I'm considering using a smaller HD for OS/programs on my Conroe build and a Samsung HD321KJ 320Gb main drive for data/storage.
For a quiet build with decent performance, what would you recommend that I use for the smaller OS/program drive? A Seagate 7200.9 80Gb Sata 3.0 drive?
Any suggestions?
Case: Antec Solo
Smallish drive for OS and programs
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
speedlever
- Posts: 128
- Joined: Sat Sep 30, 2006 1:39 pm
- Location: NC, USA
The differences between 7200 RPM drives is really pretty minor. If you want to speed things up you really have to go to 10K RPM (Raptor or SCSI) or a 15K SCSI. You can get these to be quiet with some work, but they are not silent.
FWIW, my main workstation has four SCSI drives: 3 15K and 1 10K. The fans (undervolted Panaflos) are louder than the drives.
You can get these quite inexpensively ($50 to $100) if you look around. Similarly, you can find perfectly adequate controllers for $20 to $40; decent cables cost about $20.
FWIW, my main workstation has four SCSI drives: 3 15K and 1 10K. The fans (undervolted Panaflos) are louder than the drives.
You can get these quite inexpensively ($50 to $100) if you look around. Similarly, you can find perfectly adequate controllers for $20 to $40; decent cables cost about $20.
-
speedlever
- Posts: 128
- Joined: Sat Sep 30, 2006 1:39 pm
- Location: NC, USA
SR's Performance Database: various 15krpm and 10krpm SCSI drive versus 10krpm Raptor 150 (SATA). Take a look at Single-User Suite 4.0 section aka real-life desktop performance. (Forget low-level benchmarks and server benchmarks as they have no meaning to a desktop user)
Best 15krpm SCSI drive for desktop use is no doubt Fujitsu MAU but it only manages to tie with new generation Raptor! And Raptor is much cheaper and doesn't require any addon controller to mount to your system.
Even some 7200rpm drives beat most (but not Fujitsu MAU) of 15krpm SCSI drives in desktop performance! See comparison Hitachi T7K500 & 7K500 versus 15krpm drives.
Also a comparison T7K500 vs different generations of Raptor.
My opinion: stay far away from SCSI. A Raptor is a better investment into a desktop computer. New generation 74GB Raptors (WD740ADFD) perform as about as good as Raptor 150 so I suggest you save some money there... just stay away from the old one of the same capacity (WD740GD) as some 7200rpm drive may outperform it.
Best 15krpm SCSI drive for desktop use is no doubt Fujitsu MAU but it only manages to tie with new generation Raptor! And Raptor is much cheaper and doesn't require any addon controller to mount to your system.
Even some 7200rpm drives beat most (but not Fujitsu MAU) of 15krpm SCSI drives in desktop performance! See comparison Hitachi T7K500 & 7K500 versus 15krpm drives.
Also a comparison T7K500 vs different generations of Raptor.
My opinion: stay far away from SCSI. A Raptor is a better investment into a desktop computer. New generation 74GB Raptors (WD740ADFD) perform as about as good as Raptor 150 so I suggest you save some money there... just stay away from the old one of the same capacity (WD740GD) as some 7200rpm drive may outperform it.
One simple question: have you used SCSI drives? If so, which ones?
I know the benchmarks well, but they do not correspond with my experience (or perhaps my workloads). The Raptor is of course a fine drive, and a very large step up from a 7.2K RPM drive. It should be considered if a quick, small capacity desktop drive is desired -- that is what is was designed for.
The choice between a Raptor and SCSIs is more complex. It depends on your work load -- mine favors SCSIs. It also depends on factors other than straight performance. I find I can purchase SCSI drives for a much lower cost, and they are still much more reliable. I also like the expandability of SCSI channels, where my four drives hang off a single $40 controller, and I can add quite a few more if I want.
In any event, the OP wanted a small, quick drive. Would you disagree that Raptors, SCSIs and RAID (perhaps 0+1) are the ways to do this?
I know the benchmarks well, but they do not correspond with my experience (or perhaps my workloads). The Raptor is of course a fine drive, and a very large step up from a 7.2K RPM drive. It should be considered if a quick, small capacity desktop drive is desired -- that is what is was designed for.
The choice between a Raptor and SCSIs is more complex. It depends on your work load -- mine favors SCSIs. It also depends on factors other than straight performance. I find I can purchase SCSI drives for a much lower cost, and they are still much more reliable. I also like the expandability of SCSI channels, where my four drives hang off a single $40 controller, and I can add quite a few more if I want.
In any event, the OP wanted a small, quick drive. Would you disagree that Raptors, SCSIs and RAID (perhaps 0+1) are the ways to do this?
DrJ: "One simple question: have you used SCSI drives? If so, which ones?"
I know what you're up to. You think you can discredit me just by proving that my opinion isn't based upon personal experiance. From my point of view, something being personal blurs people's vision and make people "see" thing that simply ain't there. Self-deception, that is. If I payed thousands of dollars to build a 1337 storage system with several 15krpm drives and the best server optimized hardware RAID (or alternatively a 40 dollar el Cheapo used with those expensive 15krpm drives), I'd too probably imagine all sorts of things about perceived performance difference. But in reality I would need to use a stop-watch to really notice the difference. (And like the SR benchmarks show: difference might even mean worse performance than with Raptor... or even with certain 7200rpm drives.)
