Every bittorrent client I've ever used has easy to set upstream and downstream limits. Simply set your upstream and downstream to 65% and 75% of you're max connection and you'll never be slowed down.
That may not be wise. Consider the case where you set BitTorrent to 100%, Comcast sets you to low priority, and then, say, 10% of your packets get dropped. You're still getting 90% effective throughput instead of 75%. Some real testing will be needed to determine whether you're better off throttling yourself all the time or letting Comcast throttle you some of the time.
Now on a airplane you can "bump" passengers. However in the case of bandwidth there is no bump available. The only options they have are to either put in more lines/equipment (quite often impossible due to community regulations and available space in underground cable easements) or drop customers.
Or they can drop packets, which will cause performance to gracefully degrade. And that's what Comcast is doing.
But at least on ADSL you know that the bandwidth you use has little affect on anyone but people in your household.
DSL is shared beyond the DSLAM, so packets will simply get dropped at a different location.
The throttling kicks in at 80%, so if you are on the 6 Mbps plan they throttle at 4.8 Mbps and if you are on the 8 Mbps plan they throttle at 6.4 Mbps. Paying for a better plan does get you more usable bandwidth.
If we purchase a DVD, should we not have also (included with the purchase) rights to the patent used in the product, i.e. the compression algorithms?
Consider the case where different patents apply to compression and decompression. Or consider that a DVD does not actually perform decompression; it's the DVD player software that would infringe.
In the v4 Internet, multicast exists but is usually disabled (except U-Verse). In the v6 Internet, multicast will exist but be disabled (except maybe U-Verse).
write speed issues, at least when it comes to relatively small amounts of writing, could easily be mitigated with a very long on-board RAM buffer controlled by the drive
The STEC Zeus, Memoright, and Intel "extreme" flash drives have this feature.
The clones were supposedly free-riding on Apple's R&D for mother board designs etc. Apple later figured out they weren't recouping the costs from licensing and the clone makers sold cheaper/faster hardware than Apple because they didn't have to reinvent the wheel.
Exactly. Clones didn't nearly kill Apple; giving away R&D nearly killed Apple. If Apple charged a reasonable license fee they could make money on clones.
For servers, POWER4 was released in 2001. For desktops, the Pentium D came out in mid 2005 and the PowerPC 970MP a few months later. All of these came out before Niagara.
Slashdot Mirror
You're assuming such severe congestion that starvation occurs; Comcast says they never observed starvation in their tests.
Netflix streaming is so low-bandwidth that it can't trigger Comcast's throttling.
Every bittorrent client I've ever used has easy to set upstream and downstream limits. Simply set your upstream and downstream to 65% and 75% of you're max connection and you'll never be slowed down.
That may not be wise. Consider the case where you set BitTorrent to 100%, Comcast sets you to low priority, and then, say, 10% of your packets get dropped. You're still getting 90% effective throughput instead of 75%. Some real testing will be needed to determine whether you're better off throttling yourself all the time or letting Comcast throttle you some of the time.
Now on a airplane you can "bump" passengers. However in the case of bandwidth there is no bump available. The only options they have are to either put in more lines/equipment (quite often impossible due to community regulations and available space in underground cable easements) or drop customers.
Or they can drop packets, which will cause performance to gracefully degrade. And that's what Comcast is doing.
But at least on ADSL you know that the bandwidth you use has little affect on anyone but people in your household.
DSL is shared beyond the DSLAM, so packets will simply get dropped at a different location.
The throttling kicks in at 80%, so if you are on the 6 Mbps plan they throttle at 4.8 Mbps and if you are on the 8 Mbps plan they throttle at 6.4 Mbps. Paying for a better plan does get you more usable bandwidth.
To be throttled you have to use more than 4.8 Mbps (80% of 6 Mbps); what streaming services use this much bandwidth?
If we purchase a DVD, should we not have also (included with the purchase) rights to the patent used in the product, i.e. the compression algorithms?
Consider the case where different patents apply to compression and decompression. Or consider that a DVD does not actually perform decompression; it's the DVD player software that would infringe.
Big deal. That's not a reason to adopt IPv6.
In the v4 Internet, multicast exists but is usually disabled (except U-Verse).
In the v6 Internet, multicast will exist but be disabled (except maybe U-Verse).
write speed issues, at least when it comes to relatively small amounts of writing, could easily be mitigated with a very long on-board RAM buffer controlled by the drive
The STEC Zeus, Memoright, and Intel "extreme" flash drives have this feature.
Yes, IBM shipped one server with MXT memory compression.
Volume. Microsoft sells 10 times as many OS licenses as Apple. (Windows costs more to develop than OS X, but the volume makes up for that, too.)
The clones were supposedly free-riding on Apple's R&D for mother board designs etc. Apple later figured out they weren't recouping the costs from licensing and the clone makers sold cheaper/faster hardware than Apple because they didn't have to reinvent the wheel.
Exactly. Clones didn't nearly kill Apple; giving away R&D nearly killed Apple. If Apple charged a reasonable license fee they could make money on clones.
A full OEM copy of Windows is actually cheaper than an upgrade copy, so MS provides no incentive to cheat the system.
Nope; they'll find some way to save a few cents by disabling ECC. If Intel doesn't cripple it the BIOSes will.
No, the article says a Bulldozer-based Fusion chip will be fabbed by TSMC. AMD will probably make the non-fused Bulldozer itself.
Not really, but if you enable copies=2 ZFS will replicate all data on a single disk.
Clearly Intel means C/C++ with pthreads or Win32 threads. Also, C++ is getting thread support in C++0x.
In-order cores still stall (in fact they stall more), so there's still a benefit from MT.
http://www.ubnt.com/products/sr71a.php
Nobody ever sees that article because it has no PageRank, thus no ad revenue.
Some partial fixes are possible and were discussed a while ago.
http://thread.gmane.org/gmane.comp.encryption.general/11528
Many of the problems reported by the Linux Hater are cultural; if you submit a patch the maintainer will just reject it with -ENOTABUG.
So what else should we do? Should we stick with single core and watch our computers never get any faster?
For servers, POWER4 was released in 2001. For desktops, the Pentium D came out in mid 2005 and the PowerPC 970MP a few months later. All of these came out before Niagara.