backtop


Print 51 comment(s) - last by afkrotch.. on May 30 at 12:46 AM

Western Digital gets to 334GB per plater about a year after Samsung

When hard drive manufactures move to platters with higher storage densities, it is a good thing all around -- the higher the storage density for the platters, the fewer platters are needed to reach the same capacity compared to lower density platters.

That means less power is needed and less moving parts are required to make the drive operate. Impress is reporting that Western Digital has quietly updated its Caviar GP line of drives in Asia with a platter density of 334GB in the 1TB capacity drive. Other features of the drive remain the same with a 16MB cache.

Western Digital isn’t alone in moving to 334GB per platter. This week Samsung announced a new 1TB EcoGreen hard drive aimed at surveillance and audio/video applications that use platters of 334GB in size.

Samsung claims that the EcoGreen F1 drive provides a 15% power savings compared to other low power 1TB drives and a full 50% power savings compared to traditional 1TB drives often spinning at 7200RPM. Samsung’s EcoGreen F1 spins at 5400 RPM and uses the 3Gbps SATA interface. Availability for the Samsung drive is set for Q2 2008 at $199.

In January 2008 Western Digital introduced the single platter 320GB HDD, which was its highest density platter. Samsung was first to ship a 334GB per platter 1TB drive in June of 2007.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

4.3% is news worthy
By LostInLine on 5/23/2008 12:50:33 PM , Rating: 3
how much performance can you possibly get moving from 320 to 334?

Did i miss something?




RE: 4.3% is news worthy
By jonmcc33 on 5/23/2008 12:55:28 PM , Rating: 5
Don't you know that benchmarks mean everything? That's why everyone overclocks their CPU and video card. Even if you only get an extra 30MHz that's an extra 20 3DMark points and that makes your computer faster! 3DMark and PCMark rules! /sarcasm


RE: 4.3% is news worthy
By ScottLuebke on 5/24/2008 2:57:51 AM , Rating: 2
Never heard anything more true. Preach it!


RE: 4.3% is news worthy
By masher2 (blog) on 5/23/2008 1:02:25 PM , Rating: 5
WD went to 320TB a couple months back for some other drives, but afaik the 1TB drive was still using 250GB platters.


RE: 4.3% is news worthy
By LostInLine on 5/23/2008 4:26:04 PM , Rating: 2
ok, it was my understanding (and i could be wrong) that WD was using the 320G platters but limited the drive to 1TB. Using the higher aerial density platter mostly offset the slower 5400rpm.


RE: 4.3% is news worthy
By KernD on 5/23/2008 6:44:43 PM , Rating: 2
It's not running at 5400rpm, it's a 7200 like the other WD regular drives. The 5400rpm speed stated in the article is about a Samsung drive.


RE: 4.3% is news worthy
By yonzie on 5/24/2008 1:42:56 PM , Rating: 2
Oh, but it is. They just don't want to admit it since "noone" would buy it due to it being inferior (lower power consumption notwithstanding). A bit like the Ghz myth.


RE: 4.3% is news worthy
By KernD on 5/24/08, Rating: 0
RE: 4.3% is news worthy
By KernD on 5/24/2008 2:23:07 PM , Rating: 1
The GP line is 5400-7200 RPM but all that means is it varies between both, to reduce power, heat and noise. But if you look at the review at AnandTech, with the 4 plater version I think, it still does a good job. When the demand is there it will run at 7200.


RE: 4.3% is news worthy
By Lezmaka on 5/24/2008 2:31:23 PM , Rating: 4
http://www.storagereview.com/1000.sr?page=0%2C1
quote:
For those that must know, WD admits "sub-6000 RPM operation" for the 1-TB Caviar GP


http://www.storagereview.com/1000.sr?page=0%2C2
quote:
The GP turns in a measured access time of 15.0 ms, a score that lags the 7200-RPM WD7500AAKS by a significant margin. The WD7500AAKS's measured seek time when accounting for 4.2 ms of 7200 RPM latency is 9.5 ms....

Assuming the GP also shares such a seek time, that leaves us with 15 ms [measured access time] minus 9.5 ms [assumed seek time] which equals 5.5 ms, almost exactly the rotational latency associated with a 5400 RPM spindle speed.

quote:
The Caviar GP's outer-zone score clocks in at 79.8 MB/sec and as a result lags the older, less dense WD7500AAKS by 21%. Assuming similar sector-per-track zone configurations, a 7200 RPM drive would boast a 33% advantage over a 5400 RPM unit. The difference between the GP and the WD7500AAKS is less than that, likely of course due to a density advantage on the GP's part. Nevertheless, this second low-level diagnostic again suggests a 5400 RPM spindle speed.


