Why instead of raising the spindle speeds.....

[sandman001]

Why not just give the hard-drivews like 256mb of onboard cache?

seems like it would speed them up a good bit......

 

[JDXNC]

Well if the drive can't read the data off the drive fast enough to use that amout of cache it serves to purpous. Thats my guess, sounds resonable to me.

 

[Cjwinnit]

Why not just give the hard-drivews like 256mb of onboard cache?

seems like it would speed them up a good bit......

You would still have to access the data you need from the platters, it won't speed up that. If you lose power to the computer then you might lose the cached info, so you want it written to the platters fairly quickly.

 

[sandman001]

I didn't think of the power issue. It was just an idea..........

 

[Cjwinnit]

I didn't think of the power issue. It was just an idea..........

Some systems use large caches, look at good RAID cards that have sticks of RAM..(often 64Mb EEC SDRAM i think)

 

[JDXNC]

Some systems use large caches, look at good RAID cards that have sticks of RAM..(often 64Mb EEC SDRAM i think)

I have seen a SCSI card with 512MB of ram on it.

 

[Cjwinnit]

I have seen a SCSI card with 512MB of ram on it.

Niiice

 

[Caffinehog]

Well, that cache can only be filled with what the computer THINKS it will need next. If it needs anything else, it will just have to go to another part of the drive and read without using the cache. That cache can also be emptied in about three seconds, so any longer read will lose that perk.

 

[larva]

Cache sizes generally increase as ram prices continue to decline. This is why many modern drives have 8MB of cache these days, where 32-128K was the norm when I started building PCs. But even so, don't expect a 256MB buffer when a 256MB stick of ram still costs 35 bucks. They sell complete drives for 60-100 dollars these days, so 256MB of cache is not economically feasible at this time.

Also remember how caches work. When read requests are made, the cache is checked first for the requested data, and if it does not exist in the cache it is read into the cache from the drive. Although the incredibly clever cache management strategies used manage to get the desired data in the cache before it is needed, there is no way to insure that it is always there. The time spent searching the cache for data that is not there can add up, so although we tend to think bigger would always be better this simply isn't the case.

Also bear in mind the cache is there to band-aid the physical speed of the drive mechanism. Although it does a very good job of isolating us from the (relative) slowness of the drive mechanism, if we can build a faster drive it behooves us to do so. Although caching alleviates much of the symptom, anytime technology allows to minimize the cause of the problem we have to take advantage.

 

[nikhsub1]

Not to mention that when you increase rotational speed, what you really gain is SEEK time. The new Raptors at 10K have a seek time of 5.2ms where the fastest 7200 RPM drive has a seek time of about 8ms. Cache will never be able to help in that department.