Because using botnets is highly illegal. A botnet is a network of regular machines in peoples' homes that have been infected with viruses. Spammers then remotely control these machines, instructing them to send spam.
Do the math, 50,000 emails to 1.2 million customers, that is a grand total of 60 billion spams sent. You only send that volume of email via a massive botnet. You have to, in fact, because if one server tried to send 60 billion messages to AOL's mail servers, it would very quickly be blocked. Even a small fraction of that many emails (emails in general, not just spams) would cause AOL's servers to block out the IPs. You need a large number of PCs sending a small number of emails.
Hacking into peoples' computers and using them to send spam, that is computer fraud. That is why this is illegal, and that is why the secret service is involved.
Why are virii getting through the open ports, though? IIS, if up to date with patches, will not be infected by random virii like an unpatched box will. Exactly how many ports would somebody need to open for this solution? It sounds like the only port they need open is 80, for remote access to the cameras hosted on that machine.
Why would opening a handful of specific ports (assuming other ports are needed) lead to random virus infections? It sounds like the solution to all their problems is turning on automatic update on their servers.
Don't get me wrong, I run Apache on all my servers, but IIS isn't exactly a virus magnet when properly patched.
If the US ignores the WTO over this, who would notice? They're already ignoring rulings from NAFTA/WTO (Can't remember which, I think it was NAFTA) ordering the US to stop charging illegal tarrifs on Canadian softwood lumber (and pay back the tarrifs they've already illegally collected), but the US doesn't like the ruling, so they're just going to ignore it.
So I wouldn't be surprised in the slightest if the US decided to ignore the WTO. The US only follows international agreements if the US wants to.
Nice load of crap, until recently the only choice for commercial use of mysql was the $1499 enterprise license, which wasn't year limited. Now the choices all seem to be per year.. M$ starts at free for the express edition (which is freely redistributable even for commercial use, but has limits on the number of CPU's and RAM it can use, and goes to about $4k, which you don't have to keep paying for. The base standard edition of sql server is $900 which is two years of basic mysql. Mysql is free for opensource use, but commercial use of it doesn't count as opensource if anything connecting to the box isn't opensource.
I disagree with your interpretation of the MySQL commercial license. It seems to me like you are free to use it for commercial uses so long as you do not distribute it. Since most companies looking for a database solution for in-house stuff would not need to license it. Remember, the GPL does not force you to give out your source code, you are perfectly within your rights to modify a GPL application and not share your changes. The only time you need to share your changes is if you distribute it. So for most companies, they're not going to need to pay for a MySQL license. When I was working in the QA department of a local software company, the automated testing framework used MySQL. We didn't need a license, because we weren't distributing our system to endusers.
Also, the "MySQL Network" is a support add-on.
I can't comment on the "express" edition of SQL Server, but I should point out that the opensource version of MySQL has no CPU or memory limitations that I'm aware of.
Red hat enterprise edition used to be ~1.5k per license and was node locked, while the base windows server licenses were 1k. That has changed so the basic windows license is now ~370 (ive seen them as low as $150) bucks for web edition (unlimited cal) ~550 for the standard edition (5 CAL) and $800 for the small business version which has sql server, sharepoint outlook etc already included.
Apple to oranges here. The "web" edition of Windows is locked down and you are not permitted to run (server?) software on it beyond a web server. You have to compare the standard edition of Windows, or 550 (plus CAL costs) vs RedHat's 349 (with no CAL costs). Or you can use CentOS, a functionally identical clone of RHEL (compiled from RHEL) for free. Anyhow, RHEL seems to come out to almost half the price after you add some CALs, to me.
says the RHE ES is ~349 bucks, basic edition and goes to $800, AS is ~1500 to ~2500. ES and AS have been in the past diffrent kernel versions and such, We have software where I work that _ONLY_ runs on AS forcing us to buy the $1500 versions to get the updates. This ends up costing us more per machine than the extra CALS for the windows box. In fact last year we spend nearly twice as much on the linux machines as the windows ones. We also ship a product that uses mysql as an internal database. The per year licencing costs are a very significant portion of the sale price of the machine over the expected 10 year lifetime of the products in customers locations.
Both ES and AS use the same major revision of the kernel, 2.6. Beyond that, the differences between the two kernels, if there really are any, would be minor. On top of that, nobody is forcing you to use an RHEL kernel; you can use any kernel you want. The only limitation is that you couldn't install a POWER kernel on ES, since it doesn't have the packages for the POWER processor. Essentially the only real difference between ES and AS other than the lack of support for the IBM POWER arch in ES is the support option differences between the basic and standard packages.
