-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
What's fastest? Large slow drive or small fast drive?
- Thread starter jtw99
- Start date
CubicZirconia
Diamond Member
The 7200.
CZroe
Lifer
WHAT? You guys have got to be crazy! The 5400RPM one! Think about it a second:
1 rotation can load a certain amount of data. Right?
So more rotations per second the more data can be read and transfered per second. Right?
Right, but this 5400RPM drive in the situation above has 3 times the data density.
If each rotation loads 3x the data it's much like having triple the rotational speed. 5400x3=much more than 7200
Not exactly 16200RPM performance because the seek times and such do not get better. The REAL performance measurement, sustained transfer rate, WILL nearly triple.
1 rotation can load a certain amount of data. Right?
So more rotations per second the more data can be read and transfered per second. Right?
Right, but this 5400RPM drive in the situation above has 3 times the data density.
If each rotation loads 3x the data it's much like having triple the rotational speed. 5400x3=much more than 7200
Not exactly 16200RPM performance because the seek times and such do not get better. The REAL performance measurement, sustained transfer rate, WILL nearly triple.
<< WHAT? You guys have got to be crazy! The 5400RPM one! Think about it a second:
1 rotation can load a certain amount of data. Right?
So more rotations per second the more data can be read and transfered per second. Right?
Right, but this 5400RPM drive in the situation above has 3 times the data density.
If each rotation loads 3x the data it's much like having triple the rotational speed. 5400x3=much more than 7200
Not exactly 16200RPM performance because the seek times and such do not get better. The REAL performance measurement, sustained transfer rate, WILL nearly triple. >>
That depends if the drives have the same number of platters. I'm not sure which drive models he is looking at, but it is very unlikely that the 5400 rpm will be faster than the 7200 rpm drive.
CZroe
Lifer
<< That depends if the drives have the same number of platters. >>
Not really because the overall data loaded from 1 rotation, say from an outer cylinder is still 1/3rd the density of an outer cylinder from the other drive.
Your logic does apply to drives with the same platter density, such as drives from the same series with the same RPMs. ie, 7200RPM IBM 20GB 60GXP vs 7200RPM IBM 60GB 60GXP (1 vs 3 platters)
Cylinder = 2 indentical tracks (Top and bottom) from every platter in the drive that is read at the same time. A cylinder's thickness does not matter because it is always 1 sector (512k) thick, so more cylinders does not affect much more than overall capacity performance wise.
All other things being equal, of course the 7200 rpm drive will be faster. But usually all other things are not equal. A lot depends on which generation of HD you are talking about, because platter density has invariable increased with each new generation of hd. The only real way to answer your question is to compare the specific models you are talking about through a number of different test. In other words, go to Storage Review, locate the drives in question on their database, and click on compare. That will give you the best answer possible.
WebDude😎
WebDude😎
CZroe
Lifer
<< The only real way to answer your question is to compare the specific models you are talking about through a number of different test. >>
Sorry, but that's wrong! 😀 you say
<< platter density has invariable increased >>
but it has nothing to do with platter density as I explained. It's overall density. A drive with three times the overall capacity has cylinders that are also exactly three times the capacity of the other's cylinders and therefore transfer three times as much data per rotation regardless of how many platters there are.
Its really impossible to tell like others have stated w/o knowing # of platters etc. My best advice is compare the specs for the exact drives and look at sustained thoroughput If I had to guess the 7200 20 gig is faster,BUT dont tell yourself that youll never use more than 10 gigs youd be surprised how fast you can take up space and you dont want to limit yourself because you didnt get a big enough drive.
"but it has nothing to do with platter density as I explained. It's overall density."
What on earth is overall density? You mean areal data density? We can use platter density for IDE drives because they are all 3.5" platters regardless of manufacturer or drive size/speed. SCSI uses varying sizes so you cannot make direct comparisons based on platter density.
"A drive with three times the overall capacity has cylinders that are also exactly three times the capacity of the other's cylinders and therefore transfer three times as much data per rotation regardless of how many platters there are."
That's not necessarily true. If all the increased data density was because of increased cluster density within the tracks, that would be true. But that's not the way it works. Areal data density is increased by increasing the track count and the cluster count within each track. If your theory was accurate than an IDE drive with twice the platter density would have twice the STR, and there are no real world examples of that.
