GPUs are severely limited on their types of tasks. Instead of a 1600 core GPU, pretend your CPU has as single large SIMD register that can hold 1600 floats. Now, it would be great at crunching large matrices of floats and utterly suck at everything else. That's a GPU in a nutshell. GPUs are absolutely horrible at branches. If one core takes a branch, every core in that core's group must stall and wait for the branch to finish. All cores must be working on the same instruction at the same time and branches mess with that. GPUs != CPUs
When Intel last talked about their 80 core CPUs, they talked about getting rid of cache-coherency in order to scale, AMD also recognizes this. This would mean OSs and most Apps would NOT be backwards compatible even if still using x86 instructions. Although, it would be possible to run apps in a VM that emulated cache-coherency.
"Which Microsoft has been trying to phase out entirely anyway"
MS's commandline is much better than it has ever been. Not sure where you're getting your random info from.
Speaking about OO vs MS Office, I find it hard to even get our customers to send other versions of Office files. Many times I get files in 2007 format and I can't view the files with my 2003 Office. Try telling a customer to save the file as a previous format.. good luck.
Transistor counts double every 18 months, how long until 4GB smart phones? 5 years?
With Linux taking over these parts of the market, how long until someone finds a good use to run MySQL on a smart phone? Web server? I'm sure IPv6 coupled with high bandwidth of future nation wide WiMAX/etc, we will want to connect our phones and vew data.
How long until some major breakthrough will make SSDs crazy small and we have smartphones with 2TB of storage and we bring them to our friends and want to file share?
Add a lightpeak connection, plug your phone into a 19" monitor, hook up a mouse and keyboard and play games. It won't be long.
the enterprise is moving into a virtual environment where you can live migrate images off of idle machines and consolidate them to reduce power. If the current server is getting too loaded, just migrate from that machine to another will less load.
You can already do this with current software, give it 5-10 years and this will be standard.
I mostly see a low power ARM being good for dedicated SAN/NAS/Firewall. Most other SQL/Web stuff will be on VMs where you can dynamically add/remove Memory/CPUs and migrate between physical machines without the user noticing.
nice isn't the perfect way to make things work. You may want an app running high priority, but you still need a little cpu to do basic GUI interactions.
I'm not trying to say Windows is better, but they did a few really nice things with the Win7 kernel.
Win7 counts clock cycles that each thread consumes. If a thread wakes up and goes back to sleep, it consume very few cycles. On the other hand, another thread may consume it's entire time slice. In order to keep other threads from getting starved, it will artificially increase priority for threads that consume little cpu, but wants to wake up. Win7 also has a new audio api to allow low latency audio. The idea is that audio consume very little cpu, but it needs to get processed fast and on time every few milliseconds.
Couple this stuff together and you have a responsive gui under heavy load. An easy test that I did is I loaded up prime95, set it to high priority and let her rip. Loaded up a 1080p movie and playback was flawless.
IO priority is nice to. load up defrag, set to idle priority, start virus scan(good virus scanners use new IO api options), play games. By default, idle priority threads get low priority for IO also. This kind of stuff can also be programmed via API(virus scanners rawr) so a normal priority thread can have an idle/low priority IO.
By default Win7 has 20% cpu on standby, so "starved" threads can fight over that 20%. This can be disabled and is disabled by default on Server 2K8R2 and not an option on just 2K8. This keeps things feeling peppy.
This is probably why my 1080p videos play just fine even with prime95 at high. A video/audio track can be buffered, so even though Prime95 is trying to starve the system, all my other threads still get up to 20% cpu. On and i7, 20% cpu is a lot of processing power. I would assume a realtime processing like a video game would get stuttering since it can't buffer much and getting starved for a few milliseconds is gonna to cause break up.
I would love if Linux had more options for schedulers. In another./ posting, someone mentioned that there was talk about a pluggable/module scheduler, but that got turned down. I could see an idea like this being useful so the user could have specialized scheduling, especially for desktops. May be even just modulized logic on how the scheduler should handle starved threads.
Again, I'm not trying to say Win7 > Linux or anything, I'm just recognizing a good idea that works.
Re:The problem is outsourcing not language
on
Which Language To Learn?
·
· Score: 3, Interesting
hah, outsourcing... horrible idea. well, you get what you pay for.
