Tuesday, December 03, 2013
What Happened to the Top 500?
Top 500
The last six months have been slow for this blog as well as the Top 500 supercomputers. There was only one new computer added to the top 10. The total FLOPS calculated by all 500 supercomputers increased by 12%, the lowest since the list started.
Given that the last list showed a growing divide between the top of the list and the bottom of the list, the fact that there was so little movement at the top of the list gave an opportunity for convergence. However, despite 137 new supercomputers, the 500th supercomputer can still only calculate at 0.35% the speed of Tianhe-2, a meager increase.
Wednesday, June 19, 2013
Topping out at 34 petaFLOPS
Top 500
June's list of the top 500 supercomputers has yet another leader. It seems no one is able to hold the top spot for very long. The new leader is China's Tianhe-2, calculating 34 petaFLOPS. The average of top 10 computers is 10 petaFLOPS. Tianhe-2 is faster than all 500 computers from June 2010 combined.
Overall though, it was a pretty average update. There are 182 new computers on the list and their total calculating power increased by 38%, compared to an average of 36%. Intriguingly, there seems to be a growing divide between the top and the bottom computer. The 500th computer calculates at 0.3% of the speed of Tianhe-2, the biggest difference ever. In past lists, it's closer to 1%. Still even at the bottom of the list, it's faster than the top 500 supercomputers from November 2000 combined.
June's list of the top 500 supercomputers has yet another leader. It seems no one is able to hold the top spot for very long. The new leader is China's Tianhe-2, calculating 34 petaFLOPS. The average of top 10 computers is 10 petaFLOPS. Tianhe-2 is faster than all 500 computers from June 2010 combined.
Overall though, it was a pretty average update. There are 182 new computers on the list and their total calculating power increased by 38%, compared to an average of 36%. Intriguingly, there seems to be a growing divide between the top and the bottom computer. The 500th computer calculates at 0.3% of the speed of Tianhe-2, the biggest difference ever. In past lists, it's closer to 1%. Still even at the bottom of the list, it's faster than the top 500 supercomputers from November 2000 combined.
Wednesday, April 10, 2013
Eurovision, the Dubstep Year
Eurovision 2013
First Semi-Final: May 14th, 9pm Central Europe Time (3pm EST)
Second Semi-Final: May 16th, 9pm Central Europe Time (3pm EST)
Final: May 18th, 9pm Central Europe Time (3pm EST)
Eurovision 2013 is almost here! For those of you without TV or in America, it can be watched for free on their website ESC TV. Right now, only recaps of the national competitions and previous Eurovisions are up.
This year is not quite as fun as last year, but there's a couple of good gimmicks. Dubstep has made a big entrance into Eurovision. As far as I remember, there were no dubstep songs any previous year, but this year has a couple of fully dubstep songs, one half dubstep, and a bunch of other songs with unnecessary dubstep interludes. I'm not particularly happy about this, although a couple of these songs made my list.
Here are the 10 most interesting entries. Intriguingly, two of these videos involve men being kidnapped.
10. Malta - In a world full of dubstep, one country dares to bring a ukulele
9. Romania - Wait for it...
8. France - Blues, that's what Eurovision is missing
7. Sweden - He's just so happy!
6. Switzerland - Big band, small car
5. Hungary - Video shows lyrics, which is useful if you know Hungarian
4. Greece - Fun little protest song
3. Latvia - The keytar is arguably not the most ridiculous thing in this video
2. Montenegro - Montenegrin rap is surprisingly fun
1. Finland - It's like Jenna Maroney from 30 Rock somehow got onto Eurovision
First Semi-Final: May 14th, 9pm Central Europe Time (3pm EST)
Second Semi-Final: May 16th, 9pm Central Europe Time (3pm EST)
Final: May 18th, 9pm Central Europe Time (3pm EST)
Eurovision 2013 is almost here! For those of you without TV or in America, it can be watched for free on their website ESC TV. Right now, only recaps of the national competitions and previous Eurovisions are up.
This year is not quite as fun as last year, but there's a couple of good gimmicks. Dubstep has made a big entrance into Eurovision. As far as I remember, there were no dubstep songs any previous year, but this year has a couple of fully dubstep songs, one half dubstep, and a bunch of other songs with unnecessary dubstep interludes. I'm not particularly happy about this, although a couple of these songs made my list.
Here are the 10 most interesting entries. Intriguingly, two of these videos involve men being kidnapped.