If your RHEL box is costing you so much more (if only because you are using AS for a different kernel), your money is better spent on a competant linux admin who can swap out the correct kernel onto an ES box, or perhaps even use CentOS or another distro instead, if you don't need the support options.
I must be missing something. MySQL costs $595 per year. SQL Server costs $5000 for the standard edition.
MySQL's pricing is per year, but Microsoft's price is per CPU. So on a dual xeon (not unreasonable for a database server), it is $595 vs $10,000. That is 16 years of MySQL for the price of a single SQL Server license, and something tells me most SQL servers are going to be upgraded a bit more frequently than every 16 years.
I'm also pretty sure that Win2K3 costs more than RHEL. RHEL ES (basic server edition) costs $349. Win2K3 basic is $999 plus $199 per 5 CALs.
In both cases, Microsoft charges way more, so I'd say you're wrong.
Indeed, and the article's suggestions include "Run it under WINE" for some of the programs (Including Photoshop IIRC).
They suggest a few alternatives for Photoshop, but they aren't really relative. We've already had this debate on Slashdot before, and the consensus is that even if Gimp offers the same functionality as Photoshop, unless it presents an identical interface, people aren't going to use it. Professionals using Photoshop are content to continue using Photoshop, and they're not going to switch to Linux if they have to learn Gimp (only to find missing functionality they need).
Personally I think that Gimp's interface is braindead, but that's just me. Gimpshop is a decent first step, but it only does menus. Gimp should have aimed to clone Photoshop's interface from the beginning, in my opinion.
I'm looking at this from a user perspective. I open a tab. I close a tab. The memory remains in use. I open and close many tabs, firefox's memory usage balloons. I am unsure, however, if that memoy usage is Firefox's fault, or an extension. It is so hard to tell.
My partner is in Slovenia, I'm in Montreal. 6 hour time difference. We have a well defined area of responsibility that reduces the need for communication rather nicely. In addition, there are still overlapping parts of the day where we are both awake. I try to contact him during those periods. Luckily our need for communication is often not critical.
Really, there is a large period of time where we overlap. He is available from when I wake up until about 6PM my time. Not so bad.
As for how we communicate effectively over such a large distance, three methods. When we're both at a computer, MSN messenger (or other IM). When he isn't and I need to contact him, but it isn't important, email. When he isn't and it is important, telephone.
Telephoning Slovenia is affordable. I use Skype. The quality isn't super amazing, but it beats landlines. Unfortunately, while calling a Slovenian landline is about 0.05 euros, calling a slovenian cellphone is 0.20 euros. So I still try to keep my calls short. It costs more than a normal long distance call within North America, but not so much that it breaks the bank for two or three minute calls.
Yeah, it runs linux, but only the PPE is going to be able to do anything without custom code.
That's my point. While I do not deny that there are uses where the raw number-crunching power of the SPEs is useful for certain tasks, I don't think that their uses are fairly limited.
Take the PS3 for example. From what developers are saying, the vast majority of what they have to do is limited to the single PPE. They have managed to find a few things that can run on the PPE, but not many. Physics engines are one thing. It turns out that while physics engines are highly branch oriented, the branches are so non-deterministic that a branch predictor wouldn't help anyhow.
I also think you're understating how slow branching would make the SPE. It'd be essentially useless on any code with branching compared to the PPE. I understand that compiler optimizations and software branch predictors can make that bearable for the occasional loop, but you still can't really do general purpose tasks on the thing.
So for rendering, or scientific work, sure... But what about in servers?
I think the Cell has a strong future as a DSP. I mean, as I said, it is very good at what it does... Decoding 8 simultaneous HD streams when a normal PC can just barely handle one, that is impressive. Imagine how fast a Cell processor could ENCODE video... Either eight or nine seperate streams encoded per CPU, or one stream encoded eight times faster. But what happens when you want a database server?
Perhaps you mean the PPE and it's supporting hardware, such as the cache? That'd ideally be shared among multiple PPEs.
If you look closely at the PPEs, a huge amount of their real estate seems to go to what looks like their 256KB of cache. Cache takes up a lot of space. Since the PPE's wouldn't each have dedicated cache, they're still about the same size.
They do offer a lot more than 240 lines of resolution. They offer 600 additional lines of resolution (1080 - 480), an increase of 225% in vertical resolution, and an increase of 600% in total pixel count.