The job I recently came into had some outsourced internal tools to India. These were tools that could save us time and offer useful functionality, but were not central to our job. Over time, some of these programs became fairly common. Well, these programs had a long turn around on added features and the code was huge. Easy to read, well documented, but lots of it.
Eventually, with the market down turn, my company dropped the India team. Someone else had to pick up fixing bugs when found and adding features. Eventually that person moved on and the job got passed down to me. I decided to start from scratch because the logic was hard to follow. The code was clean, but the logic was horrible.
They typically took 2-4 weeks to add features or fix bugs. In 1 week, I flow charted the program and reduced the logic to something more flexible and natural to follow. In 1 month, I had a re-write that I could debug every bug so far in under 15 minutes and added new features in under a day. The code scales crazy better to. A small dataset runs about 250 times faster, a larger dataset runs about 1200 times faster and the memory allocation is about 1/5-1/10 the amount. The server admin likes that the app doesn't bog down the servers anymore. From 45min down to 10 seconds. My code is C#.net and their code was VB.net.
Each channel individually can do up to 42.66 Mbps downstream (U.S. DOCSIS) all on its own. It isnt hard to imagine ALL provider having at least 12 free channels, which equates to 512Mbps of downstream capacity.
This is why its hard to believe the grandparents claim that 9 users downloading at the same time can prevent you from getting your email.
Adding to that is CDMA. A CMDA phone can handle hundreds to thousands of users on the same channel at the same time.
Now, cable doesn't have all the interference from air. In my case, about 20-30 users on my node. 8-12 channels, each at ~42mbps each. Toss in CDMA and you can easily have several gigabits of bandwidth.
I'm not going to have an major benefit using fiber until my ISP starts offering 200mb+ connections. Even then, DOCSIS 4.0 will come out which promises 1gb of bandwidth per user, assuming your cable operator doesn't over subscribe and your node/ISP can handle it.
I would LOVE un-capped bandwidth for local connections though.
I would *prefer* fiber, but going to fiber will give me no extra bandwidth or noticeably reduced latency.
Stay awake for the download? Bad Company 2 took under 30min for me to have up and running and I installed the instant it was released.
TF2 only took a few minutes to install. Man, I love Steam.
I really want the 60mb offered in my area, but I'm too poor for now.
I swear, Steam has special connection to my ISP. I have a 16mb package with power boost, which goes up to 38mb(max for single channel modem). But for some reason, when I download from steam, I get 38mb solid.
Fiber can be slow if they overload a node to. The issue isn't the connection from the customer to the node, but from the node to the ISP.
Since the difference between fiber and cable is customer to the node and that isn't the issue, then there is no major difference with current sub 1gb speeds.
Put 100 customers with 60mbit on a node with a 1gbit connection and ANY tech is going to have bandwidth issues.
I would still rather have fiber just because there are fewer "random" issues and issues are easier to fix, but cable works just fine if the ISP implements it correctly.
There are other forums on the internet that I watch for info about my ISP so I can stay on top of their offerings and upgrades. MANY customers claim that their 60mbit connections get 60mbit at peak hours and even burst up to 120mbit using power boost.
My ISP isn't magical in anyway, they just don't overload their nodes.
In the past year, my worst ping to Chicago has been 22ms. During off peak, I get 15ms and peak is closer to 19ms. I get 0-2ms jitter on average, 2ms jitter during peak.
I have always gotten my 16mbit speed. During peak hours, I may have to download from multiple sources to reach 16mbit, but that's not my ISP, that's the web site. I can maintain 16mbit down while playing FPS games with an in-game 20ms ping.
I really want the 60mbit package, but going from $40/month to $99/month is a bit out of my price range. The 25mbit is $55 though.. eyeing that up.
I may be an exception, but that's not an exception in technology, I'm an exception in how my ISP doesn't overload their nodes.
Again, cable isn't inherently much worse. If you have a good cable, with no nicks and no bad physical connectors, cable should be the same as fiber. Where they differ is fiber is easier to have a good line and easier to diagnose if the came is bad.
DSL is a bit different because of the dslam, but Fiber and Cable is nearly the same.