"I know the benchmarks well, but they do not correspond with my experience"
Experience is easily conceived.
"(or perhaps my workloads)"
Haha. Are you saying you do so much multitasking at you computer that it resembles more that of highly random access server-benchmark scenario? Quite a common misconception...
"Raptor ... should be considered if a quick, small capacity desktop drive is desired -- that is what is was designed for."
Exactly! It is designed for desktop computers (including gaming, video editing, etc.) and it is perfectly natural it performs better than SCSI in the environment it has been designed for. SCSI drives are generally not designed for desktop use.
The real question: why are you recommending SCSI for desktop use?
I know what you're up to. You think you can discredit me just by proving that my opinion isn't based upon personal experiance. From my point of view, something being personal blurs people's vision and make people "see" thing that simply ain't there. Self-deception, that is. If I payed thousands of dollars to build a 1337 storage system with several 15krpm drives and the best server optimized hardware RAID (or alternatively a 40 dollar el Cheapo used with those expensive 15krpm drives), I'd too probably imagine all sorts of things about perceived performance difference. But in reality I would need to use a stop-watch to really notice the difference. (And like the SR benchmarks show: difference might even mean worse performance than with Raptor... or even with certain 7200rpm drives.)
"I know the benchmarks well, but they do not correspond with my experience"
Experience is easily conceived.
"(or perhaps my workloads)"
Haha. Are you saying you do so much multitasking at you computer that it resembles more that of highly random access server-benchmark scenario? Quite a common misconception...
"Raptor ... should be considered if a quick, small capacity desktop drive is desired -- that is what is was designed for."
Exactly! It is designed for desktop computers (including gaming, video editing, etc.) and it is perfectly natural it performs better than SCSI in the environment it has been designed for. SCSI drives are generally not designed for desktop use.
The real question: why are you recommending SCSI for desktop use?
DrJ: "The choice between a Raptor and SCSIs is more complex. It depends on your work load -- mine favors SCSIs."
Sure. If you have a webserver and a very fast web connection (and lots of people who are interested in the content you are serving) you (or they) can no doubt benefit from a SCSI drive. 10krpm SCSI is better optimized of this kind of workload than 10krpm Raptor... not to mention how 15krpm SCSI pwns Raptor.
But desktop performance... no. No chance. SCSI sucks. They simply aren't designed for that kind of work.
"I find I can purchase SCSI drives for a much lower cost, and they are still much more reliable."
Are you comparing equivalent drives of SCSI vs. ATA? At least the price range "$50 to $100" is no doubt second hand stuff or refurbished (= once failed pieces of junk) ones. And can't you get a new, unused, non-refurbished 36GB Raptor at that price as well? Or add a bit more and buy a 74GB Raptor. (I admit though that 150GB Raptor has quite a bit over the top price tag...)
"In any event, the OP wanted a small, quick drive. Would you disagree that Raptors, SCSIs and RAID (perhaps 0+1) are the ways to do this?"
Sure, they are all quick, but that doesn't change the fact that Raptor is quicker than SCSI. Also, even a 7200rpm drive for OS/programs might be quick (enough). The reason to have a separate drive for OS isn't just to allow use of faster and more expensive (price/GB) drive for critical applications (and thus save money because the rest is stored more cheaply) but also to separate data and programs that read or write the data from each other, thus improving performance by reducing seeks. If seeks are reduced, rotational latency and seek time become less of meaning.
Sure. If you have a webserver and a very fast web connection (and lots of people who are interested in the content you are serving) you (or they) can no doubt benefit from a SCSI drive. 10krpm SCSI is better optimized of this kind of workload than 10krpm Raptor... not to mention how 15krpm SCSI pwns Raptor.
But desktop performance... no. No chance. SCSI sucks. They simply aren't designed for that kind of work.
"I find I can purchase SCSI drives for a much lower cost, and they are still much more reliable."
Are you comparing equivalent drives of SCSI vs. ATA? At least the price range "$50 to $100" is no doubt second hand stuff or refurbished (= once failed pieces of junk) ones. And can't you get a new, unused, non-refurbished 36GB Raptor at that price as well? Or add a bit more and buy a 74GB Raptor. (I admit though that 150GB Raptor has quite a bit over the top price tag...)
"In any event, the OP wanted a small, quick drive. Would you disagree that Raptors, SCSIs and RAID (perhaps 0+1) are the ways to do this?"
Sure, they are all quick, but that doesn't change the fact that Raptor is quicker than SCSI. Also, even a 7200rpm drive for OS/programs might be quick (enough). The reason to have a separate drive for OS isn't just to allow use of faster and more expensive (price/GB) drive for critical applications (and thus save money because the rest is stored more cheaply) but also to separate data and programs that read or write the data from each other, thus improving performance by reducing seeks. If seeks are reduced, rotational latency and seek time become less of meaning.
I don't know why I am bothering with this...
