Upgraded US Supercomputers Claim top two Spots on Top500 List:
China has more of the 500 fastest machines on the planet than ever, and the US hits an all-time low.
The US now can claim the top two machines on a list of the 500 fastest supercomputers, as Sierra, an IBM machine for nuclear weapons research at Lawrence Livermore National Laboratory, edged out a Chinese system that last year was the very fastest.
The Top500 list ranks supercomputers based on how quickly they perform a mathematical calculation test called Linpack. The top machine, IBM's Summit at Oak Ridge National Laboratory, had claimed the No. 1 spot in June with a speed of 122.3 quintillion mathematical operations per second, or 122.3 petaflops.
But an upgrade gave it a score of 143.5 petaflops on the newest list. To match that speed, each person on the planet would have to perform 19 million calculations per second. Sierra got an upgrade, too, boosting its performance from 71.6 petaflops to 94.6 petaflops and lifting it from third place to second.
The top machine on the first TOP500 list in June of 1993 was a Thinking Machines Corporation CM-5/1024 with 1,024 cores and was rated at Rpeak of 131.0 GFlop/s and Rmax of 59.7 GFlop/s. The least performant system is listed at the bottom of Page 5 of he list was a C3840 Made by Sharp of Japan which had 4 cores and had RPeak and RMax scores of 0.5 and 0.4 GFlop/s respectively. The fasted Cray Research machine in 1993 rated 9th place at 15.2/13.7 GFlop/s for RPeak and RMax.
Where on that first list would today's smartphones land?
More at Top500 and The Register.
Summit and Sierra are siblings, each using IBM Power9 processors boosted by Nvidia Tesla V100 accelerator chips and connected with Mellanox high-speed Infiniband network connections. They're gargantuan machines made of row after row of refrigerator-size computing cabinets. Summit has 2.4 million processor cores and Sierra has 1.6 million.
[...] A total of 227 of the Top500 machines are in China, compared with an all-time low of 109 for the US. The November list is the 52nd one released by a collection of academic researchers who compile it twice yearly for supercomputing conferences.
Linpack is only one speed test, though, and the Top500 has another designed to capture a broader range of performance abilities, the High-Performance Conjugate Gradient (HPCG) benchmark. On it, Summit and Sierra are head and shoulders above competing supercomputers.
Related Stories
TOP500 Becomes a Petaflop Club for Supercomputers
The 53rd edition of the TOP500 marks a milestone in the 26-year history of the list. For the first time, all 500 systems deliver a petaflop or more on the High Performance Linpack (HPL) benchmark, with the entry level to the list now at 1.022 petaflops.
The top of the list remains largely unchanged, with only two new entries in the top 10, one of which was an existing system that was upgraded with additional capacity.
Two IBM-built supercomputers, Summit and Sierra, installed at the Department of Energy's Oak Ridge National Laboratory (ORNL) in Tennessee and Lawrence Livermore National Laboratory in California, respectively, retain the first two positions on the list. Both derive their computational power from Power 9 CPUs and NVIDIA V100 GPUs. The Summit system slightly improved its HPL result from six months ago, delivering a record 148.6 petaflops, while the number two Sierra system remains unchanged at 94.6 petaflops.
The #100 system is at 2.3957 petaflops, up from 1.9661 petaflops in November 2018. The #500 system was at 0.8748 petaflops in November.
Complete list. The leading Green500 system is still "Shoubu system B" at 17.604 gigaflops per Watt.
Previously: Latest Top500 List: Upgraded US Supercomputers Claim Top Two Spots; China has Most Systems
(Score: -1, Troll) by Anonymous Coward on Tuesday November 13 2018, @10:12AM (9 children)
Are these used for anything besides (probably gigo) academic models we never hear about getting verified and spying on citizens?
(Score: -1, Troll) by Anonymous Coward on Tuesday November 13 2018, @10:34AM (3 children)
I found this cancer project but cannot find the actual grant: candle.cels.anl.gov.
The key question is whether they will have a real lockboxed holdout dataset consisting of the most recent data collected that they save until the end. This must then be used to check the performance of their model, because the cross validated, etc performance will be compromised. I have never seen an academic bio project manage to avoid leaking test dataset info into the model. Often they will even proudly describe in detail how they managed to do so.
(Score: 0) by Anonymous Coward on Tuesday November 13 2018, @10:57AM (2 children)
Well good luck on your billion dollar feel good projects if you think concerns like this are trolling. Driving people who worry about stuff like this out of bio is exactly why there is no cure for cancer in sight (they gave up and call it many diseases now).
Next depression, alzheimers, heart disease and everything else will transform into "manydiseases" too. In successful science things become simpler to understand as the underlying "laws" are figured out, we are witnessing the opposite.
(Score: 0) by Anonymous Coward on Tuesday November 13 2018, @11:10AM (1 child)
Yes, you are literally concern trolling. Academics have no need to respond to your vague and unfounded objections.
https://progressreport.cancer.gov/after/survival [cancer.gov]
(Score: 0) by Anonymous Coward on Tuesday November 13 2018, @11:18AM
Yep 5 year survival rates are increasing in populations being screen much more often. Thus the timer is started earlier in the disease process. Thanks for the perfect example of sloppy analysis and interpretation.
Also, the SEER poplation has changed a lot over time (age and location) so there are a lot of annoyances in trying to make that a usable timeseries. Dont get me wrong though, I love SEER, theres just to many questionable assumptions required for this type of analysis though.
(Score: 0) by Anonymous Coward on Tuesday November 13 2018, @10:58AM (1 child)
Mining bitcoin and deepfaking pr0n videos.
(Score: 0) by Anonymous Coward on Tuesday November 13 2018, @11:07AM
Ive been thinking about a deep fake kavanaugh orgy video. I mean labeled as such, I think it could sell.
(Score: 2) by bob_super on Tuesday November 13 2018, @11:58PM (2 children)
I've enjoyed a reliable multi-day forecast for the Santa Ana winds which kept the flames off my house this weekend.
Not having to evacuate was good. Giving early warning to the firefighters when the wind could shift was even better.
(Score: 0) by Anonymous Coward on Wednesday November 14 2018, @12:08AM (1 child)
Looks like they just use PCs for that. All the further links are dead:
https://www.predictiveservices.nifc.gov/NPSG/FMT_PredictiveServices.pdf [nifc.gov]
(Score: 2) by bob_super on Wednesday November 14 2018, @12:17AM
You're confusing tactical with strategic.
(Score: 2) by takyon on Tuesday November 13 2018, @10:43AM
June 2018, #500: 715.6 teraflops RMAX
Nov. 2018, #500: 874.1 teraflops RMAX
June 2018, #100: 1,703.3 teraflops RMAX
Nov. 2018, #100: 1,966.1 teraflops RMAX
On the Green500 list, efficiency of the #1 system is down following the addition of more cores, but the top 10 are higher:
https://www.top500.org/green500/lists/2018/06/ [top500.org]
https://www.top500.org/green500/lists/2018/11/ [top500.org]
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by MichaelDavidCrawford on Wednesday November 14 2018, @12:05AM
-r?
Espionage?
No, mining BitCoin.
Yes I Have No Bananas. [gofundme.com]
(Score: 0) by Anonymous Coward on Wednesday November 14 2018, @10:03AM
The article states a speed of 122.3 petaflops as a quintrillion operations per second. Peta = quadrillion