Google’s internal benchmarks of its own TPU, or tensor processing unit indicated that its purpose built AI board cleaned Nvidia’s clock when it came to number crunching and power consumption.
However this week Nvidia has blogged that Google’s numbers fail to take into account how wonderful its new boards are.
Google compaired its board to the older, Kepler-based, dual-GPU K80 rather than the Pascal based GPUs.
Nvidia moaned that Google’s team released technical information about the benefits of TPUs this past week but did not compare the TPU to the current generation Pascal-based P40.
While the TPU has 13x the performance of K80 is provisionally true, but there’s a snag. That 13x figure is the geometric mean of all the various workloads combined.
Nvidia’s argument is that Pascal has a much higher memory bandwidth and far more resources for inference performance than K80. As a result, the P40 offers 26x more inference performance than one die of a K80.
As Extreme Tech points out there are all sorts of things which are “unclear” about Nivida’s claims.
For example it is unclear if Nvidia’s claim takes Google’s tight latency caps into account. At the small batch sizes Google requires for its 8ms latency threshold, K80 utilization is just 37 percent of maximum theoretical performance. The vagueness of the claims make it difficult to evaluate them for accuracy.
Google’s enormous lead in incremental performance per watt will be difficult to overcome. Google said that its boffins modelled the expected performance improvement of a TPU with GDDR5 instead of DDR3, with more memory bandwidth.
Scaling memory bandwidth up by 4x would improve overall performance by 3x, at the cost of ~10% more die space. So, it is saying that it can boost the TPU side of the equation as well.
While no one is saying that the P40 is slower than the K80, but Google’s data shows a huge advantage for TPU performance-per-watt compared with GPUs, particularly once host server power is subtracted from the equation.
Basically GPU has lots of hardware that a chip like Google’s TPU simply doesn’t need.
Microsoft is touting operating system-wide power efficiencies in a recent preview of Windows 10, claiming that the technology will reduce notebook battery consumption by 11% on laptops equipped with the newest processors.
The technology, temporarily tagged as “Power Throttling,” was enabled on all copies of Windows 10 Insider build 16176, which Microsoft released Friday. Insider is the beta program Microsoft runs for both enthusiasts and businesses. The latter rely on Insider to learn how the OS will change for the next feature upgrade, as well as for testing the upgrade prior to deploying the final code when it is shipped several months later.
“With ‘Power Throttling,’ when background work is running, Windows places the CPU in its most energy efficient operating modes — work gets done, but the minimal possible battery is spent on that work,” Bill Karagounis, director of program management for Insider, said in a post to a company blog.
The CPU throttling is triggered on an app-specific basis by a detection system Microsoft integrated with the OS, said Karagounis. Like other such technologies, Microsoft’s is meant to recognize foreground tasks — such as active apps — as well as persistent applications, like music streaming applications, then give them full access to the processor. Other apps, or even individual processes within an app, that are classified as “background,” are restricted in how they impact the CPU’s power usage. For instance, they may not be allowed to kick the processor into its higher-frequency, higher-power, higher-consumption mode.
Power Throttling works only on Intel processors with that firm’s Speed Shift, a feature of sixth-generation and later CPUs, including “Skylake” and the newer “Kaby Lake.”
Recognizing that most personal computers are laptops and that battery longevity is a major factor in productivity, Microsoft has aggressively promoted Windows 10’s power savings, notably in the boosterism behind Edge, the OS’s default browser.
The Redmond, Wash. company isn’t working in a vacuum: Other operating systems also try to eke out more battery life by scaling back CPU use. Apple’s iOS, for instance, switches to a low-power mode when an iPhone or iPad battery reaches about 20% capacity. Among other things, the iOS mode halts background app refreshing and stops automatic email fetching.
Microsoft first added Power Throttling to Windows 10 in January, saying that it had turned it on for a subset of Insider-equipped devices as an experiment and promising to provide an update in mid-February. That update never appeared, hinting that Microsoft pulled it from inclusion in the then-upcoming Creators Update, the feature upgrade released April 11.
