Forgot your password?
typodupeerror

Comment: Blame the arsonists not humanity (Score 1) 379

by hardwarefreak (#47043673) Attached to: Studies: Wildfires Worse Due To Global Warming

Arsonists and idiots tossing cigarette butts start the majority of the wildfires in the West each year. But instead of blaming them, practitioners of the religion of Catastrophic Man Made Global Warming blame all of humanity for burning fossil fuels. Continuous claims of this nature make the lot of them modern day Chicken Littles (Henny Penny). If they have a dire need to blame someone other than those directly responsible, they need only look into the mirror. The liberal Western states have been sucking all of the water out of the mountains for decades for their backyard pools and lawns, preventing natural run off. This is what turns the grasslands into gasoline. Not global warming.

Comment: Re:This is a problem now? (Score 1) 128

by hardwarefreak (#46924033) Attached to: U-2 Caused Widespread Shutdown of US Flights Out of LAX

I remember hearing an anecdote about the SR71 Blackbird
A Blackbird is entering commercial airspace over CA
Pilot requests Flight level 70 (thousand Feet)
Controller laughs. If you can reach it you're welcome to it.
Blackbird pilot replies "descending from flight level 100"

Of couirse it may be an urban legend. But the SR71 could go that high

The SR-71's maximum altitude achieved in sustained level flight was 85,069 ft. It could not fly at 100,000 ft for a couple of reasons. First, the atmosphere at 100K ft lacks sufficient oxygen to keep the engines lit, both the turbine section and the ram jet section. Second, the air density is too low for the control surfaces to function at mach 3, causing directional instability, and likely the loss of the aircracft.

Comment: Re: Wikipedia ruined the internet (Score 1) 517

My god let VMS die already.

Not any time soon: http://www.hp.com/go/openvms/r...

VMS is to HP iron what Z/OS is to IBM mainframes. VMS will die when HP exits this market. And given that most of the VMS customers are banks and financial institutions, their customers aren't running out of money, and aren't looking to migrate to other platforms.

Comment: Re:We've gone beyond bad science (Score 1) 703

I guess you have not been paying attention to the drought in the Central Valley of California. You will, when food prices shoot through the roof this summer.

I grow my own vegetables in the spring/summer: asparagus, tomatoes (fruit actually), broccoli, cauliflower, peppers, green beans, lettuce (leaf and iceberg), cabbage, beats, zucchini, cucumbers, onions and potatos. I also have a black raspberry patch. I get plenty sweet corn from a couple of local farmers trading some of the above. I only rely on vegetables from California, Arizona, Florida, Mexico, and Central/South American countries in the winter. The only effect the Cali drought will have on my wallet is if I have a craving for strawberries.

Comment: Re:If only... (Score 1) 295

Why should a company that uses Linux as a server OS adopt and support the development of GPU drivers that are not useful in their business context

Apparently you forgot that Google's Android, which is a Linux distro, and Chromium ship on over 80% of all new smart phones and a large percentage of tablets. Devices that are advertised to stream HD video from the likes of Youtube (owned by Google), NetFlix, et al. Getting hardware video acceleration working on these devices is most certainly in Google's interest. If video performance sucks, people will buy iPhones/iPads and Windows powered phones and tablets. This will cost Google licensing revenue as their device partners will sell fewer Google powered devices.

Comment: Let's kick this AGW party into overdrive (Score 1) 393

I'm sitting in the central Midwest where it's currently 10F degrees and has been below zero many times this winter. Would everyone please start your car and just let it idle... for a month, start a bunch of foreset fires, tickle some volcanos into erupting, light Saudi Arabia's oil wells on fire. Let's kick this AGW party into overdrive, I'm freezing. This 0.5C degree rise per century just ain't cut'n it.

Comment: not an obvious solution (Score 1) 518

by hardwarefreak (#46021717) Attached to: Nobel Prize Winning Economist: Legalize Sale of Human Organs

Eventually, the advantages of allowing payment for organs would become obvious. At that point, people will wonder why it took so long to adopt such an obvious and sensible solution to the shortage of organs for transplant.

If it's such an "obvious and sensible solution" why did the author require 27 paragraphs to sell it?

Comment: Re:Something something online sorting (Score 1) 241

by hardwarefreak (#45804433) Attached to: Why Don't Open Source Databases Use GPUs?

They did not publish a formal TPC-H result, they simply name-dropped TPC-H. In a way that, as it can't be compared to anything, TPC themselves is "meaningless".

They used TPC-H because it's a relevant benchmark everyone in DSS is familiar with. It allows for a reference point. "Name dropping"? Please...

And the 3 guys with nvidia email addresses on the paper have no interest in how well nvidia cards sell?

You're kidding right? How many such research papers have you read in your life? Is this the first? If they'd have chosen to use AMD hardware instead of nVidia, you'd see AMD engineers at the top of the page. If one embarks on research to enhance the fuel economy of Ford's EcoTech engine, is one going to go it alone or involve Ford engineers who know everything about said engine? This is known as an "academic/industrial partnership". The bulk of all such research today is done in this manner. If you want to read a study without such commercial involvement, read pure research papers on something like particle physics.

