We have had recent stories and discussion regarding the connector on GPUs which are causing overheating and, in a small number of case, actually catching fire. :
Please cast your vote in the comments to this Meta.A valid vote should contain a single word - either "Yes" to accept the documents or "No" to reject them. A single vote is required to accept ALL of the proposed documents.TO OVERCOME THE TECHNICAL PROBLEM: Please include a single paragraph containing anything at all - it will be ignored by the software during vote counting. However, the vote "Yes" or "No" must be on a line all by itself.For ease of reference links to the documents are repeated here:
Don't worry - this will be a relatively short Meta, and it is not to explain another site outage!Community Vote on Site DocumentationIn December 2024 I released a Meta which detailed the proposed documentation for the site under the Soylent Phoenix board. This is a legal requirement resulting from the creation of a new company. I repeated the links to the documentation in January. The next step is for the community to accept or reject the proposed documentation. The previous voting software is no longer available to us but I believe that a straightforward count of comments will suffice.I will publish another Meta which will contain the links to the proposed documentation but it is not to be used for any discussion regarding the contents. Each current account in good standing (i.e. having a karma of 20+ and created on or before the publication of the December Meta (16 Dec 2024 - that is up to and including account #49487 ) will be eligible to vote. In order to cast your vote your comment should be limited to a single word - "Yes" or "No" (upper or lower case is acceptable) on a line all by itself. "Yes" will indicate your acceptance of the documentation and "No" will indicate your rejection of it. Your last comment of a maximum of 2 attempts will be the one that counts so you will have the opportunity to change your vote. Any more than 2 attempts from an account to cast a vote will be discarded. Comments may contain a single paragraph to overcome the 'lame comment' filter. The contents of the paragraph will be ignored. The vote will remain open for 1 week and will close at 23:59 (UTC) on 28 February 2025. The result will be made public once the Board are satisfied that the voting has been fair and democratic.Existing votes will remain valid and do not have to be redone.Entering into a discussion in the vote or justifying why you have voted in a particular fashion will nullify your comment. There has been a period of over 2 months for discussion and suggested changes.It is important that you cast a vote. As an extreme example, if 1 person alone votes Yes and 2 people vote No then the documentation will NOT be accepted. Not casting a vote doesn't make any statement whatsoever but may result in the majority of true community opinion being ignored.Essentially, the documentation is the same as that adopted in 2014 except it has been rewritten where necessary to clarify the meaning or intent. It also incorporates in one location changes to the rules that have been accepted by the community since 2014 (e.g. the definition of Spam which was adopted by the site in 2021).Read more of this story at SoylentNews.
looorg writes:Are noise-cancelling headphones to blame for young people's hearing problems? They are not going deaf but the brain are having difficulty processing sounds. As it has not been trained on sorting out sounds and noise due to constant headphone usage filtering out the auditory realities.
I expect that many noticed that the site went down and, if you are reading this you will also realise that it is now back up.The entire server died leaving a wake of Out-Of-Memory messages, which resulted in the site itself, IRC and our email all failing. We (and by that I really mean kolie!) have restarted the server and doubled the amount of memory available to it.Of course, that doesn't tell us why it ran out of memory, although we knew that it was a bit tight, nor what specifically happened today to push it over the edge. That will probably take a while to work out.It might take us a while to put more stories in the queue but you should be able to comment on many of today's stories that have only just appeared on your screens.We are sorry for the inconvenience and we are getting back on our feet again. As always, a big THANK YOU to kolie for his efforts.Read more of this story at SoylentNews.
fliptop writes:Chinese scientists have significantly improved the performance of supercomputer simulations using domestically designed GPUs, surpassing systems powered by Nvidia's advanced hardware:
canopic jug writes:Quanta Magazine is covering a notable advancement in a well-studied computer science algorithm for sorting books or files or database contents or other similar physical or digital objects. The foundation is a 1981 study which was followed by a significant advancement in 2004 and just recently by reaches rather close to the theoretical ideal in the list labeling problem aka the library sorting problem:
upstart writes:The European Union regulation banning the use of bisphenol A in materials that come into contact with food officially took effect on 20 January, in an attempt to minimise exposure to the harmful endocrine disruptor:
looorg writes:https://psycnet.apa.org/record/2025-66513-001?doi=1Are women more talkative than men? An analysis of gender differences in daily word use says it is so.
