World?s First Board with SerialATA Support from Iwill

[McCarthy]

Because people will mistake it for gigabytes per second and in that instance it sounds really incredibly super groovy fast.



Wish power was on the same cable too.

--Mc

 

[SexyK]

Originally posted by: Pariah
Booster, there are no SerialATA connectors on the board pictured. The standard P4E is probably the board pictured, the P4ES is the version with SerialATA.

I think the SerialATA connectors are below the ATX power connector in that picture. Little green and white things. Could be wrong though.

Kramer

 

[thorin]

"They would have greatly improved the spec if it allowed multiple devices with multiple access at the same time on the bus, similar to SCSI. "

With SATA you can access more then one drive (with full bandwidth) at the same time, it's completely point-to-point. (Each 'channel' is a 'bus' so to speak).

"One thing I'm not understanding is why the spec calls 1.5Gbps equivalent to 150MBps. 1.5Gbps is actually about 187MBps. If they're using a divider of 10 rather than the divider of 8 for a byte, then they're being misleading (and oddly, not in their own favor). If they're factoring in some sort of overhead, well, that's just stupid, because seeing the two rates beside each other makes it look like they're just converting the numbers into a more recognizable form and using bad math, rather than showing the "actual" throughput available. If there is some sort of overhead and they want to show actual throughput, then it should be 1.2Gbps and 150MBps. "

They use 10B encoding (8bits data with 2bits for error correction).

"I think the SerialATA connectors are below the ATX power connector in that picture. Little green and white things. Could be wrong though."

I had been thinking that as well however they also look like USB header connections.

Thorin

 

[Eug]

Lord Evermore wrote:

One thing I haven't understood is why SerialATA was focused on internal devices. As far as I can see, it'd be nearly ideal for external hard drives. No need for an adapter card to be built into an external enclosure using a native SerialATA drive, no need for an external power supply. Firewire can provide the same functions, but you need an adapter card which increases costs, and the throughput is limited. With an external SerialATA drive you'd get exactly the same performance with an external drive as an internal. The only downside is the lack of daisy-chaining, but just sticking two external ports into a system would be more than enough for most people to run a hard drive for backup or a CDRW drive, and anybody that needs way more than that is probably using SCSI anyway.

I like the idea in some ways of external SerialATA for external drives, but there are some limitations, some you've mentioned. 1) There is no provision for daisychaining. 2) There is no power. 3) Limitations in the number of devices.

#1 is a problem in general, and #2 is a problem for laptops when you want to use small laptop hard drives or zip drives and what not - you'll have to carry a power supply around with you for each of the drives. #3 is a huge problem in my book. I have 6 external drives/readers, and 4 of them are Firewire (and I have 2 USB). I use four of them with my laptop (3 Firewire/1 USB). Two of the ones (1 Firewire/1 USB) I use with my laptop require power over the cable to function, and I'm glad, because I'd hate to haul around 2 extra AC adaptors in my briefcase. Also, the 4 Firewire devices when used on my desktop were daisychained until recently, when I finally got a hub for it.

This brings up another question... Is it possible to use some sort of hub-like device with SerialATA? As far as I can tell, no traditional hub will exist, but I could be mistaken. Thus it means either getting another internal card with a bazillion more ports at the back, or being limited to just a couple of external devices.

Thus overall, it seems to me SerialATA is not appropriate for external devices in general, but may suit some people.

P.S. It seems that most people using multiple drives are using Firewire these days, not SCSI. If anything the numbers using Firewire and USB 2 will continue to increase, at the expense of SCSI. For the record I find external SCSI a pain in the @ss. Multiple cable and port types, termination, IDs, etc. It's no surprise that these days so many external higher-bandwidth peripherals are not going SCSI.

 

[Pariah]

SATA headers don't have any pins. Those are USB headers.

The connector on the right is the SATA headers, which would be quite difficult to pick out on that MB pic.
http://watch.impress.co.jp/pc/docs/2002/0514/3ware_11.jpg

 

[Lord Evermore]

thorin: if they use 10-bit encoding, then they're doing what I think thought they were doing, indicating the raw transfer rate in bits per second, then the "actual" throughput in MBps, which is confusing at least to me. If they're going to list the two, then they should be labelled as being different specified rates, not just two numbers listed together. Generally when you list a rate in bits per second then enclose the MBps in parentheses it indicates that it's just a unit conversion, not that it's an entirely different measurement as well as unit conversion. This is why modem manufacturers only list the raw 56k data rate, not also listing the actual throughput which most people divide by 10 to get.

 

[thorin]

Well now that you think thought you know what they're talking about it isn't confusing is it?

Personally I think it makes complete sense but I've been following SATA since the original rumours started about it, so it just seems natural 1.5Gbps/10 = 150MBps <shrug>

Also I don't think this is the SerialATA working group's doing but rather that of the journalist(s).

Anyway if you really are "... 92% parsimonious. " then you probably won't bother with the drives/technology when they come out anyway, so stop worrying so much

Thorin

 

[Lord Evermore]

The definition of parsimonious in my signature is in reference to a "quiz" rating, where it determined basically how I apply my morality (but not whether or not I'm moral compared to anyone else). It pretty much means my morals apply in nearly all situations, regardless of geographical distance, ethnicity, gender, age, et cetera. Dictionary.com has two definitions for parsimony, the second of which is "Adoption of the simplest assumption in the formulation of a theory or in the interpretation of data", which would be what applies here. Having to interpret ratings in a specification based on an unspecified assumption rather than the true definition is beyond my abilities. I honestly don't see the point of listing two different ratings like that without defining what they mean. Giving the raw transfer rate is in fact entirely useless if there is ALWAYS a 2bit reduction of the actual throughput to the user data-rate. If you can't actually use the extra two bits, then they shouldn't be listed as part of the transfer rate except if it's listed specifically as the total throughput disregarding overhead.

Ethernet and DSL and cable providers don't list transfer rates as well as "actual" rates, nor does any other computer component that I've ever seen. You lose 20% sometimes to overhead with a DSL provider, but they don't list a 1.5Mbps line as "1.5Mbps/120KBps" (where it would be theoretically be 187KBps throughput before overhead). Same thing with Ethernet. They list the actual throughput disregarding overhead. If there is any overhead on ATA33/66/100/133, it isn't listed with actual throughput AND user-throughput, you just get one number and divide by 8 if you need to convert bits to bytes or multiply by 8 for bytes to bits.