There's been a lot of talk lately about it, and a few exciting announcements...but the same could be said about February 2000. Does anyone have ideas, suggestions, or even (gasp) facts about when this new interface is going to be available?. To my knowledge, there's a few motherboards that are going to offer an SATA connector in a few months. I know the Baracuda V drives should support native SATA, and i've heard rumors about fall availability...i'm going to build my new system in about a month or so (only looking for this for my HD), and i'm wondering if it's feasible to wait on this, or ditch it and go with ATA-100 or somethin. Anyone with any more insight, please post up...thanks.
Info on Serial ATA
- Thread starter cliffa3
- Start date
Adul
Elite Member
well there is hardly any if at all real world performance benniffits. but the nice thin lines used is my biggest draw. you can get a card from hotpoint that supports it now. and they sell adapters foride drives to run on serial ata.
Performance benefits are irrelevant to me. The tiny cables are the main selling point. They're not much larger than CD-ROM audio cables for crying out loud!
I don't know how much farther we can go with rotating medium. IMO more research needs to be put into increasing static forms of memory (like flash) as a replacement for main storage. Increasing the speed of the platters lead to greater power requirements and increased heat. Increasing heat means expanding platters which become harder to compensate for as platter density increases.
I don't know how much farther we can go with rotating medium. IMO more research needs to be put into increasing static forms of memory (like flash) as a replacement for main storage. Increasing the speed of the platters lead to greater power requirements and increased heat. Increasing heat means expanding platters which become harder to compensate for as platter density increases.
wow...that's pretty dang sweet...you have one, or did you just find the link?..that's exciting, does anyone have any clue when the Baracuda V will be available to pair up with that card?.
And do you know what kind of performance that card offers?..anything noticable?.
And do you know what kind of performance that card offers?..anything noticable?.
SATA cards/mobos are useless unless u want to dish out the extra money for a 40-pin-to-SATA adapter for your hdd.
hdd manufacturers need to start coming out w/ SATA drives already...
hdd manufacturers need to start coming out w/ SATA drives already...
you can see that adapter on the link that was posted...but i'm not going to drop near 200 bucks on a nice SATA setup/converters if all it does is give me cute little cables to connect my HD to the PCI card...there has to be SOME performance benefit, i'm just wondering how much. The cables are a definite draw, but i'm not seeing how the ATA/66 HD will transmit data faster over the cable connected to a PCI card...is this possible or not?
edit:spelling
edit:spelling
there is no performance benefit - the drive technology is exactly the same and isn't even currently saturating ATA100 let alone SATA 150.
there might be a slight performance increase or decrease in performance depending on the performance of the SATA controller's chipset + drivers....but either way, it won't be anything substantial.
smaller cables are also overrated...i mean, they're nice if the cost if free, but if u're buying controllers/drives solely for having smaller cables, u're wasting your money.
there might be a slight performance increase or decrease in performance depending on the performance of the SATA controller's chipset + drivers....but either way, it won't be anything substantial.
smaller cables are also overrated...i mean, they're nice if the cost if free, but if u're buying controllers/drives solely for having smaller cables, u're wasting your money.
i agree...i'd take bigger cables if there was a performance improvement, but SATA sounds like it will (hopefully at the start, but more than likely after it develops) have performance benefits...and performance with innovation (smaller cables) is more than welcome I do agree there isn't anything pushing the limits ATA/100, or even ATA/33 in most cases, but it would be nice to have the bottleneck opened...with the new drives having 8mb of cache, it looks like they should (have a performance benefit) or that increase in buffer size shouldn't be necessary.
" but SATA sounds like it will (hopefully at the start, but more than likely after it develops) have performance benefits"
Such as?
"but it would be nice to have the bottleneck opened"
The current bottleneck is the drives, changing the interface won't fix that.
"with the new drives having 8mb of cache, it looks like they should (have a performance benefit) or that increase in buffer size shouldn't be necessary."
The benefit of extra cache is zero access time (not really 0, but compared to disk access it is) and higher burst rates. A SATA pci card and ATA133 PCI card have the exact same potential burst speed of a 133MB/s meaning there truly is no performance benefit of a SATA addin card.
The above linked Highpoint RAID controller sells for $150 at newegg. It does come with 2 adapters, but still is considerably overpriced for what you get unless you really need smaller cables.
