06-26-2011 12:03 AM
Looking for a solution instead of a workaround. eSATA gives me 150MB/s read and 149MB/s write. Using eSATA can not seriously be considered as a solution...imagine. LCD not working --> "use external LCD instead...still gets the work done" -_-
More like "I have VGA and DisplayPort/DVI/HDMI on my laptop, but I'm going to insist on using VGA."
06-26-2011 01:17 AM - edited 06-26-2011 01:21 AM
ColonelONeill wrote:Looking for a solution instead of a workaround. eSATA gives me 150MB/s read and 149MB/s write. Using eSATA can not seriously be considered as a solution...imagine. LCD not working --> "use external LCD instead...still gets the work done" -_-
More like "I have VGA and DisplayPort/DVI/HDMI on my laptop, but I'm going to insist on using VGA."
Hum...that may be...
But looking at the eSATA (II - 3Gbps), vs USB 3.0 (supposedly 5.0GT/s, but is limited to 2.5GT/s due to QM57 chipset...then now with some unknown performance issue).
The point is not about what I choose to use, or not to use. But about why isn't the USB 3.0 functioning with a decent write performance.
Plus, the reason why I want to use the USB 3.0 is because the laptop has 2 USB 3.0 ports, and I can RAID 1 the drives for the VMs I store on them. It's part of the reason why I originally chose the W510 (Ability to have large storage space, fast, cheap, software RAID 1 on the two external drives.....this, as oppose to buying 2 480GB SSD - expensive, very limited storage, and have to worry about wear level)
06-26-2011 12:17 PM - edited 06-27-2011 06:11 AM
Other reports of similar problems on ThinkPad X220 and W510:
Here's a report of a similar disparity between USB 3.0 read and write speeds on an HP Envy 17 running Windows 7, using three different benchmarks on several different drives, so the limitation doesn't appear to be unique to Lenovo:
"The other cool part about the latest USB 3.0 devices: these speeds are approaching that of the internal hard drive. Our baseline with the internal drive was able to get read speeds of between 100M and 103MBps, and write speeds of about 100MBps. The big difference between internal vs. external, however, is in the write speeds. But at the read speed level, they are pretty close. A Seagate spokesman said that buffering going on behind the scenes helps speed up the read from the drive - and that when writing, there is more overhead from the drive that tends to slow it down. Other factors include the types of components on the PC side, as well as the drivers used, the spokesman said."
So, at this point, if anyone needs to rely on fast write speeds over USB 3.0, it would be a good idea to look at benchmarks of those speeds for the machine in question, before you buy it. Not unusual for newly-adopted technologies, and a good practice to follow whenever considering the purchase of a new machine that needs to meet specific requirements.
Perhaps someone can post USB 3.0 benchmarks for the W520 in this thread, so that potential W520 owners will know what to expect.
It will be interesting to see if Lenovo is able to find the root cause of the slow writes in Chatbox's case. In any case, if anyone wants to try to diagnose the problem themselves, here's some software which may help in that process:
USBTrace : Software-only USB Protocol Analyzer
USB 3.0 Support (Super Speed USB). (New)
Captures USB 3.0 controllers and devices. Decodes 3.0 standard descriptors and requests.
(free 15-day trial, data size limitations)
I would also check for any utilities from the drive manufacturer that let you access the detailed settings of the drive itself, which would hopefully let you compare settings when the drive is connected to eSATA/SATA vs. USB 3.0. The Windows Device Manager doesn't give access to many of those properties.
And here's a google search that may yield more info, for anyone who's interested:
"usb 3.0" "mb/sec" OR "mb/s" OR "mbps" write OR writes read OR reads laptop OR notebook
06-26-2011 11:55 PM
Hi Jimbo, thank you and I really appreciate the detailed response.
I think I'll leave it to the Lenovo support engineers now. Hope they'll be able to address the matter promptly.
06-27-2011 11:37 AM
I have made several tests with USB3, eSATA and eSATAp tests and my general conclusion ist that eSATA beats them all!
08-31-2011 01:11 AM
After multiple email communications, the following is what I got back from the Technical Manager. Looks like poor performance is not an issue for Lenovo. No wonder why there are so many issues with the W510 and W520....especially overheating, GPU malfunctioning...
