Sorry Sir, but it seems that a transistor has gone in the controller circuit and your data cannot be retrieved!
If the controller goes bad you can read out the data directly from the flash chips with nothing more than an FPGA and a TSOP test clip. Then you just have to reverse-engineer the FTL to reconstruct the data. Compared to recovering data from a mangled platter, it's a piece of cake.
Well, if the drives handle sequential access well and random access not so well, and Windows puts new files in the first available space while Linux places the files in the middle of the largest area of blank space, then does that create the possibility that Windows will currently handle these drives better than Linux?
No, because a lot of write traffic is to existing files and ext2/3/4 and NTFS are write-in-place filesystems, so for most writes the OS has no choice where to place them. ZFS or Btrfs may handle crap SSDs much better because they are write-anywhere.
The only way around this is to move into the "insanely smart" wear-leveling that will actually move data when the drive is otherwise idle to re-balance the sector write counts. You wouldn't want to do this during actual write requests, as it would slow them down even more than they are now. AFAIK, no SSD does this.
Intel claims that they do this, balancing the wear across all the flash, not just the part that is frequently rewritten.
That's the problem; almost all SSDs on the market are broken and have worse write performance than hard disks. Just look at all the "JMicron rage" out there.
Traffic stopped flowing because neither Sprint nor Cogent paid for any alternate paths through the Net. This depeering reveals the fragility of the Tier 1 "people pay us, but we don't pay anyone" philosophy.
As soon as TCP notices a single packet loss, the Jacobson Algorithm kicks in and it's throttled to maybe 50-60%, and raises the limit slowly. I highly doubt that uTorrent's reworked version of UDP will play this nicely.
To fix this you can use either ZFS-style integrated filesystem and RAID or 3Par/Xiotech/XIV-style chunk RAID with checksums and unmap bad parts of the disk.
But darn it; when will someone finally offer a reasonably-priced, open-platform STB that serves as an A/V gateway to multiple Internet-based services â" one consumer-friendly, environmentally-designed, low-power gadget 'to rule them all,' if you will.
15K drives exist for a reason (at least they did until now), and they're only available in SAS or FC. I suspect the SAS version is actually the cheap one.
You don't need to reserve space for wear-leveling.
Yes you do, because flash can't be rewritten in place. You have to write into an empty block and then erase the original block. If all the space was used, you'd have no empty blocks and thus couldn't write anything. More reserved space allows better wear leveling because you can choose to write to the empty block with the least wear.
Why isn't wasting RAM installed in your computer necessary to make it addressable?
The overhead really is necessary for bad block management and garbage collection. Some "enterprise" SSDs have 20% or even 50% overhead, so you're actually getting off easy.
There is some truth to what you say; I know the national labs in particular are working on a completely open source HPC stack. But many others in the HPC world have been using proprietary compilers, debuggers, filesystems, etc. for decades.
I'm not trying to defend Apple; I think they've done the wrong thing here. Midnight Thunder implied that Apple had no choice because all computers have HDCP, and I'm saying that's wrong. Apple didn't implement HDCP last year, so they should have kept on not implementing it. Obviously people shouldn't buy old computers to avoid DRM; that's inefficient and unsustainable.
The fact that the same video will play fine on a 2007 Mac but refuse to play on a 2008 Mac proves that the copy protection is not necessary -- if it was necessary it would be applied to all computers equally.
The study shows that many people leave their PS3s on all the time, and the power difference between running F@H and idle is small, so it's really the PS3 that's killing the planet. If you're going to waste power anyway, you might as well do some folding.
However, there is no reason a console should use 100W when idle. A laptop can drop its power consumption by a factor of 10 when it's idle; why can't a game console?
Slashdot Mirror
Sorry Sir, but it seems that a transistor has gone in the controller circuit and your data cannot be retrieved!
If the controller goes bad you can read out the data directly from the flash chips with nothing more than an FPGA and a TSOP test clip. Then you just have to reverse-engineer the FTL to reconstruct the data. Compared to recovering data from a mangled platter, it's a piece of cake.
Well, if the drives handle sequential access well and random access not so well, and Windows puts new files in the first available space while Linux places the files in the middle of the largest area of blank space, then does that create the possibility that Windows will currently handle these drives better than Linux?
No, because a lot of write traffic is to existing files and ext2/3/4 and NTFS are write-in-place filesystems, so for most writes the OS has no choice where to place them. ZFS or Btrfs may handle crap SSDs much better because they are write-anywhere.
The only way around this is to move into the "insanely smart" wear-leveling that will actually move data when the drive is otherwise idle to re-balance the sector write counts. You wouldn't want to do this during actual write requests, as it would slow them down even more than they are now. AFAIK, no SSD does this.
