Western Digital uppdaterar sina Raptorer|Arkiverade nyheter|Forum|Nordichardware

oj ! tänk om de blir lika snabba som raptor 150 ?? det vore riktigt jäkla bra! 🙂

Gamblor wrote: Trevligt, de nya diskarna blir tekniskt bättre på nästan alla punkter. Har de smyguppdaterat priserna också?
150GB-versionen var vidrigt dyr.

150Gb raptorn har alltid varit dyr. >_<
Skillnaden mellan 8mb cache och 16mb cache är inte så stor enligt de recensioner jag har tittat på (på raptors).

Ondjultomte wrote: dessa "nya modeller" har ju knappt kommit ut på marknaden ännu ... fnns det verkligen recenssion ute?

öööö, Jo, 150gb modellen finns ju redan ute, och den är inte s värst mycket bättre än existerande gammla raptors.

Det jag skulle vilja se är 15krpm och dubbla (!) läshuvuden.

Ondjultomte wrote: Givetvis menade jag 36 och 74GB modellerna som nu byggs från samma grund som 150GB modellen

Skillnaden mellan dom och 150gb är ju bara plattorna, så prestandan ligger nog på ungefär samma som på 150gb. Nog för att dom beroende på hur platorna ser ut kan prestera sämmre.

Fast förrut har ju 36GB modellen, haft vanliga kullager, undrar om de har ändrat det nu med?
Vore bra isf, så den blir lika tyst som de andra.

hypno toad wrote: [quote=Ondjultomte]dessa "nya modeller" har ju knappt kommit ut på marknaden ännu ... fnns det verkligen recenssion ute?

öööö, Jo, 150gb modellen finns ju redan ute, och den är inte s värst mycket bättre än existerande gammla raptors.

Det jag skulle vilja se är 15krpm och dubbla (!) läshuvuden.

Why has no one made a drive with internal RAID 0?

nternal RAID 0 is technologically possible, but economically impractical. The drive industry is a high-volume, low-margin one (especially in the IDE realm, not so much for the SCSI realm). The only way it makes sense if drive makers lower volume is to make up for it with higher profit margins (and no OEM or other customer wants that!). For better or for worse, hard drives are now commodities. For most people, a drive is a drive is a drive.

From a consumer standpoint, folks of course would want to only have to buy a single drive instead of two, but why would a HDD firm want to sell only one drive, with the SAME profit, versus two drives with twice the profit (same % margins, I'm assuming)? I'm sure the popularity of RAID 0 amongst enthusiasts has helped increase the volumes of drives sold, and higher volumes help drive down costs.

HDD firms have to be fiscally conservative; bad decisions based on overly aggressive expectations have sunk many a drive firm (Terastor and Syquest, anyone?), not to mention many a dot-com, I suspect.

Now, from a technical standpoint, I can think of a few concerns (and I am sure I am overlooking some):

1. Head offset on a single actuator.
2. Read channel bandwidth limitations.
3. Servo complexity.
4. Interface bandwidth limitations.

Concern 1: Head offset on a single actuator.

Due to tolerances, each head is slightly offset from any other head, even though all are attached to the same actuator. Track 1000 on one surface is not the exact same radial position as Track 1000 on another surface. Thus, it is almost impossible to have all the heads line up on the same track at the same time. This head offset is exacerbated as the temperature changes, since not all heads will "shift" the exact same amount, nor even in the same direction.

One way around this is to use multiple actuators, such as seen with the Seagate Hawk drive (from several years ago). Another is to keep one actuator for moving the entire headstack, but have a smaller, secondary stage for each head. This would compensate for any head offset.

The first option takes up more space than would easily fit in the standard 3.5" form factor, at least without shrinking to smaller disks. The higher complexity and cost are likely what prevented this concept from catching on with other HDD makers.

The second option gets around the space constraints, but none of the current concepts/designs for secondary stages are cheap (e.g. piezoelectrics, shape memory alloys, microactuators). As such, until the price comes down, these are unlikely to make it to volume production.

Concern 2: Read channel bandwidth limitations.

Current channel chips, AFAIK, are designed to handle one stream of data from one head at a time. Since only one head is "reading" at any point in time anyway, this is not an issue. However, if multiple heads are reading concurrently, the channel chip would need the bandwidth to handle this.

Concern 3: Servo complexity.

WIth only one head "active" at any point in time, a servo system only needs to keep that head track-following. With multiple heads track-following, the servo system becomes more complex. This complexity takes up microprocessor time. A more powerful microprocessor (and presumably, pricier) is needed.

Concern 4: Interface bandwidth limitations.

This one is, admittedly, more of a chicken-and-egg dilemma. Why bother with an interface that has more bandwidth than current drives can offer (e.g. SATA 150 and current IDE drives)? Similarly, why make a drive that will regularly saturate the interface?

With SATA 150 (sometimes called SATA 1.0), even its higher bandwidth would get saturated if multiple heads are all transferring at the same time. For example (greatly simplified), let's assume the current crop of IDE drives can sustain 50 MB/sec transfer rate. That's using one head. If 4 heads are used, you should easily surpass 150 MB/sec transfer rate.

summasumarum, det är helt enkelt effektivare med 2 diskar i RAID0 än "inbyggd" raid0 med fler "läshuvuden" , det är ju fler läsarmar du egentligen menar.

beta wrote: Men 2 diskar i raid0 med dubbla läshuvuden då. 😀

Läste du inte? dubbla läsarmar är dyrare och svårare än 2 st hdds ..

det är effektivare att dubblera antalet hårddiskar.

och ni so är så "kåta" på raid0 borde läsa min tråd på swec
http://www.sweclockers.com/for.....did=589992

däri finns bla denna text:

SingleDrive Vs Raid0:
http://faq.storagereview.com/tiki-i...gleDriveVsRaid0

"
What about performance? This, we suspect, is the primary reason why so many users doggedly pursue the RAID 0 "holy grail." This inevitably leads to dissapointment by those that notice little or no performance gain.

Theory states that RAID 0 increases the sequential transfer rate, but how much does this really effect performance in contemporary desktop machines? As often indicated in StorageReview's forums, the answer is: Not much. STR simply does not significantly impact performance of typical desktop applications.
"

alla raptordiskar har sedan läbge fluidbeerings.