Hi Guys, Can you give me your opinion on what would be the effect on personal computers (PCs) if their hard disk systems had NO error correction? Would disk storage capacity be reduced? Would data access time be drastically increased? Thanks for your thoughts. See Ya', [-Rick-]
PC hard disk error correction question
Started by ●April 24, 2013
Reply by ●April 24, 20132013-04-24
On Wed, 24 Apr 2013 13:57:34 -0700, Rick Lyons <R.Lyons@_BOGUS_ieee.org> wrote:> >Hi Guys, > Can you give me your opinion on what >would be the effect on personal computers >(PCs) if their hard disk systems had NO >error correction? > >Would disk storage capacity be reduced? >Would data access time be drastically >increased? > >Thanks for your thoughts. > >See Ya', >[-Rick-] > >In general I think the data reliability just goes down so that there are more read faults (e.g., corrupted files, etc.). I think the error mechanisms are a moving target as the technologies change and the storage densities keep increasing, but back in the day when I looked at this many years ago, there were multiple sources of errors. Some errors are just due to a read error, e.g., a 1 is stored but a 0 is read, and some errors are due to a storage error e.g., where a 1 is stored where a 0 should have been. Error detection can fix a read error by retrying, but a storage error can't be fixed without ECC, so the statistics of read errors vs storage errors will affect the answer to your question (as will other stuff like that). If you search on terms like "hard disk soft error correction" or something like that you'll find papers like this, that may or may not be helpful to answer your question: https://engineering.purdue.edu/dcsl/publications/papers/2012/softerrorconsequences_DSN2012.pdf It seems that people are still calling "hard errors" the errors that make it past the ECC, i.e., that the ECC can't fix, whereas soft errors are anuthing that doesn't make it through the ECC and is fixed before it gets to the user. I think if you can find data about pertinent statistics regarding the frequency of soft errors as compared to hard errors in that context it may shed light on the importance of the ECC. Basically, it'll increase the frequency of hard errors if there's no ECC, so whatever affect that has will be amplified. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by ●April 24, 20132013-04-24
On 4/24/2013 6:15 PM, Eric Jacobsen wrote:> On Wed, 24 Apr 2013 13:57:34 -0700, Rick Lyons > <R.Lyons@_BOGUS_ieee.org> wrote: > >> >> Hi Guys, >> Can you give me your opinion on what >> would be the effect on personal computers >> (PCs) if their hard disk systems had NO >> error correction? >> >> Would disk storage capacity be reduced? >> Would data access time be drastically >> increased? >> >> Thanks for your thoughts. >> >> See Ya', >> [-Rick-] >> >> > > In general I think the data reliability just goes down so that there > are more read faults (e.g., corrupted files, etc.)............. Bla-bla-bla. The coding gain is going to be somewhere in 6dB ballpark. That is 4 times in power difference = 4 times in the surface area. Or, for the same physical size and spin rate, ~ 1/4 of capacity and ~1/2 of the data transfer rate. Vladimir Vassilevsky DSP and Mixed Signal Designs www.abvolt.com
Reply by ●April 24, 20132013-04-24
Rick Lyons <R.Lyons@_bogus_ieee.org> wrote: (snip)> Can you give me your opinion on what > would be the effect on personal computers > (PCs) if their hard disk systems had NO > error correction?> Would disk storage capacity be reduced? > Would data access time be drastically > increased?Search for PRML. That isn't exactly error correction, but it is related to getting the best signal off the disk even as density increases. In the early disks systems they were lucky to get enough signal off the disk at all, and so used simple coding and low density to allow it to work. As technology improved, including the ability to decode more complex bit coding methods, density could increase. Compared to CD (and CD-ROM), though, magnetic disks have relatively little error correction. CDs have to be able to read through dust and scratches and manufacturing defects that magnetic disks don't. -- glen
Reply by ●April 25, 20132013-04-25
On Wed, 24 Apr 2013 13:57:34 -0700, Rick Lyons <R.Lyons@_BOGUS_ieee.org> wrote:> >Hi Guys, > Can you give me your opinion on what >would be the effect on personal computers >(PCs) if their hard disk systems had NO >error correction? > >Would disk storage capacity be reduced? >Would data access time be drastically >increased? > >Thanks for your thoughts. > >See Ya', >[-Rick-] > >Do you have a particular situation in mind? Since ECC strategies in most current drives are proprietary (and closely held) there's no easy accurate answer for this without insider info, which is likely hard to get. I don't know what sort of statistics current drives (especially enterprise-grade drives) will report, but without some sort of empirical data or inside info regarding the ECC, the magnitudes of the effects can only be generalized in a broad sense. I would expect capacity to be reduced, but the affect on access time may be harder to predict. I can think of reasons why it might go up or down without ECC, same for transfer rates. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by ●April 27, 20132013-04-27