There are no 60GB/platter drives either, so we know right there that the 60GB drive is a multiplatter drive. The 20GB drive is probably only half a platter while the 60 is a platter and a half. Assuming relatively similar age drives the 7200 is unquestionably faster even if the platter densities aren't the same. Even if the STR was a bit behind for the 7200, the better access time would more than make up for it 90+% of the time.
What on earth is overall density? You mean areal data density? We can use platter density for IDE drives because they are all 3.5" platters regardless of manufacturer or drive size/speed. SCSI uses varying sizes so you cannot make direct comparisons based on platter density.
"A drive with three times the overall capacity has cylinders that are also exactly three times the capacity of the other's cylinders and therefore transfer three times as much data per rotation regardless of how many platters there are."
That's not necessarily true. If all the increased data density was because of increased cluster density within the tracks, that would be true. But that's not the way it works. Areal data density is increased by increasing the track count and the cluster count within each track. If your theory was accurate than an IDE drive with twice the platter density would have twice the STR, and there are no real world examples of that.
There are no 60GB/platter drives either, so we know right there that the 60GB drive is a multiplatter drive. The 20GB drive is probably only half a platter while the 60 is a platter and a half. Assuming relatively similar age drives the 7200 is unquestionably faster even if the platter densities aren't the same. Even if the STR was a bit behind for the 7200, the better access time would more than make up for it 90+% of the time.
Kingofcomputer
Diamond Member
Just look at the datasheet from manufacturer, it states the real internal transfer rate, you don't need to care about the platter density, number of platter.
you dont want to limit yourself because you didnt get a big enough drive
I already have both drives. I orginally got the 7200 rpm drive because of the speed, but I read on here earlier that the larger capacity drive would give faster data transfer. I didn't realize it was such a disputed topic! Can anybody recommend a benchmark program that would give a definitive answer?
I already have both drives. I orginally got the 7200 rpm drive because of the speed, but I read on here earlier that the larger capacity drive would give faster data transfer. I didn't realize it was such a disputed topic! Can anybody recommend a benchmark program that would give a definitive answer?
CZroe
Lifer
<< Its really impossible to tell like others have stated w/o knowing # of platters etc. >>
Wrong. Read the post right above yours 😀
<< What on earth is overall density? You mean areal data density? We can use platter density for IDE drives because they are all 3.5" platters regardless of manufacturer or drive size/speed. SCSI uses varying sizes so you cannot make direct comparisons based on platter density. >>
"Overall density" is the total capacity. And I've seen plenty of 5.25" IDE drives 🙂
<< If all the increased data density was because of increased cluster density within the tracks, that would be true. But that's not the way it works. Areal data density is increased by increasing the track count and the cluster count within each track. If your theory was accurate than an IDE drive with twice the platter density would have twice the STR, and there are no real world examples of that. >>
A track is always only one sector thick. Therefore a cylinder is only one sector thick. A cluster is a file-system definition, and isn't relevant. Yes, there are real-world examples. The 60GXP thing is one refering to using platters to increase the data per cylinder, but performance increases in the same way with triple the capacity of all three platters in a single platter drive, the capacity of a single cylinder will equal 3x the capacity of the 3 platter drive's cylinders and therefore 3x the STR performance on the outer tracks.
CZroe
Lifer
<< Internal transfer rate is a pretty useless stat which never comes into play during real world usage. >>
WHAT?!:Q The useless stat is the burst speeds because the drive can only maintain them for a split second. This refers to the maximum it can maintain, and therefore its true performance!
CZroe,
<< A drive with three times the overall capacity has cylinders that are also exactly three times the capacity of the other's cylinders and therefore transfer three times as much data per rotation regardless of how many platters there are. >>
The "data density" of a drive does not increase merely by raising its overall capacity. Manufacturers add extra platters to provide the extra space. Each of these platters have the same data density as the others, and the drive can only read from one platter at a time.