Fiber lines have a dedicated line to the local node, the local node is shared by others and connects to the ISP via fiber. Cable has as shared connection to the node, but the node is connected to the ISP via a fiber link also.
Case in point: Some posters in other forums for my cable ISP claim that they get their 60mbps during peak hours and their power boost even hits ~120mbps.
The issue isn't too many customers on the same COAX, but too many customers on the same node. If a Node has a 1gb fiber link and you have 100 customers, each with a 60mb connection, you're node is going to be overwhelmed. If you think changing the connection between the user and the node is going to make the connection from the node to ISP faster, I'd like to take what you're taking.
Obviously FTTH and FTTC is better than cable, but it's mostly better on the reduced noise on the line. DOCSIS tech is really good, but if there is something wrong with your coax, it can sometimes be hard to diagnose the issue. With fiber, there is much reduced chance of having intermittent issues and it's more likely to just not work. I would rather my connection completely fail so my ISP can fix it, than have a problem go away when my ISP shows up to fix the unknown problem.. gahh, i hate that
Most of cables bad rep comes from DOCSIS 1. v1 used time slices along with channels. This meant you may get good connection to your local node during off hours, but on peak hours you'd have to compete for time slices. Time slices suck horribly and scale badly.
DOCSIS 3.0 doesn't use time slices anymore and 2.0 it was optional. When using CDMA with 2.0/3.0, you get a pseduo-dedicated connection. Yes, your share the same line with your neighbors, *but* that don't allocate your bandwidth in any shape or form. When more people on the line does do is add noise. There will be an upper limit to the amount of people that can use a single line due to additional noise, but CDMA is quite resiliant.
An interesting note is that DOCSIS will scale to 1gbit on coax in future versions, it's already been announced.
recap. #1. you get pseduo-dedicated bandwidth with CDMA enabled 2.0 and 3.0 docsis #2. a 100gbit fiber connection to your ISP won't help at all if your ISP has too many customers and not enough bandwidth to the internet #3. the different between fiber/cable/dsl is more an exercise of theory as a practical implementation from your ISP is the biggest issue. #4. there are many choke points on any network, it just depends on your ISP to decide where yours is
The main difference between DSL and DOCSIS cable is that DSL is your personal connection. No one is sharing it. DOCSIS cable line is shared between those on the same line, so if you have active warez people in your neighbourhood or someone hosting an active server of some kind, expect much lower speeds and higher latency then advertised.
Second difference, which has been largely negated lately is latency. DSL offers slightly lower latency by advantage of design.
Tradeoff is that DSL only uses one really shitty quality copper pair, that limits distance and maximum speed far more severely then cable's coaxial. This is exacerbated by the fact that many phone lines are from times before CAT3 home cabling, which is a realistic requirement to reach even ADSL2 level of speeds, causing end user speeds to be below 10mbps even over 24mbps ADSL2+ connection.
This is VERY wrong.
#1. DSL is shared just like cable, just at a slightly different point. Several DSL customers connect to the same node and from there, they are given times slices. example. My brother has aDSL, he has ~20 people on his local node and he had an 80ms ping to his first hop. Yes, his ISP that is just down the road is an 80ms jump. If he had this dedicated connection you talked about, it would be impossible to have anything much more than 1ms to his ISP.
#2. even FTTH has local choke points. You don't connect directly to your ISP, you connect to a local node. Your local node is shared by many many people.
#3. Cable uses CDMA. You can have several people per channel talking at the same time. You share a local physical coax loop with a few of your immediate neighbors. This coax loop has a crap ton of bandwidth. You then connect from this coax loop to your local node, this node is shared by several loops.
The *only* difference between cable and other techs is that the connection between you and your node is shared, but your node to ISP is still fiber and has the same limitations of all the other techs.
My trace route to Chicago is about 700 miles long which is a 3ms ping at the speed of light in a vacuum. I get a 15-18ms pings to Chicago from my ISP , during peak hours I might add, and my ISP has over 2 million internet customers over that link.
I think I figured out what we need. We need a stated *minimum* internet speed.
example. You pay for a 60mbps i-net connection, but obviously no ISP could actually handle every user sucking down 60mbps. This is an "up to" speed. What we also need is a listed minimum speed.
So, lets say your 60mbps connection has a minimum listed speed of 8mbps.