This is because BluRay discs support 1080p. In fact, the first BluRay movie announced to be finished authoring (IIRC, Charlie's Angels: Full Throttle) was done in 1080.
Upconversion cannot create data from nothingness. You can use fancy filters for scaling up an image, but you're still only scaling up. Detail cannot be recreated. One day computer programs might be able to guess well enough to simulate that, but upconverters definately don't;)
Personally, I've only seen a 720p feed vs a 480p feed. The difference on my buddy's 50" Samsung DLP was enormous. Night and day. I can only imaging that the difference between 480p and 1080p (or 1080i) would be mind blowing.
Why go with SPEs anyhow? The whole problem with coding for the Cell involves the differences between the PPE and the SPE. The SPE doesn't have branch predictors, making it virtually useless for any sort of flow control.
Why didn't IBM just pack in a lesser number of PPEs? The PPE already seems to be a very lightweight general purpose processing core, unless I'm missing something. It is about the same size as an SPE. So why not just put 9 PPEs on a Cell chip instead of 1 PPE and 8 SPEs?
If you had 9 PPEs on the chip, any multithreaded code (servers for example) would see massive benefits without having to rewrite it to try to find aspects of the program that could run on what is effectively a DSP. While everybody else was fooling around with 2-core processors, they'd have a 9-core processor on the market. Sure, slower per-core, but 9 of them, with that number going up in the future.
Only for new releases. The wholesale price of existing movies (to which people seem to be comparing prices) is $17.95, IIRC.
This is also only Sony Pictures. Other studios may charge more or less for their movies.
In addition, if HDDVD wholesales for less, BluRay prices might drop sharply to match it.
New release DVDs retail where I am for about $22 USD retail. Some a bit more, some a bit less. Say they mark up blu-ray movies got new releases by $5 (I'm pulling this number out of my ass here). You get $22 vs $30. 36% more expensive.
Now, there is one thing people are ignoring here. BluRay movies are HD. DVDs are not. People who own an HDTV may very well feel that 36% higher costs to get the much higher quality HD copy of the movie is worth it. I would pay the extra money to get a movie in HD; SD doesn't look very good on larger HD screens, as I have noticed by playing SD XBox games on a buddy's 50" DLP HDTV. After seeing the difference between SD and HD digital cable feeds, I'd certainly pay $8 more to get a movie in HD.
A patent being invalid would mean that it never was valid. Requiring a company to pay damages for what the patent office admits was their mistake doesn't make any sense.
If the patents are ruled invalid, there shouldn't be any past infringement. Invalid means they were never valid in the first place. If the US legal system doesn't recognize that, it needs some updates itself.
We live in a world of automated updates. If RIM doesn't have an effective update solution in place, that was rather shortsited.
If they don't, I would imagine that any user could simply take their BlackBerry back to their cellular provider to have them apply the update to keep it working. Since the update would be required, the carrier would probably be obligated to provide the service at no cost.
How can RIM be required to pay damages for invalid patents?
Since RIM can't be charged ongoing royalties for a patent they no longer infringe, and since they can't be ordered to shut down a service that no longer violates patents, the only question that remains is how much lump sum they pay out, if any.
RIM can sustain such a thing. A settlement or damages payment isn't a big deal. So long as it doesn't bankrupt RIM, they take the hit and move on. A single amount of money can be replaced. It isn't nearly so damaging as having their service shut down or paying ongoing royalties.
So, from RIM's perspective, the big danger should be over. From NTP's perspective, I suppose they can still try to get some cash out of the deal, but the patents being ruled invalid, if that holds, that'd put a damper on damages. I mean, why pay damages for violating patents that are invalid?
I'm probably biased against NTP for the sole reason that I hate patent-holding companies. I see them as parasites that lurk around waiting to strike. They don't do anything productive, don't produce any products. They just sit around, wait for somebody to do the actual hard work, and then try to step in and claim a share. NTP wrote some stuff down on a piece of paper and never did anything with it. RIM came up with the same thing and spent the time and effort to turn it into a working product. If NTP actually produced a competing product, I'd feel for them. But as a patent holding company, they're really the ones stealing other peoples' work, since they don't put any effort into developing the product they want royalties for.
"Most of the sites targeted were run by small organisations and companies that do not have dedicated security workers and cannot keep up with the latest alerts and patches for vulnerabilities."
Adding "up2date -u" as a cron job is too much of a burden, is it? You need dedicated security personel to take the 30 seconds this requires?
