Support in other languages: 
Showing results for 
Search instead for 
Do you mean 
Reply
Token Ring
Chatbox
Posts: 487
Registered: ‎11-04-2010
Location: Sydney
0

Re: [W510] - USB 3.0 SSD, poor performance.


Jimbo wrote:



Of course, you want to make sure that everything within the Power Manager utility is set to highest performance mode, with the W510 plugged into AC and with a fully-charged battery installed, and I would power the drive enclosures with their own AC power cord. You also want USB 3.0 cables no longer than 9 feet.

I would also try to get someone to run similar tests on a W520, to see if the issue is specific only to the W510's generation.

Best of luck.



Hi Jimbo,

 

Yes, battery is fully charged, AC connected and powered, High Performance selected in power manager, using USB 3.0 SS cables, 3 feet long, external enclosure has it's own 12v 2Amp power supply.

 

Still no luck. :-(

Token Ring
Chatbox
Posts: 487
Registered: ‎11-04-2010
Location: Sydney
0

Re: [W510] - USB 3.0 SSD, poor performance.


hazart wrote:

USB 3.0 results:



Thanks for posting your results.

 

From what I've seen so far, The lack of decent performance is not even due to USB 3.0 transfer encoding overhead.  There seems to be something really really wrong with the USB 3.0 implementation on the W510.  Such a huge performance hit can seriously not be considered as "overhead" (write performance).

Token Ring
Chatbox
Posts: 487
Registered: ‎11-04-2010
Location: Sydney
0

Re: [W510] - USB 3.0 SSD, poor performance.

[ Edited ]

I honestly don't think I'm being unreasonable to have my expectation at around "130-150MB/s" seqential read and write performance.  That's only 25% of the design spec.  If Lenovo can't even produce a highend laptop to provide a quarter of the performance...then it really shouldn't be marketed.  Seriously.

 

Now, to be fair, Lenovo hasn't got back to me yet as to what the "ideal" configuration is.  If they can provide details as to how I can get 150MB/s write from the USB 3.0 ports, then I would be quite happy to carry out hardware changes on the external HD, enclosure, cable (consumer products, not enterprise class ones.  i.e. Don't suggest to me that I need a 5 drives hardware RAID 6 array with 512MB cache on the array).

802.11n
Jimbo
Posts: 323
Registered: ‎10-26-2008
Location: Los Angeles
0

Re: [W510] - USB 3.0 SSD, poor performance.

[ Edited ]

I assume your antivirus/security software is turned off during the tests, and that you're disconnected from the network. You might also select the USB 3.0 drive in the device manager while the drive is on and plugged in, open up its properties, and make sure that caching is set for max performance, and that there are no settings like write-verify turned on. it's just a little odd that it's working fine for reading, but slow for writing, while both are fine under eSATA.

Token Ring
Chatbox
Posts: 487
Registered: ‎11-04-2010
Location: Sydney
0

Re: [W510] - USB 3.0 SSD, poor performance.

[ Edited ]

Jimbo wrote:

I assume your antivirus/security software is turned off during the tests, and that you're disconnected from the network. You might also select the USB 3.0 drive in the device manager while the drive is on and plugged in, open up its properties, and make sure that caching is set for max performance, and that there are no settings like write-verify turned on. it's just a little odd that it's working fine for reading, but slow for writing, while both are fine under eSATA.


Antivirus off / realtime protection off.
Wireless off, wired network disconnected.

"Safe Removal" vs "Better Performance" makes no difference to the write speed.

No other external device plugged in, other than power supply, and of course, the USB 3.0 enclosure + drive.

 

I've also tried dd bs 1M count 1024 and 2048  (zeroes) while booted from an Unbuntu live CD (try to isolate Windows, drivers...etc).

Yet, still only get 65-70MB/s write.

 

Note: The enclosure that's being used is, for certain, able to handle 130MB/s read and write.  The external drive that's being used is, for certain, able to handle at least 150MB/s read and write.  Cable is SuperSpeed.

 

So, what else?  Logic suggests it's the laptop.

Guru
ColonelONeill
Posts: 7,075
Registered: ‎12-26-2009
Location: Toronto
0

Re: [W510] - USB 3.0 SSD, poor performance.

Does locking the CPU speed at maximum help?
W520: i7-2720QM, Q2000M at 1080/670/1340, 21GB RAM, 500GB HDD, FHD screen
X61T: L7500, 3GB RAM, 500GB HDD, XGA screen, Ultrabase
Token Ring
Chatbox
Posts: 487
Registered: ‎11-04-2010
Location: Sydney
0

Re: [W510] - USB 3.0 SSD, poor performance.

[ Edited ]

AFAIK, you can't get it to stay on turboboost mode on (user's) demand.

 

But I've disabled speedstep, makes no difference.

Token Ring
Chatbox
Posts: 487
Registered: ‎11-04-2010
Location: Sydney
0

Re: [W510] - USB 3.0 SSD, poor performance.

Just came across this while searching...
http://www.frescologic.com/news_show.php?pe=2&ns=34

 

So, W510 came out in Q1 of 2010...has very low efficiency?  Not even 60%? 

 

Man, we're screwed.

Bit Torrent
ThorsHammer
Posts: 2,400
Registered: ‎04-08-2011
Location: USA
0

Re: [W510] - USB 3.0 SSD, poor performance.

 


Chatbox wrote:
Looks like the USB 3.0 implementation on the W510 is a half baked job. Lenovo was trying to get (one of) the first to have "USB 3.0" on the market.

Got to say, "faster" than USB 2.0 doesn't make it USB 3.0. Having a USB 3.0 chipset in the laptop doesn't make it USB 3.0.

USB 3.0 is a technical design specification, with performance standards.

At the end of the day, I feel that what I paid for due to what's advertised is not what I'm getting.

What benchmark software are you using under Windows 7? 

802.11n
Jimbo
Posts: 323
Registered: ‎10-26-2008
Location: Los Angeles
0

Re: [W510] - USB 3.0 SSD, poor performance.

In general, USB 3.0 implementation is a bit inconsistent, even across different implementations using that same NEC controller (which was the only game in town):

http://www.tomshardware.com/reviews/nec-controller-usb-3-pd720200,2682.html

But still, the differences you're seeing are large enough and just specific to writes, so that something else appears to be going on. I can't believe that no one has tested this before on the W510.

It would be fascinating to know whether this happens on the W520 as well. Note that the W520 uses the same NEC controller, so the W520 is likely in the same boat:

http://www.ubuntu.com/certification/hardware/201103-7374/components

That would be a significant issue for anyone planning to buy a W520.