Archive for 2008

On November 17, 2008, we demonstrated a USB 3.0 Device streaming uncompressed HD video across a USB 3.0 cable at data rates exceeding 460 Megabytes per second.    To my knowledge, this is the fastest demonstration of USB 3.0 to date, and the only one demonstrating video transfers.  In addition, it is much faster than the targeted 320 Megabytes per second targeted by the USB-IF for effective throughput.  This demonstrates that the protocol is capable to moving data at these rates.  More on this later.

 

 

I’ll have pictures in a day or so when Wordpress (or my PC) is fixed so I can actually upload the photos.

CNET reported today that USB 3.0 will be “unveiled” at the USB 3.0 Developers Conference on November 17, 2008 just under 2 weeks here in San Jose.  I’m looking forward to the event. 

If you haven’t registered for the event, go to www.usb.org/usb30.  It’s for developers and it’s technical, so keep that in mind.

More devices are doing more things.  The best example of devices that have more functions are phones.

Anyone shopping for a mobile phone, will see that the newest phones have more and more functions.  Since the introduction of the iPhone, every company is introducing larger screens and touchscreens.  The larger screens require more power to operate.  More software to run more applications.  More software to manage all the applications.  More software to manage the power for the applications.  More software to manage when the WiFi and Bluetooth are on-and-off.  A touch screen that is polling all the time for data from the touch screen. And, in the case of the iPhone, an accelerometer that senses motion (like turning the phone to view pictures in the correct perspective.   For a camera phone, an image capture device and maybe even a flash.

All these feature require power.  Product makers must manage battery life or be doomed to 30 minutes of talk time.  So the choices for managing are:

1) Design the chip hardware to consume less power
2) Design the software to manage power usage for each application
3) Make the battery bigger to provide more juice
4) Accept shorter battery life (hey, you don’t get something for nothin’)

3) Bigger batteries – The iPhone is larger both because of the screen, and probably to house a larger battery.   Battery life is always relative to some other device.  For example, my Blackberry would probably run for 2 weeks as a phone, but once I add data, it’s probably 1 week, and with Bluetooth, maybe 6-7 hours of talk time, and it has to be recharged.

For 1) Change the hardware, we have done some studies with interesting results.
Some research by our own Mike Keating indicates that for some companies aggressively implementing hardware features to reduce power are nearing their limits. Phone companies are probably in the lead here.  Keating and David Flynn have published our Low Power Methodology Manual (LPMM) which you can download for free here: http://www.synopsys.com/partners/arm/lpmm/lpmm.html

If you really implement these features, you can squeeze out some extra battery life.

For me, the most interesting thing is that a lot of companies still do not employ these methods.  These are consumer devices.  There is still room on the hardware side by using our Low Power Methodology Manual.  This is absolutely clear to me, however, you must implement multiple power domains, multiple power rails, and MTCMOS, among other things.  Because of the time and effort required, many companies do not even attempt this. 

To me, this means 2 things:

A) Implement simpler hardware options for reducing power
B) Implement our Low Power Methodology.

What does this have to do with USB?

For A) the fastest, easiest hardware solution is to use HSIC to implement add-on USB functions.  HSIC uses a PHY that is 1/3 the power and area of a standard USB PHY.  If you add the USB standard Link Power Management, LPM, you add the hardware capability to use LPM.  (This is the USB standard LPM, not the Synopsys LPMM).  Add the software for LPM and you can save a lot of power, possibly up to 20% of your battery life can be recovered depending on the kind of USB device you are using.

For B) if you adopt the Synopsys LPMM, you can get additional savings, but requires a more disiplined approach.  More me, the most interesting thing about this option, is that it almost “free.”  It means, that if your engineers learn the methodology, there may be some additional silicon coast (egad Holmes) but would it be worth it to spend a few more pennies for the silicon if you could re-capture 10-40% more battery life?  Most interesting, is that if the hardware hooks are in, then the software can really drive power usage down.  The cost is minimal.  You can use your existing IP. 

Seems to me that B) offers a way to add value to raise the price of a product or sustain the product, even in potentially hard times.

Of course if you go the route of A), we have that too. (Shameless plug).

 

I received this gift from a friend.  This is a USB 2.0 Hub in the shape of a TARDIS.  When you insert a USB Device, it makes the sound of a TARDIS materializing/de-materializing.

(The light flashes too.)

You see USB everywhere and it is (can be) fun.

