If you're an investor, owning shares in a company that has almost all of, but a shrinking share of a shrinking market isn't a happy place to be, especially if they have no room for growth and are trimming their failed attempts to find new markets. Add that their flagship product is running in the single digits, their Marketingefforts are the not only the butt of much comedy but may cost more than the GDP of Haiti and you have the perfect storm.
It's more fun to be holding a company that's growing share, sales and profits too. A company that only holds 10% of its target markets. A company that can report record profits in a bloodbath holiday quarter in the middle of a dire recession? A company whose advertising is so enjoyable that it's viral. A company that's innovating and inventing new markets. That's more fun. That's a winner.
And that winner isn't MSFT. Their stock is where it was 10 years ago. Over the same period Apple is up 1000%. Unlike Microsoft they have 90% of the established market to get yet, and the prospect of undiscovered country.
/14 links? That's informative. Pretty sure you regret posting that now. Let's go again.
That's why they pay the library programmers good bucks. So they'll put down their WOW raids for a little while and spew the good libraries that abstract stuff like this.
Itanium is a turd, it's true. Larrabee isn't. I don't have any evidence to offer you, just opinion. Keep your eye on this one. And the Flash memory thing too, unless they sell it. Oh, and the Atom thing as well. All of them are good stuff if they've got the guts to push them. I think they do, but they'll have to stave off engineers screaming "cannibals are eating my sacred cow!"
We all make that mistake now and then. Once upon a forgotten era on Usenet I did that. My mistake. That's how I wound up being an anonymous contributor to your motd file, never to be forgotten. Oddly enough, the stars of the IT world do come out and opine on the issues of the day in public forums like this one. The more you know...
Bruce, you always get modded +5, so you don't need my praise, but you're so right.
A lot of people don't remember the basic stuff that Intel has made open so that we can enjoy the diverse computing environment that we have. PCI, USB, and SATA are all standards that Intel had a hand in. And then there's formfactors.org. If it weren't for Intel no doubt by now we'd all be referring to motherboards as "planar boards" and HDDs as "fixed disks". And paying $8000 for a basic word processing workstation.
On some level they "get" that they need the chaos of a diverse marketplace to make a fair bazaar in which to offer their product. That's wisdom you can't buy. Kudos to them.
There are so many problems with your post, I don't know where to start, but I'll try.
Java. Not gonna happen. Java is driven by Sun and Intel isn't going to buy into that before the actual sun goes Red Giant.
Larrabee + 4 way Intel CPU. Likewise not gonna happen. Dual Larrabee with an 8-way Intel CPU maybe. Time does funny things to specs.
General routines. Intel will almost certainly release good libraries that exploit the Larrabee CPU. They'll be written in Assembler, C and C++ because that's what people who work at that level use, and the libraries will have bindings for all the popular languages. They'll abstract the threading in a coherent and useful way because their software engineers are first rate. They'll be well documented and they'll be open source. But then at the last minute they'll balk and the demo source will be for Visual C# and.NET because the people at the top think they know how to swing a deal. The GCC port should take about 3 days and be GPL.
What Sony needs. Sony has no clue what they need. They sell a lot of TVs, so they have a lot of money. They use it to fund adventures like Mariah Carey's adventures in film and failed media formats. They'll field three or four solutions and screw them all up. Again.
They already sell more graphics chips than the competition so why bother?
They care. Their main processor vendor bought a graphics chip maker. You better believe they need to win this one.
And they can. The libraries matter a lot. If they publish a good toolchain and make a chip that does what it looks like this does, they're in the money. After that game people still have to run with it but I'm confident that ray tracing will win out over compositing in the long run if they play it right. Intel is not Sony (except as regards that Itanium beast. WTH?)
The upside is that Larrabee is way cool for stuff you would ordinarily use a cluster for. It's like GPGPU, only with real cores. A lot of sexy off-plan science engineering is going to come out of this.