To be clear, there has been no misunderstanding on my part regarding the customer's complaint. USB 3.0 has a maximum theoretical throughput of about 5Gb/s; in no way, shape or form will any single device achieve anywhere near this limit. That 5Gb/s is total bus bandwidth, not what a device will get. I have no doubt on the SSD is capable of sustained writes beyond what the number are however, however, that does not mean there is any defect. The W510's USB 3.0 controller is a first generation device and was certified at the specification that existed at the time of the machine's release. We do not know whether the drive enclosure / adapter the customer is using is affecting throughput in any way. Is this adapter or enclosure capable of 140MB/s? I have no idea. I also have no idea if the customer's drive is actually capable of this, either.
We have not made any claims as to the throughput speeds for any specific devices attached to the USB 3.0 ports. The fact is that the USB 3.0 port is running at USB 3.0 speeds, even if it's not a speed the customer would like. USB 3.0 speeds mean that throughput is faster than USB 2; it does not mean it's at any particular speed above USB 2. Customer's quote "Faster than USB 2.0 does not mean it's fulfilling its USB 3.0 specification" is incorrect, Faster than USB 2 means we're now in the USB 3 transfer speed range.
08-31-2011 01:47 AM - edited 08-31-2011 04:48 AM
It also sounds like the Manager has no idea of any testing Lenovo has done on the USB 3.0 ports to determine their throughput. Plus I don't think there was any "specification that existed at the time of the machine's release," at least none revealed publicly, other than the USB 3.0 specification itself. If there was, then it presumably would have been included in the spec sheet for the W510. Being an engineer myself, the statement "USB 3.0 speeds mean that throughput is faster than USB 2" is simply ludicrous. I would be laughed out of any engineering meeting in which I said something like that.
I don't trust any of these manufacturers any farther than I can throw them. So I don't believe anything they say until it can be tested by independent parties. Unfortunately, you've become the independent party in this case.
Pioneers often end up with arrows in their backs, Chatbox. At least, that's what my graduate advisor once told me as I was trying to pull the arrows out. It wouldn't be surprising if you, and everyone following this thread, now know more about this issue than any of the engineers at Lenovo. You certainly know more about it than that manager. If I owned a company that manufactured high-end laptops, I'd send you the prototypes to make sure they were thoroughly tested. ;-)
In the end, we vote with our wallets, as do the people who ask for our advice, as do the people who read these forums. In the end, that's what the industry understands. So our discussions contribute to that, even if the companies themselves don't read or understand those discussions. That applies to buggy computers, as well as to interfaces that don't live up to their hype.
Best of luck.
08-31-2011 01:58 PM
Hardware and software makers aren't in the habit of releasing performance results. And many of the licenses you agree to prohibit you from doing so, too. Ever wonder why?
What I really hate is when you report an issue, they insinuate you don't know what you are talking about, question your testing methodology, etc. Now granted many people do a really sloppy job of testing and documentation, but that doesn't mean you can't prove an issue exists.
The problem is that Lenovo is likely working on the W530 or whatever it's going to be called. Good luck get fixes on the W510 at this point.
09-01-2011 07:41 PM - edited 09-01-2011 07:42 PM
I agree, ThorsHamer, but I doubt Lenovo itself ever tested the sustained transfer limits of the USB 3.0 interface on the W510. When the machine was released, that kind of testing would have been difficult using off-the-shelf components. Lenovo relies on the USB 3.0 chipset's vendor for drivers, and their development kits from the vendor at that time probably included a hard drive in a USB 3.0 enclosure. That could have been used to show that the interface worked and was faster than USB 2.0, but not much else. Lenovo would have used that in their acceptance testing of the vendor's product, which would have been Lenovo's main focus in testing the interface.
So that may be the source of the comment Chatbox received from the manager, that USB 3.0 is simply any USB interface that's faster than USB 2.0. Nonetheless, it's a ridiculous statement.
And Lenovo certainly didn't test the interface's throughput in response to Chatbox's complaint. They should, however, be thankful she discovered that Lenovo's drivers were much slower than the vendor's current drivers. I didn't see any expression of thanks in the manager's comments, and Lenovo is making a mistake in not cultivating interaction with advanced users such as her. Those users are instrumental in troubleshootng bugs, and Chatbox's work in troubleshooting overheating problems in the W510 was exemplary.
When the USB 3.0 standard was being developed, Intel felt that it should be based on fiber optics instead of copper, but copper prevailed. So Intel didn't support it in their processor chipsets (hence the USB 3.0 chipset from NEC). So Intel's fiber optic version of USB 3.0, called Thunderbolt, will arrive in Wintel machines in 2012, and is already available in Macs. That's presumably what Lenovo will have in the W530, or whatever the machine will be called. It should kick the copper version of USB 3.0's butt.
But don't count on that performance in a given laptop model until an independent party has tested it.