The first opportunity most users will have to apply Power Throttling will be with 2017’s second feature upgrade. Microsoft has not revealed a release timetable, but most experts expect it to appear this fall.
Bose Corp spies on its wireless headphone owners by using an app that tracks the music, podcasts and other audio they listen to, and violates their privacy rights by selling such data without permission, a lawsuit charged.
The complaint filed by Kyle Zak in federal court in Chicago seeks an injunction to stop Bose’s “wholesale disregard” for the privacy of customers who download its free Bose Connect app from Apple Inc or Google Play stores to their smartphones.
“People should be uncomfortable with it,” Christopher Dore, a lawyer representing Zak, said in an interview. “People put headphones on their head because they think it’s private, but they can be giving out information they don’t want to share.”
Bose did not respond on Wednesday to requests for comment on the proposed class action case. The Framingham, Massachusetts-based company has said annual sales top $3.5 billion.
Zak’s lawsuit was the latest to accuse companies of trying to boost profit by quietly amassing customer information, and then selling it or using it to solicit more business.
After paying $350 for his QuietComfort 35 headphones, Zak said he took Bose’s suggestion to “get the most out of your headphones” by downloading its app, and providing his name, email address and headphone serial number in the process.
But the Illinois resident said he was surprised to learn that Bose sent “all available media information” from his smartphone to third parties such as Segment.io, whose website promises to collect customer data and “send it anywhere.”
Audio choices offer “an incredible amount of insight” into customers’ personalities, behavior, politics and religious views, citing as an example that a person who listens to Muslim prayers might “very likely” be a Muslim, the complaint said.
“Defendants’ conduct demonstrates a wholesale disregard for consumer privacy rights,” the complaint said.
Zak is seeking millions of dollars of damages for buyers of headphones and speakers, including QuietComfort 35, QuietControl 30, SoundLink Around-Ear Wireless Headphones II, SoundLink Color II, SoundSport Wireless and SoundSport Pulse Wireless.
He also wants a halt to the data collection, which he said violates the federal Wiretap Act and Illinois laws against eavesdropping and consumer fraud.
Dore, a partner at Edelson PC, said customers do not see the Bose app’s user service and privacy agreements when signing up, and the privacy agreement says nothing about data collection.
Edelson specializes in suing technology companies over alleged privacy violations.
Facebook Inc is wants to capitalize off of the technology known as augmented reality, a mix of the real and digital worlds best known from the hit smartphone game Pokemon Go, Chief Executive Mark Zuckerberg said.
Speaking at F8, the company’s annual conference for software developers, Zuckerberg said Facebook was an obvious hub for businesses to reach people and experiment with augmented reality, although he did not suggest the company was planning to make similar games itself.
Pokemon Go, jointly developed by Nintendo Co and Niantic Inc, has generated masses of followers around the world as players use their phones to capture animated characters that appear in real locations.
Other uses of augmented reality have included the ability to hang out with a hologram of “Mad Men” star Jon Hamm or assemble a virtual human brain, all on mobile devices.
A recent push by Facebook to add camera features to its suite of smartphone apps will help the company popularize similar features, Zuckerberg said.
“Even if we were a little slow to add cameras to all our apps, I’m confident that now we’re going to push this augmented reality platform forward,” he said.
For a company that began as a way for college students to see pictures of each other, Facebook’s move toward augmented reality represents another step in its long evolution. It also raises the stakes for its competition with rival Snap Inc, the maker of Snapchat that describes itself as a camera company.
Zuckerberg said people could use the technology to leave a virtual note for a friend at a bar, or to find virtual street art on a wall that in real life is blank.
“This isn’t just about finding a Pokemon in a one-block radius,” he said.
Eventually, he said, people would use augmented reality on eyewear, although he did not give any details about possible Facebook hardware.
In 2014, Facebook acquired its Oculus virtual reality goggles unit for $2 billion, although that division is a long way from making a mass-market product or contributing significantly to the company’s earnings.