Such as?
"but it would be nice to have the bottleneck opened"
The current bottleneck is the drives, changing the interface won't fix that.
"with the new drives having 8mb of cache, it looks like they should (have a performance benefit) or that increase in buffer size shouldn't be necessary."
The benefit of extra cache is zero access time (not really 0, but compared to disk access it is) and higher burst rates. A SATA pci card and ATA133 PCI card have the exact same potential burst speed of a 133MB/s meaning there truly is no performance benefit of a SATA addin card.
The above linked Highpoint RAID controller sells for $150 at newegg. It does come with 2 adapters, but still is considerably overpriced for what you get unless you really need smaller cables.
you'll be seeing more sata devices (controller and drive) in the next 12 months.
non sata devices are available in the channels right now and may not be for a few months.
non sata devices are available in the channels right now and may not be for a few months.
"or even (gasp) facts about when this new interface is going to be available?. "
? HighPoint RocketRaid 1520 (SerialATA) is already available and comes with PATA adapter. HOwever since it's the first available piece of hardware it's obviously overpriced at $150 @ newegg.
"I know the Baracuda V drives should support native SATA, and i've heard rumors about fall availability..."
? That's not rumor that's straight from Seagate.
? You may want to check out this thread and the linkage it contains. SerialATA Pics [Thread
"SATA cards/mobos are useless unless u want to dish out the extra money for a 40-pin-to-SATA adapter for your hdd."
? Unless like the HighPoint RR 1520 it comes with the adapter and then there's no extra expense ... SATA drives are coming soon anyway so no biggy.
"there is no performance benefit - the drive technology is exactly the same and isn't even currently saturating ATA100 let alone SATA 150.
there might be a slight performance increase or decrease in performance depending on the performance of the SATA controller's chipset + drivers....but either way, it won't be anything substantial."
? The performance benefit is in being able to access multiple drives at the same time (since it's point to point). The hardware is completely comapible with current OS and drivers so I don't understand your second point.
? Other benefits aside from slight performance increases. Include the obvious smaller cables (that everyone thinks about), lower electrical requirements (much lower), hot plugging, etc... visit SerialATA.org for more info.
"The current bottleneck is the drives, changing the interface won't fix that."
? Agreed.
Thorin
? HighPoint RocketRaid 1520 (SerialATA) is already available and comes with PATA adapter. HOwever since it's the first available piece of hardware it's obviously overpriced at $150 @ newegg.
"I know the Baracuda V drives should support native SATA, and i've heard rumors about fall availability..."
? That's not rumor that's straight from Seagate.
? You may want to check out this thread and the linkage it contains. SerialATA Pics [Thread
"SATA cards/mobos are useless unless u want to dish out the extra money for a 40-pin-to-SATA adapter for your hdd."
? Unless like the HighPoint RR 1520 it comes with the adapter and then there's no extra expense ... SATA drives are coming soon anyway so no biggy.
"there is no performance benefit - the drive technology is exactly the same and isn't even currently saturating ATA100 let alone SATA 150.
there might be a slight performance increase or decrease in performance depending on the performance of the SATA controller's chipset + drivers....but either way, it won't be anything substantial."
? The performance benefit is in being able to access multiple drives at the same time (since it's point to point). The hardware is completely comapible with current OS and drivers so I don't understand your second point.
? Other benefits aside from slight performance increases. Include the obvious smaller cables (that everyone thinks about), lower electrical requirements (much lower), hot plugging, etc... visit SerialATA.org for more info.
"The current bottleneck is the drives, changing the interface won't fix that."
? Agreed.
Thorin
anyone out there actually using one of those cards?..i'd like to see a review done on one to see some benches...i'm really considering grabbing one if two converter cables are included...post up if you know of anyone who's used the card/converters.
The cards don't offer any performance benefits right now, simply because there are no SATA drives available. Changing the interface does do alot; but it does next to nil when only one side is changed. Meaning exactly what I said before, there is not going to be any performance benefit when using a Parallel ATA drive with a Serial ATA controller.
But when comparing the Serial ATA interface to Parallel ATA, Serial ATA outweighs PATA in tons (IMO.. the tons part).
Because each cable for Serial ATA is only a single channel 'solution', it currently requires 1 controller (chip) per channel. On the other hand, Parallel ATA, in its current elderly state (around 20 yrs old) only requires 1 controller for up to 2 channels, under E/IDE standards.