On Thu, 25 Apr 2013 00:24:31 +0000 (UTC), glen herrmannsfeldt <gah@ugcs.caltech.edu> wrote:>Rick Lyons <R.Lyons@_bogus_ieee.org> wrote: > >(snip) > >> Can you give me your opinion on what >> would be the effect on personal computers >> (PCs) if their hard disk systems had NO >> error correction? > >> Would disk storage capacity be reduced? >> Would data access time be drastically >> increased? > >Search for PRML. > >That isn't exactly error correction, but it is related to getting >the best signal off the disk even as density increases. > >In the early disks systems they were lucky to get enough signal >off the disk at all, and so used simple coding and low density >to allow it to work. As technology improved, including the ability >to decode more complex bit coding methods, density could increase. > >Compared to CD (and CD-ROM), though, magnetic disks have relatively >little error correction. CDs have to be able to read through dust >and scratches and manufacturing defects that magnetic disks don't. > >-- glenHi glen, PRML huh? I'll search for it. Thanks, [-Rick-]
Reply by ●April 27, 20132013-04-27
On Thu, 25 Apr 2013 20:52:38 GMT, eric.jacobsen@ieee.org (Eric Jacobsen) wrote:>On Wed, 24 Apr 2013 13:57:34 -0700, Rick Lyons ><R.Lyons@_BOGUS_ieee.org> wrote: > >> >>Hi Guys, >> Can you give me your opinion on what >>would be the effect on personal computers >>(PCs) if their hard disk systems had NO >>error correction? >> >>Would disk storage capacity be reduced? >>Would data access time be drastically >>increased? >> >>Thanks for your thoughts. >> >>See Ya', >>[-Rick-] >> > >Do you have a particular situation in mind? > >Since ECC strategies in most current drives are proprietary (and >closely held) there's no easy accurate answer for this without insider >info, which is likely hard to get. I don't know what sort of >statistics current drives (especially enterprise-grade drives) will >report, but without some sort of empirical data or inside info >regarding the ECC, the magnitudes of the effects can only be >generalized in a broad sense. > >I would expect capacity to be reduced, but the affect on access time >may be harder to predict. I can think of reasons why it might go up >or down without ECC, same for transfer rates. > > >Eric JacobsenHi Eric, thanks for your replies. I have no "application" in mind. Actually what I was thinking about is quite simple. I was trying to compile a list of how the average American's life would be different if digital signal processing (DSP) did NOT exist. For example, it seems to me, with no DSP there would be: No Internet, No cell phones, No digital cameras, No musical greeting cards, etc. Then I wondered how the absence of DSP would affect the performance personal computers (PCs). And that's why I asked my "What if hard disks had no error correction" question. But know I'm thinking, if there was no Internet then 98% of all PCs would be sitting in the corner collecting dust. In which case it wouldn't matter to the average person if their PC performance was degraded by the lack of hard disk error correction. See Ya', [-Rick-]
Reply by ●April 27, 20132013-04-27
On 2013-04-24 22:57, Rick Lyons wrote:> > Hi Guys, > Can you give me your opinion on what > would be the effect on personal computers > (PCs) if their hard disk systems had NO > error correction?No error correction, but error detection, or no error handling at all?> Would disk storage capacity be reduced? > Would data access time be drastically > increased?I could easily imagine that the error correction/detection would be done in SW, by the filesystem or device driver. Which in turn would mean, probably, some lower performance. This is something already happening, BTW, since it seems people do not trust the storage devices any more. bye, -- piergiorgio
Reply by ●April 27, 20132013-04-27
Rick Lyons <R.Lyons@_bogus_ieee.org> wrote:> On Thu, 25 Apr 2013 20:52:38 GMT, eric.jacobsen@ieee.org (Eric(snip)>>Do you have a particular situation in mind?(snip)>>I would expect capacity to be reduced, but the affect on access time >>may be harder to predict. I can think of reasons why it might go up >>or down without ECC, same for transfer rates.(snip)> I have no "application" in mind. Actually what > I was thinking about is quite simple. I was trying > to compile a list of how the average American's > life would be different if digital signal processing > (DSP) did NOT exist. For example, it seems to me, > with no DSP there would be:> No Internet, > No cell phones, > No digital cameras, > No musical greeting cards, > etc.Seems to me that the Internet would exist, but not "as we know it." Well, it depends on where you draw the line between ASP and DSP. (And why there is still not comp.asp.) I remember in the floppy disk (WD 1793) days, preference for analog (PLL based) data separators over digital ones. I would say that a PLL based data separator isn't using DSP. It supplies the clock edge, and the value is either 0 or 1 at that edge. Manchester coded ethernet also uses ASP techniques, including collision detect based on analog filtering. 