As part of his conference address, Zuckerberg addressed shortcomings on another major project, Facebook’s push into video. He said the service needed to do more to prevent the spread of violent videos, such as one on Sunday of a fatal shooting in Cleveland that was visible on the site for two hours.
Companies running Firefox, and testing the browser using the “Aurora” track, will be automatically migrated to the “Beta” channel today.
“It became clear that Aurora was not meeting our expectations as a first stabilization channel,” wrote Dave Camp, director of engineering for Firefox; Sylvestre Ledru, the browser’s release manager; and Ali Spivak, head of developer marketing, in a post to a Mozilla blog.
Mozilla has offered multiple versions of each Firefox edition since 2011, when it began offering four builds — Nightly, Aurora, Beta and Release — each of which was supposed to be more stable than the previous.
“We have more modern processes underlying our [release] train model, and believe we can deliver feature-rich, stable products without the additional 6-8-week Aurora phase,” said Camp, Ledru and Spivak.
In that “train” approach, Mozilla added a new feature to the least stable version, Developer, then when the feature was ready, moved it to the next track, Aurora. As development progressed, the feature would shift to Beta and then finally to Release.
But Mozilla acknowledged that the system had sometimes failed. “The release cycle time has required that we subvert the model regularly over the years by uplifting new features to meet market requirements,” the company admitted in an accompanying FAQ, referring to times when it has had to skip one of the tracks or shorten the time a feature spent on one.
Firefox users on the Aurora channel were to be moved to Beta today, according to the FAQ. Aurora will not be updated after tomorrow, when Firefox 53 is to ship in final, or Release, form.
With Aurora’s disappearance, Mozilla will rely on Beta for the first widespread distribution of each edition of Firefox. To make up for Aurora’s absence, each beta will be rolled out in stages, just as Release has long been, with the idea that if major problems crop up, they do so early on and thus affect only a subset of customers before the spigot is turned off.
Aurora’s elimination will not increase the frequency of Release builds issued or decrease the time between each Release version; the latter will continue to range from six to nine weeks. Nor will the already-slated dates for future versions of Firefox ESR (Extended Support Release) change. That edition, designed for enterprises and other large organizations, remains stable for approximately a year. Much like Windows 10’s LTSB (Long-term Servicing Branch), ESR receives only security updates.
Ditching Aurora, however, will let Mozilla move a new feature from inception to final about six to eight weeks faster than before.
Two months ago, Microsoft had extended support for Windows 10 version 1507 — Microsoft labels feature upgrades by year and month — from March to May, but did not specify the date in the latter month.
The May 9 retirement was quietly announced on several support documents, including the “Windows lifecycle fact sheet,” which lists several kinds of deadlines for various versions of the operating system.
Another document put it plainly. “The time has now come to end servicing for version 1507,” the support document read.
Stopping support for Windows 10 editions — Microsoft released the fourth on Tuesday — is an important part of the company’s software-as-a-service model. The company has pledged to support an individual edition, such as 1507, not for 10 years, as policy required for, say, Windows 7 or 8.1, but only for 18 months or so. That mandate insured Microsoft would not need to craft security patches, fix other bugs or add new features for an increasing number of versions.
By the time Windows 10 1507 slips off the list, it will have been supported for about 21 months. Part of the reason it lasted longer than Microsoft’s stated norm was because the firm issued just one feature upgrade — v. 1607 — in 2016.
The next Windows 10 edition, v. 1511, could be purged from support as soon as early October. That’s because Microsoft has committed to simultaneously supporting just two Current Branch for Business (CBB) builds. At the release of N+2 onto CBB, the company starts a 60-day-or-so countdown. At the end of the 60 days, N drops off the support list. N+1 then becomes N and N+2 morphs into N+1.
Under that policy, N would be 1511, N+1 version 1607 (released in August 2016) and N+2 1703 (this month’s feature upgrade). Version 1703 will likely be promoted to the CBB in four months, or August; two months more would put 1511’s support demise in October.
Users running 1507 must have upgraded to 1511, 1607 or 1703 by May 10 to receive future security patches, and other fixes or enhancements. Windows 10 1507 will not suddenly fail to boot, however, or degrade, as do copies that have not been activated with a product key.
