New hdd, speed improvement?
- Thread starter wm51
- Start date
Let's put it this way - if the speed doesn't scale up with platter/head count (platters being the same density), that means that the disk is bottlenecking on the drive electronics. Not all manufacturers' electronics can keep up with the mechanics, sadly.
Soldato
- Joined
- 2 Dec 2006
- Posts
- 8,204
so basicaly your saying you cant prove you self because the manufacturers are inadiquit?
So what manufacturers do show this sort of scaling?
Your thesis would only be true if all data was split into 5 parts on a 5 platter HD and one fifth placed on each, that way each head could work on 1 part of the file and it would be close to 5 times the speed - that's not the case at all.
what he said.
The drive reads sequentially down through the platters before moving inward towards the slow part of the drive. As a result a 640Gb two platter drive will be faster *at the 320Gb point* than a 320Gb single platter drive at the 320Gb point.
However, and this is the crucial thing here, the maximum, minimum and average transfer rates will be the same for both drives. Check the benchmark thread for evidence of that.
That's not how it works these days. The order of geometry tuple, major to minor, is (cylinder, sector, head). That means that if sequential sectors are read, for each (cylinder, sector) heads are switched in turn. This is basically a free optimization of drive's performance that also nullifies the need to have separately actuated heads for each platter (which would be prohibitively difficult and expensive, and pointless if you just do the geometry in the order mentioned).
As to why modern drives don't regularly deliver on the extra performance that should be available from additional heads/platters, I don't know. My best guess is that the drive electronics can't handle it and improving sustained transfers is not as much of an issue overall performance-wise compared to seek times. The fact that Samsung seem to quote 175MB/s sustained transfers vs. Seagate's 120MB/s with same RPM while Seagate has more platters seems to corroborate this - the bottleneck isn't the media but the drive logic. When you think about it, decoding over 100MB/s of data is no trivial task. I haven't done any tests on disks in the past 7-8 years or so (damn, I'm feeling old now...), but the disks I tested back then showed an improvement along the lines of what I described.
*shrug*
I guess some of the ancient wisdom has been deprecated. It also explains why disks haven't kept pace in terms of speed relative to their size - we'd need a 100W TDP signal processor on disk to keep up with the media transfer speeds, and having a 3.5in disk with a 6in heatsink isn't deemed practical.
what disks are there your working on pal? i still have some diamondmax 740dx drives and even a fer 1gb samsung and a 450mb quantum fireball lying around. i can run benchmarks on those and prove that your talking out your rear end.
the fact that you even believe manufacturers specifications
shows that you dont know jack pal. fastest sata drives are topping out at the 120mb/s mark with 334gb per platter and 7200rpm spin speed. even the 10k rpm valociraptor's peak is around 120mb/s.
but for a laugh could you share you views on SSD drives with us please?
the fact that you even believe manufacturers specifications
shows that you dont know jack pal. fastest sata drives are topping out at the 120mb/s mark with 334gb per platter and 7200rpm spin speed. even the 10k rpm valociraptor's peak is around 120mb/s.
but for a laugh could you share you views on SSD drives with us please?
I'm going by the manufacturers published figures, true, but unless you learn to be polite I probably shouldn't acknowledge your post as worthy of a response.
The 1TB 7200.11 Seagates I have in a server I built 2 days ago quite happily push about 110MB/s sustained, at the front of the disk. You are mistaking a drive's sustained speed with access time. Velociraptors have a faster access time due to higher spindle speed, but if you are saying they can't handle a faster sustained rate than the 7200rpm Seagates then that's down to WD's electronics.
How's about this: my 500GB Samsung 501s can handle about 50MB/s. The 1TB Seagates are pushing double that, as mentioned.
Just because you have self-taught how to run a benchmark clearly doesn't mean you know jack about hard disk design. Some of us, OTOH, went to school.
i think your trying to change the subject here since access times havent been mentioned by me at all. you stated that the more platters a drive has the faster it is. which is clearly false since its been proven everywhere that drives of a same generation with the same platter density but with different numbers of platters have the same transfer rate from start to end. thats they the transfer rate of a 320gb single platter samsung f1 starts at around 120mb/s which is the same as the 3 platter 1tb version of the samsung f1 and both have a similar end transfer rate too. have a look at the reviews pal.
going by the claims you have made in this thread i think everyone knows who went to school or not.
going by the claims you have made in this thread i think everyone knows who went to school or not.
oh and let me be the first one to post some personal proof here:
2 drives tested, both seagate 7200.10 series IDE, one is a 500gb drive using 3 x 188gb platters, other drive is a 750gb using 4 x 188gb platters:
as you can see the performance is the same, the 750gb drive has the extra platter but its not any faster than the 3 platter drive.
gordon your lucky this is a family forum and heavily moderated, you posted your crap anywhere else and you would get blasted big time.
2 drives tested, both seagate 7200.10 series IDE, one is a 500gb drive using 3 x 188gb platters, other drive is a 750gb using 4 x 188gb platters:
as you can see the performance is the same, the 750gb drive has the extra platter but its not any faster than the 3 platter drive.
gordon your lucky this is a family forum and heavily moderated, you posted your crap anywhere else and you would get blasted big time.
gordon speaks good BS though, very convincing. Maybe he should write the tech babble for startrek or something. He must have got his info from www.howstuffdoesntwork.com.
High praise.
Indeed, I see that ancient wisdom doesn't apply to modern disks. Having spoken today to someone who writes HDD firmware for a living, I'm told that these days the tracks are so dense that head-to-head switch time (1-1.5ms) actually exceeds track-to-next-track seek time (0.5-0.8ms), which means that modern LBA disks write cylinders in consecutive groups before switching to the next head down.
So, I lose this round. Next time, Gadget.
this is why the internet should also have upper age limits.
Yeah, yeah. I'll get my zimmerframe.
("During the war ..."
)don't take this the wrong way..
maybe edit your other posts, so people don't read into what's been said (put a disclaimer in them )
like when that guy jokingly said "lite-on's website says, you should only burn 5 discs a day max, or the chip will burn out" which people seemed to really believe for years!
maybe edit your other posts, so people don't read into what's been said (put a disclaimer in them
)like when that guy jokingly said "lite-on's website says, you should only burn 5 discs a day max, or the chip will burn out" which people seemed to really believe for years!
