• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

2 Part Question on Harddrive.

jazzhound

Banned
I've 2 60GB Maxtor which perform quite nicely and I'm happy, but faster is always better right? Which leads to my questions. I have been hearing alot about the WD 8MB cache HD's almost showing parallel performance of SCSI drives. I've never used SCSI before, only info I have is that it spins in excess of 10,000RPM to 15,000 RPM! Compare that to your standard 7200RPM it translates to FAST! The plan is, sometime after the New Year, I want to buy a 80-120GB hdd sans 8MB cache (difference of $50), use it as my storage drive (currently my 2 Maxtor's are fulfilling that role nicely) and RAID the 2 Maxtor to get a theoretical 14,000RPM performance. (correct me if there are any omissions).

If the 8MB cache is so superior to regular drives and at a fraction of cost compared to SCSI why aren't Maxtor (though I've seen a 2MB cache version), Seagate, Quantum, IBM, et al. adopting this ? So far I've only seen WD and Maxtor have this. Is there also some techinical limitation on how much can be used per capacity? i.e 120GB has 8MB, why not more ?

Thx
 
RAID the 2 Maxtor to get a theoretical 14,000RPM performance. (correct me if there are any omissions).

RAID'ing 2 drives would be somewhat equivalent to doubling the areal data density, not doubling the RPM's. RAID improves transfer rates which are usually increased by increasing areal data density on the platters. Higher RPM's reduces latency and because higher RPM drives need to decrease the diameter of the platter also reduce seek times which leads to lower access times, which RAID does not do (not much anyway).

If the 8MB cache is so superior to regular drives and at a fraction of cost compared to SCSI

8MB cache benchmarks very well, but you would be hard pressed to distinguish the difference in a blind test. Don't believe that 8MB cache puts ATA drives on par with SCSI (which have 8MB as well). When you compare current gen to current gen, there is no competition between the two.

Seagate, Quantum, IBM, et al. adopting this ?

Seagate has an 8MB SATA drive out, IBM has an 8MB 180GXP out and Quantum was bought out by Maxtor last year.

120GB has 8MB, why not more ?

IBM, Maxtor and WD all have 8MB cache drives of at least 180GB.
 
I have an 8mb cache model... the WD800JB to be exact. And on average, I get 9 MB/s uncached transfer rate. The drive is quiet contrary to what some people say. I just replaced a Maxtor 60 GB drive with this one... the Maxtor died.

You asked why not more than 8mb? Well, I guess some reason would have to do with the cost... but I have heard talk about new hard drives coming out soon with 16mb of cache.

The reason more cache is beneficial is because your hard drive is somewhat intuitive. If certain data is called on repeatedly, it puts it in cache memory so if it's called on again, it can be retreived without reading it from the physical disk. The more cache you have, the more room there is to store frequently used data, thus increasing the POTENTIAL for your computer to use something your hard drive has cached. More cache doesn't guarantee faster speeds, it just makes it possible... however, it is almost guaranteed that you will see a performance gain like I have.

I had tested my Maxtor 60 GB Diamond Max Plus drive on pcpitstop.com many times, and the best transfer rate I got was 3 MB per second of uncached data... the first test I did with this WD drive I got over 9 MB/s of uncached data.

For some reason I haven't been able to explain though, the Maxtor had a higher cached transfer rate, I cannot figure out why, or how... the Maxtor's maximum cached transfer rate was about 475 MB/s and my WD's max cached transfer rate is under 180 MB/s. I'm confused as to why this is possible at all, seeing as how the PCI bus is only capable of 133 MB/s, so even the WD drive's measurement shouldn't be possible. Someone said it's probably because the drive cached some data in RAM... which would explain the 475 MB/s, and MAYBE since the WD drive has more cache, it didn't put much of anything in RAM, so the numbers look like the Maxtor is faster, but in reality, it's not.

By the way... the highest my Maxtor scored in PCMark 2001 was 860... I ran the test on this WD drive without defragging it and got over 1,000 😀
 
Originally posted by: jazzhound
I have been hearing alot about the WD 8MB cache HD's almost showing parallel performance of SCSI drives.

Someone lied to you, unless they were comparing 8meg ATAs to 5 year old SCSIs.

I've never used SCSI before, only info I have is that it spins in excess of 10,000RPM to 15,000 RPM! Compare that to your standard 7200RPM it translates to FAST!

Yes, data platters on SCSI drives spins faster than on your ATA drives. Does that mean you will see much of a difference? Nope, if you are just counting RPMs. Performance comes from many different factors, as state here in the storage review.

The plan is, sometime after the New Year, I want to buy a 80-120GB hdd sans 8MB cache (difference of $50), use it as my storage drive (currently my 2 Maxtor's are fulfilling that role nicely)

If your current HDs are fulfilling the roll nicely, why are you getting a new hd for it? 8meg cache for storage HD (mp3s, movies, blah) is a huge waste of money. You'll be playing your mp3s at .... their regular speed. Your movies? Why, they'll be playing at... movie speed. If you want mp3s or movies to load up faster, get a better CPU and RAM, those are the bottlenecks. The only reason you will need more cache for a HD is for games and ..... games. And then you will notice about a 2 sec load time difference.

and RAID the 2 Maxtor to get a theoretical 14,000RPM performance. (correct me if there are any omissions).