The only exceptions will be customers whose devices are running v. 1507 from the Long-term Servicing Branch (LTSB), a special release track available only to organizations using Windows 10 Enterprise.
U.S. video streaming service provider Netflix is engaged in negotiations with Indonesia’s top telecom firm PT Telekomunikasi Indonesia Tbk (Telkom) to offer its service in the country, a spokesman at the Indonesian company said.
The U.S. company has made an aggressive push globally, but faced problems such as tough local competition and regulatory hurdles in several major Asian markets. In Indonesia, a country of 250 million people, Netflix ran afoul of the film censorship board last year for carrying content deemed inappropriately violent or sexual.
The communications ministry of Indonesia, home to the world’s largest Muslim population, had also demanded that Netflix set up a office in the country and pay local taxes.
While state-controlled Telkom had blocked Netflix, the service was still available in Indonesia via WiFi connections and other carriers.
Telkom is now negotiating a partnership agreement with Netflix and hopes to complete the process next month, Arif Prabowo, vice president for corporate communication at Telkom, said in a text message.
Telkom was previously concerned that Netflix carried “content that has a negative element”, Prabowo said.
“If we work together, that means we would know and can be responsible for the content broadcast by Netflix.”
Teaming up with Netflix would expand Telkom’s content offering, Prabowo added. “The choices for our customers will be more varied.”
A Netflix spokeswoman declined to comment.
A US news station, which normally chants Apple mantras with the rest of them claimed Apple was the focus of a mini firestorm which is about as scathing as the Tame Apple Press gets.
At the centre of the problem is Apple’s aging Mac Pro desktop line which was due for a refresh to bring the rubbish bin PC into the internet age. You would think after not improving a computer for a since 2013 you could add a few improvements.
Apple decided that the best thing to do was jack up the price – after all you get what you pay for right?
Even MacWorld thought that Apple was taking the Nintendo.
But there’s nothing new about what Apple did. “The two available Mac Pro configurations aren’t new, they’re just newly priced,” MacWorld pointed out.
The entry-level $2,999 Mac Pro model now has 6 Intel Xeon processor cores –versus 4 cores on the previous configuration – with dual AMD FirePro D500 graphics chips. And the $3,999 Mac Pro gets an 8-core Intel Xeon processor with dual AMD
FirePro D700 graphics silicon. Woop!
Apple marketing VP Phil Schiller reportedly said that the Mac Pro had heat (aka “thermal”) issues that “restricted” a user’s ability to upgrade and that Apple is “sorry to disappoint customers”.
Apparently Apple had a meeting where it was claimed that Apple said it is “completely rethinking” the Mac Pro model. And, as a result, the company acknowledged “that its flashy 2013 Mac Pro redesign was a mistake”.
It has apparently taken them four years at least to have worked that out, and even longer before Apple comes up with a solution. In fact the overhaul will not happen this year.
9to5Mac insists that Apple is trying to assuage any perceived user frustration, and this is the closest thing that Jobs Mob has got to an apology.
“The very fact that Apple felt compelled to hold [Monday’s] meeting in the first place is evidence of just how much it thinks it screwed-up here. The company that has always taken the view that ‘people don’t know what they want until you show it to them’ has clearly had to face the fact that, in the pro market at least, that’s not the case.”
In January, Toshiba officially announced it would seek to sell a portion of its flash memory business, including the SSD business of the Storage & Electronic Device Solutions Division, to a not-yet-named buyer.
The Nikkei Asian Review has reported that Toshiba may sell a 20% stake in the memory business for between $1.77 billion and $2.65 billion, “while retaining a majority stake and keeping the new company in group earnings.”
Toshiba’s solvency and fundraising ability are presently in doubt because of a $1.9 billion accounting scandal and a huge loss related to the purchase of a U.S. nuclear plant business. The company, which invented NAND flash in the early 1980s, had been considering spinning off its semiconductor operations and selling a partial stake to Western Digital (WD) and others, as it tries to cope with a massive impairment loss in its U.S. nuclear power unit.
