supercomputing Race with China
11–20 of 82 posts
Re: U.S. Invests $258M in Supercomputing Race with China
#12Does betteridge's law apply to conference talks?
http://sc14.supercomputing.org/schedule/event_detail-evid=pa...
Re: U.S. Invests $258M in Supercomputing Race with China
#13That's like 1/5 the cost of one F35 Flying Turkey?
And a mere 2% the cost of the USS Gerald Ford (which would probably be destroyed within 24 hours of active war with China)
Re: U.S. Invests $258M in Supercomputing Race with China
#14It'll be a shame if they don't dump a chunk of that money on the Mill. However there may very well be some string attached unsuitable to Ivan Godard et al.
I wonder if it being completely open source would have an impact on that.
Re: U.S. Invests $258M in Supercomputing Race with China
#15Re: U.S. Invests $258M in Supercomputing Race with China
#16It'll be a shame if they don't dump a chunk of that money on the Mill. However there may very well be some string attached unsuitable to Ivan Godard et al.
The Mill isn't really very well suited for supercomputing workloads. It tries to do well on poorly written C++ code (aka "general purpose code") that OoO machines have historically don well on. By contrast, VLIW machines are perfectly suitable for supercomputing, and indeed, I think that Itanium did well in the HPC world until it was discontinued.
Something like the Rex Computing Neo would be more suitable for this.
Re: U.S. Invests $258M in Supercomputing Race with China
#17This is chump change, we should have invested $2.58 billion or more. China isn't taking half measures in their state's science and technology investments. They realize we're in a second cold war, just one where the two combatants have more economic interdependence. They are making investments like the ones that helped us win the last war.
It's not like the DoE are the only ones with expensive computers: The NSA has their underground lair.
There's a commercial need for expensive distributed processing now, with Google/AWS/Azure spending a lot on video cards. The ML craze may make government spending on supercomputers less important, especially if people prefer to distribute calculations on many interchangeable computers.
I wouldn't mind a few NSF or DoD grants for novel computing architectures, but I'm not sure I buy that having massive supercomputers for economic forecasting is "strategically important for nations".
Re: U.S. Invests $258M in Supercomputing Race with China
#18Here's the same content at AnandTech (w/o paywall):
http://www.anandtech.com/show/11547/us-dept-of-energy-awards...
Re: U.S. Invests $258M in Supercomputing Race with China
#19This is chump change, we should have invested $2.58 billion or more. China isn't taking half measures in their state's science and technology investments. They realize we're in a second cold war, just one where the two combatants have more economic interdependence. They are making investments like the ones that helped us win the last war.
> They realize we're in a second cold war...
For God's sake, don't start another cold war. This is an established incumbent, who has monopolized world power for decades, feeling threatened by a new kid in town. Any progress China has made has been labeled as threat, cold war, or dismissed as useless, or blamed on China's stealing US secrets. As an instance, China bought many Intel chips to create supercomputers. Then the U.S. government banned Intel from selling high end chips to China[1]. China had to create homemade chips for world's fastest supercomputer[2].
[1] http://wccftech.com/us-government-bans-intel-nvidia-amd-chip... [2] http://www.pcworld.com/article/3086107/hardware/chinas-secre...
Re: U.S. Invests $258M in Supercomputing Race with China
#20That's like 1/5 the cost of one F35 Flying Turkey?
And a mere 2% the cost of the USS Gerald Ford (which would probably be destroyed within 24 hours of active war with China)
Your premise is comical. Any war that involves destroying a US carrier and killing thousands of soldiers in doing so, will move to a nuclear stand-off almost immediately thereafter. China and Russia have always understood that. For China, the ability to sink a US carrier only matters in instances of absolute war, where hundreds of millions of people will die. Being able to sink a carrier if you're a major military competitor to the US, is low on your list of actual concerns in the end (the threat is worth drastically more than the actual ability; if you use the ability and succeed, it's WW3). Carriers are for force projection, some intimidation, and fighting nations not named China or Russia.