There's also the cable length issue, current in servers, and possibly full-tower cases:
Current Parallel ATA requirements can only allow cables up to 18 inches in length, per channel.
Serial ATA's requirements are a lot more lenient: it allows the cable to be up to 1 meter (approx. 3 feet) in length, per channel.
IMHO, Serial ATA is much more than just a cable improvement over Parallel ATA..
But until Serial ATA hard drives start showing up on the market (no more than 2 months from now), all Serial ATA is going to be is a smaller cable...
And don't worry.. I know I didn't get everything. Those are just the most important features at this time, and imo, again. Now all we need is a better HDD technology (preferably something like flash RAM), and a better I/O bus (PCI Express, formerly 3GIO).
-Tarmax
Edit for readability and some other stuff
But when comparing the Serial ATA interface to Parallel ATA, Serial ATA outweighs PATA in tons (IMO.. the tons part).
- Max burstable transfer rate
- Parallel - 133MB/s theoretical (currently.. anything faster is economical suicide for mass market. ie. SCSI)
- Serial - 150MB/s theoretical, only the first iteration of Serial ATA (further iterations could allow up to 600MB/s bandwidth).
- Power requirements
- Parallel - 5V transceivers (1 per wire), which prevents higher bandwidth without adding technology like LVD (Low Voltage Differential signaling) - which basically raises the prices to those around SCSI's
- Serial - 500mV max, peak to peak
- Pin count
- Parallel - 80 wires per cable (26 per channel + power/ground), effectively 5cm wide
- Serial - 7 wires per cable (1 channel), effectively 8mm wide
Because each cable for Serial ATA is only a single channel 'solution', it currently requires 1 controller (chip) per channel. On the other hand, Parallel ATA, in its current elderly state (around 20 yrs old) only requires 1 controller for up to 2 channels, under E/IDE standards.
There's also the cable length issue, current in servers, and possibly full-tower cases:
Current Parallel ATA requirements can only allow cables up to 18 inches in length, per channel.
Serial ATA's requirements are a lot more lenient: it allows the cable to be up to 1 meter (approx. 3 feet) in length, per channel.
IMHO, Serial ATA is much more than just a cable improvement over Parallel ATA..
But until Serial ATA hard drives start showing up on the market (no more than 2 months from now), all Serial ATA is going to be is a smaller cable...
And don't worry.. I know I didn't get everything. Those are just the most important features at this time, and imo, again. Now all we need is a better HDD technology (preferably something like flash RAM), and a better I/O bus (PCI Express, formerly 3GIO).
-Tarmax
Edit for readability and some other stuff
anyone got price estimates for the cards+converters? My 6 HD's in my case make it terribly messy to work with.
"Max burstable transfer rate
Parallel - 133MB/s theoretical (currently.. anything faster is economical suicide for mass market. ie. SCSI)
Serial - 150MB/s theoretical, only the first iteration of Serial ATA (further iterations could allow up to 600MB/s bandwidth)."
Burst rate stats are completely worthless. With SATA each drive gets dedicated bandwidth meaning the theoretical max is so far beyond the capability of current drives it shouldn't even be considered a feature. At the present rate that IDE HD STR is increasing it will be some time in 2004-05 that IDE drives will surpass ATA133. I don't consider improving on a standard that won't be surpassed for over 2 years of much importance. Also, most people will probably add SATA compatibility initially via PCI card which means the 150MB/s theoretical gets cut down to 133MB/s by the PCI bus. So it doesn't have any performance advantage at all in that scenario.
"Pin count
Parallel - 80 wires per cable (26 per channel + power/ground), effectively 5cm wide
Serial - 7 wires per cable (1 channel), effectively 8mm wide"
The smaller cables are all the interface currently has going for it, and is the only thing that people familiar with the current state of storage technology would care about.
"On the other hand, Parallel ATA, in its current elderly state (around 20 yrs old)"
The original ATA spec was not submitted for ANSI approval until 1990. And it wasn't until 1994 that the inital spec was given approval. So at oldest the concept of P-ATA is 12 years old with the actual initial standard only 8 years old. SCSI is considerably older, with its initial spec approved in 1986.
"Current Parallel ATA requirements can only allow cables up to 18 inches in length, per channel."