Yes, your HDs, in the perfect world, should be 14k if you RAID 0 them. But, in the this world, you won't notice much of a difference. That is because even if your RPMs is 14k (which it won't be), you still have 2 HDs fighting for a very limited bandwidth. This is where SCSI excels, and this is mainly why SCSI RAID is much faster than IDE RAID.

If the 8MB cache is so superior to regular drives and at a fraction of cost compared to SCSI why aren't Maxtor (though I've seen a 2MB cache version), Seagate, Quantum, IBM, et al. adopting this ? So far I've only seen WD and Maxtor have this. Is there also some techinical limitation on how much can be used per capacity? i.e 120GB has 8MB, why not more?

Cause they all know that the 8meg cache is just a marketing gimmick, HDs benefit more from faster transfer rates than from more cache. That is why there is a push to get rid of ATA and move on to SATA. And to answer your question about why there isn't higher capacity 8 meg cache hds, there is. You don't have to worry, they are always after your money.

Edit--
This quote from THD. "Thanks to a huge buffer memory (8 MB instead of the typical 2 MB), the WD1200JB performs similarly to high-performance SCSI drives." Translation -> Thanks to a huge buffer memory, the WD1200JB performs similarly to other 7200 RPM SCSI drives. I don't think they even make those drives anymore. No wonder people think Tom sleeps with his suppliers, freaking WD is using a quote similiar to that (WD1200JB outperforms SCSI) to advertise their drives! :disgust:
 
Originally posted by: Pariah
RAID the 2 Maxtor to get a theoretical 14,000RPM performance. (correct me if there are any omissions).

RAID'ing 2 drives would be somewhat equivalent to doubling the areal data density, not doubling the RPM's. RAID improves transfer rates which are usually increased by increasing areal data density on the platters. Higher RPM's reduces latency and because higher RPM drives need to decrease the diameter of the platter also reduce seek times which leads to lower access times, which RAID does not do (not much anyway).

If the 8MB cache is so superior to regular drives and at a fraction of cost compared to SCSI

8MB cache benchmarks very well, but you would be hard pressed to distinguish the difference in a blind test. Don't believe that 8MB cache puts ATA drives on par with SCSI (which have 8MB as well). When you compare current gen to current gen, there is no competition between the two.

Seagate, Quantum, IBM, et al. adopting this ?

Seagate has an 8MB SATA drive out, IBM has an 8MB 180GXP out and Quantum was bought out by Maxtor last year.

120GB has 8MB, why not more ?

IBM, Maxtor and WD all have 8MB cache drives of at least 180GB.

Listen to Pariah...this is all correct, and well done 🙂
 
Yes, your HDs, in the perfect world, should be 14k if you RAID 0 them. But, in the this world, you won't notice much of a difference. That is because even if your RPMs is 14k (which it won't be), you still have 2 HDs fighting for a very limited bandwidth. This is where SCSI excels, and this is mainly why SCSI RAID is much faster than IDE RAID.

No, this is incorrect.
They should not be 14k... they will still act like 7200 RPM drives, but your transfer rates will be almost double. RPM doesn't have as much to do with transfer rates as data density. RPM has more to do with seek time and latency. On a 7200 RPM drive, maximum latency (the time you have to wait for a section of the disk to sping back around to the head to be read or written to) is 4.2 milliseconds. So if you measure your seek times using a benchmark utility, and subtract 4.2, that will give you the track to track seek time. Which is what they're rated in.

Also, not all SCSI interfaces have more bandwidth available than an ATA133 controller. The only SCSI interfaces I know of that have more bandwidth than ATA133 is Ultra 3 and Ultra 4... with bandwidths of 160 MB/s and 320 MB/s respectively.

ALSO... a good SCSI 15k RPM drive maxes out at about 60 MB/s transfer rate... so... even 2 SCSI drives in RAID couldn't saturate the an ATA133 bus if it was capable of connecting to it... true, you'd need about 3 ATA133 drives in RAID 0 to equal the performance of 2 SCSI drives... but that's ONLY in terms of maximum transfer rate.

The main reason SCSI drives are faster than IDE is because of their seek time. A 15k SCSI drive can have seek times below 2 milliseconds, whereas the lowest IDE drives are right around 8.
 
Good information guys! I have a friend who did IDE RAID and he said Windows/Games are lightning fast! I'm going to get a regular drive for backup (about 100GB worth of mp3s/appz/movies) and RAID the two existing drives.

Seems like every few months we get a bump in CPU clockspeed, yet the biggest bottleneck (as it's been reiterated many times by pundits) lies in harddrive performance. You folks seem to be on top of things as far as what hits the market first. In your estimation when will we see this SATA standard? Hell we've done away with Serial and went on to better things like USB, USB 2.0, 1394 (albeit on a limited scale), it's time the industry address the problem.. hmm how old is ATA anyways?