The report, authored by researchers at Rice University and Texas A&M University's Texas Transportation Institute, showed crashes increased slightly at intersection approaches where cameras had been installed. The number of crashes, however, rose dramatically at unmonitored lanes of those same intersections, leading the study authors to conclude that the cameras had kept collisions lower than they would have been without the devices.
Astonishing. More accidents in lanes with cameras, crashes up dramatically in unmonitored lanes of the same intersections. Basically, the cameras increase accidents. And so of course...
They also could add to the 70 cameras now placed at 50 intersections around the city.
If Linux for the desktop is ever going to really be a viable option, someone needs to come out with a distro with the goal of, "absolutely, positively, 100% Windows Compatible" via Wine or similar technologies.
It would probably be more productive for software developers to code for Linux natively. It's not like it's hard.
I had that happen with the W7 Beta. Go to YouTube and start a video. Click the full screen button and seize up hard. No mouse, no keyboard, no three finger salute would bring it around. Happened every time.
It's just that slapping some RAM and a little power storage on a SSD might be a nice way to improve...
Agreed. They're actually building this into next generation spindle drives too. Pretty slick stuff. I'm looking forward to it. You've got another really good point in there about designing a special purpose all-ram card without using DIMMS. I'm sure that could deliver some really high end performance. Of course it would cost an arm and a leg, but the people who buy in that world don't really think too much about price. You could probably put a main card and two daughters with RAM per add-in card and reserve some space on the end for an exhaust fan like a graphics card. That should get you enough surface area for about 360GB. I hate to think what that would cost, but it could saturate a PCIe 2.0 X16 connection for sure.
It was all bullshit speculation anyway.
Actually, not as much as you probably think. You would be surprised what can happen.
There are costs and costs. You can get PC2-6400 for under $10 a gig. But then you have to buy cards for it and they're not free. There's no way you're going to fit 40 of those on one card, so you're going to need a lot more cards, and best case you can't scale up to the 320GB of the top end card here without filling the entire box with RAM, and you can fit multiple SSD cards in one server too. If you buy the more capacious sticks, the price goes up exponentially for both the sticks and the cards. And then there's the watts. Idle RAM still requires power, but idle flash does not. Part of the point is to shrink the server parts so you can fit more servers in one rack and improve power efficiency and reduce costs by running more virtual servers on one real server.
Forgive me for digressing but we're off the track. The common folk don't see the software in the server room. The server software isn't representative of the company and vice versa to them. It's harder to attach a symbol to something they can't see. Harder, but not impossible.
Something that's in front of them every day like desktop software, especially when it shows them both symbols together each morning when they start their PC, that's easy. It says "Microsoft" and "Vista" right on it, doesn't it? They must have some relation. It's not that hard to push it the extra inch and say the one is the other - especially when afterward the software itself will reinforce the connection every morning.
I could, indeed, write a piece of code that would wear out parts of your SSD within days.
Could you? Would you fill the disk almost completely full, and then write and delete the last block over and over with random data 100,000 times per second?
Gee, if they had though of that they might have done something really clever like include a RAM cache and a thousand extra blocks you can't see, and happily report the block written and deleted when it really wasn't, or actually write it to a different physical block each time. They might have had a stroke of genius and included logic to move least-written data to the heavily used blocks and let you bang on fresh ones now and then. It would take a real men of genius engineer to predict this pathological case and include a special purpose computer onboard to deal with it. At least it would if the engineers didn't read slashdot where we've discussed these problems to death for years and years.
They would have done it transparently in the device logic without even telling you because the device is solid state and every bit is as close as every other bit so latency is not a problem. But no, if they were that clever they would have also included some spare bytes in every block and a map so that if a bit in the middle went bad it wouldn't knock out the whole block and some sort of error detection and correction mechanism. It a fit of brilliance they might even have planned for a heavily worn block with too many burned out bits to borrow unused spare bits from another block. Gee, if they were practically omniscient they might have included programmable firmware in case they needed to push out a cure for pathological case they hadn't considered yet.
