Sata vs. Esata

[wakkoguy]

I have googled this and couldn't find an answer. Is internal Sata the same speed as eSata?

 

[groberts101]

For an HDD?.. yes about the same dependant on the bridge chip used in the external enclosure.

For SSD?(especially 6G variations).. internal ports running off the native sata chip will be much faster.

So, it really depends on what sata chip the different ports run off of. Most boards have 3rd party chips running the eSATA whereas the native Intel/AMD chips internal ports will be faster.

It can be so pronounced at times that even a native 3G port can actually be faster overall than a 3rd party 6G sata chip. Like Marvell and ASMedia which are typically used most often for eSATA these days.

 

[Elixer]

eSATA enclosures don't need a bridge chip, just a pass-through.

But yeah, eSATA should be just as fast as the internal ports, with that caveat that groberts says, some MB makers use a 3rd party chip, and that might be slightly slower than native, but you won't really notice it that much. I haven't seen anything that is lower than about 1Mb/sec from native.

 

[taltamir]

For an HDD?.. yes about the same dependant on the bridge chip used in the external enclosure.

For SSD?(especially 6G variations).. internal ports running off the native sata chip will be much faster.

eSATA can run off of the native SATA. And in many cases does.
Check your inividual mobo to see if it uses an additional chip for that (sub par) or the intel's built in SATA controller (best in class). AMD SATA controller is a little better then your average 3rd party chip but is still no match for intel's.

That being said, its not THAT much slower. Just noticeably lower on benchmarks.

In terms of actual functionality, there is no difference between SATA and eSATA except for eSATA having a "eject drive" button appear in windows.

 

[corkyg]

... In terms of actual functionality, there is no difference between SATA and eSATA except for eSATA having a "eject drive" button appear in windows.

One major usage difference for me is that eSATA can be configured for hot plugging by turning off write caching. I use it for accessing backup externals on 2 machines, one a laptop. IOW, for eSATA to really function it needs to be connected to a genuine eSATA port. If connected to a regular SATA port on the mobo, it will be seen as an internal drive with respect to write caching (grayed out.)

 

[taltamir]

One major usage difference for me is that eSATA can be configured for hot plugging by turning off write caching. I use it for accessing backup externals on 2 machines, one a laptop. IOW, for eSATA to really function it needs to be connected to a genuine eSATA port. If connected to a regular SATA port on the mobo, it will be seen as an internal drive with respect to write caching (grayed out.)

I am pretty sure that is an OS thing or a mobo firmware thing rather then a SATA standard thing...

That being said, it IS a thing that you will encounter in real life usage so thank you for pointing this out.

BTW, you can hotplug internal SATA ports... if the controller elects to implement hotplug (it is optional)... I believe this "elected not to implement hotplug to save some dough" is part of the reason for the issue you described. It was stupid to make it an optional feature rather then mandatory.

 

[wakkoguy]

Awesome, great info thank you all very much!

 

[Cerb]

For an HDD?.. yes about the same dependant on the bridge chip used in the external enclosure.

For SSD?(especially 6G variations).. internal ports running off the native sata chip will be much faster.

What about eSATA running off whatever port you want? My mobo allows that, and I don't intend to replace it with one that doesn't (all eSATA requires is tighter voltage and current tolerances, since every mobo controller supports it).

One major usage difference for me is that eSATA can be configured for hot plugging by turning off write caching. I use it for accessing backup externals on 2 machines, one a laptop. IOW, for eSATA to really function it needs to be connected to a genuine eSATA port. If connected to a regular SATA port on the mobo, it will be seen as an internal drive with respect to write caching (grayed out.)

If you use the safely remove hardware feature, you can have your cake and eat it, too 🙂. My eSATAs are hooked up to my internal Intel ports, running AHCI, using MSAHCI. Windows doesn't know or care about the 'e', and write caching can be toggled per drive.

 

[Old Hippie]

What about eSATA running off whatever port you want? My mobo allows that,

You can run eSATA from any port with the correct conversion cables. It's simply an outside SATA connection.

AFAIK there's no "tighter voltage and current tolerances" but just a little more shielding on the external cables.

 

[groberts101]

It's true that an eSATA enclosures interface itself won't have bearing on the speeds supported since it should just be cabling (no bridge with controller) for passing through.

But, while an eSATA enclosure doesn't "need" a bridge chip.. many include them as gateways for the other interfaces included on the enclosure and they are certainly there if the enclosure supports multiple drives. In fact, raid capable enclosures can actually even slow you down with single drives compared to the native sata chips speeds because they create the next weakest link now that they are included in the interface chain. But I do agree that it's rarely an issue with slower single HDD as they just don't tax most solutions hard enough to bring out the weakness. I have definitely lost far more than 1MB/s on some enclosures though and it's good to do your homework because the way the thing is config'd and what interface/hardware it leverages internally can make a difference. Not earth shattering.. but you may regret some purchases when compared to other available options.

And while Intel and AMD can and do support the eSATA capability.. motherboard mfgrs will rarely force you to share bandwidth or allow reduction in the amount of internal ports that can be populated while an eSATA device is connected. That's why they use 3rd party chips to separate the I/O and keep all the fastest native sata ports available. Many sata chips support port multiplier capability by allowing you to attach multiple drives per port, but you do so at the cost of channel throughput. I bet it's very tough to find even 2 current mobo's these days that would leverage the native sata chip for eSATA duty. The primary chips used for additional internal/external sata ports today are the Marvell, ASMedia, JMicron and Silicone Image chips. Would probably have to be a damned cheap board if eSATA were configured to hog any native sata channels. And of course, if it were that cheap?.. it probably wouldn't have an included eSATA port anyways.