100baseTX is close to the line, but 1000baseT definitely uses DSP techniques, especially for the echo cancellation. I would agree for cell phones and digital cameras. Musical greeting cards generating square waves dividing down from a reference frequency I might say aren't using DSP, but they might have improved since then.> Then I wondered how the absence of DSP would affect > the performance personal computers (PCs). And that's > why I asked my "What if hard disks had no error correction" > question.The original HD technology was pretty similar to floppy, but faster. There was CRC, but that isn't DSP and isn't correction in most cases. Much of the improvement in disk technology comes from closer head-disk spacing, which depends on mechnical and aerodynamics. But computational fluid dynamics isn't DSP, so some of that could have been done, anyway. I probably wouldn't have called the Reed-Solomon coding DSP related, but it seems that it can be explained either in mathematics terms or in DSP (frequency space) terms.> But know I'm thinking, if there was > no Internet then 98% of all PCs would be sitting > in the corner collecting dust. In which case it > wouldn't matter to the average person if their > PC performance was degraded by the lack of > hard disk error correction.I believe the 103 modem uses ASP techniques, but 212 should probably be considered using DSP techniques. But the real development of the internet was done on 56K leased lines. I am not sure how the modems for those work. 10broad36, the original broadband ethernet, could have been used for home cable modem technology. As well as I know it, that uses ASP techniques (filters). Much of DSP technology has allowed for things to be cheaper and more reliable, and also for economy of scale that wouldn't otherwise have existed. Digital filters are a lot easier to mass produce than hand tuned analog filters. -- glen
Reply by ●April 28, 20132013-04-28
On Sat, 27 Apr 2013 04:52:53 -0700, Rick Lyons <R.Lyons@_BOGUS_ieee.org> wrote:>On Thu, 25 Apr 2013 20:52:38 GMT, eric.jacobsen@ieee.org (Eric >Jacobsen) wrote: > >>On Wed, 24 Apr 2013 13:57:34 -0700, Rick Lyons >><R.Lyons@_BOGUS_ieee.org> wrote: >> >>> >>>Hi Guys, >>> Can you give me your opinion on what >>>would be the effect on personal computers >>>(PCs) if their hard disk systems had NO >>>error correction? >>> >>>Would disk storage capacity be reduced? >>>Would data access time be drastically >>>increased? >>> >>>Thanks for your thoughts. >>> >>>See Ya', >>>[-Rick-] >>> >> >>Do you have a particular situation in mind? >> >>Since ECC strategies in most current drives are proprietary (and >>closely held) there's no easy accurate answer for this without insider >>info, which is likely hard to get. I don't know what sort of >>statistics current drives (especially enterprise-grade drives) will >>report, but without some sort of empirical data or inside info >>regarding the ECC, the magnitudes of the effects can only be >>generalized in a broad sense. >> >>I would expect capacity to be reduced, but the affect on access time >>may be harder to predict. I can think of reasons why it might go up >>or down without ECC, same for transfer rates. >> >> >>Eric Jacobsen > >Hi Eric, thanks for your replies. > > I have no "application" in mind. Actually what >I was thinking about is quite simple. I was trying >to compile a list of how the average American's >life would be different if digital signal processing >(DSP) did NOT exist. For example, it seems to me, >with no DSP there would be: > >No Internet, >No cell phones, >No digital cameras, >No musical greeting cards, >etc.I think a big home for DSP in the last couple of decades has been image processing, for computer graphics (especially games), medical imaging, radar, movies, etc., etc. Music (including unfortunate things like autotuners) and CDs and MP3s are also common things that people can relate to.>Then I wondered how the absence of DSP would affect >the performance personal computers (PCs). And that's >why I asked my "What if hard disks had no error correction" >question.FWIW, I think of ECC and FEC separately from DSP, since they're more closely related to Information Theory and may be coding information that isn't related to a signal of any kind in the sense that we normally use for DSP. Channel coding (ECC/FEC) and source coding (compression) certainly have a lot of overlap in theories, though.> But know I'm thinking, if there was >no Internet then 98% of all PCs would be sitting >in the corner collecting dust.There were a lot of PCs and various flavors of home computers before the internet got popular. I think connectivity would be different without the internet, but 'puters and stuff would still be pretty siginificant.> In which case it >wouldn't matter to the average person if their >PC performance was degraded by the lack of >hard disk error correction. > >See Ya', >[-Rick-]Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