I really want to open it up and see whose chip is inside the TARDIS, but I’m not going to risk wrecking it.

(I bet you were expecting some insight or something on USB)  Maybe next week.

So I looked at Best Buy and Target Ads in my local newspaper this weekend. 

Of the 6 Camcorders in the Best Buy Ad:
1 uses tape
1 uses DVD
1 uses flash
3 use hard drives

So Jay and Mick I win.  Okay, I just stated the obvious.   The camcorders use hard drives mostly, and not necessarily for HD.   Flash is coming, so we will see when the crossover point is from hard drives to flash.   I think that hard drives still dominate the market in some form because they will just be cheaper.  I’m sure some expert out there knows.

So what I want to do is set up to camcorders with 20GB of memory each on 2 corners of my kids soccer field. I  want to be able to start them up at the same time, and when I’m done,  plug both into my PC, have the video download to my PC, synchronize, and build a DVD with 2 angles of video.  I can switch back and forth and my kids can watch themselves play soccer.  This would require the camcorder guys to develop a combined hardware and software solution to make this happen, but it might drive up camcorder sales because I would then be buying 2 instead of just 1 camcorder.  For me, I’d spend maybe $1000 for this package, although in a survey I’d probably lie and say I’d pay $1200 for a 2 camera system.  But I’d probably wait for the system to have USB 3.0.

Alright, so I need to back up my PC at home.  So what do I do?  I buy a hard drive from Costco or Best Buy or New Egg.  For me it is the small 320GB Toshiba Hard Drive.  This is great because it’s bus powered and small.  This means only one small USB cable.   I have about 20 GB of photos so I dump that with my few megabytes of important documents.  Unplug it. Stick it in a small bag, or a clean sock, and stick it in a drawer with my Compact Flash cards, SD cards, and Flash drives.  In 1 month, rinse, repeat.

If I buy the Terabyte drive it requires a power cable and a USB cable, and a place to put it.  I have about 6 drives like this.  3 I built back in the days when 40 GB was a big hard drive, and 3 more I purchased pre-built on sale.  I never use them. They have backup data and sit with their power cables in unused boxes in my unused wet bar.  (I don’t drink, I really should remove that wet bar.)

The other thing that is funny about back-ups is that I have 2 friends that each had RAID based systems. Theres are the Redudant Array’s of Independent Drives.  With a RAID system of 2, 3, or 4 drives, you should be able to rebuild all the data on all the drives because the data is spread-out in a way to make it redundant. 

I like to call these RUD’s.  Why?  Both of my friends had failures of these systems at home.  For one it was weeks of photos.  For the other it was months and months of programming. Redundant Undependable Drives.  On top of this, this happened at my old company too on an R&D drive.  

Okay, this doesn’t happen all the time.  I’m sure there are 100 examples of where RAID works.
Let’s face it though.  For my dad or mom, they don’t want to deal with the complexity of a RAID system.  They can copy files to and from a portable drive, and put it in a drawer.  They can alternate drives if they have 2.  This is simple. (Like me)

So the obvious application for USB 3.0 is Hard Drives.  These are Direct Attached Storage drives that everyone pickup and buy at Target or Costco or Walmart.   Not just FNAC, Yodobashi Camera, Suning, or Best Buy.

Of course, there are Solid-State Drives also. These are around 128GB now and will go to a Terabyte withing 4-5 years assuming a doubling each year.  These offer faster 3x+read/write speeds than rotating hard drives and 1/3 or less power consumption.  This is an almost 10 fold increase in performance. (based on my 3rd grade arithmatic.  So the likelihood is that these will replace hard drives in laptops to extend battery life further.  At some point they will be cheaper than rotating hard drives.  This means that they will also be used in Direct Attached Storage products also.

The other thing would be flash drives.  These flash drives, sometimes call thumb drives, were the killer app for USB 2.0.   I found a 4GB one at the park the other day.  (I almost took it home when I realized that this would be akin to finding a chocolate bar on the ground and bringing it home to share with the family.  Ew.)

My guess is that these will be at least 500GB or 1/2 terabyte within 5 years also.  These have the possibility of displacing the Direct Attached Storage as viable backup up for most people like me with tons of photos, and an itty bitty amount of personal data.  But they could be used for caring movies and music also for viewing on TV’s through DVD players that have host ports that read these files.

