If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#122
|
|||
|
|||
"Neil Maxwell" wrote in message ... On Thu, 4 Nov 2004 01:21:39 -0000, "half_pint" wrote: I can wander into any PCfashionvictim store and click on a file, it will half_pint stumbles and wobbles, swinging wildly, clearly ready to go down for the count. The crowd would go wild, but it got bored and wandered off in search of a fair fight some time ago. The locals circle lazily, wondering if it's worth the effort of a few more punches... I think it's about time you quit taking hallucination inducing drugs as it appears to have become a permenant condition. -- Neil Maxwell - I don't speak for my employer |
#123
|
|||
|
|||
|
#124
|
|||
|
|||
Imagine two hard drives - both single platter single-sided, same rotational speed. One is 1GB, the other is 4GB. In order to get 4GB in the same space as 1GB, the larger drive has twice as many tracks, and each track holds twice as much data as a track on the 1GB drive. Now imagine that you want to read a file that occupies 3/4 track on the 4GB drive - this will occupy 1 1/2 tracks on the 1GB drive, so while the 4GB drive can read it in a single revolution, the 1GB drive will need to make up to two revolutions to read the same amount of data. So, all other things being equal (which they are not) the 4GB drive is up to twice as fast as the 1GB drive. Are you with me so far? Yes!! Now imagine a file which occupies 1 degree of the track (on *my* drive) and it is 180%s away (%=degree here) the disk has to spin 181%s to get the data, however on your 'faster' drive it only has to spin 180.5 degrees, wow!!!!! thats a great improvement!!!!! your drive is 181/180.5 faster than mine, that is 1.00277 or 0.277% (back to real percents now). So you are a quarter of a percent faster than me!! Big deal!!!!!!! you would never notice it!!!! Actually nobody will notice that because the likelyhood of above situation is extremely small. By the way, what you describe is more likely a random, not sequential access pattern. As we all know random access depends mostly on seek time and latency. For a single track that would mean latency only (assuming the same settle time, command overhead and others). So reading very small chunks of data randomly will be done with the same speed. I guess your hard drive contains only very small files. And all of them are placed on the single track. The rest of the disk is empty. |
#125
|
|||
|
|||
"Peter" wrote in message ... Imagine two hard drives - both single platter single-sided, same rotational speed. One is 1GB, the other is 4GB. In order to get 4GB in the same space as 1GB, the larger drive has twice as many tracks, and each track holds twice as much data as a track on the 1GB drive. Now imagine that you want to read a file that occupies 3/4 track on the 4GB drive - this will occupy 1 1/2 tracks on the 1GB drive, so while the 4GB drive can read it in a single revolution, the 1GB drive will need to make up to two revolutions to read the same amount of data. So, all other things being equal (which they are not) the 4GB drive is up to twice as fast as the 1GB drive. Are you with me so far? Yes!! Now imagine a file which occupies 1 degree of the track (on *my* drive) and it is 180%s away (%=degree here) the disk has to spin 181%s to get the data, however on your 'faster' drive it only has to spin 180.5 degrees, wow!!!!! thats a great improvement!!!!! your drive is 181/180.5 faster than mine, that is 1.00277 or 0.277% (back to real percents now). So you are a quarter of a percent faster than me!! Big deal!!!!!!! you would never notice it!!!! Actually nobody will notice that because the likelyhood of above situation is extremely small. By the way, what you describe is more likely a random, not sequential access pattern. As we all know random access depends mostly on seek time and latency. For a single track that would mean latency only (assuming the same settle time, command overhead and others). So reading very small chunks of data randomly will be done with the same speed. I guess your hard drive contains only very small files. And all of them are placed on the single track. The rest of the disk is empty. I am not 100% sure of the term seek time and latency in this context. One is changing track anf the other waiting for the disk to spin? Reading large files is not generally a problem as their processing is generally slower than the read. It is the constant switching between files etc... (seeking?) which slows things down considerabley. Web browsing, for istance involves a lot of very small files, scattered "all over the shop", reading the data is not a problem, it is the finding the data to read which is the big problem. |
#126
|
|||
|
|||
"Rob Morley" wrote in message t... In article , "half_pint" says... "Rob Morley" wrote in message t... snip Let's try this using really basic concepts and small words, shall we? Imagine two hard drives - both single platter single-sided, same rotational speed. One is 1GB, the other is 4GB. In order to get 4GB in the same space as 1GB, the larger drive has twice as many tracks, and each track holds twice as much data as a track on the 1GB drive. Now imagine that you want to read a file that occupies 3/4 track on the 4GB drive - this will occupy 1 1/2 tracks on the 1GB drive, so while the 4GB drive can read it in a single revolution, the 1GB drive will need to make up to two revolutions to read the same amount of data. So, all other things being equal (which they are not) the 4GB drive is up to twice as fast as the 1GB drive. Are you with me so far? Yes!! Now imagine a file which occupies 1 degree of the track (on *my* drive) and it is 180%s away (%=degree here) the disk has to spin 181%s to get the data, however on your 'faster' drive it only has to spin 180.5 degrees, wow!!!!! thats a great improvement!!!!! your drive is 181/180.5 faster than mine, that is 1.00277 or 0.277% (back to real percents now). So you are a quarter of a percent faster than me!! Big deal!!!!!!! you would never notice it!!!! Are you still with me? Indeed. Now take into account that the head of the 1GB drive in my example has to move from one track to the next before it can read the second track - by the time it gets there it might have missed the start of the data, in which case it would need another revolution before it was