Since you seem know so much about how I work, could you please tell me why you believe it is different?
So the answer is "no." You go on this whole diatribe with no experience at all.whiic wrote:DrJ: "One simple question: have you used SCSI drives? If so, which ones?"
I know what you're up to...
Sigh. The four drives cost me a total of $250, and the "cheapo" card is an LSIU160. It is fine for the throughputs I need. It may be inexpensive, but it is reliable and well-supported. And yes, I do use server-class hardware, including an E-ATX motherboard, dual processors (the old style) and ECC/Reg memory. I also run a server-class operating system.If I payed thousands of dollars to build a 1337 storage system with several 15krpm drives and the best server optimized hardware RAID (or alternatively a 40 dollar el Cheapo used with those expensive 15krpm drives), I'd too probably imagine all sorts of things about perceived performance difference.
So if defense of someone else's benchmark without real-world experience.Experience is easily conceived."I know the benchmarks well, but they do not correspond with my experience"
Yes, that is exactly what I am claiming. When I push the drives, that is exactly what happens. Other times, of course, it does not, but then there is no benefit to any particular drive over another.Haha. Are you saying you do so much multitasking at you computer that it resembles more that of highly random access server-benchmark scenario?
Since you seem know so much about how I work, could you please tell me why you believe it is different?
Of course they are refurbished. I have had one out-of-box failure that was replaced under warranty, but that happens. It happens for all electronics. The age of those refurbished drives varies from 10 to 30,000 hours hours. None have failed. They are far from "junk" if they perform as expected and over the long term.Are you comparing equivalent drives of SCSI vs. ATA? At least the price range "$50 to $100" is no doubt second hand stuff or refurbished (= once failed pieces of junk) ones.
Because SCSI is a valid option to the question posed originally, as is the Raptor or RAID.The real question: why are you recommending SCSI for desktop use?
-
speedlever
- Posts: 128
- Joined: Sat Sep 30, 2006 1:39 pm
- Location: NC, USA
Guys, I appreciate the replies.
The Raptor suggestion is more along the lines of what I was thinking.. but given this is the spcr forum, I am not sure how the Raptor fits into the quiet part of the equation.
Reading reviews of the Raptor over on the 'Egg, some say it's hot and noisy and others say it's quiet and cool.
Okaaaay. Guess I better do a search here and see what I turn up on spcr.
For $140, I can get the 74Gb Raptor mentioned above from the 'Egg. A bit pricey, but the performance may be worth it for a new build.
I'm trying to build as much cool and quiet performance as I can, within reason and trying to be cost effective too. Really, I'm talking about degrees of overkill here. But since it's a new build, I'm taking some latitude with various components and spec'ing more than I really need.
For the record:
Antec Solo with 2x92mm Nexus in the front.
Corsair HX520w PSU
Asus P5B-E
e6600 CPU
Gigabyte 7600GT SP II
Noctua NH-U12F HSF
Samsung HD321KJ 320Gb HD
Still working these items:
74 Gb Raptor for OS/programs?
Ram: DDR2-800 of some sort to be determined
internal card reader of some sort (Enermax multi-function panel?)
2x optical drives
The Raptor suggestion is more along the lines of what I was thinking.. but given this is the spcr forum, I am not sure how the Raptor fits into the quiet part of the equation.
Reading reviews of the Raptor over on the 'Egg, some say it's hot and noisy and others say it's quiet and cool.
Okaaaay. Guess I better do a search here and see what I turn up on spcr.
For $140, I can get the 74Gb Raptor mentioned above from the 'Egg. A bit pricey, but the performance may be worth it for a new build.
I'm trying to build as much cool and quiet performance as I can, within reason and trying to be cost effective too. Really, I'm talking about degrees of overkill here. But since it's a new build, I'm taking some latitude with various components and spec'ing more than I really need.
For the record:
Antec Solo with 2x92mm Nexus in the front.
Corsair HX520w PSU
Asus P5B-E
e6600 CPU
Gigabyte 7600GT SP II
Noctua NH-U12F HSF
Samsung HD321KJ 320Gb HD
Still working these items:
74 Gb Raptor for OS/programs?
Ram: DDR2-800 of some sort to be determined
internal card reader of some sort (Enermax multi-function panel?)
2x optical drives
You don't really need 74GB for your application. 36GB, or even 18GB, would be plenty, as long as you have a large IDE drive to hold you bulk data (say, 300 or 400GB).
You need the bulk drive only if you work with media files, like music or videos. Without those, disk capacity just does not fill up that fast.
(I just checked the pricing at NewEgg. If that is representative, then you may not get enough savings from the smaller Raptor to make it worthwhile. Damn, these are expensive!).
The thing with higher-RPM drives is that the vibrate a lot in comparison with the usual IDE drives. They can couple with cases which can "sing along" -- and that is a very irritating noise. So it depends a lot on the case that is used. Unless the drive mounting method and case are mentioned, I would not put too much stock in the New Egg reports. There may also be drive variability, but my guess is that it is the skill of the builder that matters more.