It is you who seems confused, or at least naive. This was nvidia marketting. The company I just quit from used to do exactly the same - they all do.

I think everyone who has read this exchange knows who is confused and/or naive, and who has an axe to grind for some illogical reason. Take any 4th year or post grad marketing class and tell me where on the ROI list partnering on academic research papers falls. It's not there. Why? Marketing is an overt exercise designed for maximum exposure. A few employees' names on an academic research paper that will be seen by a few hundred people is the ultimate in obscurity.

If you want to poo-poo something at least have a few facts to back up your arguments. You've stated none in this exchange. I think you should exit this thread at this point. You simply dig yourself a deeper hole with every reply. However it is clear that you are the type who cannot be humbled no matter how many times you're proven wrong. So, please, continue making illogical arguments. Some may find it humorous.

Comment: Re:Something something online sorting (Score 1) 241

by hardwarefreak (#45797269) Attached to: Why Don't Open Source Databases Use GPUs?

No, the point of the TPC benchmarks is to sell hardware and software. Period. Always has been, always will be. Look at the results. Every system configuration is exact down to the last wire and transistor, all necessary software SKUs, licenses, etc. One can order an identical system directly from the PDF as every part# is there. The TPC organization is funded by fees paid for the test suite, verification of and publication of vendor results. The big vendors spend tens of millions on confgiuration engineering and testing trying to stay at the top of each list.

You seem to be confused here. The researchers did not publish a TPC-H result. They published a method for running SQL queries on a GPU. The TPC-H benchmark was simply the workload they chose to demonstrate that it not only works, but works pretty well. This is research, not a product launch. They didn't submit their results to TPC. They're not selling a computer system or software. Whether their test rig has the size or throughput of a system configuration published on the TCP website is simply not relevant.

They proved the concept. Others will pick up the torch just as happened with GPUs in the HPC space. In the not too distant future we'll see commercial DSS systems, from the same vendors on the current lists, with a GPU (or 2 or 4) in each cluster node.

Comment: Re:Something something online sorting (Score 1) 241

by hardwarefreak (#45792407) Attached to: Why Don't Open Source Databases Use GPUs?

Read the paper - page 7 (which bizarrely doesn't render clearly for me at all, and I can't copy/paste)
"Scale Factor 1 (SF 1) ... data fits in GPU memory"

They ran the TPC-H ("H"="Huge") with a dataset that was ABSOLUTELY FUCKING TINY.

"H" has no definition. The TPC benchmark series started well over a decade ago. "A", and each subsequent benchmark created over the years, was labeled more or less sequentially, with few exceptions. TCP-W is a "web" benchmark. The two most widely used, TPC-C and TPC-H are, respectively, emulating "transaction processing" and "decision support" workloads. The "C" and "H" have no meaning to anyone outside those who created the benchmarks. Best guess for the origin of "H" is that "decision support" is also called "data warehousing". So maybe the "H" stands for "house". Again, there is no official description of what the letters represent.

With that correction out of the way, I'll point out that nVidia currently offers the Tesla K40 GPU accelerator which has 12GB of RAM. Decision support database systems have been mostly cluster architectures for many years now because they cost less than big monolithic servers and yield superior performance on DSS workloads, at least for problem sets of any size. An 8 node cluster with 12GB Teslas provides 96GB of RAM. If/when APIs are available to allow parallel DSS databses to use the GPUs, such clusters would run circles around traditional CPU based clusters for many DSS workloads whose datasets are not significantly larger than the memory size. And even then it is possible to make more system memory accessible to the GPU.

These guys have shown significant speedups for DSS workloads. It's probably only a matter of time now before we see database engines supporting GPUs. Prior to this I think most people saw GPUs as floating point only accelerators for HPC. This paper, and those preceding it, demonstrate integer workloads can benefit as well.

Comment: Re:The Solution is Obvious (Score 1) 829

by hardwarefreak (#45769315) Attached to: Microsoft's Ticking Time Bomb Is Windows XP

Microsoft should extend support for XP...but only on a cash-for-patch basis. Sell patches at $5 a pop for XP user's, or a one Year Security Update Subscription for $20.

It's a win-win situation....

Nobody would pay $5 per patch. That would be hundreds per year and Microsoft would abuse it with extraneous patches that don't address vulnerabilities. Users would have no way to verify if they needed a given patch, so they'd pay through the nose.

I think the $20/year would be reasonable, but if they did it I'd bet they'd ask for more, probably $35-50/yr, to incentivize folks to upgrade. It would be a cash cow for Microsoft given the number of XP machines still kick'n. If MS did this, and only 20M of the tens of millions of XP users signed up, that's an extra $400M in revenue per year. Maybe $400M/year means nothing to the execs at Microsoft which did $75B in revenue last year. The net profit on that $400M would likely be well over $350M. Regardless of his company's size or revenue stream, any other CFO on the planet would sacrifice his offspring for an extra $400M/$350M/year in revenu/profits. And he'd face a firing squad if/when his investors found out he took a pass on it.