That's what the spec says, but you can buy 24in and longer cables that will work fine.
"But until Serial ATA hard drives start showing up on the market (no more than 2 months from now), all Serial ATA is going to be is a smaller cable..."
Won't make the slightest bit of difference. The next generation of HD's will released in PATA and SATA formats and there won't be a bit of difference between the performance of the 2.
Parallel - 133MB/s theoretical (currently.. anything faster is economical suicide for mass market. ie. SCSI)
Serial - 150MB/s theoretical, only the first iteration of Serial ATA (further iterations could allow up to 600MB/s bandwidth)."
Burst rate stats are completely worthless. With SATA each drive gets dedicated bandwidth meaning the theoretical max is so far beyond the capability of current drives it shouldn't even be considered a feature. At the present rate that IDE HD STR is increasing it will be some time in 2004-05 that IDE drives will surpass ATA133. I don't consider improving on a standard that won't be surpassed for over 2 years of much importance. Also, most people will probably add SATA compatibility initially via PCI card which means the 150MB/s theoretical gets cut down to 133MB/s by the PCI bus. So it doesn't have any performance advantage at all in that scenario.
"Pin count
Parallel - 80 wires per cable (26 per channel + power/ground), effectively 5cm wide
Serial - 7 wires per cable (1 channel), effectively 8mm wide"
The smaller cables are all the interface currently has going for it, and is the only thing that people familiar with the current state of storage technology would care about.
"On the other hand, Parallel ATA, in its current elderly state (around 20 yrs old)"
The original ATA spec was not submitted for ANSI approval until 1990. And it wasn't until 1994 that the inital spec was given approval. So at oldest the concept of P-ATA is 12 years old with the actual initial standard only 8 years old. SCSI is considerably older, with its initial spec approved in 1986.
"Current Parallel ATA requirements can only allow cables up to 18 inches in length, per channel."
That's what the spec says, but you can buy 24in and longer cables that will work fine.
"But until Serial ATA hard drives start showing up on the market (no more than 2 months from now), all Serial ATA is going to be is a smaller cable..."
Won't make the slightest bit of difference. The next generation of HD's will released in PATA and SATA formats and there won't be a bit of difference between the performance of the 2.
"Because each cable for Serial ATA is only a single channel 'solution', it currently requires 1 controller (chip) per channel. "
No No NO This is incorrect. One controller (chip) multiple devices, it's up to the manufacturers to decide which particular controller (chip) they implement (and how many ports it can handle).
Directly from the WhitePaper (Implementing SerialATA Technology):
"The Serial ATA interface replaces today?s 80 pin ribbon cable with a 4 conductor cable. Rather than sending out data in parallel, the data from the controller is serialized, and sent out as a differential signal pair to the target disk device. The disk also sends data on a differential pair back to the host controller. Simultaneous transmission occurs on both channels, from host to disk and disk to host. Because of this, Serial ATA is a point-to-point link, and only supports a single device per controller interface, in contrast to the primary and secondary support of two drives from a single parallel ATA port. Controllers can address multiple devices, but each device requires a separate, dedicated port."
Read this thread and/or visit The SerialATA Working Group for more info.
Thorin
No No NO This is incorrect. One controller (chip) multiple devices, it's up to the manufacturers to decide which particular controller (chip) they implement (and how many ports it can handle).
Directly from the WhitePaper (Implementing SerialATA Technology):
"The Serial ATA interface replaces today?s 80 pin ribbon cable with a 4 conductor cable. Rather than sending out data in parallel, the data from the controller is serialized, and sent out as a differential signal pair to the target disk device. The disk also sends data on a differential pair back to the host controller. Simultaneous transmission occurs on both channels, from host to disk and disk to host. Because of this, Serial ATA is a point-to-point link, and only supports a single device per controller interface, in contrast to the primary and secondary support of two drives from a single parallel ATA port. Controllers can address multiple devices, but each device requires a separate, dedicated port."
Read this thread and/or visit The SerialATA Working Group for more info.
Thorin
"No No NO This is incorrect. One controller (chip) multiple devices, it's up to the manufacturers to decide which particular controller (chip) they implement (and how many ports it can handle)."
Like I said:
Because each cable for Serial ATA is only a single channel cable (solution, if you will), it currently requires 1 controller per channel.