Tell me about your RAID experiences.  Does your DVD player have a USB port on it?  Do you have a portable hard drive?  What is the probability that Flash Drives will replace Direct Attached Storage?

This quick article is to correct some press impressions that may set expectations to a different level than what is actually planned.  I am a great fan of USB 3.0, so my views are biased.  I just don’t want the critics beating up USB because they expected.

The USB 3.0 Promoters are targeting a 350 Megabytes per second effective throughput for USB 3.0. This is 10x faster than the effective throughput of USB 2.0 (about 32 Megabytes per second). 

The actual signalling rate of USB 3.0 is actually higher, something like 600 Megabytes per second, however, because of the protocol overhead, hardware, device, operating system, and driver latencies, the effective throughput lower.  This is true of all devices you and I own today, not just USB.  It is the reason why USB 2.0 goes at 350 Megabits per second of effective throughput instead of 480 Megabits per second (again the electrical signalling rate).  Of course, the USB 3.0 Promoters are doing everything they can to minimize this.

This is cool because the USB 3.0 Promoters are actually targetting an EFFECTIVE throughput, rather than the more engineering oriented signalling rate.  Also, I think it’s important to set expectations so that when data rates actually meet the expected effective throughput, the critics won’t be saying “you didn’t meet your goals” when really the goals were met.

I think USB 3.0 is going to be fantastic.  I can use the bandwidth.  Yeah, I’m a USB Chauvanist.

I’m heading up to IDF today to set up for the show.  At this show, there will be seminars on USB 3.0 and Wireless USB.

We will be demonstrating our Wireless USB and our HS OTG controller in the USB-IF community.

I look forward to this show every year because I get to see the latest technology and plans for products around handhelds and PCs.    If you are attending IDF, stop by our booth.  If you are not, check the business section for a run down of events. (or the internet, that works too).

JayC commented that flash is a viable option.  I think this is interesting.

At CostCo I can buy two (2) 2 GB SD cards for $27.  If this trend continues, I could by two (2) 4 GB SD cards for $30 in 12 months.  It might be interesting if you could do the same with a 32GB that sells for $130 plus.

This makes sense for short videos, not videos of longer events, but if your kid is singing 2-3 songs in show, this works well. Videos up to 5-10 minutes at a NTSC or slightly better resolution.

As long as SD can keep up with the read/write speeds then this might be viable.  The issues for SD memory is speed and cost.  SD makers can speed up the rate for read/write by putting more banks in parallel. However, this adds cost.

Can the flash keep up with the needs of pro-sumers?  Can it keep up with HD video Camcorders?
How do the read/write speeds of flash compare to rotating media like hard drives? 

Wikipedia http://en.wikipedia.org/wiki/Secure_digital says:

The first column is rating, the second is speed.  I’m pretty sure that SATA is faster, and that hard drives are faster.  SD may have a power advantage. I know that SSD drives do. However, the 64GB drives are something like $500.  So an SSD with SATA in a camcorder makes more sense to me from a speed perspective.

If you know something about this, please post a comment.

Also, let us know what you think the first products will be?

I’ll respond to Mick’s incendiary comments another day.  He’s always drawing to draw attention to SATA for some reason.

I know what the first USB 3.0 Consumer Products should be. 

The Digital Camcorder.

The Camcorder will have a hard drive (like a laptop drive) and will likely shoot high definition.  It will have a 80 GB drive minimum, and will sell for $800-$1000.  A good example of an existing hard drive camcorder can be found here.

This targets the Prosumer market, the people that are used to paying $800 or more for a camcorder.  They are serious enough that, today, these people already either use tapes and take the 1 hour to transfer and another 1 hour to convert these to an MPEG file.  Or they do nothing and accumulate the tapes.

I don’t know about you, but since I had kids, and then 2 kids, I have no free time to spend converting hours of video to hours of DVDs. 

With USB 3.0, you will be able to transfer 27GB of data in about 70 seconds.  This makes a new business viable. 

For example, for digital still cameras, at CostCo or other places, you can plug in you photo memory card, and print pictures.  If you want to print all, the process is less than 3 minutes of transfer and checkout. Then you just shop and pick up your photos after you check-out.  This is not viable with todays tapes or USB 2.0 speeds.

With Super Speed USB, you will be able to drop off your camera with the attendant, pick you DVD menu, and when you check out you will have a DVD (or Blu-Ray Disk) with your videos.  This makes shooting video much, much more compelling that today’s process.