But don't let any of that take away from your supreme confidence that it's a 7200 rpm drive...


RE: 4.3% is news worthy
By KernD on 5/24/08, Rating: -1
RE: 4.3% is news worthy
By Ringold on 5/25/2008 5:23:06 AM , Rating: 5
quote:
but the benchmark did give good result compared to the Seagate and Deskstar 1TB drive,


If by "good result" you mean "it sucked", then I agree..

Take just the FarCry test;
Deskstar: 877
WD Caviar GP: 671

The Deskstar is about 30% faster.


RE: 4.3% is news worthy
By rs2 on 5/25/2008 6:20:53 PM , Rating: 3
320 TB? Now that would be impressive.


RE: 4.3% is news worthy
By Lonyo on 5/23/2008 1:03:16 PM , Rating: 2
Um, 3x3320 = 960GB. Current 1TB drives are made up of 4 platters, meaning you're only using 250GB of the 320GB available, or just different platters which have 250GB available, I'm not exactly sure how they work it.
The move to 334GB platters means 1TB drives become 3 platter drives instead of 4, reducing power requirements, and meaning there are increased densities (1TB split over 3 platters gives you greater density than 1TB split over 4).

It's more than "oh the biggest platters are now 4.3% larger", it's "now we have 3 platter 1TB drives with up to 25% increase in density per platter".


RE: 4.3% is news worthy
By rninneman on 5/23/08, Rating: -1
RE: 4.3% is news worthy
By mcnabney on 5/23/2008 10:18:25 PM , Rating: 2
Re-read the post you responded to.

Th WDs were 320GB per platter, which didn't get them to the 1TB level since three of them would be 960. Now that they moved from 320 to 340 they can get to 1TB with a couple to spare. And in theory be faster due to higher densities.


RE: 4.3% is news worthy
By Lezmaka on 5/24/2008 2:44:24 PM , Rating: 5
Um, no.

The original 1TB GP drive used 4 250GB platters. WD came out with a single-platter 320GB drive in February (see http://www.anandtech.com/storage/showdoc.aspx?i=32... ), then a two-platter 640GB drive in March (see http://www.anandtech.com/storage/showdoc.aspx?i=32... )
From that second link:
quote:
While the WD 640GB drive does not fit in with the industry-standard capacity sizes, we fully understand Western Digital's rationale behind this move. This allows WD to use economies of scale with their new 320GB per-platter design and allows a natural progression up to the 1TB~1.3TB level by simply increasing platter count for each logical step. Of course, unless you use sub-prime mortgage mathematics, three 320GB platters only equals 960GB of capacity. WD engineering told us they can easily stretch the areal density of the current platter design to get to the magical 1TB capacity to match their competitors


RE: 4.3% is news worthy
By nirolf on 5/23/2008 1:04:01 PM , Rating: 2
From what I remember current GP drives use 4 x 250 GB platters.


Even more importantly
By masher2 (blog) on 5/23/2008 12:37:29 PM , Rating: 5
> "That means less power is needed and less moving parts are required to make the drive operate"

An even larger factor is that a higher areal density translates into higher drive performance, as it means more bits are moving past the head each second.




RE: Even more importantly
By Oregonian2 on 5/23/2008 1:59:50 PM , Rating: 1
Unless it was done by squeezing in more tracks where each track is of the same capacity.


RE: Even more importantly
By masher2 (blog) on 5/23/2008 3:23:50 PM , Rating: 2
No, that increases performance as well. It doesn't increase bandwidth per rotation, but it decreases the mean seek time (as the head has to move a shorter distance between tracks).


RE: Even more importantly
By Oregonian2 on 5/27/2008 3:02:49 PM , Rating: 2
Actually, the disk size would be the same so the average seek spec for a disk of the same physical size would be about the same. The average seek distance in terms of physical distance would be the same so the servos would take about as long (even if the track-to-track spec is the same, most seeks aren't one-track seeks and a 100-track seek doesn't take 100 times the time of a single track seek). So there may be some improvement, but it'd be relatively minor -- unlike increasing the data per rotation method of increasing capacity.


RE: Even more importantly
By Oregonian2 on 5/27/2008 3:07:02 PM , Rating: 2
P.S. - Because there would be more tracks for the same overall capacity of the drive (with fewer platters) one might argue that the small change in performance when having more tracks and fewer platters would result in a slightly slower drive.


RE: Even more importantly
By DeepBlue1975 on 5/23/2008 4:34:41 PM , Rating: 2
In theory that's true.
In practice it is true most of the time, but in "real world" tests and in many cases, some drives with lower density platters perform as well as the higher density ones, and in even better in some cases.

Take a look at WD's own SE16 750gb which has a less than 200gb/platter density (it has 4 platters, 188gb/platter), tested against WD's SE16 640gb one, which sports a 320gb/platter density (only 2 platters).

We are talking about a larger than 60% difference in areal density, that accounts for peformance differences which are not anywhere near that, in fact only in few cases the new drive is better than the older one by a 30% margin (mainly file copy tests)

Just take a look at this review, in which both drives (amonng many others) are included.

http://techreport.com/articles.x/14380/3

Most of the time the new drive only scores around 3-4% better than the older one, and in some cases, like system boot time, game level loading and multi user simulated scenarios, the older is better.

So, the most important factor in better areal density is just the benefit of less moving parts, which equates to better realiability, less noise and power requirements (power requirements are not that better comparing the 2 platter drive to the older 4 platter one, but noise is way better handled on the 2 platter one).

What you said about acces times is a valid point, but only talking about track to track seek time. Nevertheless, in access times on an HDD, mechanical characteristics of the actuator, precision of the head and rotational speed are way more important than any other factor, and I'd tend to say that the 8% better random access time in the new drive is mostly due to improved mechanical characteristics and caching scheme than areal density (for example the single platter WD 320gb disk scores an awful 16+ ms access time, which is amongst the worst you can get, and the areal density is the same as the 640gb drive sporting 2 of those same 320gb platters instead of one)

In the case of the article, it talks about 334gb/platter. The previous offering for wd was 320gb/platter, that is, 4.3% better density which won't surely buy you any benefit at all, at least not if we talk about performance.


RE: Even more importantly
By masher2 (blog) on 5/23/2008 11:58:43 PM , Rating: 2
> "but in "real world" tests and in many cases, some drives with lower density platters perform as well as the higher density ones"

Of course, because different drives use different heads, stepping motors, seek algorithms, cache sizes, and many other factors. However, all else being equal , a 50% increase in areal density translates to a 50% increase in STR (sustained transfer rate).

The other factor here is that areal density has no effect on rotational latency (RPM rate determines that) and none on mean seek times (head stepping speed controls that).

The fact remains that areal density and RPM rate are the two most important factors by far for overall drive performance.


RE: Even more importantly
By DeepBlue1975 on 5/24/2008 5:34:44 AM , Rating: 2
50% better areal density should equate to aroun 22% increase in str, as data is read in a linear fashion, while areal density refers... well, to an area which is the product of two linear dimensions :) in fact that`s about what tests show.
Anyway, the greenpower line is not performance oriented and because of that I wouldn`t expect it`s new configuration to be significantly faster than the previous one. Just take a look at the review I`ve linked before and you`ll see that not even in performance oriented drives a significantly bigger density coupled with equal or slightly better mechanics can`t bring such a great benefit in most scenarios (file cooy operations usually benefit the most, while the other tests just get marginally better results and even worse in some cases).
I`m talking about the 750g
b, 4 platter drive against the 640gb, 2 platter one from wd`s SE16 line.


RE: Even more importantly
By masher2 (blog) on 5/24/2008 10:29:48 AM , Rating: 2
> "50% better areal density should equate to aroun 22% increase in str, as data is read in a linear fashion"

No, this isn't true, as one can easily understand with the following thought experiment. An increase in areal density doesn't affect the time to read the entire platter (1/RPM * # of tracks). Therefore, in reading that entire platter, a 50% areal density increase means 50% more total bits will be passed over in the same amount of time. The mean STR therefore increases in the same proportion.


RE: Even more importantly
By masher2 (blog) on 5/24/2008 10:33:25 AM , Rating: 2
> "you`ll see that not even in performance oriented drives ...can`t bring such a great benefit in most scenarios..."

All you're proving there is that most applications aren't STR-bound, e.g. limited by raw drive bandwith. Most applicatoins are some combination of CPU, memory, or disk latency bound.

By the way, this is the exact same reason most desktop (not server) applications see little or no benefit from RAID, and some even see a small performance decrease. A RAID array increases STR, but also increases rotational latency.


RE: Even more importantly
By DeepBlue1975 on 5/24/2008 4:51:59 PM , Rating: 2
Ooops. I must admit you are right about the areal density.
It's actually a very basic point, but well, after a 4 day long fever I am even finding myself having a hard time remembering my own name :)

Anyway, what you said was actually my point. General performance doesn't grow linearly with sequential transfer rates.