Before you get all updates-must-be-tested-first on me, realize that something like RHEL doesn't do big updates. The only updates they provide are the small security-fix variety. The chance of the entire server going down because of one of these updates is much smaller than the chance of somebody exploiting one of the security flaws.
RIM's other markets were never at risk. RIM isn't even based in the US anyhow. They're based in Waterloo, Canada, IIRC. And we don't seem to have as broken a patent system as the US does.
The "article" seems to indicate that the changes are to the backend only, and absolutely nothing will change for the user. Everything is identical in the GUI and usage. So this should be a kick in the pants for NTP. NTP now has two options:
1) Settle for a paltry amount, far less than what they could have accepted before. 2) Press forward to get an injunction, and have the case die when RIM uses the workaround, without NTP seeing a dime.
All of NTP's patents (All of them now) have now been invalidated (at least preliminarily), and even if NTP succeeds in getting an injunction, those invalid patents won't even apply anymore. NTP doesn't exactly have a strong case anymore, and they have to know it.
Of course, NTP didn't stop when the US government stepped in to warn that an injunction would compromise the nation's security, so I'm pretty sure they fall into the evil-but-stupid category that is too both too evil to abandon the case, and too stupid to realize it is a lost cause.
Raptor SATA can break 88 MBps. Thats 704 Gbps. Add a little bit of network overhead, even for something as raw-metal as ATA-over-IP, you're still talking "damned close". Thats for _one_ hard drive, how passe.
I think you mean 0.704 Gbps. I don't know about gigabit NICs, but with 100mbit NICs I usually see about 80mbit of real-world transfer speed using regular Windows filesharing. Still a bit of leeway there, if the percentage holds true.
Besides, that 88MB/s figure is the peak sequential transfer (at the edge of the platter) for the new 150GB model. With file fragmentation, realworld random access performance is going to be lower.
Sure, with a nice RAID array you can break the gigabit level but even then a two-disk RAID-0 array isn't going to be much over a gigabit in realworld (random access, fragmented files) performance. Any more than that and you're moving beyond consumer-level performance anyhow.
Of course 10 gb is not for "consumers". At least not now, not for what we've got. But some day, if homes really are shuffling around 1080p streams (or bigger) like candy, even your average consumer home might need 10 gb backbones. If 1 gigabit is one single modern hard drive, 10 gigabit is not unreasonable for even a home backbone.
IIRC, 720p is only 25 megabits (MPEG-2). I guess 1080p would be around 50 megabits. You could do fine with 100 megabit ethernet (3 streams of 720p, 1 of 1080p), let alone gigabit ethernet (30 streams of 720p, 15 streams of 1080p). What sort of device would ever need to handle more than 30 HD streams? With a gigabit switch that had a decent backplane bandwidth, you'd be able to handle more than 30 streams between multiple devices, the only time that the 30 streams becomes a bottleneck is if ONE device has to handle it all (or you have more than one switch). I really don't think that will ever be a problem with HD.
Another consideration is that HD is not limited to MPEG-2. Certain HD providers use MPEG-4, or even h.264/AVC (Several US satellite providers use this), which require a lot less bandwidth.
I think it is also safe to say that, considering the half and half approach HD is taking towards MPEG-2 and newer codecs, any future better-than-HD would certainly use one of these newer codecs as a minimum, with the possible option of an even better as of yet uninvented codec.
So I see 1 gigabit ethernet as being nearly overkill for current HD, and more than sufficient for beyond-HD.
For enterprises over 100 employees doing data intensive work, I can picture 10 gigabit being a problem. They need someplace to go. In the meanwhile, the only way anyone else is going to be afford 10 gig is by there being some higher realm start the downpush of commodification.
I don't see 10 gigabit as being a problem. There are so very few enterprises that would need MORE than 10 gigabits per second of networked access (Local caching does wonders), since virtually no enterprises currently even have 10 gigabit.
Besides, there would probably be a greater benefit from taking a 10 gigabit network and adding 100 gigabit inter-switch backbone connections rather than upgrading every port to 100 gigabits. It'd certainly cost a lot less.
Slashdot Mirror
Errm? 50,000 emails per user. Sent to 1.2 million users. 60 billion total.
Because using botnets is highly illegal. A botnet is a network of regular machines in peoples' homes that have been infected with viruses. Spammers then remotely control these machines, instructing them to send spam.
Do the math, 50,000 emails to 1.2 million customers, that is a grand total of 60 billion spams sent. You only send that volume of email via a massive botnet. You have to, in fact, because if one server tried to send 60 billion messages to AOL's mail servers, it would very quickly be blocked. Even a small fraction of that many emails (emails in general, not just spams) would cause AOL's servers to block out the IPs. You need a large number of PCs sending a small number of emails.
Hacking into peoples' computers and using them to send spam, that is computer fraud. That is why this is illegal, and that is why the secret service is involved.
Why are virii getting through the open ports, though? IIS, if up to date with patches, will not be infected by random virii like an unpatched box will. Exactly how many ports would somebody need to open for this solution? It sounds like the only port they need open is 80, for remote access to the cameras hosted on that machine.
Why would opening a handful of specific ports (assuming other ports are needed) lead to random virus infections? It sounds like the solution to all their problems is turning on automatic update on their servers.
Don't get me wrong, I run Apache on all my servers, but IIS isn't exactly a virus magnet when properly patched.
Using Hymn, which can decrypt legally purchased music.
It only runs on OSX and Windows, but since the same applies to iTunes, that should not pose a problem.
If the US ignores the WTO over this, who would notice? They're already ignoring rulings from NAFTA/WTO (Can't remember which, I think it was NAFTA) ordering the US to stop charging illegal tarrifs on Canadian softwood lumber (and pay back the tarrifs they've already illegally collected), but the US doesn't like the ruling, so they're just going to ignore it.
So I wouldn't be surprised in the slightest if the US decided to ignore the WTO. The US only follows international agreements if the US wants to.
Nice load of crap, until recently the only choice for commercial use of mysql was the $1499 enterprise license, which wasn't year limited. Now the choices all seem to be per year.. M$ starts at free for the express edition (which is freely redistributable even for commercial use, but has limits on the number of CPU's and RAM it can use, and goes to about $4k, which you don't have to keep paying for. The base standard edition of sql server is $900 which is two years of basic mysql. Mysql is free for opensource use, but commercial use of it doesn't count as opensource if anything connecting to the box isn't opensource.
I disagree with your interpretation of the MySQL commercial license. It seems to me like you are free to use it for commercial uses so long as you do not distribute it. Since most companies looking for a database solution for in-house stuff would not need to license it. Remember, the GPL does not force you to give out your source code, you are perfectly within your rights to modify a GPL application and not share your changes. The only time you need to share your changes is if you distribute it. So for most companies, they're not going to need to pay for a MySQL license. When I was working in the QA department of a local software company, the automated testing framework used MySQL. We didn't need a license, because we weren't distributing our system to endusers.
Also, the "MySQL Network" is a support add-on.
I can't comment on the "express" edition of SQL Server, but I should point out that the opensource version of MySQL has no CPU or memory limitations that I'm aware of.
Red hat enterprise edition used to be ~1.5k per license and was node locked, while the base windows server licenses were 1k. That has changed so the basic windows license is now ~370 (ive seen them as low as $150) bucks for web edition (unlimited cal) ~550 for the standard edition (5 CAL) and $800 for the small business version which has sql server, sharepoint outlook etc already included.
Apple to oranges here. The "web" edition of Windows is locked down and you are not permitted to run (server?) software on it beyond a web server. You have to compare the standard edition of Windows, or 550 (plus CAL costs) vs RedHat's 349 (with no CAL costs). Or you can use CentOS, a functionally identical clone of RHEL (compiled from RHEL) for free. Anyhow, RHEL seems to come out to almost half the price after you add some CALs, to me.
says the RHE ES is ~349 bucks, basic edition and goes to $800, AS is ~1500 to ~2500. ES and AS have been in the past diffrent kernel versions and such, We have software where I work that _ONLY_ runs on AS forcing us to buy the $1500 versions to get the updates. This ends up costing us more per machine than the extra CALS for the windows box. In fact last year we spend nearly twice as much on the linux machines as the windows ones. We also ship a product that uses mysql as an internal database. The per year licencing costs are a very significant portion of the sale price of the machine over the expected 10 year lifetime of the products in customers locations.
Both ES and AS use the same major revision of the kernel, 2.6. Beyond that, the differences between the two kernels, if there really are any, would be minor. On top of that, nobody is forcing you to use an RHEL kernel; you can use any kernel you want. The only limitation is that you couldn't install a POWER kernel on ES, since it doesn't have the packages for the POWER processor. Essentially the only real difference between ES and AS other than the lack of support for the IBM POWER arch in ES is the support option differences between the basic and standard packages.
If your RHEL box is costing you so much more (if only because you are using AS for a different kernel), your money is better spent on a competant linux admin who can swap out the correct kernel onto an ES box, or perhaps even use CentOS or another distro instead, if you don't need the support options.
As for using MySQL as
I must be missing something. MySQL costs $595 per year. SQL Server costs $5000 for the standard edition.
MySQL's pricing is per year, but Microsoft's price is per CPU. So on a dual xeon (not unreasonable for a database server), it is $595 vs $10,000. That is 16 years of MySQL for the price of a single SQL Server license, and something tells me most SQL servers are going to be upgraded a bit more frequently than every 16 years.
I'm also pretty sure that Win2K3 costs more than RHEL. RHEL ES (basic server edition) costs $349. Win2K3 basic is $999 plus $199 per 5 CALs.
In both cases, Microsoft charges way more, so I'd say you're wrong.
Indeed, and the article's suggestions include "Run it under WINE" for some of the programs (Including Photoshop IIRC).
They suggest a few alternatives for Photoshop, but they aren't really relative. We've already had this debate on Slashdot before, and the consensus is that even if Gimp offers the same functionality as Photoshop, unless it presents an identical interface, people aren't going to use it. Professionals using Photoshop are content to continue using Photoshop, and they're not going to switch to Linux if they have to learn Gimp (only to find missing functionality they need).
Personally I think that Gimp's interface is braindead, but that's just me. Gimpshop is a decent first step, but it only does menus. Gimp should have aimed to clone Photoshop's interface from the beginning, in my opinion.
I'm looking at this from a user perspective. I open a tab. I close a tab. The memory remains in use. I open and close many tabs, firefox's memory usage balloons. I am unsure, however, if that memoy usage is Firefox's fault, or an extension. It is so hard to tell.
The problem is when you close tabs, Firefox doesn't seem to free the memory leak. It seems to stay resident until you kill Firefox. Not cool.
This GPU has nothing to do with the GeForce 5xxx series. This is a GoForce not a GeForce. Totally different, only the numbers are similiar.
This type of chip would be used in cellphones or PDAs, not motherboards, laptops, or graphics cards.
They're even more expensive than Skype for calls to Slovenian mobiles, unfortunately.
My partner is in Slovenia, I'm in Montreal. 6 hour time difference. We have a well defined area of responsibility that reduces the need for communication rather nicely. In addition, there are still overlapping parts of the day where we are both awake. I try to contact him during those periods. Luckily our need for communication is often not critical.
Really, there is a large period of time where we overlap. He is available from when I wake up until about 6PM my time. Not so bad.
As for how we communicate effectively over such a large distance, three methods. When we're both at a computer, MSN messenger (or other IM). When he isn't and I need to contact him, but it isn't important, email. When he isn't and it is important, telephone.
Telephoning Slovenia is affordable. I use Skype. The quality isn't super amazing, but it beats landlines. Unfortunately, while calling a Slovenian landline is about 0.05 euros, calling a slovenian cellphone is 0.20 euros. So I still try to keep my calls short. It costs more than a normal long distance call within North America, but not so much that it breaks the bank for two or three minute calls.
Yeah, it runs linux, but only the PPE is going to be able to do anything without custom code.
That's my point. While I do not deny that there are uses where the raw number-crunching power of the SPEs is useful for certain tasks, I don't think that their uses are fairly limited.
Take the PS3 for example. From what developers are saying, the vast majority of what they have to do is limited to the single PPE. They have managed to find a few things that can run on the PPE, but not many. Physics engines are one thing. It turns out that while physics engines are highly branch oriented, the branches are so non-deterministic that a branch predictor wouldn't help anyhow.
I also think you're understating how slow branching would make the SPE. It'd be essentially useless on any code with branching compared to the PPE. I understand that compiler optimizations and software branch predictors can make that bearable for the occasional loop, but you still can't really do general purpose tasks on the thing.
So for rendering, or scientific work, sure... But what about in servers?
I think the Cell has a strong future as a DSP. I mean, as I said, it is very good at what it does... Decoding 8 simultaneous HD streams when a normal PC can just barely handle one, that is impressive. Imagine how fast a Cell processor could ENCODE video... Either eight or nine seperate streams encoded per CPU, or one stream encoded eight times faster. But what happens when you want a database server?
From the images of the core that I've seen, PPEs are virtually the same size as the SPEs.
i ghlight.jpg
I refer you to this image:
http://images.anandtech.com/reviews/cpu/cell/ppeh
Perhaps you mean the PPE and it's supporting hardware, such as the cache? That'd ideally be shared among multiple PPEs.
If you look closely at the PPEs, a huge amount of their real estate seems to go to what looks like their 256KB of cache. Cache takes up a lot of space. Since the PPE's wouldn't each have dedicated cache, they're still about the same size.
They do offer a lot more than 240 lines of resolution. They offer 600 additional lines of resolution (1080 - 480), an increase of 225% in vertical resolution, and an increase of 600% in total pixel count.
;)
This is because BluRay discs support 1080p. In fact, the first BluRay movie announced to be finished authoring (IIRC, Charlie's Angels: Full Throttle) was done in 1080.
Upconversion cannot create data from nothingness. You can use fancy filters for scaling up an image, but you're still only scaling up. Detail cannot be recreated. One day computer programs might be able to guess well enough to simulate that, but upconverters definately don't
Personally, I've only seen a 720p feed vs a 480p feed. The difference on my buddy's 50" Samsung DLP was enormous. Night and day. I can only imaging that the difference between 480p and 1080p (or 1080i) would be mind blowing.
Why go with SPEs anyhow? The whole problem with coding for the Cell involves the differences between the PPE and the SPE. The SPE doesn't have branch predictors, making it virtually useless for any sort of flow control.
Why didn't IBM just pack in a lesser number of PPEs? The PPE already seems to be a very lightweight general purpose processing core, unless I'm missing something. It is about the same size as an SPE. So why not just put 9 PPEs on a Cell chip instead of 1 PPE and 8 SPEs?
If you had 9 PPEs on the chip, any multithreaded code (servers for example) would see massive benefits without having to rewrite it to try to find aspects of the program that could run on what is effectively a DSP. While everybody else was fooling around with 2-core processors, they'd have a 9-core processor on the market. Sure, slower per-core, but 9 of them, with that number going up in the future.
Or am I missing something here?
Only for new releases. The wholesale price of existing movies (to which people seem to be comparing prices) is $17.95, IIRC.
This is also only Sony Pictures. Other studios may charge more or less for their movies.
In addition, if HDDVD wholesales for less, BluRay prices might drop sharply to match it.
New release DVDs retail where I am for about $22 USD retail. Some a bit more, some a bit less. Say they mark up blu-ray movies got new releases by $5 (I'm pulling this number out of my ass here). You get $22 vs $30. 36% more expensive.
Now, there is one thing people are ignoring here. BluRay movies are HD. DVDs are not. People who own an HDTV may very well feel that 36% higher costs to get the much higher quality HD copy of the movie is worth it. I would pay the extra money to get a movie in HD; SD doesn't look very good on larger HD screens, as I have noticed by playing SD XBox games on a buddy's 50" DLP HDTV. After seeing the difference between SD and HD digital cable feeds, I'd certainly pay $8 more to get a movie in HD.
A patent being invalid would mean that it never was valid. Requiring a company to pay damages for what the patent office admits was their mistake doesn't make any sense.
If the patents are ruled invalid, there shouldn't be any past infringement. Invalid means they were never valid in the first place. If the US legal system doesn't recognize that, it needs some updates itself.
We live in a world of automated updates. If RIM doesn't have an effective update solution in place, that was rather shortsited.
If they don't, I would imagine that any user could simply take their BlackBerry back to their cellular provider to have them apply the update to keep it working. Since the update would be required, the carrier would probably be obligated to provide the service at no cost.
How can RIM be required to pay damages for invalid patents?
Since RIM can't be charged ongoing royalties for a patent they no longer infringe, and since they can't be ordered to shut down a service that no longer violates patents, the only question that remains is how much lump sum they pay out, if any.
RIM can sustain such a thing. A settlement or damages payment isn't a big deal. So long as it doesn't bankrupt RIM, they take the hit and move on. A single amount of money can be replaced. It isn't nearly so damaging as having their service shut down or paying ongoing royalties.
So, from RIM's perspective, the big danger should be over. From NTP's perspective, I suppose they can still try to get some cash out of the deal, but the patents being ruled invalid, if that holds, that'd put a damper on damages. I mean, why pay damages for violating patents that are invalid?
I'm probably biased against NTP for the sole reason that I hate patent-holding companies. I see them as parasites that lurk around waiting to strike. They don't do anything productive, don't produce any products. They just sit around, wait for somebody to do the actual hard work, and then try to step in and claim a share. NTP wrote some stuff down on a piece of paper and never did anything with it. RIM came up with the same thing and spent the time and effort to turn it into a working product. If NTP actually produced a competing product, I'd feel for them. But as a patent holding company, they're really the ones stealing other peoples' work, since they don't put any effort into developing the product they want royalties for.