Neither Toshiba nor WD have confirmed a potential sale, however.
Earlier this year, Toshiba took a writedown of $6.56 billion against its struggling U.S. nuclear equipment operations and it’s hoping to rebound from that loss with a sale.
“Its financial problems were a major drag on the growth of its memory business,” Sean Yang, research director of DRAMeXchange, said in an earlier interview.
Several potential buyers have been identified in reports, including Apple, according to Bloomberg’s news service. Apple is considering investing several billion in Toshiba’s memory business, according to the report.
“It seems like they are selling the Golden Goose and keeping the money pit,” said Jim Handy, an analyst with Objective-Analysis, referring to Toshiba.
If Apple were to purchase a stake in Toshiba’s semiconductor business, it would be a departure for a company that has historically outsourced most of its parts and labor, Handy said.
“Seagate and Western Digital used to believe that vertical integration was necessary in order to compete in the SSD market, although Seagate appears to have changed its tune,” Handy said. “A captive source of supply is a good thing to have during a shortage, but can be a millstone during an oversupply.”
Semiconductor Manufacturing Co., a major supplier to Apple, reportedly said it was not participating in talks after reviewing a possible deal “with interest.”
According to one report, Western Digital (WD) is none too happy about Toshiba’s plans to sell its memory business. WD reportedly sent a letter to Toshiba telling it the proposed sale breaches a joint-venture agreement as part of the FlashAlliance to build flash fabrication plants in Japan, which are operated by Toshiba. WD’s SanDisk holds a 49.9% share in the FlashAlliance and a Toshiba has a 50.1% share.
Any potential sale by Toshiba might be on hold for now as it deals with WD’s concerns.
If Toshiba does sell a major stake in its memory business — or the entire unit — it would do little to effect the memory market as a whole from the perspective of supply and demand, according to Handy.
“From the perspective of national security there are significant concerns that Japan will lose control of an important technology, and that it will be owned by a company from a country that has a difficult history with Japan,” Handy said, referring to China and Foxxcon. “From WD’s perspective it’s really strange, since they have a very good working relationship and understanding with Toshiba, but not necessarily with the buyer.
“I like to think of it as your spouse coming in and saying: ‘Here’s somebody new for you to be married to!’ then walking off.”
Apple Inc has been granted a permit to test autonomous vehicles in California, increasing the likelihood that it is working on self-driving car technology in a crowded arena of companies hoping to offer those cars to the masses.
The permit allows it to conduct test drives in three vehicles with six drivers, the state Department of Motor Vehicles said on Friday. The vehicles are all 2015 Lexus RX450h, according to the DMV.
Although it has never openly acknowledged it is looking into building an electric car, Apple has recruited dozens of auto experts in recent years, and the permit pulls the curtain back a bit on any possible plan.
“This does confirm what’s long been rumored: that Apple is at least toying with the idea of getting into the autonomous game in some capacity,” said Chris Theodore, president of consultancy Theodore & Associates, and a former vice president at Ford Motor Co and Chrysler.
The permit does not mean Apple is definitely building a car. “This is not necessarily automobiles as initially rumored, but software or possibly hardware associated with autonomous technology,” Theodore said.
An Apple spokesman declined to comment directly on the filing, pointing back to a statement the company made in November when it wrote to the U.S. National Highway Traffic Safety Administration (NHTSA) on the subject of regulating self-driving vehicles.
“The company is investing heavily in the study of machine learning and automation, and is excited about the potential of automated systems in many areas, including transportation,” Apple’s director of product integrity, Steve Kenner, wrote in that five-page letter.
Apple executives have been coy about their interest in cars. Chief Executive Tim Cook has suggested that Apple wants to move beyond integration of Apple smartphones into vehicle infotainment systems.
Apple joins a growing list of traditional carmakers, technology companies, and small start ups to test drive cars in California – all vying to be the first to have commercially viable vehicles on the roads.