OK, now what? Now we need some form of QoS, even as simple as High and Normal. Lets say you can assign QoS on your local network. I assign My video games to be high priority. Now, up to that 8mbit, is all mine. So, if I start playing video games and some other user is downloading bit-torrent, his first 8mbit will be all his, but packets past the 8mbit will be lower priority than my first 8mbit.
I still get my low pings, and other users still get their high bandwidth. My "high" priority data can only get processed before someone else's data if the packet in question is past their allotted minimum speed.
My ISP has power boost and it already does something very similar to this. Powerboost allows me to use any "free" network bandwidth. I have a 16mbps connection, but it will burst up to 38mbps(max speed for my non-bonded connection) if network usage is low. People on my ISP with bonded cable channels are reporting their 16mbps connection bursts upwards of 120mbit during non-peak hours.
But in the case of my above, local LAN QoS will make sure that the packets I want will get in that first 8mbps of my guaranteed connection, while the rest of the lower priority packets will get shoved into the "what ever is free" portion.
Most latency sensitive applications are fairly low bandwidth, so it's not like you'd need a high minimum bandwidth.
Because of the way networks work, this would only be useful if the client and the cable modem had some sort of QoS. You need a way to distinguish which packets are part of that 8mbit, and the only real way is to have QoS so those packets get to the front of the line in the modem's buffer.
I would assume there would still be a decent amount of jitter, but the overall average latency would be lower for data in that minimum speed.
So, if I steal half a dozen cars from the local car dealer and give them to people, I get no punishment at all. My "proceeds" are exactly 0. In fact, there is some cost out of my pocket; do you think maybe the car dealer should have to reimburse me for expenses?
Really, there is the concept of damages and punitive damages in existing law. If all you got reimbursed for was the actual amount of damages, then people who commit the offences would, on average, profit, because they, on average, don't get caught all the time.
That's not saying that $1.5 mil is a good judgement, just that "proceeds" is a very poor way to levy fines or jury awards in civil cases.
Let me fix that for you
"So, if I copy half a dozen cars[...]"
I still agree with your summary, but you could've made a better analogy:-)
Chrome doesn't support freaking cut-and-paste within Slashdot comments. Seriously WTF?
Ditto
And for some reason, when I spell-check a word on/. it takes several seconds it to update. Weird. Gotta be some javascript that's messing with stuff.
I also get it some times where I click on a child response and instead of expanding, it loads me into a new page. If I middle click, it expands the current page, but then loads a new tab with just that response and it's children.
Won't somebody think of the children?!.... responses
"So it's okay for MS to have large profit margins but not Apple? Besides, I thought the "Free Software Mantra" was that MS should give the software away and sell support?"
Typically, companies with high R&D charge more for their stuff. MS is a pure software development company and as such has lots of R&D. iPad may be lots of R&D and iPhone to, but selling commodity hardware at exuberant costs with no benefit other than it runs a custom UI on top of BSD?
They have no reason to price gouge on their PC hardware. Really, an Apple computer is just a status symbol, you can afford the ridiculous extra cost.
"So how much has all the R&D money that MS spent actually benefited the users? Or even Windows?"
Are you implying that a bunch of people with PHDs and experimental hardware doing pure research and communicating their findings to other experts haven't contributed anything?
"Find one laptop that is equivalent to a Macbook and has better battery life for half the price"
Well, I did find a decent amount of i5 notebooks for very close to 1/2 the price of the Macbook core duos. Pretty sure the i5(32nm+core parking) has a decent lead on the duos(45nm) for power savings.
An i5 at 45nm uses less power than a duo, while having more performance. now add on the 32nm and it's an obvious lead.
Slashdot Mirror
"Lazy CPU designers" hah!
GPUs are severely limited on their types of tasks. Instead of a 1600 core GPU, pretend your CPU has as single large SIMD register that can hold 1600 floats. Now, it would be great at crunching large matrices of floats and utterly suck at everything else. That's a GPU in a nutshell. GPUs are absolutely horrible at branches. If one core takes a branch, every core in that core's group must stall and wait for the branch to finish. All cores must be working on the same instruction at the same time and branches mess with that. GPUs != CPUs
When Intel last talked about their 80 core CPUs, they talked about getting rid of cache-coherency in order to scale, AMD also recognizes this. This would mean OSs and most Apps would NOT be backwards compatible even if still using x86 instructions. Although, it would be possible to run apps in a VM that emulated cache-coherency.