The point that makes me doubt a bit, aside from theory, is that holding on to the example of the 750gb, 4 platter wd vs. the 640gb, 2 platter one, the increase in STR is only a bit higher than 20%, not anywhere near the 60%+ increase in areal density.

The 2 platter drive even has better access times... I doubt if it has something to do with the other drive having 8 heads instead of four, but taking into account the aggregated latencies caused by head and cylinder switching, that should not provide any significant advantage to masquerade that 60% theoritical advantage.

I'll search for detailed WD specs on those drives so I can calculate those factors, too...
I might as well wait for my deplorable status to go away or I could come up with transfer rates of many pettabytes / millisecond.

Sorry for the pointless discussion!


RE: Even more importantly
By Fnoob on 5/26/2008 4:42:23 PM , Rating: 2
"....more bits are moving past the head each second. "

Do you think that the Sata 3G interface will every become fully utilized before it's replaced? With the latest greatest SSD's managing only 160-200Mb/s, what will it take?


Stuck at 1TB
By Doormat on 5/23/2008 1:43:20 PM , Rating: 2
I want to know when we'll see much larger capacities? We've been stuck at 1TB for a while now - I can remember back as mid/late 2005 for when the first 1TB HDDs were announced. Two and a half years later we haven't moved up in capacity. The platters have decreased, bit density and performance have increased, but they haven't surpassed 1TB in capacity...




RE: Stuck at 1TB
By bespoke on 5/23/2008 2:04:17 PM , Rating: 2
We will see 1.2TB drives very soon, as drive makers can use 4 of these 320/340GB platters.


RE: Stuck at 1TB
By walk2k on 5/23/2008 5:00:52 PM , Rating: 5
How much porn do you need to download anyway?


RE: Stuck at 1TB
By amanojaku on 5/23/2008 5:49:01 PM , Rating: 2
What the hell kind of question is that? ALL OF IT! Except the kiddie stuff. And the animals. And the scat. And the vomit. And the preggers. And the gay porn (lez is OK!)

I guess he only need a few yottabytes, after all...


RE: Stuck at 1TB
By retrospooty on 5/24/2008 10:48:14 AM , Rating: 2
You forgot "Brazilian fart porn"... Unless that was left off on purpose =)


RE: Stuck at 1TB
By Saosin on 5/26/2008 5:02:27 AM , Rating: 2
Isn't five platters feasible? That would mean 1670GB, quite a lot bigger than the 1TB drives we have today.


RE: Stuck at 1TB
By PenguinTech on 5/26/2008 1:04:54 PM , Rating: 2
Agreed! My 25 years of home videos/photos and my collection of 1920's/30's/40's movies is at 2.5 TB and growing. I keep two backups, one home and one off-site, for a total of 7.5 TB. (And not a single torrent file on them!) There are real reasons for larger drives, and being stuck at 1 TB for several years is a problem.

Add the fact that all new HD camcorders are AVCHD disk or flash-based, and more people are going to need more storage. You can't just throw the camcorder tape into a drawer anymore, you need a big disk to store your HD memories.


RE: Stuck at 1TB
By ImSpartacus on 5/26/2008 2:20:01 PM , Rating: 2
It is actually cheaper to use $99 640GB drives. Just put 5 of them in RAID 5 and call it a day. That's $500 and 2560GB.

You don't need larger drives, just more of the same capacity.


Just a marketing move
By DeepBlue1975 on 5/23/2008 8:00:11 PM , Rating: 2
As they already have 320gb/platter disks with up to 2 platters, stretching it to 334 to yield 1tb is just a marketing stretch as they could as well sell 960gb drives.

I guess that the extra 4% didn't even need a lot of work over the existing plates they were using, and for the little difference I guess the mechanical characteristics of this "new" drive remain 100% equal to those on the 6400aaks.

But well, after all, WD is by now the only big name in the HDD industry that's not selling terabyte drives as the others are, so it must be really a marketing milestone to achieve, so, from that standpoint is a very valid move.

Technologically speaking, this is not interesting at all.




RE: Just a marketing move
By Fulcrum77 on 5/23/2008 8:36:33 PM , Rating: 2
I was under the impression that the previous 320GB platters were actually the 334GB platters with 14GB disabled for performance reasons. This is probably why the mechanical performance characteristics are identical.


