[bradleykavin]

what is a WD Raptor 10000RPM 74GB...i would think that having an 80 gigabyte harddrive wouldnt be that great, as it is only 80 gigabytes. but i see the 100000 rpm and im thinking there has to be some advantage to having this...could someone explain it to me..?

[TETENAL]

I guess the advantage of a 10.000 RPM disk drive over a 7.200 RPM one is the higher speed.

[TiDual]

Originally posted by TETENAL:
I guess the advantage of a 10.000 RPM disk drive over a 7.200 RPM one is the higher speed.

What a helpful reply. I assume the question was "to what extent does the internal speed of the HD translate into an advantage to the user". This is an empirical question best answered by someone with experience with both kinds of drives. Anyone out there?

[mikellanes]

for the price I would get 2 80GB 7200RPM drives, RAID 0 those puppy's and have some real speed.

Then there is the whole data loss if the raid dies, but I would argue to get 2 good (like seagate) drives. The Raptor is a fairly new drive so them failing probably has a higher chance then even 2 drives...

No to diss in the raptor though the benchmarks are pretty good fir a single drive solution.

I guess it all depends what you are using it for... will this just be your main system drive? If so I have seen some impressive benchmarks from 200GB WD's they have higher density then the raptor so almost catch up in speed, probably around the same price.

I would say a higher density raptor would be killer once they are established... I think that is a fight against the 10K speeds as the higher the density the more precise things need to be, i could be wrong.

Hope some of this helped? I guess it just depends what you are planning on using this for.

[mikellanes]

Looking at benchmarks the gain over a fast (Maxtor DiamondMax Plus 9) drive is only around 6%... from ars..

"Our tests have also shown us that the 10,000RPM Raptor can offer a noticeable, but not dramatic, performance improvement over the current generation 7200RPM 8MB cache drives. While the performance improvement is there, it's not as significant as the synthetic tests would have you believe."

I would say a Maxtor DiamondMax Plus 9, in the highest capacity you can afford would be the best bet.

Raptor is $177

DimondMax 10 250GB is only $168

(both prices drive only from newegg.com)

[aaanorton]

I use these (and the smaller 36 GB models) as single partition Photoshop scratch drives. This works very well for me. I recently did a Helios LAN test to one of them and was quite impressed with the results.

[UnixMac]

I use both a 74gb raptor and a maxtor 250gb, with OS X and photoshop (and other apps, like FCP) on the raptor and my large files and .jpgs on the maxtor and the benefit of SATA it's like multithreaded software.. you can lanch photoshop and load the large .jpg at a much faster rate than if you did both of the same drive.

[reader50]

Ok, I'll take a stab at a solid answer.

Hard drive performance is controlled by several factors:

Platters
These are the number of magnetic disks inside a hard drive. Usually, data is recorded on both sides of each platter. More platters give higher capacity (more GBs), but tend to increase noise and manufacturing cost. Since platters are moving parts, more platters also tends to decrease reliability. Within a hard drive family, the lowest capacity starts out with one platter and goes up across the line to the maximum number of platters. Manufacturers have gone both up and down with the maximum number of platters in each successive drive family. I've seen numbers from 3 to 5.

Heads
The number of read/write heads per platter. More of these moves data faster on/off the drive. They are moving parts, so more also decreases reliability and increases noise. More heads equals more mass that has to be moved during track-to-track seeks, so increasing the heads either increases the track-to-track seek time, or increases the power/noise of track-to-track seeks (if more power is used to overcome the greater inertia). The number of heads per platter seems to have remained constant for some time. I don't recall hearing of them going up or down during the last several years.

Recording Density
This goes up with each HD generation. It is the drive capacity per platter, which you can get from the manufacturer's specs. Increasing the density increases sequential read and write speeds, since more data moves beneath the heads per second at a given rotational speed. To a lesser extent, it also improves track-to-track seek time, since individual tracks are closer together - but it also increases the number of tracks. If your data is close together and not heavily fragmented, it will improve the average track-to-track seek time for your data - this is the usual case. Recording density does not affect sector-to-sector seek time.

Rotational Speed
This improves sequential read/write speed, since more data moves beneath the heads per second at the higher rotational speed. It also improves sector-to-sector seek time. It will decrease bearing lifespan - or require more expensive bearing construction to compensate. It increases noise and the drive's required operating power. More expensive drive construction may compensate for the noise, but you are stuck with the increased power demand unless you decrease the number of platters. Lowering the mass in motion can compensate for having to spin the mass at a higher speed, but fewer platters lowers your drive's capacity.