If your theory was correct, a 60G drive would have three times the transfer speed of a 20G drive, which doesn't happen. Check the benchmarks. In fact, the overall capacity of a drive is so irrelevant to its performance that competent reviewers like StorageReview simply pick an arbitrary capacity from a drive family and test that; they know that all drives from the same family will perform roughly the same, regardless of their overall capacity.
To answer the original question, assuming the drives were manufactured within two years of each other, the 7200rpm unit should be faster. Platter densities increase over time, so that a modern 5400rpm might have a higher transfer rate than an older 7200rpm unit. But if they were manufactured close together, the 7200rpm drive will definitely be faster, in both access time and throughput.
Modus
<< A drive with three times the overall capacity has cylinders that are also exactly three times the capacity of the other's cylinders and therefore transfer three times as much data per rotation regardless of how many platters there are. >>
The "data density" of a drive does not increase merely by raising its overall capacity. Manufacturers add extra platters to provide the extra space. Each of these platters have the same data density as the others, and the drive can only read from one platter at a time.
If your theory was correct, a 60G drive would have three times the transfer speed of a 20G drive, which doesn't happen. Check the benchmarks. In fact, the overall capacity of a drive is so irrelevant to its performance that competent reviewers like StorageReview simply pick an arbitrary capacity from a drive family and test that; they know that all drives from the same family will perform roughly the same, regardless of their overall capacity.
To answer the original question, assuming the drives were manufactured within two years of each other, the 7200rpm unit should be faster. Platter densities increase over time, so that a modern 5400rpm might have a higher transfer rate than an older 7200rpm unit. But if they were manufactured close together, the 7200rpm drive will definitely be faster, in both access time and throughput.
Modus
CZroe
Lifer
<< the drive can only read from one platter at a time >>
Nope. It reads one CYLINDER at a time. A cylinder consists of 1 track from each side of each platter. It's in any A+ Cert. book
<< If your theory was correct, a 60G drive would have three times the transfer speed of a 20G drive, which doesn't happen. Check the benchmarks. >>
I think you are forgetting, that I am the one saying that cylinder count doesn't make a difference between these two, but a 60GB drive DOES have nearly 3x the sustained transfer of a 20GB drive in the same series because the 2 extra 20GB platters make for a cylinder with 3x the data. Therefore it is still a cylinder capacity issue like I pointed out, not a platter count one. This is the primary misconception in drive benchmarks, and *ALL* IBM 60GXP retailers always list the specs for the highest capacity one, which furthers the misconception. I've noted several IBM 60GXP benchmarks in specific that prove this. Besides, what I'm trying to prove is that platters make no difference, because the one with the higher capacity STILL has a higher capacity per cylinder, no matter how many platters it has.
Nope. It reads one CYLINDER at a time. A cylinder consists of 1 track from each side of each platter. It's in any A+ Cert. book
<< If your theory was correct, a 60G drive would have three times the transfer speed of a 20G drive, which doesn't happen. Check the benchmarks. >>
I think you are forgetting, that I am the one saying that cylinder count doesn't make a difference between these two, but a 60GB drive DOES have nearly 3x the sustained transfer of a 20GB drive in the same series because the 2 extra 20GB platters make for a cylinder with 3x the data. Therefore it is still a cylinder capacity issue like I pointed out, not a platter count one. This is the primary misconception in drive benchmarks, and *ALL* IBM 60GXP retailers always list the specs for the highest capacity one, which furthers the misconception. I've noted several IBM 60GXP benchmarks in specific that prove this. Besides, what I'm trying to prove is that platters make no difference, because the one with the higher capacity STILL has a higher capacity per cylinder, no matter how many platters it has.