But now that I reread that (while not intoxicated).. it's not how it was meant to be said.. or something like that. Let's try something along these lines:
Because the only current on board (read: on mobo) controller is that Marvel chip, it is currently limited to 1 device per controller. Even if other Serial ATA controllers are capable of much more than that, that's all that is being manufactured currently.
You just jealous, cuz I tell it how it iz and you tell it how it might be...
/end Ghetto Thug Puffy script
-Tarmax
PS. I apologize if it makes little or no sense.. 4:30 at time of writing, and I'm dozin quick =)
Like I said:
Because each cable for Serial ATA is only a single channel cable (solution, if you will), it currently requires 1 controller per channel.
But now that I reread that (while not intoxicated).. it's not how it was meant to be said.. or something like that. Let's try something along these lines:
Because the only current on board (read: on mobo) controller is that Marvel chip, it is currently limited to 1 device per controller. Even if other Serial ATA controllers are capable of much more than that, that's all that is being manufactured currently.
You just jealous, cuz I tell it how it iz and you tell it how it might be...
/end Ghetto Thug Puffy script
-Tarmax
PS. I apologize if it makes little or no sense.. 4:30 at time of writing, and I'm dozin quick =)
"Because the only current on board (read: on mobo) controller is that Marvel chip, it is currently limited to 1 device per controller. Even if other Serial ATA controllers are capable of much more than that, that's all that is being manufactured currently. "
Actually I'm sorry but I've found out that that is also incorrect (not that the truth is any better). As you can see in this picture the RocketRaid 1520 has HighPoints usual ATA133 RAID controller (HPT372A) and two little Marvell chips however these aren't SerialATA controllers they're SerialATA to Parallel converter chips. So the Highpoint 1520 isn't actually a SerialATA solution it is currently a ParallelATA solution with SerialATA support***. (This would also apply to the motherboards you're talking about).
Check here for more info:
RocketRaid 1520 Review
(***See page 5 of the review for information on the PATA-SATA-PATA conversion latency etc...)
Thorin
Actually I'm sorry but I've found out that that is also incorrect (not that the truth is any better). As you can see in this picture the RocketRaid 1520 has HighPoints usual ATA133 RAID controller (HPT372A) and two little Marvell chips however these aren't SerialATA controllers they're SerialATA to Parallel converter chips. So the Highpoint 1520 isn't actually a SerialATA solution it is currently a ParallelATA solution with SerialATA support***. (This would also apply to the motherboards you're talking about).
Check here for more info:
RocketRaid 1520 Review
(***See page 5 of the review for information on the PATA-SATA-PATA conversion latency etc...)
Thorin
"Actually I'm sorry but I've found out that that is also incorrect (not that the truth is any better)."
Ehh.. c'est la vie.
Oh well. 2 more months
Ehh.. c'est la vie.
Oh well. 2 more months
>I don't know how much farther we can go with rotating medium. IMO more research needs to be put into increasing >static forms of memory (like flash) as a replacement for main storage. Increasing the speed of the platters lead to >greater power requirements and increased heat. Increasing heat means expanding platters which become harder to >compensate for as platter density increases.
There was a point a while back when non-rotating storage, in the form of magnetic bubble memory, seemed to have potential to compete with the rotating storage, but that didn't last long. Rotating storage so far outclasses anything else in cost per bit it is pathethetic.
They say the end is not in sight for HDs just using current technology. Rotating the disks faster is not needed. The amount of storage on a platter usually goes down with higher rpm. What they do is increase the bit density by reducing the strength of the magnetic domains, then increase the signal strength by making the platter flatter and flying the head closer to the surface.
But research in any kind of feasible storage has money being poured into it. Practically any form has a use somewhere somehow.
There was a point a while back when non-rotating storage, in the form of magnetic bubble memory, seemed to have potential to compete with the rotating storage, but that didn't last long. Rotating storage so far outclasses anything else in cost per bit it is pathethetic.
They say the end is not in sight for HDs just using current technology. Rotating the disks faster is not needed. The amount of storage on a platter usually goes down with higher rpm. What they do is increase the bit density by reducing the strength of the magnetic domains, then increase the signal strength by making the platter flatter and flying the head closer to the surface.
But research in any kind of feasible storage has money being poured into it. Practically any form has a use somewhere somehow.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter