Faster USB 3.0 Performance: Examining UASP And Turbo Mode

[Zingam_Duo]

Yey! Finally a professional article on this site just like the old days!

I am totally tired of Jane's reporting on cool iCrApps, that teach you how to wipe your ass!

 

[itarticle]

thanks for the post. Is have any price changes?
by:- http://www.itarticle.net

 

[voyager1]

I added a NEC/Renasas PCI USB3 card to allow me to use a Thermaltake BlacX5G to make large 100GB+ transfers from my internal SATA HDDs to external SATA HDDs for archival storage.
The transfers take place in the 75 to 90MB/s range. Almost exactly the same rate I get between my internal SATA HDDs.
A huge improvement over eSATA where I would get corruption during the large GB transfers and slow USB2 performance.
My assumption is that the holdup is due to the performance limitations of the disk drives.
Until something better comes along I'm happy.

 

[weegidy]

[citation][nom]forestie[/nom][citation][nom]jimmysmitty[/nom]One thing with USB is that it was never designed for massive large data thoroghputs like eHDDs and larger flash drives.[/citation]

But USB 3.0 IS designed for peripherals such as large HDD. For example, I have a 3tb USB Raid 0 box I setup, and it has USB 3.0, as well as eSATA. But I get better transfer rates from the USB3.0.

USB 3.0 was designed with multiple upload and download connections in mind,

 

[palladin9479]

I bought one of these esata adapters a while back

http://i.imgur.com/RN3nu.jpg

but instead of the standard connections, it simply has one end as esata and the other end the normal sata used for internal drives. and it uses a normal sata power cable so you can take a regular hard drive without an enclosure and just connect it to the PC.

I have it installed on my work system and at home, so I can take video footage, then put it on my 90GB SSD and edit directly on the drive and get the full sata 6gbit speeds with sequential reads of nearly 500MB/s

For other devices that I have tried that actually use USB 3, (a few new DSLR's and external drives) the speed never comes close to even 200MB/s so a faster controller wont really help much with those devices)

I feel that esata has some serious potential if they could improve the connector or at least use the same design as USB but change the shape of the plastic thing in the connector to keep actual USB devices from connecting to it.

SATA and eSATA and the same protocol, from the computer's point of view there is nothing difference between them. The only real difference is that eSATA is speced to require shielding, supports higher voltage and longer cables. That's why the cable was different, it has space and connections for shielding.

 

[Guest]

"There is no word on the status of UAS support in OS X."

AnandTech's review of the retina MBP says that it supports UAS. (He doesn't say how he knows this, but he stated it pretty confidently, so perhaps a friend within Apple told him?)

 

[Cyberat_88]

USB never performed for HDDs from get go and all performance claims by vendors were gimmicks. To Joyfully add more to the failure of USB3 and the cheating business hype of performance speed, Benchmarked by HD Tune on my Acer Laptop directly plugged in to the port:
1) SeaGate FA GoFlex Desk 2TB - Average Transfer rate 43.5 MB/sec., Burst rate 26.4 MB/sec.. (eSATA native docked to a powered USB3 converter by SeaGate).
Advertised Speed - 625 MB/sec.. - This is where the huge gap and big scam exists.
2) Hitachi HDP725050GLA360 500MB - Average Transfer rate 25.7 MB/sec., Burst rate 18.9 MB/sec.. (IDE drive in a USB2.0/eSata enclosure connected via USB 2)
Advertised Speed - 60 MB/sec..
3) SeaGate ST3400633A 400GB - Average Transfer rate 25.8 MB/sec., Burst rate 19.1 MB/sec.. (IDE drive in USB2.0 enclosure).
Advertised Speed - 60 MB/sec..
When all is said and done, reasons for lower speed are but EXCUSES, we need to stop buying USB products until the business comes clean.

 

[Cyberat_88]

IDE drives installed in a P4 desktop were performing at an average of 42 MB/sec., surely you're not blind to the lack of speed of new HDD technology and the numbers written on a box no longer have a meaning. HDDs are the MAIN bottleneck of today's computers and it's wall to wall corporate ripoff for the nerd wannabe geek and the pseudo scientist not bothered with facts or actual performance tests. SELL OUTS !

 

[Crashman]

[citation][nom]Cyberat_88[/nom]When all is said and done, reasons for lower speed are but EXCUSES, we need to stop buying USB products until the business comes clean.[/citation]Tossing out all my thumb drives now, that's for the advice!

 

[Cyberat_88]

I said STOP buying USB technology, not throw out what you already spent money on.

 

[palladin9479]

@Cyber ...

Really don't get what your trying to say there. Those numbers are just max bandwidth per channel not your actual transfer speed. You have encoding to worry about, then latency, then protocol overhead, then finally system I/O overhead, all conspiring to lower your transfer speeds. USB protocol was never designed for bulk transfers, down the years they've created a few enhanced transfer modes, and that's helped somewhat, but ultimately the protocol as envisioned was designed for hot-plunging peripherals. It acts as a all-in-one connector / protocol / stack for Keyboards / Mice / Printers / Fax machines / scanners / camera's and other generic peripherals, to achieve that ubiquity meant they couldn't specialize in fast efficient bulk data transfers. No false advertising there, it does exactly what it supposed to.