CZroe
Lifer
Personally, I don't buy Maxtors because they do not provide a decent *hardware* diagnostic utility that can interface with the drive controller itself to detect and remap bad sectors to spares. I've had enough trouble with Maxtor, Toshiba, Quantum, and others that are nearly impossible to diagnose as a bad drive or bad sector because the issues can not even be detected without a spontaneous restart as soon as the error is encountered, even with the supplied "software" diagnostics. With these errors, you can't mark a cluster as bad in the file system because you can't access it without a restart unless you have a hardware diag utility. I've seen so many people format their drives after this occurs only to see it crash in the middle of formatting and then have to resort to trial and error ways to get partitions made that do not use these areas of the disk. It's HELL and it's a problem on nearly every laptop over 3 years old with a Toshiba drive in it (Try finding a replacement for one that cost less than a new laptop). Things are a little better on the desktop side of things, but still, why woould you buy a new drive if you didn't know what was wrong with the old one yet? Or if it was possible to fix... Actually, IBM and Western Digital are the only manufacturers that make such diagnostic tools, so they are the only ones I would consider buying. Even if a Maxtor is less likely to fail, you won't really be able to tell when it starts failing! Many times, the only indication I have that anything is wrong, despite thourogh ScanDisking and other diagnostics, is the hardware diag reports. So if you think a Maxtor or whatever is perfectly fine just because you ran Norton DiskDoctor on it, I wouln't trust that if I were you.
I always have to bring this up 😀
I always have to bring this up 😀
CZroe, I was going to respond to your other posts, but with that last one claiming drives within the same series with 2 platters have 2x the transfer rates of one platter, I know you have no idea what you are talking about. The only thing you were right about was my misnaming of sectors as clusters, I shouldn't post at 7AM on a Sunday.
"WHAT?! The useless stat is the burst speeds because the drive can only maintain them for a split second. This refers to the maximum it can maintain, and therefore its true performance!"
Internal transfer rate is not the same thing and STR. Go look up the definition and get back to us.
Only one platter can be read from at once. If you don't beleive that, go to StorageReview and look at some benchmarks.
45GB IBM 75GXP - 3 platters - STR - (37200KB/s - 19700KB/s)
75GB IBM 75GXP - 5 platters - STR - (36967KB/s - 19933KB/s)
"WHAT?! The useless stat is the burst speeds because the drive can only maintain them for a split second. This refers to the maximum it can maintain, and therefore its true performance!"
Internal transfer rate is not the same thing and STR. Go look up the definition and get back to us.
Only one platter can be read from at once. If you don't beleive that, go to StorageReview and look at some benchmarks.
45GB IBM 75GXP - 3 platters - STR - (37200KB/s - 19700KB/s)
75GB IBM 75GXP - 5 platters - STR - (36967KB/s - 19933KB/s)
CZroe
Lifer
Looks like you guys are right about the platters (But I've read that in so many places!). Scott Meuller himself says:
"To calculate the true transfer rate of a drive, you need to know several important specifications. The two most important specifications are the true rotational speed of the drive (in RPM) and the average number of physical sectors on each track [notice, tracks, not cylinders]...When you know these figures, you can use the following formula to determine the maximum transfer rate in millions of bits per second (Mbps): Maximum Data Transfer Rate (Mbps) = SPT x 512 bytes x RPM / 60 seconds / 1,000,000 bits For example, the ST-12551N 2G 3 1/2-inch drive spins at 7,200 RPM and has an average of 81 sectors per track. The maximum transfer rate for this drive is figured as follows: 81 x 512 x 7,200 / 60 / 1,000,000 = 4.98Mbps Using this formula, you can calculate the true maximum sustained transfer rate of any drive." Although that last bit of info was wrong (It is calculating MBps, not Mbps), I consider him the authority. Now to go track down those misleading 60GXP reviews I read about a year ago! :|
"To calculate the true transfer rate of a drive, you need to know several important specifications. The two most important specifications are the true rotational speed of the drive (in RPM) and the average number of physical sectors on each track [notice, tracks, not cylinders]...When you know these figures, you can use the following formula to determine the maximum transfer rate in millions of bits per second (Mbps): Maximum Data Transfer Rate (Mbps) = SPT x 512 bytes x RPM / 60 seconds / 1,000,000 bits For example, the ST-12551N 2G 3 1/2-inch drive spins at 7,200 RPM and has an average of 81 sectors per track. The maximum transfer rate for this drive is figured as follows: 81 x 512 x 7,200 / 60 / 1,000,000 = 4.98Mbps Using this formula, you can calculate the true maximum sustained transfer rate of any drive." Although that last bit of info was wrong (It is calculating MBps, not Mbps), I consider him the authority. Now to go track down those misleading 60GXP reviews I read about a year ago! :|
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