"Which Microsoft has been trying to phase out entirely anyway"
MS's commandline is much better than it has ever been. Not sure where you're getting your random info from.
Speaking about OO vs MS Office, I find it hard to even get our customers to send other versions of Office files. Many times I get files in 2007 format and I can't view the files with my 2003 Office. Try telling a customer to save the file as a previous format.. good luck.
ummm.. what?
Transistor counts double every 18 months, how long until 4GB smart phones? 5 years?
With Linux taking over these parts of the market, how long until someone finds a good use to run MySQL on a smart phone? Web server? I'm sure IPv6 coupled with high bandwidth of future nation wide WiMAX/etc, we will want to connect our phones and vew data.
How long until some major breakthrough will make SSDs crazy small and we have smartphones with 2TB of storage and we bring them to our friends and want to file share?
Add a lightpeak connection, plug your phone into a 19" monitor, hook up a mouse and keyboard and play games. It won't be long.
I remember my dad paying $800 for 16MB of edo-dram
the enterprise is moving into a virtual environment where you can live migrate images off of idle machines and consolidate them to reduce power. If the current server is getting too loaded, just migrate from that machine to another will less load.
You can already do this with current software, give it 5-10 years and this will be standard.
I mostly see a low power ARM being good for dedicated SAN/NAS/Firewall. Most other SQL/Web stuff will be on VMs where you can dynamically add/remove Memory/CPUs and migrate between physical machines without the user noticing.
int i = 0;
while(i infinite)
{
i++;
}
---
Whatever computer finishes first is clearly the fastest supercomputer.
while(++i)
nice isn't the perfect way to make things work. You may want an app running high priority, but you still need a little cpu to do basic GUI interactions.
I'm not trying to say Windows is better, but they did a few really nice things with the Win7 kernel.
Win7 counts clock cycles that each thread consumes. If a thread wakes up and goes back to sleep, it consume very few cycles. On the other hand, another thread may consume it's entire time slice. In order to keep other threads from getting starved, it will artificially increase priority for threads that consume little cpu, but wants to wake up. Win7 also has a new audio api to allow low latency audio. The idea is that audio consume very little cpu, but it needs to get processed fast and on time every few milliseconds.
Couple this stuff together and you have a responsive gui under heavy load. An easy test that I did is I loaded up prime95, set it to high priority and let her rip. Loaded up a 1080p movie and playback was flawless.
IO priority is nice to. load up defrag, set to idle priority, start virus scan(good virus scanners use new IO api options), play games. By default, idle priority threads get low priority for IO also. This kind of stuff can also be programmed via API(virus scanners rawr) so a normal priority thread can have an idle/low priority IO.
By default Win7 has 20% cpu on standby, so "starved" threads can fight over that 20%. This can be disabled and is disabled by default on Server 2K8R2 and not an option on just 2K8. This keeps things feeling peppy.
This is probably why my 1080p videos play just fine even with prime95 at high. A video/audio track can be buffered, so even though Prime95 is trying to starve the system, all my other threads still get up to 20% cpu. On and i7, 20% cpu is a lot of processing power. I would assume a realtime processing like a video game would get stuttering since it can't buffer much and getting starved for a few milliseconds is gonna to cause break up.
I would love if Linux had more options for schedulers. In another ./ posting, someone mentioned that there was talk about a pluggable/module scheduler, but that got turned down. I could see an idea like this being useful so the user could have specialized scheduling, especially for desktops. May be even just modulized logic on how the scheduler should handle starved threads.
Again, I'm not trying to say Win7 > Linux or anything, I'm just recognizing a good idea that works.
hah, outsourcing... horrible idea. well, you get what you pay for.
The job I recently came into had some outsourced internal tools to India. These were tools that could save us time and offer useful functionality, but were not central to our job. Over time, some of these programs became fairly common. Well, these programs had a long turn around on added features and the code was huge. Easy to read, well documented, but lots of it.