Buffer Size
The RAM cache on the hard drive. Usually 2MB, often 8MB. Drives with 16MB caches are appearing, such as the Maxtor DiamondMax 10 series. Larger buffer size improves apparent read/write speed for small data transfers since the data is read/written to the drive's cache, rather then directly to disk. The cache usually has the entire track that you last requested a block from, possibly nearby tracks, and probably some blocks you requested in the past. If you request a block near the last block requested, it is likely that you can get it from the drives's cache, and skip the seek time. Large sequential data transfers such as video read/writes make the cache size irrelevant, since a 1GB+ data transfer will smash even a 16MB cache. In the event of a power failure, modern drives save enough power to finish writing all data from cache to disk. Then they auto-park on unused locations of the platters. Very old hard drives didn't do this, they could lose data and crash during a power failure.

Interface Speed
This is ATA-33, ATA-66, ATA-100, ATA-133, SATA-1 (1500), SATA-2 (3000), SCSI-160, SCSI-320, FireWire-400, FireWire-800, etc. The numbers are maximum transfer speeds over the data cable, in megabytes per second (except the SATA & FireWire numbers, which are in megabits per second). Control signals and error-check packets have to be sent over the cable also, taking up some of the bandwidth. ATA has a control overhead of about 1/4 of the rated capacity - ie, ATA-66 typically gives ~50MB/sec transfer speeds. I'd assume the same ratio is true for SATA, but I've not heard that for sure. SCSI has a much lower overhead, the drives are smarter and handle themselves, so you approach the rated interface speed. FireWire behaves like SCSI with minimal control overhead. FireWire 400 will give ~ATA-66 speeds to external drives, FireWire 800 will give ~ATA-133 speeds, and should be somewhat slower than SATA-1 speeds. Interface speed matters for small writes (and small reads to a lesser extent) because you get the interface speed to the HD's cache. For larger data transfers, higher interface speed matters only if the drive can keep up. The best ATA drives can manage 60-75MB/sec to the beginning of the drive, and fall off to 1/2 speed by the end of the drive. RAID arrays put multiple drives together to increase speeds, a RAID array can benefit much more from a faster interface than a single drive does.

Native Command Queuing
This reorders block read/write commands into optimum order for the disk's operation. It improves multiple parallel read/writes quite a bit, but has minimal effect on single data transfers. This makes this feature more suited to server applications and heavy multitaskers than the average home user. Two or more file operations have to be going on at the same time for NCQ to be very helpful, the average home user has that happen only when they duplicate files on disk. SCSI drives have had native command queuing for some time, it is only just appearing on ATA drives. The Raptor has it*, so does the DiamondMax 10. The interface chipset on the motherboard or PCI card has to support command queuing for it to work.

* The Raptor actually has Tagged Command Queuing (TCQ), which is a bit less effective than Native Command Queuing (NCQ). TCQ reorders block access commands based solely on block number, NCQ takes the platter's current rotational position into account as well as block number.
-----------------------------------------------
It looks to me like Western Digital took a SCSI 10,000 RPM drive, and replaced the control board with a SATA board. That explains the metal frame which looks like a SCSI drive, where ATA drives normally have a plastic frame. It also explains the 37/74 GB sizes, which are SCSI size increments rather than the usual ATA sizes for that range: 40/80 GB.

The improvements you get from higher rotational speed are a lot like those you get from higher recording density. However, higher density gives you larger drive capacity as well as less noise and less power consumption, vs getting the same performance boost from rotational speed.

The Raptors were the speed champs when they were released, as well as having the highest prices and only so-so capacity. They are overdue for a density/capacity boost, since normal 7200-RPM ATA drives have caught up and passed the Raptors in raw speed, and have a heck of a lot more capacity

Finally, Raptors are only available in SATA. If you have a Mac with parallel ATA, the price of a SATA card is about the same as a mid-range drive. So for the price of adding a SATA drive + SATA card, you could get dual ATA drives. If you have a G5 with native SATA, Raptors aren't nearly as bad a deal. You still get higher capacity for the price with normal SATA drives, but you don't get penalized by the price of the interface card. Note, SATA interface cards can be gotten for a lot less on the PC ($22) than on the Mac ($58), so the interface card penalty may be small enough to ignore on a PC.