10. Malta - In a world full of dubstep, one country dares to bring a ukulele
9. Romania - Wait for it...
8. France - Blues, that's what Eurovision is missing
7. Sweden - He's just so happy!
6. Switzerland - Big band, small car
5. Hungary - Video shows lyrics, which is useful if you know Hungarian
4. Greece - Fun little protest song
3. Latvia - The keytar is arguably not the most ridiculous thing in this video
2. Montenegro - Montenegrin rap is surprisingly fun
1. Finland - It's like Jenna Maroney from 30 Rock somehow got onto Eurovision
Friday, April 05, 2013
Eurovision Winning
Eurovision is coming up soon, so it's been on my mind. I've been curious for awhile which countries have been the most successful and which have been the least successful. Of course, one could just look at number of wins, but that is skewed because countries have competed for a different number of years and there was less competition prior to 1992. Therefore, I decided to look at the predicted number of wins a country should have based on how many times it competed and the likelihood of winning each years competition. The graph below shows the results.
Click to enlarge
The graph shows the difference in actual wins versus predicted wins. So, for example, based on the number of times Portugal has competed, it should have won 2.3 times by now, but it has not, so its score is -2.3. Denmark is the closest to where it should be, having won twice against a prediction of 1.99 wins. At the other end, Ireland's predicted wins are 2.3, the same as Portugal, but has actually won 7 times. Germany has the highest predicted wins of 3.1, having competed all but once.
The next three worse scores after Portugal are Belgium, Finland, and Austria. This is interesting because these are all countries that have won before, but have been in the competition for a very long time. Finland would be in last place if they hadn't won with Lordi in 2006. The second best score goes to Luxembourg which quit participating in 1993, just as the number of participants was increasing. In fact, with the exception of Sweden, the top 7 scores belong to countries which haven't won any of the last 14 competitions. Of course, 3 of those 7 won during the infamous 1969 contest when four countries tied and there was no tie-breaking rule.
While it would be nice to see Portugal bring its score up, they have decided not to compete this year due to financial reasons. So here's hoping that Belgium wins this year.
Click to enlarge
The graph shows the difference in actual wins versus predicted wins. So, for example, based on the number of times Portugal has competed, it should have won 2.3 times by now, but it has not, so its score is -2.3. Denmark is the closest to where it should be, having won twice against a prediction of 1.99 wins. At the other end, Ireland's predicted wins are 2.3, the same as Portugal, but has actually won 7 times. Germany has the highest predicted wins of 3.1, having competed all but once.
The next three worse scores after Portugal are Belgium, Finland, and Austria. This is interesting because these are all countries that have won before, but have been in the competition for a very long time. Finland would be in last place if they hadn't won with Lordi in 2006. The second best score goes to Luxembourg which quit participating in 1993, just as the number of participants was increasing. In fact, with the exception of Sweden, the top 7 scores belong to countries which haven't won any of the last 14 competitions. Of course, 3 of those 7 won during the infamous 1969 contest when four countries tied and there was no tie-breaking rule.
While it would be nice to see Portugal bring its score up, they have decided not to compete this year due to financial reasons. So here's hoping that Belgium wins this year.
Wednesday, March 13, 2013
Rare Movies 2
Nearly 4 years ago, I made a list of the 30 most obscure films I had seen (Rare Movies). This list was based loosely on how many people using the movie site Criticker had seen them. I figured by now I've seen a few more obscure films, or maybe films that had been obscure four years ago had gained a following, so it was time for an update. This time I'm just going to list the movies based on the number of rankings on Criticker (minus myself).
From most to least obscure:
0 - Polite People - Icelandic movie I saw while flying on Icelandair.
1 - Taarka - Estonian movie about the Seto people.
3 - The Last Relic - Medieval film from Soviet Estonia.
5 - Names in Marble - Film about Estonia's fight for independence during WWI.
10 - Passione: A Musical Adventure - Musical documentary about Naples.
10 - Ulzhan - Boring French movie.
12 - Tibet: Cry of the Snow Lion - Documentary about the Chinese occupation of Tibet.
17 - Reckless Kelly - Has Hugo Weaving kicking a wombat.
23 - The Baker - Cute British movie I saw on a flight.
25 - The Celestine Prophecy - I have no memory of this movie.
26 - Burn Notice: The Fall of Sam Axe - Do made-for-TV movies count?
28 - Benji the Hunted - Traumatic, would not recommend.
30 - The Big Tease - Cute Craig Ferguson film.
32 - Krabat - German fantasy film.
34 - Silent Star - Saw on Mystery Science Theater 3000.