 

[Cyberat_88]

I am aware of the procedure the data goes through and it makes no difference to me as an IT professional, I couldn't care less why. The main subject I am trying to convey is that those averages ARE the ACTUAL transfer speeds, files do not move any faster, I've had movies transferred and large numbers of small files. There IS false advertising, it's called 5 megabits per second or 625 megabytes per second, THAT is your FALSE advertising, why are you trying to cover that up and protect bad business practices. I've never denied the use for the all-in-one connector, just the claim to great speeds. I do not want to see a more direct and fast external connection such as eSATA dumped by unscrupulous businessmen, pseudo-geeks and other people with interests vested in failed devices. USB3 is useless, you do not need more speed for kb/mouse/printer/scanner than already exists in USB2. The proof is in the numbers, not excuses or beliefs. Deliver the 625MB/sec. ACTUAL speed or go back to the drawing board.

 

[Cyberat_88]

5 gigabits per second that is and yes it is advertised for "fastest" external connector for bulk transfer and being sold as such to people of less knowledge as you can see above. Few MB manufacturers from Z68 to X79 embed eSATA in their boards because of the USB2/3 myths of speed.

 

[Guest]

I just found this article so sorry for the late comment.
I wold like to remind everyone about something really great and request that it be added to this review in an update.
http://www.delock.de/produkte/G_61862/merkmale.html?setLanguage=en

This is a front panel or bracket connector that allows you to use not only USB 2.0 and eSATAp but also USB 3.0. (they also offer the receptacle itself for soldering to motherboards.
I have a strong opinion that for consumer non-video, non-network, non-audio connectivity this is the ideal host port.
Unfortunately manufacturers don't seem to care.

Also I request you test this
http://www.vantecusa.com/gl/product/view_detail/471
It is the only true SATA 3.0 6Gbps to USB 3.0 adapter I know of and it would be interesting to see if it makes any difference.
According to http://www.cnet.com.au/can-we-stop-lying-about-numbers-now-339327230.htm
it shouldn't.

 

[Cyberat_88]

I have a vantec nexstar usb 2.0, it's in the same slow speed bracket.
Our only weapon to fight greed and nerds combined is to return products to the store based on under performance or non-performing the speed advertised.

 

[Kyle Sebion]

The ASM1051 controller does not have UAS support. The ASM1051E, on the Thermaltake Blacx 5G, and ASM1051U do. I have a device that uses the ASM1051 controller and have experienced the BOT bottleneck.

ASM1051: http://www.asmedia.com.tw/eng/e_show_products.php?cate_index=97&item=90

 

[PTNLemay]

I'm worried, I got a new external hard drive and it's download speeds are holding steady at 5 to 10 MB/sec. And yes, I'm sure I'm using one of the blue USB 3.0 ports.

I've tried using Asus's USB 3.0 Turbo Boost but I don't think it's actually done anything.

 

[Diceman2037]

Article is WRONG

054C00C1 is a pre-existing reg key that is predefined to work around issues in a particular usb storage device

For this change to function, you must CREATE a subkey for the device you intend it to be used with by converting the VIDPID (in VVVVPPPP) of the device into hex (and adding 0's to the front to make the key 8 characters)
for example, for my kingston DT R500, the keypath would be

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\usbstor\009122e6

you would then enter the MaximumTransferLength dword under this key.

 

[Diceman2037]

Sorry, the VID and PID must be converted seperately, the correct path would be

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\usbstor\03B70676

 

[Diceman2037]

actually according to

http://www.sevenforums.com/news/180768-boost-windows-7-sp1-usb-storage-devices-performance-increasing-maxi-5.html#post2303593

you just combine the VID and PID obtained from the device manager (or tool like usbdeview) as it is already in hex. and use this as the usbstor subkey.

 

[nononono4t67]

could have taken screenshots once the estimated time remaining was calculated..

 

[jmichae3]

according to http://www.asmedia.com.tw/eng/e_show_products.php?cate_index=98&item=143 only the ASM1042A controller chip (not the one without A, and thus not the syba controllers) support UAS. note the lack of UAS on this ASM1042 chip:
in my ASUS p9x79 deluxe 1.0 mobo, it lists an ASMedia xCHI USB controller. but this could be either one, and I am suspecting because mine is an older version, that it's not the A chip. USB 3.0 Boost caused flakiness, and the USB firmware update tool broke my USB firmware apparently or it's just buggy, nothing fixed yet. apparently the firmware for the controller on my mobo has been bricked, because support says I need to have my motherboard replaced. I hope a BIOS update fixes this, it could. http://forums.anandtech.com/showthread.php?t=2351620

hard drives appear and disappear with the Boost driver I am using in win7. so I uninstalled it or just turned boost off and things became more stable I think.

newegg.com has this 4-port HighPoint ASM1042A usb 3.0+UASP controller card: http://www.newegg.com/Product/Product.aspx?Item=N82E16816115144
http://www.newegg.com/Product/Product.aspx?Item=N82E16816115138
latter is for the mac.