read. This makes the 1GB drive a third of the speed of the 4GB drive. Now lets look at the combined effect of seek time and access time. Assume an average seek time of 11 milliseconds for both disks. For a 5400RPM drive one revolution takes 11 milliseconds. So for the scenario in my example you have for the 1GB disk: 11 + 11 + 11 + 11 = 44mS for the 4GB disk: 11 + 11 = 22 mS That's the worst-case scenario on an unfragmented disk. As you said, the best-case scenario sees very little difference. So on average we might expect the 4GB drive to have completed its read in 66% of the time that it takes the 1GB drive. Remember we're talking a 4x difference in arial density - the drives you were originally comparing were ISTR 5GB per platter versus 60 GB per platter, which gives a difference in linear density of around 330%, while my example used 200%. Now tell me that the small drive is as fast as the big one - show your workings. I will do that tomorrow when I have more time :O) |
#127
|
|||
|
|||
half_pint wrote:
"Isaac" wrote in message .. . On Wed, 03 Nov 2004 23:41:43 -0500, J. Clarke wrote: half_pint wrote: . Ok, so my drive spins at 7200 rpm, the same as yours. How big are your platters? Lets be VERY generous, and say the full 5 gig capacity of your drive is on a single platter. My smallest drive is 180 gigs - lets say there are 3 platters there. My platters therefore hold 60 gigs each, despite being the same physical size as your platters. Therefore the data density on my platters is 12 times greater than on yours. Therefore, for each revolution of the platter, my drive can read 12 times more data. That`s 12 times the amount of data in the same amount of time, making the data transfer rate 12 times greater. Is that simple enough for you, or is it still too complicated for you to understand? You have demonstrated how stupid you are, you have no idea how a computer works, statistacially the data will be on the other side of the drive and it will take your drive just as long to assess it as mine. (aprox bearing in mind your marginally faster spin speed). Actually, _statistically_ it will be halfway to the "other side of the drive", if in fact it is truly being sought at random. That model is sometimes valid but not always--sometimes sequential access going from one track to the next with interleaving to reduce latency to near zero is the correct model. In fact for situations where you are burning data to a CD or DVD, most likely the next needed chunk of data is right next to the last accessed piece unless your drive is extremely fragmented. On average the transfer rate is not determined by the average time to get to a randomly located piece of data. The average transfer rate is going to predominately related to how fast the drive can reach adjacent chunks. At this point you have to assume that Half's obtuseness is deliberate. Well not really, I do actually have a point but I should mention we are talking about hard drives rather than CDs or DVD. My point is that that there are two aspects to drive speed, data transfer rate and 'seek time' (is that the correct expression?). Seek times have not improved much at all as they depend on spin speed (and other *mechanical* things which have not improved by much at all (I doubt changing track has improved much either)). Data transfer time is pretty negligle compared to seek time for most applications so I am basically correct (as usual). You don't even know the correct terminology and yet you blithely assert that "data transfer time is pretty negligible compared to seek time for most applications" as if by asserting this you make it true. Isaac -- --John Reply to jclarke at ae tee tee global dot net (was jclarke at eye bee em dot net) |
#128
|
|||
|
|||
Yes!! Now imagine a file which occupies 1 degree of the track
(on *my* drive) and it is 180%s away (%=degree here) the disk has to spin 181%s to get the data, however on your 'faster' drive it only has to spin 180.5 degrees, wow!!!!! thats a great improvement!!!!! your drive is 181/180.5 faster than mine, that is 1.00277 or 0.277% (back to real percents now). So you are a quarter of a percent faster than me!! Big deal!!!!!!! you would never notice it!!!! Actually nobody will notice that because the likelyhood of above situation is extremely small. By the way, what you describe is more likely a random, not sequential access pattern. As we all know random access depends mostly on seek time and latency. For a single track that would mean latency only (assuming the same settle time, command overhead and others). So reading very small chunks of data randomly will be done with the same speed. I guess your hard drive contains only very small files. And all of them are placed on the single track. The rest of the disk is empty. I am not 100% sure of the term seek time and latency in this context. One is changing track anf the other waiting for the disk to spin? Reading large files is not generally a problem as their processing is generally slower than the read. It is the constant switching between files etc... (seeking?) which slows things down considerabley. Web browsing, for istance involves a lot of very small files, scattered "all over the shop", reading the data is not a problem, it is the finding the data to read which is the big problem. Latency [ms] = 30000 / SpindleSpeed [RPM] |
#129
|
|||
|
|||
|
#130
|
|||
|
|||
"Toshi1873" wrote in message
.. . In article , says... Latency [ms] = 30000 / SpindleSpeed [RPM] Where does the 30000 figure come from? If SpindleSpeed is in rpm, we get: Time for one revolution in minutes = 1 / SpindleSpeed There are 60000 ms in a minute, so this is equivalent to: Time for one revolution in ms = 60000 / SpindleSpeed On average, rotational latency is the time for half a revolution, hence: Latency in ms = (1/2) * 60000 / SpindleSpeed = 30000 / SpindleSpeed HTH, Alex |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
my new mobo o/c's great | rockerrock | Overclocking AMD Processors | 9 | June 30th 04 08:17 PM |
Sata and Data Corruption | Robert Neville | General | 7 | April 25th 04 11:02 AM |
Sata and Data Corruption | Robert Neville | Homebuilt PC's | 7 | April 25th 04 11:02 AM |
Cost of blank CDs versus DVDs | Doug Ramage | Cdr | 12 | April 17th 04 07:31 PM |
Backup Small Office Data | Jim Turner | General | 6 | August 17th 03 09:31 PM |