Companies that have been issued permits also include Alphabet Inc’s Google unit, Ford Motor Co, Volkswagen AG, Daimler AG, Tesla Motors Inc and General Motors Co.
Many companies have said the first cars will launch in 2020 but some experts believe it may take much longer due to regulatory challenges.
Severe cyber security breaches, such as those having legal or regulatory consequences, involve the loss of hundreds of thousands of records and hurt the firm’s brand, caused share prices to fall on average 1.8 percent on a permanent basis, the analysis of 65 companies affected since 2013 globally has found.
Investors in a typical FTSE 100 firm would be worse off by an average of £120 million after such a breach, the report said. Overall the cost to shareholders of these 65 companies would be in excess of 42 billion pounds ($52.40 billion).
CGI’s analysis compared each company’s share price against a cohort of similar companies to isolate the impact of cyber breaches from other market movements, during incidents detailed in a breach index compiled by Dutch security firm Gemalto.
Two-thirds of firms had their share price adversely impacted after suffering a cyber breach. Financial firms were the worst affected, followed closely by communications firms.
“Financial services experience the greatest burden in terms of impact, reflecting the high levels of regulation, the importance of customer confidence and the potential for financial fraud to be a facet of the breach,” the report said.
hose least affected were retail, hospitality and travel companies.
Hacking attacks and other cyber security breaches have impacted companies across the world in recent years, from retailer Target in the United States in 2013 to British communications firm TalkTalk in 2015.
This is not an extremely late April 1st, and we admit that it is a little early given that its replacement has not shown up yet, but we predict that it will go the way of the dodo, the Norwegian Blue, the bleeper and the Crackberry.
OK it is probably a few years off, but the technology is so persuasive that its death will be longer than the exit of a hero in a South American soap opera.
For a while now smartphone sales have slowed. Basically the structure developed by Nokia and stolen by Apple and copied everywhere has run out of places to go. There is no more innovation in smartphones any longer, despite what is claimed particularly by the Tame Apple Press. Chip speeds have increased slightly and are about as fast as they are going to get. Even if someone gets a chip to the speeds of a PC it is not going to make a hell of a lot of difference.
What is coming next is being sorted out by the likes of Microsoft, Facebook, Amazon, Google along with Elon Musk. Apple of course is waiting for the next biggest thing to be developed by others before it takes a risk.
So what will get rid of it? While the Tame Apple Press think it will be something more like the Amazon Echo, Sony PlayStation VR, and the SmartWatch that is mostly because that is pretty much Apple’s current agenda.
No doubt AR and VR could be the way it is going. Certainly some sort of interface which projects detailed 3D images straight into your eyes while you interact with your environment. So instead of typing this on a screen I will be typing it on a nice egonomic bit of rubber while the words are appearing before my eyes. A more portable version would put a keyboard onto any surface.
Microsoft thinks that is the way things will go and the tech will replace the smartphone, the TV, and sex, and anything else with a screen with sounds going in through a headphone.
As artificial intelligence systems like Apple’s Siri, Amazon’s Alexa, Samsung’s Bixby, and Microsoft’s Cortana get smarter, there is going to be a rise not just in talking to computers, but having them talk back.
All this makes the smartphone redundant and limited. Sure it will be a good decade before this brave new world takes off and it will be a slow slide rather than anything great, but we are seeing the change start happening now. The world is bored with smartphones and they are just not having the impact they used to.
The first report about the attacks came from antivirus vendor McAfee after the company’s researchers analyzed some suspicious Word files spotted a day earlier. It turned out that the files were exploiting a vulnerability that affects “all Microsoft Office versions, including the latest Office 2016 running on Windows 10.”
The flaw is related to the Windows Object Linking and Embedding (OLE) feature in Microsoft Office that allows documents to embed references and links to other documents or objects, the McAfee researchers said in a blog post.
When the rogue documents used in this attack are opened, they reach out to an external server and download an HTA (HTML Application) file that contains malicious VBScript code. The HTA file is disguised as an RTF (Rich Text Format) document and is automatically executed.