That is, if they were clever (pdf) that way (pdf).
And if you're trying that hard to break it, a spinning disk won't hold up long at all either.
I have to admit that I was concerned about this too as I discussed this issue with an engineer five years ago. However, even the palimpsest of Archimedes survives to this day. With digital technologies we can do better. It turns out that by providing wear levelling and planning for the predictable degradation of your media, you can design a controller that provides reliable access to written data transparently to the user, despite the fallibility of the media. With sufficient parallel redundancy you can do so without even alerting the user to your difficulty, and in the extreme case you can degrade gracefully. When the devices are solid state, you can build predictable reliable performance for a specific time/use. That's what these devices do. Your concern is unwarranted. We figured that out.
Microsoft is Vista. Vista is Microsoft. These failures are by design. They're deliberate. They're intentional. They couldn't be otherwise -- Microsoft has the entire history of systems design to draw from, the best minds on H1B visas from around the world. They've been in the business for a long time. They know what they're doing and they're failing on purpose.
"What purpose?" Is an open question. I confess I don't know it. But they must have a reason.
Whatever it is, my guidance is still the same: If you can't bear Linux, get a Mac.
Slashdot Mirror
If you're an investor, owning shares in a company that has almost all of, but a shrinking share of a shrinking market isn't a happy place to be, especially if they have no room for growth and are trimming their failed attempts to find new markets. Add that their flagship product is running in the single digits, their Marketing efforts are the not only the butt of much comedy but may cost more than the GDP of Haiti and you have the perfect storm.
It's more fun to be holding a company that's growing share, sales and profits too. A company that only holds 10% of its target markets. A company that can report record profits in a bloodbath holiday quarter in the middle of a dire recession? A company whose advertising is so enjoyable that it's viral. A company that's innovating and inventing new markets. That's more fun. That's a winner.
And that winner isn't MSFT. Their stock is where it was 10 years ago. Over the same period Apple is up 1000%. Unlike Microsoft they have 90% of the established market to get yet, and the prospect of undiscovered country.
/14 links? That's informative. Pretty sure you regret posting that now. Let's go again.
Whoosh.
Oh?
if you program it directly
That's why they pay the library programmers good bucks. So they'll put down their WOW raids for a little while and spew the good libraries that abstract stuff like this.
Itanium is a turd, it's true. Larrabee isn't. I don't have any evidence to offer you, just opinion. Keep your eye on this one. And the Flash memory thing too, unless they sell it. Oh, and the Atom thing as well. All of them are good stuff if they've got the guts to push them. I think they do, but they'll have to stave off engineers screaming "cannibals are eating my sacred cow!"
Never before have I had the opportunity to get so much karma for saying so little. I'm going to savor the moment for a bit, ok?
.
/
o
.
/
o
.
/
o
.
/
o
You're retarded.
You do know that's the real Bruce, right?
We all make that mistake now and then. Once upon a forgotten era on Usenet I did that. My mistake. That's how I wound up being an anonymous contributor to your motd file, never to be forgotten. Oddly enough, the stars of the IT world do come out and opine on the issues of the day in public forums like this one. The more you know...
Bruce, you always get modded +5, so you don't need my praise, but you're so right.
A lot of people don't remember the basic stuff that Intel has made open so that we can enjoy the diverse computing environment that we have. PCI, USB, and SATA are all standards that Intel had a hand in. And then there's formfactors.org. If it weren't for Intel no doubt by now we'd all be referring to motherboards as "planar boards" and HDDs as "fixed disks". And paying $8000 for a basic word processing workstation.
On some level they "get" that they need the chaos of a diverse marketplace to make a fair bazaar in which to offer their product. That's wisdom you can't buy. Kudos to them.
/ no, I don't work for Intel.
Isn't the article. It's the discussion. There's lots of bright minds here. This is why most people don't read TFA.
There are so many problems with your post, I don't know where to start, but I'll try.