38 - Days and Nights in the Forest - Cool old Indian film.
39 - April - By Italian Woody Allen.
41 - The King and the Clown - Historical drama from Korea.
45 - Animals Are Beautiful People - Disney nature "documentary".
46 - Hellbound - Chuck Norris film.
52 - Felix the Cat: The Movie - Absurdity.
57 - How to Succeed in Business Without Really Trying - Fun old musical.
62 - The Princess Blade - Emo Ninjas.
64 - Farce of the Penguins - Failure of a parody.
72 - Undisputed III: Redemption - MMA film seen at the behest of a Ukrainian boy.
74 - Undisputed II: Last Man Standing - MMA film seen at the behest of a Ukrainian boy.
75 - Volunteers - Movie about Peace Corps with Tom Hanks.
80 - Laserblast - Saw on Mystery Science Theater 3000.
85 - Things to Come - Really old sci-fi film.
96 - So Close - Chinese action film.
Apparently I've been watching a lot more obscure movies as 20 of these were not on my previous list. Otherwise, there doesn't seem there was much movement. All the other movies on the previous list seem to have remained fairly obscure. Of these films that I think more people should see, I say Names in Marble, Reckless Kelly, The Baker, The King and the Clown, and Animals Are Beautiful People.
From most to least obscure:
0 - Polite People - Icelandic movie I saw while flying on Icelandair.
1 - Taarka - Estonian movie about the Seto people.
3 - The Last Relic - Medieval film from Soviet Estonia.
5 - Names in Marble - Film about Estonia's fight for independence during WWI.
10 - Passione: A Musical Adventure - Musical documentary about Naples.
10 - Ulzhan - Boring French movie.
12 - Tibet: Cry of the Snow Lion - Documentary about the Chinese occupation of Tibet.
17 - Reckless Kelly - Has Hugo Weaving kicking a wombat.
23 - The Baker - Cute British movie I saw on a flight.
25 - The Celestine Prophecy - I have no memory of this movie.
26 - Burn Notice: The Fall of Sam Axe - Do made-for-TV movies count?
28 - Benji the Hunted - Traumatic, would not recommend.
30 - The Big Tease - Cute Craig Ferguson film.
32 - Krabat - German fantasy film.
34 - Silent Star - Saw on Mystery Science Theater 3000.
38 - Days and Nights in the Forest - Cool old Indian film.
39 - April - By Italian Woody Allen.
41 - The King and the Clown - Historical drama from Korea.
45 - Animals Are Beautiful People - Disney nature "documentary".
46 - Hellbound - Chuck Norris film.
52 - Felix the Cat: The Movie - Absurdity.
57 - How to Succeed in Business Without Really Trying - Fun old musical.
62 - The Princess Blade - Emo Ninjas.
64 - Farce of the Penguins - Failure of a parody.
72 - Undisputed III: Redemption - MMA film seen at the behest of a Ukrainian boy.
74 - Undisputed II: Last Man Standing - MMA film seen at the behest of a Ukrainian boy.
75 - Volunteers - Movie about Peace Corps with Tom Hanks.
80 - Laserblast - Saw on Mystery Science Theater 3000.
85 - Things to Come - Really old sci-fi film.
96 - So Close - Chinese action film.
Apparently I've been watching a lot more obscure movies as 20 of these were not on my previous list. Otherwise, there doesn't seem there was much movement. All the other movies on the previous list seem to have remained fairly obscure. Of these films that I think more people should see, I say Names in Marble, Reckless Kelly, The Baker, The King and the Clown, and Animals Are Beautiful People.
Thursday, February 07, 2013
Static States Disadvantage
I've falling way behind on posts for this blog. Peace Corps tends to keep my attention elsewhere. Also answering questions on Quora has been eating into my "research for the sake of research" time.
This is a post I've been wanting to make for awhile, but wanted to wait until official numbers for the 2012 election were complete. I like playing around with prediction models and election trends, so of course, I made several posts about the subject: 2008 Election Trends, Surrendering the Big States, My 2008 Election Predictor, 2012 Predictions, Presidential Election Statistics, Hispanic Election.
The reason I bring this up is because I started to notice two patterns going into 2012. One was that since 2000 the electoral map had been surprisingly static. The other was that Republicans were increasingly being put at a disadvantage.
I mention the static electoral maps phenomenon in the Hispanic Election post. I found that if I alter election results from previous elections to make them so that it all comes down to one state or a tie, a pattern starts to form in 1984 and coalesces in 2000.