fliptop writes:Surprise surprise, we've done it again. We've demonstrated an ability to compromise significantly sensitive networks, including governments, militaries, space agencies, cyber security companies, supply chains, software development systems and environments, and more:
day of the dalek writes:NASCAR's first points race of 2025 is the Daytona 500, which is on February 16. The Daytona 500 is NASCAR's most prestigious race, with a unique style of racing known as pack racing. This is characterized by many cars running at very high speeds in large packs, and often massive wrecks referred to as the "Big One". However, many drivers and fans have been critical of rules changes in recent years leading to racing at NASCAR's biggest oval tracks that they describe as boring.Bobby Allison's 210 mph crash at the 1987 Winston 500 forever changed how NASCAR races at superspeedways. Allison's car became airborne, severely damaged the catch fence along the frontstretch at Talladega, and almost flew into the stands. NASCAR decided the speeds had become too fast at their two largest and highest-banked ovals, Daytona and Talladega, and implemented restrictor plates at those tracks starting in 1988. Restrictor plates reduce the air intake into the engine, reducing both horsepower and speeds.Although drafting had always been powerful at superspeedways, the changes caused the cars to race in large packs, often with 20 or 30 cars within a couple seconds of each other. Although NASCAR says that this is necessary to prevent the worst wrecks, often leads to large multi-car wrecks. Drivers also complain that winning restrictor plate races is influenced too heavily by luck, though this is disputed.When a car drives forward, it displaces the air with its nose, creating high pressure at the front of the car, and low pressure behind the car in its turbulent wake. The combination of high pressure in front and low pressure behind the car creates a rearward pointing pressure gradient force (PGF), which is drag and slows the car down. If another car rides in the wake of the lead car, it experiences lower pressure on its nose, reducing the drag and allowing the car to go faster. However, if the trailing car puts its nose right behind the rear bumper of the lead car, it increases the pressure behind the lead car, reducing the lead car's drag as well. When the cars are in close proximity, both leading and trailing cars benefit from the draft When NASCAR reduced the horsepower at superspeedways, the draft became particularly powerful, and the fastest way around the track was now in a group of cars driving bumper-to-bumper.The result is often a large pack of cars, two or three wide, driving around the track at full throttle with speeds around 190 mph. One of the best ways to pass in pack racing is for a car to back up to the bumper of the car behind it, get pushed forward to increase its speed, and then get out of line to try to move forward. For this strategy to work, either that car has to get back in the draft soon before drag slows it down too much, or it needs other cars to also get out of line and start a new line. The result is a style of racing that leads to cars making aggressive moves at high speeds, and it can be spectacular to watch. However, in recent years and especially since the introduction of NASCAR's next-gen Cup Series car in 2022, the racing at superspeedways has been criticized as boring.Although it's difficult to find detailed historical engine specs, for much of the restrictor plate era, cars might have 750 horsepower at most tracks but be limited to 450 horsepower at superspeedways. More recently, NASCAR has been increasing the power at superspeedways while adding more aerodynamic drag to slow the cars down. However, this means the drag is more severe when a car gets out of line, and a single car will drop back quickly. This makes it much more difficult for cars to pass without multiple cars getting out of line at once.Driver Denny Hamlin also said that higher drag in the next-gen car leads to poorer fuel mileage, leading to slower speeds to conserve fuel, and less passing. Instead of making aggressive moves to pass, cars tend to ride around in line for much of the race leading to a style of racing that many describe as boring. Suggestions to improve the racing include reducing drag, lowering horsepower in the engines, and either adjusting the lengths of race stages or eliminating stage racing altogether at superspeedways.Original SubmissionRead more of this story at SoylentNews.
day of the dalek writes:Tornado strength is rated from 0 (weakest) to 5 (strongest) on the Enhanced Fujita (EF) scale, with roughly 2% of tornadoes being rated EF4 or EF5. The EF scale replaced the older Fujita scale to provide much more fine-grained detail in determining a tornado's rating. The EF5 rating corresponds to estimated peak winds of 200+ mph. However, it is purely a damage scale, from which the peak winds in the tornado are later estimated. Although meteorologists often discuss the wind speeds in tornadoes, measured wind speeds are never a factor in rating tornadoes.This distinction was made apparent on April 26, 1991 when the Andover, Kansas tornado was rated F5 while the Red Rock, Oklahoma tornado was rated F4 despite likely being the stronger tornado. A mobile radar from the University of Oklahoma measured 270+ mph winds in the Red Rock tornado, well into the F5 range, and the strongest tornado winds that had ever been measured to date. However, because the Red Rock tornado remained over mostly rural areas unlike the Andover tornado, there was little opportunity for it to do severe enough damage to be rated F5. This distinction remains true with the EF scale, where the 2013 El Reno, Oklahoma tornado was originally rated EF5 on the basis of mobile radar observations, then downgraded to EF3 based on the lack of EF4 or EF5 damage in damage surveys.A new article in the Bulletin of the American Meteorological Society discusses the current "drought" in EF5 tornadoes, with that rating being most recently assigned to the 2013 Moore, Oklahoma tornado that happened just 11 days before the 2013 El Reno tornado. The lack of EF5 tornadoes for over 11 years has raised questions of why, and if the EF5 rating is essentially obsolete.The journal paper argues that the lack of EF5 tornadoes for 11 years is roughly 0.3%, and it's very unlikely that there have been zero EF5 tornadoes during that period. Instead, it's probable that this is due to stricter application of the EF scale standards, and several tornadoes were estimated to have peak winds of 190+ mph during that period. If those tornadoes were reclassified to EF5, it would be statistically consistent with the previous climatology of EF5 tornadoes. The authors note that some of the previous EF5 ratings such as the 2011 Joplin Missouri tornado were based on damage indicators that were not part of the EF scale specifications.One of the biggest reasons for not assigning an EF5 rating is the presence of areas with limited damage very close to near-total devastation. However, the strongest tornadoes are generally multi-vortex tornadoes, where the strongest winds are found within small vortices embedded within a broader tornadic circulation. This could explain the proximity of extreme damage to areas with much less damage. The damage severity also depends on how long structures are exposed to extreme winds, an example of which is the 1997 Jarrell, Texas tornado, which was rated F5 but damage was more severe due to the tornado moving slowly and exposing buildings to the tornado winds for a longer than usual time. This raises the question of whether the EF5 rating is obsolete based on how the EF scale is currently applied, and if it's time to again revise how meteorologists rate tornado strength.Original SubmissionRead more of this story at SoylentNews.
The Beeb decided to test some LLMs to see how well they could summarize the news https://www.bbc.com/news/articles/c0m17d8827ko Turns out the answer is, "not very well".