Java. Not gonna happen. Java is driven by Sun and Intel isn't going to buy into that before the actual sun goes Red Giant.
Larrabee + 4 way Intel CPU. Likewise not gonna happen. Dual Larrabee with an 8-way Intel CPU maybe. Time does funny things to specs.
General routines. Intel will almost certainly release good libraries that exploit the Larrabee CPU. They'll be written in Assembler, C and C++ because that's what people who work at that level use, and the libraries will have bindings for all the popular languages. They'll abstract the threading in a coherent and useful way because their software engineers are first rate. They'll be well documented and they'll be open source. But then at the last minute they'll balk and the demo source will be for Visual C# and .NET because the people at the top think they know how to swing a deal. The GCC port should take about 3 days and be GPL.
What Sony needs. Sony has no clue what they need. They sell a lot of TVs, so they have a lot of money. They use it to fund adventures like Mariah Carey's adventures in film and failed media formats. They'll field three or four solutions and screw them all up. Again.
They already sell more graphics chips than the competition so why bother?
They care. Their main processor vendor bought a graphics chip maker. You better believe they need to win this one.
And they can. The libraries matter a lot. If they publish a good toolchain and make a chip that does what it looks like this does, they're in the money. After that game people still have to run with it but I'm confident that ray tracing will win out over compositing in the long run if they play it right. Intel is not Sony (except as regards that Itanium beast. WTH?)
The upside is that Larrabee is way cool for stuff you would ordinarily use a cluster for. It's like GPGPU, only with real cores. A lot of sexy off-plan science engineering is going to come out of this.
Sony's real mistake with the PS3 was expecting the Cell processor to be the most incredible computing device ever.
No, the Cell processor is the most incredible computing device yet. Sony's mistake was thinking that was all that mattered.
Hint: the processor isn't the only technology in the machine. Technology doesn't make the game "fun".
I just read this astonishing report in Houston.
The report, authored by researchers at Rice University and Texas A&M University's Texas Transportation Institute, showed crashes increased slightly at intersection approaches where cameras had been installed. The number of crashes, however, rose dramatically at unmonitored lanes of those same intersections, leading the study authors to conclude that the cameras had kept collisions lower than they would have been without the devices.
Astonishing. More accidents in lanes with cameras, crashes up dramatically in unmonitored lanes of the same intersections. Basically, the cameras increase accidents. And so of course...
They also could add to the 70 cameras now placed at 50 intersections around the city.
You can't make this stuff up.
If Linux for the desktop is ever going to really be a viable option, someone needs to come out with a distro with the goal of, "absolutely, positively, 100% Windows Compatible" via Wine or similar technologies.
It would probably be more productive for software developers to code for Linux natively. It's not like it's hard.
is always no. The internet is not for sissies.
To get the full functionality of this site you need silverlight.
But you can click through to use just the limited flash version. Does that count?
I'm pretty sure AOL can beat you. They've got more CDs than even you can handle.
I had that happen with the W7 Beta. Go to YouTube and start a video. Click the full screen button and seize up hard. No mouse, no keyboard, no three finger salute would bring it around. Happened every time.
It's just that slapping some RAM and a little power storage on a SSD might be a nice way to improve ...
Agreed. They're actually building this into next generation spindle drives too. Pretty slick stuff. I'm looking forward to it. You've got another really good point in there about designing a special purpose all-ram card without using DIMMS. I'm sure that could deliver some really high end performance. Of course it would cost an arm and a leg, but the people who buy in that world don't really think too much about price. You could probably put a main card and two daughters with RAM per add-in card and reserve some space on the end for an exhaust fan like a graphics card. That should get you enough surface area for about 360GB. I hate to think what that would cost, but it could saturate a PCIe 2.0 X16 connection for sure.
It was all bullshit speculation anyway.
Actually, not as much as you probably think. You would be surprised what can happen.