RE: Just a marketing move
By teldar on 5/23/2008 9:49:21 PM , Rating: 2

quote:
But well, after all, WD is by now the only big name in the HDD industry that's not selling terabyte drives as the others are, so it must be really a marketing milestone to achieve, so, from that standpoint is a very valid move.


Since when are they not selling terabyte drives? They have had them for several months now. Hello? The terabyte greenpower? This is simply a change to fewer platters. Now, instead of 4 platters as previously or 5 platters like Hitachi has, they use 3 platters like the Samsung has since the F1 series was first announced.


RE: Just a marketing move
By DeepBlue1975 on 5/24/2008 5:41:21 AM , Rating: 2
You are right. I was talking about performance tb drives, and yet there I`ve got confused as the article talks about the gp line. So, WD still has no performance offering with 1TB. Sorry for the confusion!


Platter Wars????
By bapcorp on 5/26/2008 2:04:38 AM , Rating: 2
I cant believe that people dont realise that the platter size was always bigger than the market size.
250 GB is really about 266 GB
320 GB is really about 333-340 GB
It is the marketing guys who said 250 GB drives sound better than a 266 GB drive etc.
Plus the fact they dont use the inner tracks as much so this increases overall performance.
1.2 TB vs 1.3 TB I would put my money on a 1.3 TB drive first, unless they get 400+ GB platters out soon. This will then probably be the last you see of the good'ol conventional hard drive as SSD Drives are catching on and memory densities are increasing faster then platter densities. Not to mention access times are better, sustained read and writes are better. The only thing stopping the SSD revolution is the price, and this will decrease in the very short term as what happens to all new technologies. ( who paid over $200 for a DVD burner when they first came out ??? dont be shy put your hand up boys and girls )




RE: Platter Wars????
By PenguinTech on 5/26/2008 12:52:17 PM , Rating: 2
I paid $600 for the first HP DVD+RW burner about 5 years ago. It was obsolete within a year when the "+" consortium introduced the DVD+R and HP couldn't add that functionality with a firmware upgrade. Oh well, pioneers get the arrows in their backs, etc.


RE: Platter Wars????
By afkrotch on 5/30/2008 12:46:58 AM , Rating: 2
A DVD+RW could burn DVD+Rs. They couldn't do anything with a DVD-R or DVD-RW.


capacity vs speed
By ilkhan on 5/24/2008 1:15:27 AM , Rating: 3
Id rather have an inexpensive, 2TB drive, even if it means spinning at 5400 or even slower rpm. My application cares about capacity, not speed. And I doubt Im alone.

Stop making bigger faster drives and give us some BIG cheap drives.




RE: capacity vs speed
By DeepBlue1975 on 5/24/2008 5:53:51 AM , Rating: 2
I personally care quite a bit about performance, but in spite of that I think that more options should exist for those who care more about space than speed. A new bigfoot could be welcome by many. Lower rotational speeds could help squeezing out better recording density and hence pave the way for even higher density platters made ad hoc for those slower drives. Though I have to admit I don`t really know wether by the moment the most difficult thing is to increase platter density or to get disk heads to be able to fetch data in less time.


T'would be teh sweetness...
By larson0699 on 5/23/2008 1:20:03 PM , Rating: 2
...if they made the 3x334 transition on their RE2 GP drives as well.

5 year warranty + green + RAID ops FTW




RE: T'would be teh sweetness...
By bryanW1995 on 5/23/2008 3:34:05 PM , Rating: 2
1tb sounds really cool for marketing purposes. 1.2 tb...not so much, plus it is much more epensive to make it b/c of the extra platter. so they would need to get a lot more money for not much more storage. They probably won't come out with larger hd's until 1.5 gb is attainable at reasonable prices.


not quite true
By PLaYaHaTeD on 5/23/2008 10:55:31 PM , Rating: 3
quote:
Western Digital gets to 334GB per plater about a year after Samsung


Samsung have announced their drives last summer, but they didn't arrive until dec/jan. And to top it off, those drives were known to have a particularly high failure rate. So saying they were a year late is not accurate at all.




less = cheaper
By Suomynona on 5/23/2008 7:47:01 PM , Rating: 2
Besides the added performance benefits that increased density brings to the table--when the samsung 750gb first came to the market a lot of people were impressed at how close it came to the WD raptors in some real world benchmarks--taking a platter out of a hard drive reduces the moving parts in the drive making it a a lot cheaper to produce. It's these advances in tech that have driven the price of consumer storage down to 16¢ per GB. Fewer platers also have some other benefits such as the noted cooler operating temperatures and lower power consumption.




"And boy have we patented it!" -- Steve Jobs, Macworld 2007














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki