If your server starts to swap, you've lost the battle
Not necesarally. A smart pre-emptive swapper can swap out things even if you have 50% free memory. If some amount of memory is only accessed extremely rarely (how often does '/sbin/getty 38400 tty2' really get used, or if a program starts leaking memory) the swapper can stick it on disk. Now this dead block of memory can be used for more cache, which is always a good thing.
Sun could have put that $6000 to good use. That would have been enough to pay an intern for the summer...
Erm, $6k is about 1 month's direct salary as an intern, not including all the hiring and related overhead. But you're just a trolling AC anyway, so who cares about facts.
Do we really need this many articles from Ars posted on/.? Don't get me wrong, I love Ars and it's on my daily read list. It's just that since I've already read Ars, do I really need to see it again? A quick search shows 5 articles in the past 3 days, and an average of ~1 article/day for the last week or two.
Haven't you heard? AMD is going to buy out ATI based on the "PC food chain" (load of shit). ATIs been doing more for cell phone designs (true) so it totally makes sense for them to buy RIM (I'm making this up). And once they have RIM it would be totally synergistic for them to partner to with Apple on a new product, like wow stock market price! Could you just imagine, you'd have a Crackberry with iTunes, a spiffy graphics card, and a 64-bit CPU! How much more synergistic could you get?!
Really, these Wall Street types are annoying. I can make up stuff just as well as they can, and I have just as good chance of being right.
Not really, X-bit has some of the best, in-depth reviews out there. Sure they're long but they have lots of meat to them. It's not/new page/ like Tom's/new page/ reviews which/new page/ are completely useless.
This could be an instance of "if you can't beat them, join them." There's going to be typosquatting no matter what. Since it's not going away Google might as well a) make some money off of it, and b) know where all these fake sites are to remove them from their listings.
Not saying it's the right thing to do, just an idea.
If you're a small college why on earth do you need an OC-12? Also if you're connecting to a GigaPop you pay them for some portion of their connector fee. It even says that: "A Participant that is not also a Connector will not see this fee directly, but should expect to pay to its Connector its appropriate share of this fee (at the discretion of the Connector)." Overall you're talking something on the order of $50-100k for a small college. Considering that a T3 costs on the order of $100-150k/year, if you have any amount of traffic going across the I2 link instead of your comodity link it'll actually be *cheaper*. Plus by peering with a POP you can usually hit a lot of local in-town sites across the "free" POP peering connections.
In conclusion you are either trolling, or have amazingly stupid accountants and IT staff.
Re:Please don't joke about this.
on
ATI's 1GB Video Card
·
· Score: 2, Informative
Sorry, but it's almost certinaly talking about 4x32bit SP floats. It's simply marketing multiplying out numbers for a bigger one.
> Really, wouldn't it be better to stick with a known system and, you know, do your job as a sysadmin by fixing any security holes?
A lot easier said than done for a number of windows-based "solutions." I'm always amused by how often we kick the PoS (point of sale or piece of shit, take your pick) systems in our building offline because some new virus comes around and infects them all. As he pointed out you can isolate them through layers of external protection, but it's a hassle and it would be a lot nicer if they just didn't suck to begin with.
Did you even read the document? "But when -0 is mishandled..." The paper is entirely about java's handling of special-case conditions like -0, it has nothing to do with precision.
It's good that you have no idea what you're talking about. IEEE Single-Precision floating point numbers have a 23-bit mantissa (fraction). Because of the implied 1 that is not stored you actually have a 24 bit number. 2^24 = 16,777,216 so you can precisely represent integers between 1 and 16 million with no error.
(10 kilometers) / (1 centimeter) = 1,000,000. 1 million is much less than 16 million. Please play again.
Anyway, if all you want is 1 centimeter precision why not use standard integer arithmatic? It'll be orders of magnitude faster than floating point arithmatic and (assuming signed numbers) you get a range of (2^31) * 1 centimeter = 21,474.8365 kilometers (for reference Earth's diameter is 12,756.3 km).
With all that said you actually want far greater than centimeter precision and there are good reasons to use floating point numbers, but you wouldn't know any of them. And the parent is correct, if your game depends on dual-precision, you're doing something wrong.
A hack for DRM10 (what's on WMVs) has been floating around for about a year now. Basically you play it normally and it intercepts the license. It then uses the intercepted license to decode the video.
The point was that if they spent the time working with this format and re-implementing it in WINE, they should have seen the potential exploit. Instead they blindly implemented it without analyzing the format (which you can't really blame them, considering how much crap they have to parse through).
Slashdot Mirror
Not necesarally. A smart pre-emptive swapper can swap out things even if you have 50% free memory. If some amount of memory is only accessed extremely rarely (how often does '/sbin/getty 38400 tty2' really get used, or if a program starts leaking memory) the swapper can stick it on disk. Now this dead block of memory can be used for more cache, which is always a good thing.
Erm, $6k is about 1 month's direct salary as an intern, not including all the hiring and related overhead. But you're just a trolling AC anyway, so who cares about facts.
Ars Technica already has posted a follow-up to the original story that says this isn't actually needed.
From what I've read that's basically it. Oh, and probably some magic drivers to make it all work.
Who among you is pathetic enough to post a comment on a slashback article on a Friday night?
Those of us still at work, "being productive".
Availability of Treo 650 smartphone without a camera
I know I saw it listed on sprint's website a month or two ago, but now they are only listing the 700p.
ClickMonkeys worked for Pets.com and MSN, certainly it can work for you!
I wonder where an online OS fits between the hourses and lottery balls flushing Ted Steven's pipes. (Last night's Daily Show on Net Neutrality)
Do we really need this many articles from Ars posted on /.? Don't get me wrong, I love Ars and it's on my daily read list. It's just that since I've already read Ars, do I really need to see it again? A quick search shows 5 articles in the past 3 days, and an average of ~1 article/day for the last week or two.
Haven't you heard? AMD is going to buy out ATI based on the "PC food chain" (load of shit). ATIs been doing more for cell phone designs (true) so it totally makes sense for them to buy RIM (I'm making this up). And once they have RIM it would be totally synergistic for them to partner to with Apple on a new product, like wow stock market price! Could you just imagine, you'd have a Crackberry with iTunes, a spiffy graphics card, and a 64-bit CPU! How much more synergistic could you get?!
Really, these Wall Street types are annoying. I can make up stuff just as well as they can, and I have just as good chance of being right.
I'm the one that hosted those. I took them down because you can get PDF's from AMD's site now of the presentation: http://www.amd.com/us-en/Corporate/InvestorRelatio ns/0,,51_306_14047,00.html
> It should make absolutely no difference whether the product being promoted is yours or a third party's.
Unless you're a convicted monopoly.
Company was 3dfx. The library was Glide.
Not really, X-bit has some of the best, in-depth reviews out there. Sure they're long but they have lots of meat to them. It's not /new page/ like Tom's /new page/ reviews which /new page/ are completely useless.
This could be an instance of "if you can't beat them, join them." There's going to be typosquatting no matter what. Since it's not going away Google might as well a) make some money off of it, and b) know where all these fake sites are to remove them from their listings.
Not saying it's the right thing to do, just an idea.
If you're a small college why on earth do you need an OC-12? Also if you're connecting to a GigaPop you pay them for some portion of their connector fee. It even says that: "A Participant that is not also a Connector will not see this fee directly, but should expect to pay to its Connector its appropriate share of this fee (at the discretion of the Connector)." Overall you're talking something on the order of $50-100k for a small college. Considering that a T3 costs on the order of $100-150k/year, if you have any amount of traffic going across the I2 link instead of your comodity link it'll actually be *cheaper*. Plus by peering with a POP you can usually hit a lot of local in-town sites across the "free" POP peering connections.
In conclusion you are either trolling, or have amazingly stupid accountants and IT staff.
Sorry, but it's almost certinaly talking about 4x32bit SP floats. It's simply marketing multiplying out numbers for a bigger one.
> Really, wouldn't it be better to stick with a known system and, you know, do your job as a sysadmin by fixing any security holes?
A lot easier said than done for a number of windows-based "solutions." I'm always amused by how often we kick the PoS (point of sale or piece of shit, take your pick) systems in our building offline because some new virus comes around and infects them all. As he pointed out you can isolate them through layers of external protection, but it's a hassle and it would be a lot nicer if they just didn't suck to begin with.
> And fixed-point isn't integer, bozo.
Yes it is, as long as you're willing to put a few seconds of thought into it (or just google for the answer).
Did you even read the document? "But when -0 is mishandled..." The paper is entirely about java's handling of special-case conditions like -0, it has nothing to do with precision.
It's good that you have no idea what you're talking about. IEEE Single-Precision floating point numbers have a 23-bit mantissa (fraction). Because of the implied 1 that is not stored you actually have a 24 bit number. 2^24 = 16,777,216 so you can precisely represent integers between 1 and 16 million with no error.
(10 kilometers) / (1 centimeter) = 1,000,000. 1 million is much less than 16 million. Please play again.
Anyway, if all you want is 1 centimeter precision why not use standard integer arithmatic? It'll be orders of magnitude faster than floating point arithmatic and (assuming signed numbers) you get a range of (2^31) * 1 centimeter = 21,474.8365 kilometers (for reference Earth's diameter is 12,756.3 km).
With all that said you actually want far greater than centimeter precision and there are good reasons to use floating point numbers, but you wouldn't know any of them. And the parent is correct, if your game depends on dual-precision, you're doing something wrong.
And that's amazingly useless. Number of transistors and all that means absolutely nothing for final performance.
Actually it's better suited for a quicky. Anyone remember those?
A hack for DRM10 (what's on WMVs) has been floating around for about a year now. Basically you play it normally and it intercepts the license. It then uses the intercepted license to decode the video.
The point was that if they spent the time working with this format and re-implementing it in WINE, they should have seen the potential exploit. Instead they blindly implemented it without analyzing the format (which you can't really blame them, considering how much crap they have to parse through).