-----------------------------------------------
The standard disclaimer: I don't work in the HD industry. If you do and something here is obviously wrong, post. Interested geeks want to know the right answers.

Edit: the Raptor is indeed still faster than the Diamondmax 10 series in most tests. Barefeats apparently updated their graphs after I first read the review, and the storagereview.com graphs prove it. However, the performance gap is much smaller than it was when the Raptor was first released, it really is time for the Raptor's size to go up.

[tooki]

Originally posted by mikellanes:
for the price I would get 2 80GB 7200RPM drives, RAID 0 those puppy's and have some real speed.

The goals of increasing rotational speed are:
1. reducing rotational latency (the amount of time it takes for the desired sector to spin under the heads) -- "snappiness"
2. increase the data transfer rate. -- shorter read/write times

RAID 0 (which is disastrous for reliability) does increase transfer rate, but cannot improve rotational latency (and can in fact increase latency, which is bad).

A drive with a higher rotational speed feels "snappier" than a slower drive.

tooki

[OreoCookie]

Originally posted by bradleykavin:
what is a WD Raptor 10000RPM 74GB...i would think that having an 80 gigabyte harddrive wouldnt be that great, as it is only 80 gigabytes. but i see the 100000 rpm and im thinking there has to be some advantage to having this...could someone explain it to me..?

For the average user, it's little to none. (Rather: too little to be worth it.) The main advantages are (i) SCSI-like robustness of the drive and (ii) faster seek times. So if you work with many little files and need the extra robustness, then you might consider it.

Larger drives have a higher data density, so the raw transfer rates are not 25 % higher or something. I recently bought a Hitachi 250 GB@7200 for �130. Really a nice fast drive.

[Macanoid]

Anyone here doing audio with one of those 10.000 rpm drives. I'm told that for audio those drives are really worth the money.

[tooki]

Audio people love to claim they have higher drive needs than anyone else. This was only true before video began to be edited on computers, because video's needs blow audio's needs out of the water.

Realistically, HD video is by far the most demanding.

Audio certainly enjoys fast drives, but any drive on the market is fast enough for most audio work.

tooki

[OreoCookie]

Originally posted by tooki:
Audio people love to claim they have higher drive needs than anyone else. This was only true before video began to be edited on computers, because video's needs blow audio's needs out of the water.

Realistically, HD video is by far the most demanding.

Audio certainly enjoys fast drives, but any drive on the market is fast enough for most audio work.

tooki

Well, I think originally it was asked about `average user' performance benefit. I don't think this includes HD video. Even my iPod was fast enough for simple FCP DV editing.

So I think the issue is pretty much settled for `average applications'.

[UnixMac]

Originally posted by reader50:

The Raptors were the speed champs when they were released, as well as having the highest prices and only so-so capacity. They are overdue for a density/capacity boost, since normal 7200-RPM ATA drives have caught up and passed the Raptors in raw speed, and have a heck of a lot more capacity.

Not sure I agree with this statement, however the rest of your info was quite nice and informative.

The barefeats (which is not the standard for HD benchmarks btw) info shows the raptor as still faster than the 7200rpm Maxtor..

There are several HD benchmark sites that show the raptor is still the faster (data transfer) drive out there including the 15000RPM Chetas. I seem to remeber it being 70mbs+ sustained I believe.

[OreoCookie]

Originally posted by UnixMac:
Not sure I agree with this statement, however the rest of your info was quite nice and informative.

The barefeats (which is not the standard for HD benchmarks btw) info shows the raptor as still faster than the 7200rpm Maxtor..

There are several HD benchmark sites that show the raptor is still the faster (data transfer) drive out there including the 15000RPM Chetas. I seem to remeber it being 70mbs+ sustained I believe.

Benchmarks are always a difficult story. How synthetic is it, what importance is put on which part of the results, etc.

(BTW the Cheetahs have smaller platters so that they have faster seek times.)

[SalBaker]

Originally posted by tooki:
Audio people love to claim they have higher drive needs than anyone else. This was only true before video began to be edited on computers, because video's needs blow audio's needs out of the water.

Realistically, HD video is by far the most demanding.

Audio certainly enjoys fast drives, but any drive on the market is fast enough for most audio work.

tooki

Actually it is MIDI sample streaming that benefits from the 10K drives. Sample streaming is much more demanding than HD video because of the number of simultaneous streams of data coming off the drive at one time.