Heck, even the 3rd party chips have limitations in that you cannot simultaneously run 2 devices via the eSATA ports while leveraging the included internal sata ports that run to that same chip. They are interconnected with a max channel/port capability and when you tie up a channel inside or out?.. it eliminates another port internally or externally in the process. Most mfgrs realize that many users do not want to trade ports/hog bandwidth like that and therefore implement 3rd party chips for that very reason.

Anywho.. that's far more than I intended to share on the subject and I'll just shut up and leave it to the experts now. lol

 

[Cerb]

And while Intel and AMD can and do support the eSATA capability.. motherboard mfgrs will rarely force you to share bandwidth or allow reduction in the amount of internal ports that can be populated while an eSATA device is connected. That's why they use 3rd party chips to separate the I/O and keep all the fastest native sata ports available.

Most with eSATA are the other way around, that I've seen: add-on chip internal, dedicated eSATA port from AMD or Intel.

AM3+, 1xeSATA: http://www.newegg.com/Product/Produc...=1&srchInDesc=

1155, 1xeSATA: http://www.newegg.com/Product/Produc...=1&srchInDesc=

 

[groberts101]

yeah.. but if you dig deeper into those spec's you'll see that they drop 2 Intel 3G internal ports out of the mix to also run that eSATA off the native chip. Back to the bandwidth hogging we go. This one even appears to rob an internal Intel 3G port for use as a mini-SATA port. Can't seem to have our cake and eat it too.

From the ASUS Maximus GENE
Intel® Z77 chipset : *4
2 x SATA 6Gb/s port(s), red
2 x SATA 3Gb/s port(s), black
1 x eSATA 3Gb/s port(s), red
1 x mini-SATA 3Gb/s port(s), black

All you need to do is look at the mobo mfgrs site to see how they are typically config'ing eSATA these days and most will not kill internal ports to do it. 3rd party chips dominate the eSATA configs for many midrange on up mobo's. And it's mandatory to have a 3rd party chip when we get into the eSATA 6G spec's.

Although I should have included an escape clause earlier for my lack of knowledge about AMD's config's. Maybe I should have used more "probably".. and "some".. or "many" in there too as there is certainly much lost in translation/interpretation of the written word. lol

PS.. I was rather surprised at how expensive some of those lopped up mobo's are. I'd be spending the extra 40 bucks to get more ports and connectivity. To each his won though, I guess.

 

[corkyg]

If you use the safely remove hardware feature, you can have your cake and eat it, too 🙂. My eSATAs are hooked up to my internal Intel ports, running AHCI, using MSAHCI. Windows doesn't know or care about the 'e', and write caching can be toggled per drive.

And that is exactly what I don't want to have to do. It is another layer and process. The way I have it set up is instant on and off, any time I want, without being bothered by that Safely Remove Hardware layer. With write caching off, it is optimized for quick removal. Been doing it that way for a few years now - and what started in XP is now done by 7. If I connect to an internal SATA port, then that layer is added. That is not having my cake and eating it too - it is a process delay. And, I am not running AHCI.

 

[Cerb]

PS.. I was rather surprised at how expensive some of those lopped up mobo's are. I'd be spending the extra 40 bucks to get more ports and connectivity. To each his won though, I guess.

Depends on what you want/need. When you're spending $800+ on a new PC, the extra $30-60 doesn't look as bad as just comparing mobos.

 

[Elixer]

I am pretty sure that is an OS thing or a mobo firmware thing rather then a SATA standard thing...

It is a BIOS (most have enable/disable hotplug) + driver (must be AHCI) + OS (must support hotplug events) thing. AFAIK, it is in the SATA I specs that they must support hotplug.

And that is exactly what I don't want to have to do. It is another layer and process. The way I have it set up is instant on and off, any time I want, without being bothered by that Safely Remove Hardware layer. With write caching off, it is optimized for quick removal. Been doing it that way for a few years now - and what started in XP is now done by 7. If I connect to an internal SATA port, then that layer is added. That is not having my cake and eating it too - it is a process delay. And, I am not running AHCI.

When you do the 'Safely Remove...', it also powers down the device.
With just doing the write caching off, it doesn't power anything down, and doesn't seem to be as safe a way to do it than using the official way. I also think AHCI is required for hot-plugging.

 

[Old Hippie]

Anywho.. that's far more than I intended to share on the subject and I'll just shut up

LOL!

I knew it was too good to be true.

That slience lasted @ 39 mins. LOL!

 

[corkyg]

...When you do the 'Safely Remove...', it also powers down the device. With just doing the write caching off, it doesn't power anything down, and doesn't seem to be as safe a way to do it than using the official way. I also think AHCI is required for hot-plugging.

1. The eSATA is optimized for uick Removal by Windows (Dev Mgr, Disk Policies.)

2. It is turned on and off readily any time I choose using a power box switch. I believe it is powered down when off. 🙂

3. AHCI is not required to do this in XP Pro/SP3 or Win 7 Ultimate X86. I know this to be true because I have done it every day for more than 3 years.

 

[groberts101]

LOL!

I knew it was too good to be true.

That slience lasted @ 39 mins. LOL!

LOL.. yeah.. I made the mistake of leaving that browser tab buried and then just got too curious again and had to F5 before I logged off.

You should be blaming Cerb for always trying to engage the conversation and "learn-n-teach" stuff.. about stuff.

Corky and Elixer's dialogue kind of pull folks in too. 🙂

 

[Old Hippie]

You should be blaming Cerb for always trying to engage the conversation and "learn-n-teach" stuff.. about stuff.

Corky and Elixer's dialogue kind of pull folks in too.

LOL!

I knew someone made you do it! 😀