“The successful exploit closes the bait Word document, and pops up a fake one to show the victim,” the McAfee researchers said. “In the background, the malware has already been stealthily installed on the victim’s system.”
By searching back through its data, McAfee has tracked down attacks exploiting this vulnerability to late January.
Following McAfee’s report, security researchers from FireEye also confirmed that they’ve been aware of these attacks and exploit for several weeks and have coordinated disclosure with Microsoft.
According to FireEye, the malicious Word documents are sent as email attachments. The company hasn’t provided examples of the malicious emails, but because this is a previously undisclosed, zero-day vulnerability, the attacks are likely targeted toward a limited number of victims.
Both McAfee and FireEye noted that the exploit can bypass most memory-based mitigations included in Windows. That’s because the vulnerability is a logic bug rather than a programming error.
Microsoft is scheduled to release its monthly security updates on Tuesday, but it’s not clear if a patch for this vulnerability will be included. The company did not immediately respond to a request for comment.
In the meantime, users should be wary of documents received from untrusted sources and should enable the Office Protected View mode because it can block this attack.
A computer hack triggered all the emergency sirens in Dallas for about 90 minutes overnight in one of the largest known breaches of a siren warning system, officials in the Texas city said on Saturday.
Dallas’ 156 sirens, normally used to warn of tornadoes and other dangerous weather, were triggered at 11:42 p.m. CDT on Friday. The wailing did not end until 1:17 a.m. CDT on Saturday when engineers manually shut down the sirens’ radio system and repeaters, city Emergency Management Director Rocky Vaz said.
“At this point, we can tell you with a good deal of confidence that this was somebody outside of our system that got in there and activated our sirens,” he told reporters.
The breach in the city of 1.6 million people was believed to have originated in the area, city spokeswoman Sana Syed said in an emailed statement.
Vaz cited industry experts as saying the hack was among the largest ever to affect emergency sirens, with most breaches triggering one or two. “This is a very, very rare event,” he said.
Engineers are working to restart the system and should have it restored by late on Sunday, he said. Until the sirens are running, Dallas will rely on local media, emergency 911 phone calls, and a federal radio alert system, Vaz said.
The hack is being investigated by system engineers and the Federal Communications Commission has been contacted, but police have not been involved, he said.
The sirens went through 15 cycles of a 90-second activation before they were shut down, he said.
The department has found “systemic compensation disparities against women pretty much across the entire workforce,” Labor Department Regional Director Janette Wipper testified in a court in San Francisco on Friday, according to a report by The Guardian. Janet Herold, the department’s regional solicitor, told the Guardian that pay discrimination against women was extreme.
Wipper said that the Labor Department found pay disparities in a snapshot of salaries from 2015, according to the Guardian.
Wipper’s testimony was part of a hearing about a lawsuit that the Labor Department brought against Google to force the company to hand over salary information. The department is authorized to conduct audits of Google’s employment practices because the company gets government contracts. It says Google hasn’t been cooperating.
The agency has asked the Office of Administrative Law Judges, a special court for Labor Department programs, to cancel all of Google’s government contracts and keep it from getting future contracts if it doesn’t comply with the request for data.
Google vehemently disagrees with the department’s assertion, the company said in an emailed statement.
“Every year, we do a comprehensive and robust analysis of pay across genders and we have found no gender pay gap,” the statement read. “Other than making an unfounded statement which we heard for the first time in court, the DoL hasn’t provided any data, or shared its methodology.”
At the time it filed the lawsuit, the Labor Department characterized the request for information as routine, but Google says the agency has cast too wide a net. (In a statement earlier this year, the company said it provided “hundreds of thousands of records” to the DoL as part of the audit.)
That argument appears to hold some water for Steven Berlin, the administrative law judge overseeing the case.
Last month, Berlin denied the Labor Department’s motion for summary judgment, which would have immediately concluded the case in its favor. He said the department’s request for the data was “unreasonably burdensome, given its extremely limited relevance.”
The reported testimony on Friday came three days after Google said in a tweet that it had “closed the gender pay gap globally.” The company also published a guide to doing the same at other companies.