Serial ATA II

[godmare]

The Inquirer

THE FIRST Serial ATA devices are only just starting to make themselves known. Only a very brave motherboard manufacturer would dare launch a product with SATA as the only drive interface. Which makes it a bit of a surprise that Marvell has already started sampling SATA II chips.
The new SATA II bridge chips as Marvell calls them pump data back and forth at 3gigabits pre second. That's twice the speed of the current SATA devices and chips. It should be noted that, while the speed increase is undoubtedly a good thing, there aren't any hard drives that come even close to delivering that kind of constant throughput.

Marvell claims it was the first to market with a SATA I device in 2001. The company makes interfaces for both sides of the cable, motherboard/card based host controllers and drive-side.

 

[Adul]

drool

how many WD raptor drives would we chain up in raid 0

 

[Stealth1024]

lol @ Adul <---- you try it then give them to me for "further testing"

 

[Lord Evermore]

They don't specify how many ports the chips actually support. So if it only supports 2, well then it doesn't matter that they're faster. There's really not any sense to even having more bandwidth since the 150MBps bandwidth is already per-port, so having 300MBps per port means nothing. Not until the "hub" design is available that lets you connect multiple drives to a single port.

 

[Ilmater]

Marvell claims it was the first to market with a SATA I device in 2001.

Well, look at their past track record. Considering that they brought the first SATA device in 2001, but they didn't hit the market in any decent volume until 2003, if they're going to have the first SATA II device in 2003, then we should see this some time in 2005, which is nothing special.

 

[Lonyo]

Originally posted by: Lord Evermore
They don't specify how many ports the chips actually support. So if it only supports 2, well then it doesn't matter that they're faster. There's really not any sense to even having more bandwidth since the 150MBps bandwidth is already per-port, so having 300MBps per port means nothing. Not until the "hub" design is available that lets you connect multiple drives to a single port.

I think it's the daisy-chain idea that would be the most obvious thing for needing to increase the throughput of SATA controllers, yeah, so 1 drive might not hit even 70mbps, let alone 300, but if you could link 4 drives, each at 60 or so, then that's 240, making 300 something actually useful.
Although whether that will actually happen is another thing totally.

 

[Lord Evermore]

SerialATA is really of very limited use until the daisy-chain/hub is available. Only 2 native ports in any chipset, even the next versions from all the makers, so only two devices without using an add-in controller which is then limited by the PCI bus, unless we start to see PCI-X or PCI-Express before the hub design or more than 2 native ports.

 

[bgeh]

the most interesting part about SATA to me is the ability to do hot-swapping
but can this chip do that?

 

[thorin]

Originally posted by: Lord Evermore
SerialATA is really of very limited use until the daisy-chain/hub is available. Only 2 native ports in any chipset, even the next versions from all the makers, so only two devices without using an add-in controller which is then limited by the PCI bus, unless we start to see PCI-X or PCI-Express before the hub design or more than 2 native ports.

I don't think I agree with that statement. With today's HD capacities SATA is just fine (do you need more then 400 to 600GB of storage on a home system?). Hell even if you had a Raptor to boot from a then a 200GB or 300GB storage drive that's way more then "most" people need. If you need TOP of the line performance then RAID two Raptors as your boot drive and then put a high performance PATA drive on your 100/133 channels (or again even raid them for excellent storage drive performance).

Also lets keep in mind that 2 channels/ports/drives isn't a limit of SATA but rather a limit of the controllers the manufacturers are choosing to implement in their products.

Thorin

 

[Lord Evermore]

I know it's just a limit of the implementation, I was just pointing out that that IS how it's being implemented for now and for any announced products or rumors.

For the "average" person, yes two ports is enough for a hard drive and a CDROM (if anybody ever DOES bother to announce native SATA CDROMs or even ones with a bridge-chip integrated), but for that sort of person, parallel ATA works just fine as well, they don't care about the added airflow capacity and smaller cables, or that the bandwidth per-port is higher than PATA. Until native optical devices are available of course, even two ports will be enough for most enthusiasts to use their hard drives (I don't need half a terabyte of storage, but apparently a lot of people here think they do). But just putting in SATA hard drives and then having parallel optical drives just seems like half-assing it. Why bother until everything can be moved to SATA? By the time SATA optical devices are out of course, maybe chipsets will have more than 2 ports integrated, or adapters will get cheaper so they can be used on add-in ports since they're not as bandwidth sensitive. But there's no news of anything like that happening.

 

[thorin]

Originally posted by: Lord Evermore
I know it's just a limit of the implementation, I was just pointing out that that IS how it's being implemented for now and for any announced products or rumors.

That wasn't aimed at you directly but more generally toward the entire conversation/thread.

For the "average" person, yes two ports is enough for a hard drive and a CDROM (if anybody ever DOES bother to announce native SATA CDROMs or even ones with a bridge-chip integrated)

I predict they'll start coming out early next year (maybe even Q4 if we're lucky).

but for that sort of person, parallel ATA works just fine as well, they don't care about the added airflow capacity and smaller cables, or that the bandwidth per-port is higher than PATA.

Agreed.

Until native optical devices are available of course, even two ports will be enough for most enthusiasts to use their hard drives (I don't need half a terabyte of storage, but apparently a lot of people here think they do).

Again I agree, and even if they do need half a TB they can have it since there are 200, 250, and 300GB drives becoming more and more popular (even SATA).

But just putting in SATA hard drives and then having parallel optical drives just seems like half-assing it.

Again I agree ... manuf's should have been faster with their adoption, but it is coming along nicely and gaining speed.

Why bother until everything can be moved to SATA?

Well I don't know if I agree but alot of people cite ease of use as the main reason for adoption (small cables, long cables, no jumpers, etc ....)

By the time SATA optical devices are out of course, maybe chipsets will have more than 2 ports integrated, or adapters will get cheaper so they can be used on add-in ports since they're not as bandwidth sensitive.

I don't think that's beyond the realm of believability.

But there's no news of anything like that happening.

Good things to those who wait? I dunno what to say here, I haven't read anything specific for you but this does seem like the natural progression of the technology (ie: more/cheaper).

Thorin