They are more challenging to get quiet, but it can be done. So part of what you have to decide is how hard you want to work at silencing the computer. If you want an easy computer to build, then go with a 7.2K IDE. If you want higher performance, and are willing to work at getting things quiet, then go with the Raptor.
Since SCSIs are off the table, I won't mentioned them (*ahem*).
You need the bulk drive only if you work with media files, like music or videos. Without those, disk capacity just does not fill up that fast.
(I just checked the pricing at NewEgg. If that is representative, then you may not get enough savings from the smaller Raptor to make it worthwhile. Damn, these are expensive!).
The thing with higher-RPM drives is that the vibrate a lot in comparison with the usual IDE drives. They can couple with cases which can "sing along" -- and that is a very irritating noise. So it depends a lot on the case that is used. Unless the drive mounting method and case are mentioned, I would not put too much stock in the New Egg reports. There may also be drive variability, but my guess is that it is the skill of the builder that matters more.
They are more challenging to get quiet, but it can be done. So part of what you have to decide is how hard you want to work at silencing the computer. If you want an easy computer to build, then go with a 7.2K IDE. If you want higher performance, and are willing to work at getting things quiet, then go with the Raptor.
Since SCSIs are off the table, I won't mentioned them (*ahem*).
DrJ: "So the answer is "no." You go on this whole diatribe with no experience at all."
Exactly. The point is, I don't need the experience as I don't live with the delusion that I would have the ability to notice a fraction of a second difference in computer's reactiveness. (With the exception that booting up will take always several seconds.)
Talking about your experiences, how many seconds did boot-up take before and after? What was your "before" drive? And what drive (manufacturer, model, generation, capacity) was the "after"? Did configuration remain otherwise unchanged (same hardware, bit-by-bit cloning of original OS install (i.e keeping the same degree of fragmentation)?
If you don't make the HDD the only variable, even a stop-watch can't really tell what effect the HDD swap had.
Also, do remember that while HDDs (with all interfaces) have evolved quite a bit. Compare an old ATA against newer SCSI and the ATA one is destined to be beaten. But in reality drives created for exterprise segment cost more, so drives that end up in gaming (etc) computers are already old - where as most ATA drives are bought new. Price may be about the same. Thus, in reality it's more "fair" to compare old SCSI vs new ATA than the other way. (The other way may be possible when upgrading the system to newer and more 1337.)
"Yes, that is exactly what I am claiming. When I push the drives, that is exactly what happens."
To you I copy-paste the following from SR's Raptor 150 review:
"A Word of Caution to Power Users
It is all too common for an enthusiast to believe that his or her usage pattern is closer to that of a server's rather than a desktop's. This idea arises from a variety of sources- "I multitask a lot," "I hear the hard drive grinding away," "I deal with lots of huge files," etc. The truth is, however, that even the heaviest, grinding multitasker experiences disk access patterns that are far more localized in nature than the truly random access that servers undergo. Individuals who choose a hard drive based on its prowess in IOMeter with the belief that their usage habits mimic a server simply do themselves a disservice. It is measures such as the SR Office and High-End DriveMarks that most accurately depict a non-server's response, whether it be the sheer speed experienced under intense disk access or the "snap and feel" associated with intermittent but bursty operations."
So you should understand that having multiple (for example 2 to 5 data streams) doesn't make it random access. It will cause a LOT of seeks but they are not random. They files may be located on different parts of the HDD but the seeks themselves are not random as the read command is likely to continue where it was interrupted by another read. Thus read-ahead caching allows reading sectors that have not been requested by the system (yet) and later returning then straight from the cache. Read-ahead caching requires some predictability to work but even with multiple data streams, there's predictability.
Sure, there's read-ahead caching (prefetching) on OS/software level as well, but the real benefit of hardware read-ahead is that HDD itself knows when it is beneficial to read ahead and when it is not. It all depends on latency. Access time = seek time + latency. Latency is waiting till the requested sector is under the R/W head. This time can be used effectively by using read-ahead cache: before seeking to new track, HDDs controller reads some number of sectors of the previous data stream (but only for a time that is equal or shorter than predicted latency after the seek). Only HDD can make this prediction since only HDD knows it's physical geometry and it's seek performance.
"Of course they are refurbished."
And you probably know that statistically refurbished drives are more than twice prone to failure than new ones? They have already worn out some (even though the reason for their failure wasn't likely mechanical), they have been shipped around a lot (shipping damage is one of the most common reason for HDD deaths: some are DOA, some develop over time), they might have handled without proper ESD protection, etc.
"The age of those refurbished drives varies from 10 to 30,000 hours hours. None have failed."
What the hell is that supposed to prove? Do you have thousands of them or why do you believe there's statistical significance in your anecdotal evidence?
No, you have four of them. Haha. Oldest is more than 3 years old. None failed (except one DOA). I have a dozen ATAs. Oldest one is 13 years and still working without a single bad sector... doesn't even have noisy bearings. One ATA drive I have returned due to bad sectors and even that might have passed diagnostics if it wasn't USB drive (i.e manufacturer didn't offer any diagnostics for the drive).
So what significance does your experience have?