Thx again!
 
SATA is on motherboards now, Harddrives expected within a month or 2. And maybe I'm just reading your post about USB 1.1/2.0 and Firewire wrong, but those aren't better technologies for harddrives, and they are also both serial interfaces. The ATA standard is holding up just fine, SATA will hold on for a good number of years just fine...
 
Yes, your HDs, in the perfect world, should be 14k if you RAID 0 them.

Not exactly, as Jeff and I explained.

That is because even if your RPMs is 14k (which it won't be), you still have 2 HDs fighting for a very limited bandwidth.

If the drives are on seperate channels, which is always recommended, this is a non-issue. Even if they are on the same channel, it's an overstated myth that performance will be heavily reduced.

Edit--
This quote from THD. "Thanks to a huge buffer memory (8 MB instead of the typical 2 MB), the WD1200JB performs similarly to high-performance SCSI drives." Translation -> Thanks to a huge buffer memory, the WD1200JB performs similarly to other 7200 RPM SCSI drives. I don't think they even make those drives anymore. No wonder people think Tom sleeps with his suppliers, freaking WD is using a quote similiar to that (WD1200JB outperforms SCSI) to advertise their drives!

This is a completely unnecessary shot at THG which isn't even completely accurate. Tom didn't right the article, so leave him out altogether. Let's look at the quote right above the one from THG on the WD retail box:

"...delivers the desktop performance of a good 10K RPM drive."-StorageReview.com

Going to claim SR is sleeping with their suppliers as well? It's a valid statement to say 8MB drives benchmark as well as SCSI in certain applications. It doesn't mean they are on par with SCSI in actual use and across the board.

a good SCSI 15k RPM drive maxes out at about 60 MB/s transfer rate.

15k drives are up to 76MB/s now. Even the current gen 10k drives are around 70MB/s now, with 3 at 69MB/s or higher.

A 15k SCSI drive can have seek times below 2 milliseconds

The lowest SCSI seek times are around 4.8ms on average. Latency for 15k drives is 2ms, added to seek totals an average access time of 6.8ms.

In your estimation when will we see this SATA standard

Controllers are readily available, and if you look hard enough Seagate has drives out. Reports from Japan say Maxtor drives are available there.
 
I don't see Serial ATA as any big deal... if I understand it correctly, it's as big a deal as ATA100 vs. ATA133. You'll HAVE to have a RAID configuration to use the bandwidth... and as Pariah said, 2 hard drives on the same channel is no big deal considering that most IDE drives can't sustain more than 40 MB/s if that much... which means you'd need 3 drives in RAID 0 to come close to saturating the PCI bus.
 
But cabling and reducing line noise are huge steps, so why not make a possible speed jump while fixing that? Hotswapping, longer/thinner cables, individual cables/channels (a plus for 2 drives not near each other), and the bonus of the same connector for 2.5" drives as 3.5"...

Those are tangible benifits to the std
 
Originally posted by: Pariah
Even if they are on the same channel, it's an overstated myth that performance will be heavily reduced.
Straight from storage review, "However, when using large numbers of drives with RAID, suddenly things become a bit different. Well, under IDE/ATA they aren't different, because IDE/ATA can only handle transfers from one drive at a time. With SCSI though, it's a very different story. On the SCSI bus, multiple drives can be transferring data simultaneously." And how does that not heavily reduce performance?

Edit--
This quote from THD. "Thanks to a huge buffer memory (8 MB instead of the typical 2 MB), the WD1200JB performs similarly to high-performance SCSI drives." Translation -> Thanks to a huge buffer memory, the WD1200JB performs similarly to other 7200 RPM SCSI drives. I don't think they even make those drives anymore. No wonder people think Tom sleeps with his suppliers, freaking WD is using a quote similiar to that (WD1200JB outperforms SCSI) to advertise their drives!

This is a completely unnecessary shot at THG which isn't even completely accurate. Tom didn't right the article, so leave him out altogether. Let's look at the quote right above the one from THG on the WD retail box:

I agree, Tom didn't "right" the article. But it is on his review site. I don't see strangers writing unrequested reviews on the website. Try reading the article, I bet you can't find "SCSI" mentioned anywhere but the title and the summary box. It's so nice of them to include proof backing up that claim. Plus, it's such an outrageous claim. It's like claiming a Boxer outperforms a McLaren F1.

"...delivers the desktop performance of a good 10K RPM drive."-StorageReview.com

Going to claim SR is sleeping with their suppliers as well?

Well, SR doesn't have an outrageous quote. It's somewhat doubtful, but it's nice and specific. Instead of being like Tom, SR doesn't claim that the WD SE outperforms higher RPM drives, they state exactly which RPM they are talking about.

It's a valid statement to say 8MB drives benchmark as well as SCSI in certain applications. It doesn't mean they are on par with SCSI in actual use and across the board.

If you were a consumer and just read that quote, would you think that the WD SE is only better than SCSI at certain applications, or it's just plain better than SCSI? The perfect marketing quote.
 
Back
Top