Eventually, with the market down turn, my company dropped the India team. Someone else had to pick up fixing bugs when found and adding features. Eventually that person moved on and the job got passed down to me. I decided to start from scratch because the logic was hard to follow. The code was clean, but the logic was horrible.
They typically took 2-4 weeks to add features or fix bugs. In 1 week, I flow charted the program and reduced the logic to something more flexible and natural to follow. In 1 month, I had a re-write that I could debug every bug so far in under 15 minutes and added new features in under a day. The code scales crazy better to. A small dataset runs about 250 times faster, a larger dataset runs about 1200 times faster and the memory allocation is about 1/5-1/10 the amount. The server admin likes that the app doesn't bog down the servers anymore. From 45min down to 10 seconds. My code is C# .net and their code was VB .net.
Well, that's my experience anyway.
My Blizz authenticator uses an RSA spec commercial grade dongle, and that only costs $6.50
Uses the exact same open algorithm used by other RSA dongles.
Behold, the power of mass production.
Each channel individually can do up to 42.66 Mbps downstream (U.S. DOCSIS) all on its own. It isnt hard to imagine ALL provider having at least 12 free channels, which equates to 512Mbps of downstream capacity.
This is why its hard to believe the grandparents claim that 9 users downloading at the same time can prevent you from getting your email.
Adding to that is CDMA. A CMDA phone can handle hundreds to thousands of users on the same channel at the same time.
Now, cable doesn't have all the interference from air. In my case, about 20-30 users on my node. 8-12 channels, each at ~42mbps each. Toss in CDMA and you can easily have several gigabits of bandwidth.
I'm not going to have an major benefit using fiber until my ISP starts offering 200mb+ connections. Even then, DOCSIS 4.0 will come out which promises 1gb of bandwidth per user, assuming your cable operator doesn't over subscribe and your node/ISP can handle it.
I would LOVE un-capped bandwidth for local connections though.
I would *prefer* fiber, but going to fiber will give me no extra bandwidth or noticeably reduced latency.
Stay awake for the download? Bad Company 2 took under 30min for me to have up and running and I installed the instant it was released.
TF2 only took a few minutes to install. Man, I love Steam.
I really want the 60mb offered in my area, but I'm too poor for now.
I swear, Steam has special connection to my ISP. I have a 16mb package with power boost, which goes up to 38mb(max for single channel modem). But for some reason, when I download from steam, I get 38mb solid.
OpenID + dongle = awesome
Rainbow tables don't work against a salted hash. I should hope everyone salts their hashes.. and it makes them taste better.
When you can't get money from mom, you gotta buy second hand.
I border poverty, I get no assistance from the state/government, and I can afford new games. Not sure why someone would have to purchase a used game.
May be saving that $10-$20 gives them electricity for 6 months.
A station wagon to carry reel-to-reel tape and a large social network.
Fiber can be slow if they overload a node to. The issue isn't the connection from the customer to the node, but from the node to the ISP.
Since the difference between fiber and cable is customer to the node and that isn't the issue, then there is no major difference with current sub 1gb speeds.
Put 100 customers with 60mbit on a node with a 1gbit connection and ANY tech is going to have bandwidth issues.
I would still rather have fiber just because there are fewer "random" issues and issues are easier to fix, but cable works just fine if the ISP implements it correctly.
There are other forums on the internet that I watch for info about my ISP so I can stay on top of their offerings and upgrades. MANY customers claim that their 60mbit connections get 60mbit at peak hours and even burst up to 120mbit using power boost.
My ISP isn't magical in anyway, they just don't overload their nodes.
In the past year, my worst ping to Chicago has been 22ms. During off peak, I get 15ms and peak is closer to 19ms. I get 0-2ms jitter on average, 2ms jitter during peak.
I have always gotten my 16mbit speed. During peak hours, I may have to download from multiple sources to reach 16mbit, but that's not my ISP, that's the web site. I can maintain 16mbit down while playing FPS games with an in-game 20ms ping.
I really want the 60mbit package, but going from $40/month to $99/month is a bit out of my price range. The 25mbit is $55 though.. eyeing that up.
I may be an exception, but that's not an exception in technology, I'm an exception in how my ISP doesn't overload their nodes.
Again, cable isn't inherently much worse. If you have a good cable, with no nicks and no bad physical connectors, cable should be the same as fiber. Where they differ is fiber is easier to have a good line and easier to diagnose if the came is bad.
DSL is a bit different because of the dslam, but Fiber and Cable is nearly the same.
Fiber lines have a dedicated line to the local node, the local node is shared by others and connects to the ISP via fiber. Cable has as shared connection to the node, but the node is connected to the ISP via a fiber link also.
Case in point: Some posters in other forums for my cable ISP claim that they get their 60mbps during peak hours and their power boost even hits ~120mbps.
The issue isn't too many customers on the same COAX, but too many customers on the same node. If a Node has a 1gb fiber link and you have 100 customers, each with a 60mb connection, you're node is going to be overwhelmed. If you think changing the connection between the user and the node is going to make the connection from the node to ISP faster, I'd like to take what you're taking.
I shouldn't have started my rebuttal with "This is VERY wrong".
I'm sorry.. :*(
I should've stated something more like "DSL being "dedicated" is a common misconception" or something.
Wanted to add a few more things.
Obviously FTTH and FTTC is better than cable, but it's mostly better on the reduced noise on the line. DOCSIS tech is really good, but if there is something wrong with your coax, it can sometimes be hard to diagnose the issue. With fiber, there is much reduced chance of having intermittent issues and it's more likely to just not work. I would rather my connection completely fail so my ISP can fix it, than have a problem go away when my ISP shows up to fix the unknown problem.. gahh, i hate that
Most of cables bad rep comes from DOCSIS 1. v1 used time slices along with channels. This meant you may get good connection to your local node during off hours, but on peak hours you'd have to compete for time slices. Time slices suck horribly and scale badly.
DOCSIS 3.0 doesn't use time slices anymore and 2.0 it was optional. When using CDMA with 2.0/3.0, you get a pseduo-dedicated connection. Yes, your share the same line with your neighbors, *but* that don't allocate your bandwidth in any shape or form. When more people on the line does do is add noise. There will be an upper limit to the amount of people that can use a single line due to additional noise, but CDMA is quite resiliant.
An interesting note is that DOCSIS will scale to 1gbit on coax in future versions, it's already been announced.
recap.
#1. you get pseduo-dedicated bandwidth with CDMA enabled 2.0 and 3.0 docsis
#2. a 100gbit fiber connection to your ISP won't help at all if your ISP has too many customers and not enough bandwidth to the internet
#3. the different between fiber/cable/dsl is more an exercise of theory as a practical implementation from your ISP is the biggest issue.
#4. there are many choke points on any network, it just depends on your ISP to decide where yours is
#5. fiber rocks.
The main difference between DSL and DOCSIS cable is that DSL is your personal connection. No one is sharing it. DOCSIS cable line is shared between those on the same line, so if you have active warez people in your neighbourhood or someone hosting an active server of some kind, expect much lower speeds and higher latency then advertised.
Second difference, which has been largely negated lately is latency. DSL offers slightly lower latency by advantage of design.
Tradeoff is that DSL only uses one really shitty quality copper pair, that limits distance and maximum speed far more severely then cable's coaxial. This is exacerbated by the fact that many phone lines are from times before CAT3 home cabling, which is a realistic requirement to reach even ADSL2 level of speeds, causing end user speeds to be below 10mbps even over 24mbps ADSL2+ connection.
This is VERY wrong.
#1. DSL is shared just like cable, just at a slightly different point. Several DSL customers connect to the same node and from there, they are given times slices. example. My brother has aDSL, he has ~20 people on his local node and he had an 80ms ping to his first hop. Yes, his ISP that is just down the road is an 80ms jump. If he had this dedicated connection you talked about, it would be impossible to have anything much more than 1ms to his ISP.
#2. even FTTH has local choke points. You don't connect directly to your ISP, you connect to a local node. Your local node is shared by many many people.
#3. Cable uses CDMA. You can have several people per channel talking at the same time. You share a local physical coax loop with a few of your immediate neighbors. This coax loop has a crap ton of bandwidth. You then connect from this coax loop to your local node, this node is shared by several loops.
The *only* difference between cable and other techs is that the connection between you and your node is shared, but your node to ISP is still fiber and has the same limitations of all the other techs.