There are costs and costs. You can get PC2-6400 for under $10 a gig. But then you have to buy cards for it and they're not free. There's no way you're going to fit 40 of those on one card, so you're going to need a lot more cards, and best case you can't scale up to the 320GB of the top end card here without filling the entire box with RAM, and you can fit multiple SSD cards in one server too. If you buy the more capacious sticks, the price goes up exponentially for both the sticks and the cards. And then there's the watts. Idle RAM still requires power, but idle flash does not. Part of the point is to shrink the server parts so you can fit more servers in one rack and improve power efficiency and reduce costs by running more virtual servers on one real server.
There's no way to solve this problem with RAM.
It's ok. They're treating you better now, I see. I knew it was inevitable. I just didn't know it was going to be now.
My, how the Mac has grown up. Sniff. It seems like just yesterday...
Forgive me for digressing but we're off the track. The common folk don't see the software in the server room. The server software isn't representative of the company and vice versa to them. It's harder to attach a symbol to something they can't see. Harder, but not impossible.
Something that's in front of them every day like desktop software, especially when it shows them both symbols together each morning when they start their PC, that's easy. It says "Microsoft" and "Vista" right on it, doesn't it? They must have some relation. It's not that hard to push it the extra inch and say the one is the other - especially when afterward the software itself will reinforce the connection every morning.
Oh, and SharePoint's security model? That's an interesting bit of work. Have you had a look at how that's actually deployed in the field? Doesn't that concern you just a wee bit? Nothing says due diligence like a press release along the lines of "a Security Group employee downloaded documents from the server and fled. We don't know what he got yet, but we will soon because he's using them as evidence in a lawsuit."
I could, indeed, write a piece of code that would wear out parts of your SSD within days.
Could you? Would you fill the disk almost completely full, and then write and delete the last block over and over with random data 100,000 times per second?
Gee, if they had though of that they might have done something really clever like include a RAM cache and a thousand extra blocks you can't see, and happily report the block written and deleted when it really wasn't, or actually write it to a different physical block each time. They might have had a stroke of genius and included logic to move least-written data to the heavily used blocks and let you bang on fresh ones now and then. It would take a real men of genius engineer to predict this pathological case and include a special purpose computer onboard to deal with it. At least it would if the engineers didn't read slashdot where we've discussed these problems to death for years and years.
They would have done it transparently in the device logic without even telling you because the device is solid state and every bit is as close as every other bit so latency is not a problem. But no, if they were that clever they would have also included some spare bytes in every block and a map so that if a bit in the middle went bad it wouldn't knock out the whole block and some sort of error detection and correction mechanism. It a fit of brilliance they might even have planned for a heavily worn block with too many burned out bits to borrow unused spare bits from another block. Gee, if they were practically omniscient they might have included programmable firmware in case they needed to push out a cure for pathological case they hadn't considered yet.
That is, if they were clever (pdf) that way (pdf).
And if you're trying that hard to break it, a spinning disk won't hold up long at all either.
I have to admit that I was concerned about this too as I discussed this issue with an engineer five years ago. However, even the palimpsest of Archimedes survives to this day. With digital technologies we can do better. It turns out that by providing wear levelling and planning for the predictable degradation of your media, you can design a controller that provides reliable access to written data transparently to the user, despite the fallibility of the media. With sufficient parallel redundancy you can do so without even alerting the user to your difficulty, and in the extreme case you can degrade gracefully. When the devices are solid state, you can build predictable reliable performance for a specific time/use. That's what these devices do. Your concern is unwarranted. We figured that out.
Microsoft is Vista. Vista is Microsoft. These failures are by design. They're deliberate. They're intentional. They couldn't be otherwise -- Microsoft has the entire history of systems design to draw from, the best minds on H1B visas from around the world. They've been in the business for a long time. They know what they're doing and they're failing on purpose.
"What purpose?" Is an open question. I confess I don't know it. But they must have a reason.
Whatever it is, my guidance is still the same: If you can't bear Linux, get a Mac.