2000
2004
2008
2012
Over four elections, only two states have switched parties, Nevada and New Hampshire and that is largely a result of redistricting, not changes in the electorate. No similar sustained pattern arises before this.
The advantage of this is that it makes elections ridiculously easy to predict. Every election US Election Atlas allows members to make prediction maps (note: party colors are reversed on the site). For 2012, I made a map on February 14th based solely on the 2008 results and slightly decreasing the victory margin. I did not change the prediction for the rest of the year and got every state correct.
The disadvantage is all for Republicans. In two previous posts I predicted that in order to win in 2012, the Republicans would need to win by a sizable margin in the popular vote otherwise they risked losing the electoral vote. This prediction was based on the 2008 election wherein even if McCain had won the popular vote, he could still fall short of winning Iowa, which he needed to win the electoral vote. This is called the "tipping point" and it is something Nate Silver also noticed and predicted there was a 5.3% chance that Romney would win the popular vote while losing the electoral college. To compare, I made tipping point charts for all elections after 1964 and posted the results on Tipping Point.
In the last 3 elections, Republicans went in with a disadvantage according to the charts. Bush overcame the meager 0.36% disadvantage in 2004, but in 2008, McCain would have needed to beat Obama by 2.27%. Romney narrowed this a bit in 2012, but he still would have needed to win by at least 1.53%. Republicans are facing a disadvantage because their voters are heavily concentrated. Romney won 11 states by margins greater than 20%, but only 1 state by a margin less than 7%. Obama won 10 states by less than 7%. This concentration of support makes national polls deceptive. Republicans could have a large share of the popular vote, but because Republican voters are concentrated in a few states, it doesn't translate into electoral vote gains.
The static maps and the concentration of voters combined should be deeply troubling for Republicans. The concentration of voters means that Republicans have to win more than 50% of the popular vote to win the electoral vote while the static electoral map means that this situation is unlikely to change in the near future.
This is a post I've been wanting to make for awhile, but wanted to wait until official numbers for the 2012 election were complete. I like playing around with prediction models and election trends, so of course, I made several posts about the subject: 2008 Election Trends, Surrendering the Big States, My 2008 Election Predictor, 2012 Predictions, Presidential Election Statistics, Hispanic Election.
The reason I bring this up is because I started to notice two patterns going into 2012. One was that since 2000 the electoral map had been surprisingly static. The other was that Republicans were increasingly being put at a disadvantage.
I mention the static electoral maps phenomenon in the Hispanic Election post. I found that if I alter election results from previous elections to make them so that it all comes down to one state or a tie, a pattern starts to form in 1984 and coalesces in 2000.
2000
2004
2008
2012
Over four elections, only two states have switched parties, Nevada and New Hampshire and that is largely a result of redistricting, not changes in the electorate. No similar sustained pattern arises before this.
The advantage of this is that it makes elections ridiculously easy to predict. Every election US Election Atlas allows members to make prediction maps (note: party colors are reversed on the site). For 2012, I made a map on February 14th based solely on the 2008 results and slightly decreasing the victory margin. I did not change the prediction for the rest of the year and got every state correct.
The disadvantage is all for Republicans. In two previous posts I predicted that in order to win in 2012, the Republicans would need to win by a sizable margin in the popular vote otherwise they risked losing the electoral vote. This prediction was based on the 2008 election wherein even if McCain had won the popular vote, he could still fall short of winning Iowa, which he needed to win the electoral vote. This is called the "tipping point" and it is something Nate Silver also noticed and predicted there was a 5.3% chance that Romney would win the popular vote while losing the electoral college. To compare, I made tipping point charts for all elections after 1964 and posted the results on Tipping Point.
In the last 3 elections, Republicans went in with a disadvantage according to the charts. Bush overcame the meager 0.36% disadvantage in 2004, but in 2008, McCain would have needed to beat Obama by 2.27%. Romney narrowed this a bit in 2012, but he still would have needed to win by at least 1.53%. Republicans are facing a disadvantage because their voters are heavily concentrated. Romney won 11 states by margins greater than 20%, but only 1 state by a margin less than 7%. Obama won 10 states by less than 7%. This concentration of support makes national polls deceptive. Republicans could have a large share of the popular vote, but because Republican voters are concentrated in a few states, it doesn't translate into electoral vote gains.
The static maps and the concentration of voters combined should be deeply troubling for Republicans. The concentration of voters means that Republicans have to win more than 50% of the popular vote to win the electoral vote while the static electoral map means that this situation is unlikely to change in the near future.
Subscribe to:
Posts (Atom)