"Most of the sites targeted were run by small organisations and companies that do not have dedicated security workers and cannot keep up with the latest alerts and patches for vulnerabilities."
Adding "up2date -u" as a cron job is too much of a burden, is it? You need dedicated security personel to take the 30 seconds this requires?
Before you get all updates-must-be-tested-first on me, realize that something like RHEL doesn't do big updates. The only updates they provide are the small security-fix variety. The chance of the entire server going down because of one of these updates is much smaller than the chance of somebody exploiting one of the security flaws.
RIM's other markets were never at risk. RIM isn't even based in the US anyhow. They're based in Waterloo, Canada, IIRC. And we don't seem to have as broken a patent system as the US does.
The "article" seems to indicate that the changes are to the backend only, and absolutely nothing will change for the user. Everything is identical in the GUI and usage. So this should be a kick in the pants for NTP. NTP now has two options:
1) Settle for a paltry amount, far less than what they could have accepted before.
2) Press forward to get an injunction, and have the case die when RIM uses the workaround, without NTP seeing a dime.
All of NTP's patents (All of them now) have now been invalidated (at least preliminarily), and even if NTP succeeds in getting an injunction, those invalid patents won't even apply anymore. NTP doesn't exactly have a strong case anymore, and they have to know it.
Of course, NTP didn't stop when the US government stepped in to warn that an injunction would compromise the nation's security, so I'm pretty sure they fall into the evil-but-stupid category that is too both too evil to abandon the case, and too stupid to realize it is a lost cause.
Raptor SATA can break 88 MBps. Thats 704 Gbps. Add a little bit of network overhead, even for something as raw-metal as ATA-over-IP, you're still talking "damned close". Thats for _one_ hard drive, how passe.
I think you mean 0.704 Gbps. I don't know about gigabit NICs, but with 100mbit NICs I usually see about 80mbit of real-world transfer speed using regular Windows filesharing. Still a bit of leeway there, if the percentage holds true.
Besides, that 88MB/s figure is the peak sequential transfer (at the edge of the platter) for the new 150GB model. With file fragmentation, realworld random access performance is going to be lower.
Sure, with a nice RAID array you can break the gigabit level but even then a two-disk RAID-0 array isn't going to be much over a gigabit in realworld (random access, fragmented files) performance. Any more than that and you're moving beyond consumer-level performance anyhow.
Of course 10 gb is not for "consumers". At least not now, not for what we've got. But some day, if homes really are shuffling around 1080p streams (or bigger) like candy, even your average consumer home might need 10 gb backbones. If 1 gigabit is one single modern hard drive, 10 gigabit is not unreasonable for even a home backbone.
IIRC, 720p is only 25 megabits (MPEG-2). I guess 1080p would be around 50 megabits. You could do fine with 100 megabit ethernet (3 streams of 720p, 1 of 1080p), let alone gigabit ethernet (30 streams of 720p, 15 streams of 1080p). What sort of device would ever need to handle more than 30 HD streams? With a gigabit switch that had a decent backplane bandwidth, you'd be able to handle more than 30 streams between multiple devices, the only time that the 30 streams becomes a bottleneck is if ONE device has to handle it all (or you have more than one switch). I really don't think that will ever be a problem with HD.
Another consideration is that HD is not limited to MPEG-2. Certain HD providers use MPEG-4, or even h.264/AVC (Several US satellite providers use this), which require a lot less bandwidth.
I think it is also safe to say that, considering the half and half approach HD is taking towards MPEG-2 and newer codecs, any future better-than-HD would certainly use one of these newer codecs as a minimum, with the possible option of an even better as of yet uninvented codec.
So I see 1 gigabit ethernet as being nearly overkill for current HD, and more than sufficient for beyond-HD.
For enterprises over 100 employees doing data intensive work, I can picture 10 gigabit being a problem. They need someplace to go. In the meanwhile, the only way anyone else is going to be afford 10 gig is by there being some higher realm start the downpush of commodification.
I don't see 10 gigabit as being a problem. There are so very few enterprises that would need MORE than 10 gigabits per second of networked access (Local caching does wonders), since virtually no enterprises currently even have 10 gigabit.
Besides, there would probably be a greater benefit from taking a 10 gigabit network and adding 100 gigabit inter-switch backbone connections rather than upgrading every port to 100 gigabits. It'd certainly cost a lot less.