Comment: The big lies (Score 1) 582

by hardwarefreak (#45570959) Attached to: The Dismantling of POTS: Bold Move Or Grave Error?

The first lie: “Our current infrastructure has served us well for almost a century but it no longer meets the needs of America?s consumers,” AT&T senior executive vice president Jim Cicconi said

Analog phones and analog dial tones perfectly serve everyone needing a voice call with another party, and they always will.

The second and 3rd lies: Cleland, a former White House telecom policy adviser, said that even if people wanted to keep the old system, “they are not making the switches anymore for this. And the engineers they need to keep it alive are retiring.”

The POTS system is fully mature, fully built. And there are fewer customers on POTS. So there is no need for new production switches, only parts to maintain the currently installed units. Which is why the industry stopped producing new units. But they haven't stopped producing parts. This argument is a red herring.

And when anyone says "nobody knows how to do it anymore, they're all retiring" you KNOW the entire argument is bullsh-t. This argument is literally the equivalent of saying "it's impossible to train new people to do this job". Ahem, if the job pays well, people will gladly learn to do it.

Make no mistake my fellow slashdot readers, the push by the telcos to switch the last mile of analog copper to digital has nothing to do with any of these 3 lies. It is all about profit motive. They have all been losing money as many customers have switched entirely to cell phones for voice, and cable TV for internet service. The city centric telcos want digital to the home phone so they can charge more for additional mandatory bundled services, generate more revenue by displaying ads on the new phones with big multi-line displays that people will be forced to buy, etc, etc. Want proof from the article itself?

'though the transition should not be harmed by “burdensome economic regulations,” such as mandates or price caps.'

This says the ILECs (Incumbent Local Exchange Carriers) should be allowed to charge whatever they want for the new digital phone service. Every person but one they interviewed for that article is a paid shill for the ILECs.

"Anna-Maria Kovacs, a visiting scholar at Georgetown?s Center for Business and Public Policy, stressed that phone companies “must be allowed to repurpose the capital that is currently deployed to support their obsolete circuit-switched networks” during the switch to guarantee a competitive edge."

Since when does a public policy scholar shill the position of a corporate entity? "MUST be allowed... to guarantee a competitive edge."? When she and her study are funded by these corporate interests, not by the taxpayers.

Switching the last mile of copper from analog to digital is a big loser for the consumer from both a reliability of infrastructure and monthly cost perspective. Digital only phone service over copper will be inherently less reliable and will cost significantly more than POTS. Switching to digital won't save the telcos any money. They're banking on charging customers more for it, for extra mandatory services nobody wants or needs.

Comment: Re:Misleading summary (Score 1) 114

by hardwarefreak (#45556921) Attached to: FOIA: NSA Contracts Stored In Paper Files, Unsearchable, Unindexed

In practice, a judge might decide they should be able to do the search in a reasonable amount of time, and force them to comply.

Judges don't force the NSA to do anything. Kiddie porn puts a judge in the slammer as quickly as anyone else. The "That's not mine, I didn't download that! And those logs must have been fabricated! I'm being setup!" doesn't work any better for judges than anyone else, when it comes to kiddie porn.

Comment: Re:...and (Score 1) 182

by hardwarefreak (#45556733) Attached to: SSD Manufacturer OCZ Preparing For Bankruptcy

SandForce was purchased by LSI quite some time ago. LSI wanted directional control of the silicon to make sure future enhancements fit their needs. That and the fact LSI has always been a storage centric IP/ASIC company, making such an acquisition a perfect fit with their core business. The SF22xx controllers are used in all of LSI's enterprise PCIe SSDs, including the $35K, 3.2TB, model BFH8-3200:

http://www.lsi.com/products/flash-accelerators/pages/nytro-warpdrive-bfh8-3200.aspx#tab/tab0
http://www.cdw.com/shop/products/LSI-Nytro-WarpDrive-BFH8-3200-solid-state-drive-3.2-TB-PCI-Express-3/3072358.aspx

If the SandForce controller is good enough for a $35K enterprise device it's more than good enough for your needs. Your shunning of the SF controllers, simply due to low OCZ product quality, is unwarranted. They're one of the two best (Samsung being the other) SSD flash controllers on the planet, hands down. OCZ's problems weren't due to the SF controllers.

Comment: Re:Belgium is a NATO member (Score 1) 264

by hardwarefreak (#45523385) Attached to: NSA Infected 50,000 Computer Networks With Malicious Software

Yes it does matter that some countries do have moral standards. Unlike, as displayed by the article, USA and UK.

To not spy is immoral. Spying on enemies prevented WWIII. Spying prevented the Cold War from going hot and killing billions of people. Spying on friends in the current era keeps them honest, so to speak.

Kleeneness is next to Godelness.

Working...