"Because SCSI is a valid option to the question posed originally"
No it is not. SCSI = server. "OS and programs" = desktop.
speedlever: "The Raptor suggestion is more along the lines of what I was thinking.. but given this is the spcr forum, I am not sure how the Raptor fits into the quiet part of the equation."
Since Raptors have a quite many non-professional uses (as they provide better real-life non-server performance than SCSI) they have less obtrusive acoustics than SCSI drives. Seeks are of course noisy, but idle noise shouldn't be that awful.
Anyway, I have no first-hand experience with these either so I cannot tell how noisy it is. "Less noisy" might still be too noisy. When comparing to SCSI drives, the difference comes due to focus (SCSI drives are not intended to be used in rooms inhabited by people) but when comparing Raptor to 7.2krpm SATA drive the focus is the same: quiet(ish) drive. Even though Raptor has higher rpm, it has one benefit: smaller diameter platters allowing better balancing and less turbulence. (Same does apply to SCSI drives. But the fact that they are not intended to be silent does more harm.)
Exactly. The point is, I don't need the experience as I don't live with the delusion that I would have the ability to notice a fraction of a second difference in computer's reactiveness. (With the exception that booting up will take always several seconds.)
Talking about your experiences, how many seconds did boot-up take before and after? What was your "before" drive? And what drive (manufacturer, model, generation, capacity) was the "after"? Did configuration remain otherwise unchanged (same hardware, bit-by-bit cloning of original OS install (i.e keeping the same degree of fragmentation)?
If you don't make the HDD the only variable, even a stop-watch can't really tell what effect the HDD swap had.
Also, do remember that while HDDs (with all interfaces) have evolved quite a bit. Compare an old ATA against newer SCSI and the ATA one is destined to be beaten. But in reality drives created for exterprise segment cost more, so drives that end up in gaming (etc) computers are already old - where as most ATA drives are bought new. Price may be about the same. Thus, in reality it's more "fair" to compare old SCSI vs new ATA than the other way. (The other way may be possible when upgrading the system to newer and more 1337.)
"Yes, that is exactly what I am claiming. When I push the drives, that is exactly what happens."
To you I copy-paste the following from SR's Raptor 150 review:
"A Word of Caution to Power Users
It is all too common for an enthusiast to believe that his or her usage pattern is closer to that of a server's rather than a desktop's. This idea arises from a variety of sources- "I multitask a lot," "I hear the hard drive grinding away," "I deal with lots of huge files," etc. The truth is, however, that even the heaviest, grinding multitasker experiences disk access patterns that are far more localized in nature than the truly random access that servers undergo. Individuals who choose a hard drive based on its prowess in IOMeter with the belief that their usage habits mimic a server simply do themselves a disservice. It is measures such as the SR Office and High-End DriveMarks that most accurately depict a non-server's response, whether it be the sheer speed experienced under intense disk access or the "snap and feel" associated with intermittent but bursty operations."
So you should understand that having multiple (for example 2 to 5 data streams) doesn't make it random access. It will cause a LOT of seeks but they are not random. They files may be located on different parts of the HDD but the seeks themselves are not random as the read command is likely to continue where it was interrupted by another read. Thus read-ahead caching allows reading sectors that have not been requested by the system (yet) and later returning then straight from the cache. Read-ahead caching requires some predictability to work but even with multiple data streams, there's predictability.
Sure, there's read-ahead caching (prefetching) on OS/software level as well, but the real benefit of hardware read-ahead is that HDD itself knows when it is beneficial to read ahead and when it is not. It all depends on latency. Access time = seek time + latency. Latency is waiting till the requested sector is under the R/W head. This time can be used effectively by using read-ahead cache: before seeking to new track, HDDs controller reads some number of sectors of the previous data stream (but only for a time that is equal or shorter than predicted latency after the seek). Only HDD can make this prediction since only HDD knows it's physical geometry and it's seek performance.
"Of course they are refurbished."
And you probably know that statistically refurbished drives are more than twice prone to failure than new ones? They have already worn out some (even though the reason for their failure wasn't likely mechanical), they have been shipped around a lot (shipping damage is one of the most common reason for HDD deaths: some are DOA, some develop over time), they might have handled without proper ESD protection, etc.
"The age of those refurbished drives varies from 10 to 30,000 hours hours. None have failed."
What the hell is that supposed to prove? Do you have thousands of them or why do you believe there's statistical significance in your anecdotal evidence?
No, you have four of them. Haha. Oldest is more than 3 years old. None failed (except one DOA). I have a dozen ATAs. Oldest one is 13 years and still working without a single bad sector... doesn't even have noisy bearings. One ATA drive I have returned due to bad sectors and even that might have passed diagnostics if it wasn't USB drive (i.e manufacturer didn't offer any diagnostics for the drive).
So what significance does your experience have?
"Because SCSI is a valid option to the question posed originally"
No it is not. SCSI = server. "OS and programs" = desktop.
speedlever: "The Raptor suggestion is more along the lines of what I was thinking.. but given this is the spcr forum, I am not sure how the Raptor fits into the quiet part of the equation."