 

[cat1092]

Just purchased on the Newegg site a USB enclosure with the UASP technology for $20, and ordered another right after seeing the performance of the first one. With a last gen (SATA-2) WD Caviar Black, was able to create backup images of all three of my Windows installs on my USB 3.0 equipped Dell XPS 8700 in under 10 minutes each, including the time needed to create the task, point the image to a folder, and also checked the 'verify' function (which checks the image for errors).

The device also ran very cool, and is on the Newegg site, still on promo for $25 (Rosewill RX307-PU3-35B -3.5"). Am not sure if the site forbids links to products (some does), so have provided the model number of the product as shown on the Newegg site, though any that carries Rosewill branded products should carry the same product. Simple to assemble, aluminum body with plastic ends (only the end where the HDD goes in need to be slid out).

If I has a SATA-3 HDD to install in that enclosure, it's possible that 5 minutes total would have been needed for each OS, on the other hand, not a not of 'real world' difference between SATA-3 & SATA-3 HDD's on MB's shipped with preassembled PC's.

Thanks to the hard working folks at Tom's for the article that helped to explain 'why' the fast backups, and in fact have a unused, still in the package StarTech USB 3.0 cable for 2.5" drives with UASP. Didn't understand the significance of it, now surely the cable will soon be used.

And BTW, the reviews on this site has led to my purchasing decisions several times. These folks knows their hardware.

Cat

 

[JuanPC]

Samsung 850 Pro is capable of 500MB/s, R&W, but...
My Sata6Gbps port only goes to 372MB/s "with HDTune."?
Anyway... with Thermaltake 5G BlacX st0019 "single" USB3.0 Superspeed with ASMT2105 goes to 262MB/s "Options Benchmark Accurate 8MB block size."
I've seen the same Thermaltake 5G BlacX perform much faster in a MacBookPro 2013 pr 2014 with USB3.0.
Also that MacBookPro does Boot in USB3.0, better to have an external SSD to boot, than the internal Sata port. The internal cable gets damaged easy...
I can't remember the speed of the USB3.0 in OSX but I remember it was a lot more than 262MB/s of Windows7x64sp1.
Tested 2 different USB3.0 cables, and two different st0019, same result.
Maybe the AC power is too noisy, and the TT 5Gx does not have enough coupling capacitors, "big enough." Maybe I will solder a 220uF or 330uF 63v 105°C rating somewhere inside. "Everybody has USB3.0 faster than me." Kind of suspicious.

 

[jimmysmitty]

Samsung 850 Pro is capable of 500MB/s, R&W, but...
My Sata6Gbps port only goes to 372MB/s "with HDTune."?
Anyway... with Thermaltake 5G BlacX st0019 "single" USB3.0 Superspeed with ASMT2105 goes to 262MB/s "Options Benchmark Accurate 8MB block size."
I've seen the same Thermaltake 5G BlacX perform much faster in a MacBookPro 2013 pr 2014 with USB3.0.
Also that MacBookPro does Boot in USB3.0, better to have an external SSD to boot, than the internal Sata port. The internal cable gets damaged easy...
I can't remember the speed of the USB3.0 in OSX but I remember it was a lot more than 262MB/s of Windows7x64sp1.
Tested 2 different USB3.0 cables, and two different st0019, same result.
Maybe the AC power is too noisy, and the TT 5Gx does not have enough coupling capacitors, "big enough." Maybe I will solder a 220uF or 330uF 63v 105°C rating somewhere inside. "Everybody has USB3.0 faster than me." Kind of suspicious.

The problem with USB in general is how it was designed. Back when it was designed top speeds were 1.5/12Mbps. It was mostly designed as a all in one peripheral replacement instead of having PS/2, serial, COM, parallel and LPT1 which all worked for specific devices. The use of external storage was never thought of in the original design and it was designed with a 8b/10b encoding scheme which means that 20% of bandwidth is dropped off the top no matter what.

Another part is that a single USB host controller splits bandwidth between all the USB ports on that controller. That means the more devices using a USB host controller, the less speed each device has.

USB is great for keyboards, mice, printers and phones but it is not as great for external drives. Take a USB 3.0 dock and a eSATA dock and do tests for data transfers and read/writes. You will find that a eSATA dock will be faster and more consistent, i.e. it will maintain a better average speed. That is because SATA was designed specifically for storage purposes.

Better is Thunderbolt which takes the idea of USB and has upgrades that allow it to perform better than USB in many ways. It has a top speed of 40Gbps vs USB 3.1 Gen 2 10Gbps and supports all kinds of protocols such as PCIe 3.0, DP 1.2 and HDMI 2.0. It can also use various connection interfaces and V3 runs on the new USB Type-C connection.

As for your 850 Pros speed, it depends on the controller you have. I have a Intel 520 and my wife has a 850 Evo. I have a Intel Z87 chipset and she has a AMD 970 chipset. Both drives have nearly the same top R/W but mine gets closer to its top R/W and IOPS than her 850 Evo does because Intel has better SATA controllers than AMD.