My trace route to Chicago is about 700 miles long which is a 3ms ping at the speed of light in a vacuum. I get a 15-18ms pings to Chicago from my ISP , during peak hours I might add, and my ISP has over 2 million internet customers over that link.
I think I figured out what we need. We need a stated *minimum* internet speed.
example. You pay for a 60mbps i-net connection, but obviously no ISP could actually handle every user sucking down 60mbps. This is an "up to" speed. What we also need is a listed minimum speed.
So, lets say your 60mbps connection has a minimum listed speed of 8mbps.
OK, now what? Now we need some form of QoS, even as simple as High and Normal. Lets say you can assign QoS on your local network. I assign My video games to be high priority. Now, up to that 8mbit, is all mine. So, if I start playing video games and some other user is downloading bit-torrent, his first 8mbit will be all his, but packets past the 8mbit will be lower priority than my first 8mbit.
I still get my low pings, and other users still get their high bandwidth. My "high" priority data can only get processed before someone else's data if the packet in question is past their allotted minimum speed.
My ISP has power boost and it already does something very similar to this. Powerboost allows me to use any "free" network bandwidth. I have a 16mbps connection, but it will burst up to 38mbps(max speed for my non-bonded connection) if network usage is low. People on my ISP with bonded cable channels are reporting their 16mbps connection bursts upwards of 120mbit during non-peak hours.
But in the case of my above, local LAN QoS will make sure that the packets I want will get in that first 8mbps of my guaranteed connection, while the rest of the lower priority packets will get shoved into the "what ever is free" portion.
Most latency sensitive applications are fairly low bandwidth, so it's not like you'd need a high minimum bandwidth.
Because of the way networks work, this would only be useful if the client and the cable modem had some sort of QoS. You need a way to distinguish which packets are part of that 8mbit, and the only real way is to have QoS so those packets get to the front of the
line in the modem's buffer.
I would assume there would still be a decent amount of jitter, but the overall average latency would be lower for data in that minimum speed.
someone needs to mod this up.. fuuuunny :-)
So, if I steal half a dozen cars from the local car dealer and give them to people, I get no punishment at all. My "proceeds" are exactly 0. In fact, there is some cost out of my pocket; do you think maybe the car dealer should have to reimburse me for expenses?
Really, there is the concept of damages and punitive damages in existing law. If all you got reimbursed for was the actual amount of damages, then people who commit the offences would, on average, profit, because they, on average, don't get caught all the time.
That's not saying that $1.5 mil is a good judgement, just that "proceeds" is a very poor way to levy fines or jury awards in civil cases.
Let me fix that for you
"So, if I copy half a dozen cars[...]"
I still agree with your summary, but you could've made a better analogy :-)
+1 Agreed.
Chrome doesn't support freaking cut-and-paste within Slashdot comments. Seriously WTF?
Ditto
And for some reason, when I spell-check a word on /. it takes several seconds it to update. Weird. Gotta be some javascript that's messing with stuff.
I also get it some times where I click on a child response and instead of expanding, it loads me into a new page. If I middle click, it expands the current page, but then loads a new tab with just that response and it's children.
Won't somebody think of the children?!.... responses
"So it's okay for MS to have large profit margins but not Apple? Besides, I thought the "Free Software Mantra" was that MS should give the software away and sell support?"
Typically, companies with high R&D charge more for their stuff. MS is a pure software development company and as such has lots of R&D. iPad may be lots of R&D and iPhone to, but selling commodity hardware at exuberant costs with no benefit other than it runs a custom UI on top of BSD?
They have no reason to price gouge on their PC hardware. Really, an Apple computer is just a status symbol, you can afford the ridiculous extra cost.
"So how much has all the R&D money that MS spent actually benefited the users? Or even Windows?"
Are you implying that a bunch of people with PHDs and experimental hardware doing pure research and communicating their findings to other experts haven't contributed anything?
"Find one laptop that is equivalent to a Macbook and has better battery life for half the price"
Well, I did find a decent amount of i5 notebooks for very close to 1/2 the price of the Macbook core duos. Pretty sure the i5(32nm+core parking) has a decent lead on the duos(45nm) for power savings.
An i5 at 45nm uses less power than a duo, while having more performance. now add on the 32nm and it's an obvious lead.