Since Raptors have a quite many non-professional uses (as they provide better real-life non-server performance than SCSI) they have less obtrusive acoustics than SCSI drives. Seeks are of course noisy, but idle noise shouldn't be that awful.
Anyway, I have no first-hand experience with these either so I cannot tell how noisy it is. "Less noisy" might still be too noisy. When comparing to SCSI drives, the difference comes due to focus (SCSI drives are not intended to be used in rooms inhabited by people) but when comparing Raptor to 7.2krpm SATA drive the focus is the same: quiet(ish) drive. Even though Raptor has higher rpm, it has one benefit: smaller diameter platters allowing better balancing and less turbulence. (Same does apply to SCSI drives. But the fact that they are not intended to be silent does more harm.)
I will just offer a quick comment, and then leave this topic.
My oldest computer I bought in 1980, before there were practical computer hard drives. My use for the desktop is with multithreaded applications used to search substantial databases (which I do locally) and for computational fluid mechanics, with a heavy searching part to fit theory to data. Those really are server-like loads.
Boot time is irrelevant, since I don't shut off the Unix boxes (or Unix-derived, if you want to split hairs, since I use FreeBSD primarily).
If you want to lecture me on scientific methods, put your scientific bonafides on the table, and I will consider them against my own.
My sample size of SCSIs is not significant statistically, but I do have 20 drives spread over the box I mentioned, two other development machines and two servers (with a few hot spares). The lab computers all run straight IDEs.
God, what a presumptuous idiot. He sounds like my granddaughter who thinks all truth is told to her by her fifth-grade teacher.
My oldest computer I bought in 1980, before there were practical computer hard drives. My use for the desktop is with multithreaded applications used to search substantial databases (which I do locally) and for computational fluid mechanics, with a heavy searching part to fit theory to data. Those really are server-like loads.
Boot time is irrelevant, since I don't shut off the Unix boxes (or Unix-derived, if you want to split hairs, since I use FreeBSD primarily).
If you want to lecture me on scientific methods, put your scientific bonafides on the table, and I will consider them against my own.
My sample size of SCSIs is not significant statistically, but I do have 20 drives spread over the box I mentioned, two other development machines and two servers (with a few hot spares). The lab computers all run straight IDEs.
God, what a presumptuous idiot. He sounds like my granddaughter who thinks all truth is told to her by her fifth-grade teacher.
whiic, SPCR is built on people coming together to share their experiences in a friendly and welcoming environment to hopefully further the art of quiet computing; you MUST always assume that other posters have good intentions, unless it is conclusively proven otherwise. In this thread you have been very rude to DrJ, and while it is probably too much to expect you to apologise (as any honourable person would) you should at least acknowledge that this kind of conduct is not in the spirit in which SPCR expects its users to conduct themselves. I realise your native language is not English, so part of this is probably simply misunderstanding, but I don't think this is only a language issue.I know what you're up to. You think you can discredit me just by proving that my opinion isn't based upon personal experiance
...
What the hell is that supposed to prove?
...
No, you have four of them. Haha.
...
Exactly. The point is, I don't need the experience as I don't live with the delusion
To mods, maybe this thread should be locked.
DrJ: "I will just offer a quick comment, and then leave this topic."
Yeah, because you just have to say the "final word".
"God, what a presumptuous idiot."
And since it's final, why wouldn't you make it insulting? Creative thinking. Bravo.
"He sounds like my granddaughter who thinks all truth is told to her by her fifth-grade teacher."
Does that mean you were a fifth-grade drop-out? Or was there any school at your village when you grew up during year Stick and Stone?
Computer using grandpas are quite funny. They know everything. And teachers know nothing... if they don't teach the same thing using the same methods that was used 50 years ago. (Off-topic example: the idea of an antiperspirant killing bacteria that creates the smell "of sweat" is something my father will likely never believe. He will likely think all deodorants are merely Eau de Cologne intended for use in armpits and functions only by adding a stronger smell to overpower the smell "of sweat". And to him, I'm merely an idiot who accept marketing BS without thinking.)
"My oldest computer I bought in 1980, before there were practical computer hard drives."
What has computers without HDDs have to do with this topic? Probably as much as an antiperspirants.
"My use for the desktop is with multithreaded applications used to search substantial databases (which I do locally) and for computational fluid mechanics, with a heavy searching part to fit theory to data. Those really are server-like loads."
You just said they are threaded... not random. Random is like infinite number of threads.
"Boot time is irrelevant, since I don't shut off the Unix boxes (or Unix-derived, if you want to split hairs, since I use FreeBSD primarily)."
That's true. It's pretty much irrelevant. If you were to run a search inside a database of substantial size using two different HDDs cloned so that the database and it's fragmentation would be retained on both identical copies of dataset, then you could use a stop-watch to find some hard evidence on effect of changing the HDD. But for some strange reason you have not provided any actual measurements. Why the heck should anyone listen to your claim on improved performance when you have no way of truely being sure?
"If you want to lecture me on scientific methods, put your scientific bonafides on the table, and I will consider them against my own."
Two important things:
- objective measurements (stop-watch, etc.)
- causality (keeping the HDD the only variable).