For example, creating a symphonic score on a G5: The high-end sample libraries are comprised of all instruments, and grouping of instruments, in an orchestra. Each note of, say a french horn, is recorded in stereo at 16 or 24-bit with no looping. This is a real stereo recording of a musician playing the horn, one note at a time, and playing the notes in multiple playing styles. As long as you hold down the MIDI key on the keyboard (or through a sequencer) the note of that "real" french horn sustains until the original player runs out of wind. When you release the key, another stereo sample of that horn's actual reverb in a hall is fired giving a very real sound to the instrument.

Now add 3 more more french horns (3 more stereo streams and 3 more reverb release tails). Now add the rest of the orchestra one instrument, or small group of instruments at a time. Now add the ability to cross-fade from a mellow horn note to a loud brassy note, which adds expression to the instrument--this requires even more streams per note.

One can easily end up with a composition made with over a hundred instruments or groups, all hitting notes in rapid secession, with each stereo note (some instrument samples are actually recorded in 5.1 surround) streaming from the drive in real-time. Only the first fraction of a second of each sample is played-back from RAM, the disk then catches up and streams the rest.

Surprisingly, 7200RPM drives work pretty well up to a point, but more demanding work requires the extra edge and quick access time the 10k drives offer. Too bad the 10K drives are so small.

[Macanoid]

Originally posted by SalBaker:
Actually it is MIDI sample streaming that benefits from the 10K drives. Sample streaming is much more demanding than HD video because of the number of simultaneous streams of data coming off the drive at one time. (edit)

Would the same be true for Virtual Instruments??

[BurpetheadX]

I have the 36 GB 10,000 RPM version as my boot drive, and my home folder on my other 7,200 RPM.

Boy, does it fly. The 10,000 Raptors are about 33% faster than the average 7,200 RPM drive, and OS X eats it up.

Even for basic use, you will love one of these drives. It runs cooler and more quiet than my 7,200 250 Gig!!

[SalBaker]

Originally posted by Macanoid:
Would the same be true for Virtual Instruments??

Yes, but it's the total number of voices that make the difference. Most virtual instruments are made from streaming samples and are neatly wrapped up in a proprietary player. A single virtual instrument or two should work fine with 7200RPM drives. I have been able to sustain 60 streaming stereo voices (120 individual streams) on my 7200RPM 250gb drive before choking the system. It's only when you start building large sample orchestras before the 10K drives become an advantage. Many composers use 2 or 3 computers to handle the load when using 24-bit 5.1 surround orchestra samples.

Of course it also depends on how full your drives are. A friend of mine uses a 10k RPM drive for his samples. When he filled it up to about 70% full, it started running slower than his 250gb 7200RPM which was only 30% full. So the 10K drives become even smaller if you can't fill them up much more than 50% before they slow down. Although this probably wouldn't be apparent in less demanding applications.

[UnixMac]

Originally posted by BurpetheadX:
I have the 36 GB 10,000 RPM version as my boot drive, and my home folder on my other 7,200 RPM.

Boy, does it fly. The 10,000 Raptors are about 33% faster than the average 7,200 RPM drive, and OS X eats it up.

Even for basic use, you will love one of these drives. It runs cooler and more quiet than my 7,200 250 Gig!!

All so true... so true.

This this runs like a G5 3.0 would have on one drive!

[Cadaver]

Originally posted by BurpetheadX:
I have the 36 GB 10,000 RPM version as my boot drive, and my home folder on my other 7,200 RPM.

Boy, does it fly. The 10,000 Raptors are about 33% faster than the average 7,200 RPM drive, and OS X eats it up.

Even for basic use, you will love one of these drives. It runs cooler and more quiet than my 7,200 250 Gig!!

I boot my G5 off a 74GB Raptor, and yes, it flies. Boots in a flash and apps launch instantly. Much faster than the stock 160GB.
While the newer 300GB and 400GB drives may be nearly as fast, they didn't exist when I bought the Raptor. Even so, fast is fast.

[macaddict0001]

has anyone raid 0'd two raptors?

[UnixMac]

Originally posted by macaddict0001:
has anyone raid 0'd two raptors?

Sailfish over at the www.G5support.com forum as and it apparently smokes!

[macaddict0001]

sweeet.