Intel claims that they do this, balancing the wear across all the flash, not just the part that is frequently rewritten.
Unless the design is broken
That's the problem; almost all SSDs on the market are broken and have worse write performance than hard disks. Just look at all the "JMicron rage" out there.
With all those people being cut off, what about the claim the internet "heals itself" and routes around damage?
That hasn't been true for many years if it ever was.
If there's any possible route between me and http://slashdot.org/ I want the system to find it, dammit!
If a possible route exists but no one has paid for it, traffic won't flow over that route. Your suggestion amounts to bandwidth socialism.
Traffic stopped flowing because neither Sprint nor Cogent paid for any alternate paths through the Net. This depeering reveals the fragility of the Tier 1 "people pay us, but we don't pay anyone" philosophy.
As soon as TCP notices a single packet loss, the Jacobson Algorithm kicks in and it's throttled to maybe 50-60%, and raises the limit slowly. I highly doubt that uTorrent's reworked version of UDP will play this nicely.
You're right; uTP is actually nicer than TCP.
RAID is built on the assumption that any hard disk has checksumming built right in.
Too bad that assumption is wrong. Despite the ECC in disks, corruption still sneaks in.
http://www.usenix.org/event/fast08/tech/full_papers/bairavasundaram/bairavasundaram_html/index.html
To fix this you can use either ZFS-style integrated filesystem and RAID or 3Par/Xiotech/XIV-style chunk RAID with checksums and unmap bad parts of the disk.
But darn it; when will someone finally offer a reasonably-priced, open-platform STB that serves as an A/V gateway to multiple Internet-based services â" one consumer-friendly, environmentally-designed, low-power gadget 'to rule them all,' if you will.
http://www.neurostechnology.com/neuros-link
Also runs Linux and a Web browser with Flash so it can access all the TV sites like Hulu.
Is this article a clever plant?
Sorry, that was a typo. I meant you can get .99+.99 Gbps out of Gigabit Ethernet.
15K drives exist for a reason (at least they did until now), and they're only available in SAS or FC. I suspect the SAS version is actually the cheap one.
You might want to invest in better equipment, because other people are getting 9.9+9.9 Gbps out of their Gigabit Ethernet.
You don't need to reserve space for wear-leveling.
Yes you do, because flash can't be rewritten in place. You have to write into an empty block and then erase the original block. If all the space was used, you'd have no empty blocks and thus couldn't write anything. More reserved space allows better wear leveling because you can choose to write to the empty block with the least wear.
Why isn't wasting RAM installed in your computer necessary to make it addressable?
Because DRAM can be rewritten in place.
The overhead really is necessary for bad block management and garbage collection. Some "enterprise" SSDs have 20% or even 50% overhead, so you're actually getting off easy.
It has 256GiB of raw capacity but some capacity is used for overhead so only 256GB is left.
There is some truth to what you say; I know the national labs in particular are working on a completely open source HPC stack. But many others in the HPC world have been using proprietary compilers, debuggers, filesystems, etc. for decades.
It requires a combination of hardware and software.
I'm not trying to defend Apple; I think they've done the wrong thing here. Midnight Thunder implied that Apple had no choice because all computers have HDCP, and I'm saying that's wrong. Apple didn't implement HDCP last year, so they should have kept on not implementing it. Obviously people shouldn't buy old computers to avoid DRM; that's inefficient and unsustainable.
Good luck finding a computer without it.
A 2007 MacBook.
The fact that the same video will play fine on a 2007 Mac but refuse to play on a 2008 Mac proves that the copy protection is not necessary -- if it was necessary it would be applied to all computers equally.
The study shows that many people leave their PS3s on all the time, and the power difference between running F@H and idle is small, so it's really the PS3 that's killing the planet. If you're going to waste power anyway, you might as well do some folding.
However, there is no reason a console should use 100W when idle. A laptop can drop its power consumption by a factor of 10 when it's idle; why can't a game console?
It supports MPEG-1, MPEG-2, H.264, and VC-1. No MPEG-4 (aka XVID/DIVX).
ftp://download.nvidia.com/XFree86/vdpau/doxygen/html/group___vdp_decoder.html
Google modified their home page to try to load a URL over IPv6; if it works then the client supports IPv6.
OpenCL will be the standard; it should support real processors, ATI, NVidia, and maybe Cell if someone bothers to write a backend.
Plenty of people seem to be getting serious work done with NVidia's proprietary Linux CUDA drivers.
The PS3 has a G7x GPU which never supported CUDA in the first place. At best you could do shader-based GPGPU if Sony allowed it.