One of those is missing: everything will become complete bullsh*t.
Example on causality (or lack thereoff): I just can't change to dual-core and upgrade my OS HDD to antoher brand/interface/rpm (making a non-fragmented installation of OS), then claim that drive made 50% reduction in time required to run antivirus check.
After those two are OK, the rest is jsut tweaking for the most accurate simulation of "real-world" use scenario. Remember: because we want measurable and invariable results, we cannot measure "snappiness" or time to perform an operation which requires user intervention during the time of testing. Human is always a variable that has to be eliminated.
"My sample size of SCSIs is not significant statistically, but I do have 20 drives spread over the box I mentioned, two other development machines and two servers (with a few hot spares). The lab computers all run straight IDEs."
Still far from being relevant. And you still haven't shown any proof of ultimate SCSI performance. Always remember that comparisons are drive vs. drive. All drives having the same interface aren't the same: there's much differences between the models. And unlike you say, there's a LOT of variance between 7200rpm drives.
Yeah, because you just have to say the "final word".
"God, what a presumptuous idiot."
And since it's final, why wouldn't you make it insulting? Creative thinking. Bravo.
"He sounds like my granddaughter who thinks all truth is told to her by her fifth-grade teacher."
Does that mean you were a fifth-grade drop-out? Or was there any school at your village when you grew up during year Stick and Stone?
Computer using grandpas are quite funny. They know everything. And teachers know nothing... if they don't teach the same thing using the same methods that was used 50 years ago. (Off-topic example: the idea of an antiperspirant killing bacteria that creates the smell "of sweat" is something my father will likely never believe. He will likely think all deodorants are merely Eau de Cologne intended for use in armpits and functions only by adding a stronger smell to overpower the smell "of sweat". And to him, I'm merely an idiot who accept marketing BS without thinking.)
"My oldest computer I bought in 1980, before there were practical computer hard drives."
What has computers without HDDs have to do with this topic? Probably as much as an antiperspirants.
"My use for the desktop is with multithreaded applications used to search substantial databases (which I do locally) and for computational fluid mechanics, with a heavy searching part to fit theory to data. Those really are server-like loads."
You just said they are threaded... not random. Random is like infinite number of threads.
"Boot time is irrelevant, since I don't shut off the Unix boxes (or Unix-derived, if you want to split hairs, since I use FreeBSD primarily)."
That's true. It's pretty much irrelevant. If you were to run a search inside a database of substantial size using two different HDDs cloned so that the database and it's fragmentation would be retained on both identical copies of dataset, then you could use a stop-watch to find some hard evidence on effect of changing the HDD. But for some strange reason you have not provided any actual measurements. Why the heck should anyone listen to your claim on improved performance when you have no way of truely being sure?
"If you want to lecture me on scientific methods, put your scientific bonafides on the table, and I will consider them against my own."
Two important things:
- objective measurements (stop-watch, etc.)
- causality (keeping the HDD the only variable).
One of those is missing: everything will become complete bullsh*t.
Example on causality (or lack thereoff): I just can't change to dual-core and upgrade my OS HDD to antoher brand/interface/rpm (making a non-fragmented installation of OS), then claim that drive made 50% reduction in time required to run antivirus check.
After those two are OK, the rest is jsut tweaking for the most accurate simulation of "real-world" use scenario. Remember: because we want measurable and invariable results, we cannot measure "snappiness" or time to perform an operation which requires user intervention during the time of testing. Human is always a variable that has to be eliminated.
"My sample size of SCSIs is not significant statistically, but I do have 20 drives spread over the box I mentioned, two other development machines and two servers (with a few hot spares). The lab computers all run straight IDEs."
Still far from being relevant. And you still haven't shown any proof of ultimate SCSI performance. Always remember that comparisons are drive vs. drive. All drives having the same interface aren't the same: there's much differences between the models. And unlike you say, there's a LOT of variance between 7200rpm drives.
to jaganath:
I don't thing DrJ has any intention of intentionally misleading others. I believe he just doesn't know better. Thus, I feel that I should question his methods of proving SCSI better performing than ATA (that includes PATA and SATA) drives so that others would not accept his opinions just because they were the only.
Sure, I'm not natively English speaking, thus small-talk is not my best skill. My way of putting thing to words are sometimes rude, but only to make my opinion obvious. Though calling other people idiots is not something I consider necessary (like some other people do).
I'm going to comment of why I wrote those lines you quoted:
"I know what you're up to. You think you can discredit me just by proving that my opinion isn't based upon personal experiance"
It was quite obvious (considering the context) that when a person asks if other has personal experience it is to disqualify his beliefs. What else could it be? I though at this point that it would be better to emphasize the fact that none of my previous statements were based on first-hand experience. And I don't think they have to. A dog can learn from it's mistakes and an ape may learn from other ape's mistake... human can prevent a mistake from being made by reasoning. I don't know know if I qualify as a human but at least I'm not a dog. I don't need to experience everything myself.
"What the hell is that supposed to prove?
...
No, you have four of them. Haha."
When a person says SCSI is more reliable (which is true in general) and says he has 4 of them and none of them failed, I consider that something so irrelevant that if those sentences are put next to each other it has to be either intentional fraud or lack of thinking. Sorry, that's about as politically correct I can express it.
"Exactly. The point is, I don't need the experience as I don't live with the delusion"
You blame be because I assumed he tried to discredit me (#quote1). And the quote#4 is be replying to him. (His words: "So the answer is "no." You go on this whole diatribe with no experience at all.")
So was a wrong when I assumed his intentions were to discredit me? Though I do not know what relevance does it have whether I have first-hand experience or second-hand evidence. Evidence beats experience even if second-hand. Evidence is proof, experience is hunch. No doubt can DrJ build a SCSI chain in just a few minutes while I would most likely be googling for a tutorial or reading manufacturers documentation. I don't doubt that. But if DrJ never cared to obtain the undisputable proof to back up his hunch, I would consider StorageReview Single-User Suite 4.0 more valid way of ranking HDDs. This has nothing to do with my SCSI experience so I think it's unreasonable to ask it.
OK. I admit, I don't have much experience with SATA either (one drive, preinstalled by OEM). Not only zero experience with SCSI but SAS as well. And let's not forget the good old cluncking MFM drives. And all the drives of 50's to 80's... the washing machines.
I don't thing DrJ has any intention of intentionally misleading others. I believe he just doesn't know better. Thus, I feel that I should question his methods of proving SCSI better performing than ATA (that includes PATA and SATA) drives so that others would not accept his opinions just because they were the only.
Sure, I'm not natively English speaking, thus small-talk is not my best skill. My way of putting thing to words are sometimes rude, but only to make my opinion obvious. Though calling other people idiots is not something I consider necessary (like some other people do).
I'm going to comment of why I wrote those lines you quoted:
"I know what you're up to. You think you can discredit me just by proving that my opinion isn't based upon personal experiance"
It was quite obvious (considering the context) that when a person asks if other has personal experience it is to disqualify his beliefs. What else could it be? I though at this point that it would be better to emphasize the fact that none of my previous statements were based on first-hand experience. And I don't think they have to. A dog can learn from it's mistakes and an ape may learn from other ape's mistake... human can prevent a mistake from being made by reasoning. I don't know know if I qualify as a human but at least I'm not a dog. I don't need to experience everything myself.
"What the hell is that supposed to prove?
...
No, you have four of them. Haha."
When a person says SCSI is more reliable (which is true in general) and says he has 4 of them and none of them failed, I consider that something so irrelevant that if those sentences are put next to each other it has to be either intentional fraud or lack of thinking. Sorry, that's about as politically correct I can express it.
"Exactly. The point is, I don't need the experience as I don't live with the delusion"
You blame be because I assumed he tried to discredit me (#quote1). And the quote#4 is be replying to him. (His words: "So the answer is "no." You go on this whole diatribe with no experience at all.")
So was a wrong when I assumed his intentions were to discredit me? Though I do not know what relevance does it have whether I have first-hand experience or second-hand evidence. Evidence beats experience even if second-hand. Evidence is proof, experience is hunch. No doubt can DrJ build a SCSI chain in just a few minutes while I would most likely be googling for a tutorial or reading manufacturers documentation. I don't doubt that. But if DrJ never cared to obtain the undisputable proof to back up his hunch, I would consider StorageReview Single-User Suite 4.0 more valid way of ranking HDDs. This has nothing to do with my SCSI experience so I think it's unreasonable to ask it.
OK. I admit, I don't have much experience with SATA either (one drive, preinstalled by OEM). Not only zero experience with SCSI but SAS as well. And let's not forget the good old cluncking MFM drives. And all the drives of 50's to 80's... the washing machines.
-
Ralf Hutter
- SPCR Reviewer
- Posts: 8636
- Joined: Sat Nov 23, 2002 6:33 am
- Location: Sunny SoCal
There is a review of the Raptor on the main site and lots of people have posted experiences on them on the board. My personal opinion & experience: a Raptor adds more seek noise to your system, even though it is very quiet when idling. This means that you will hear it seeking, given your component list. Since seeking involves airborne noise, suspension won't do much for it even though it does reduce vibration noise obviously. I think that enclosures might help a bit, but I have no experience with them. This noise penalty gives you a relatively small gain in noticable performance under normal use, which IMO can hardly be called cost effective.speedlever wrote:The Raptor suggestion is more along the lines of what I was thinking.. but given this is the spcr forum, I am not sure how the Raptor fits into the quiet part of the equation.
(snip)
A bit pricey, but the performance may be worth it for a new build. I'm trying to build as much cool and quiet performance as I can, within reason and trying to be cost effective too.
(snip)
(Enermax multi-function panel?)
Having said that, I have 2 Raptors in my main machine because I want that extra speed margin and because I could justify the cost. I like their speediness and it is definately more noticable than expensive overclockable memory (for my usage anyway).
My experience with Enermax MF panels is lousy (ugly, bad quality), but obviously YMMV.
-
speedlever
- Posts: 128
- Joined: Sat Sep 30, 2006 1:39 pm
- Location: NC, USA