Fresco Logic launches the FL1009 USB 3.0 host controller

We sit down for a chat with the CTO

Nov 30, 2010 by

ALTHOUGH FRESCO LOGIC is still a USB-IF certification short of a sellable product, the company has finally launched its two port xHCI 1.0 FL1009 USB 3.0 host controller. Earlier today we had a chance to sit down for a chat with Bob McVay, the CTO of Fresco Logic and he gave us an insight into Fresco Logic and why it has taken so long time to bring this product to market, among many things.

Fresco Logic has a strong R&D team and prior to Fresco Logic, many of the engineers and the senior management at the company worked on the development of standards such as Gigabit Ethernet, InfiniBand and PCI Express. The company is based out of Beaverton, Oregon, but also has a branch office in Taipei, Taiwan. Despite its engineering expertise, Fresco Logic has run into a few bumps along the road to the final FL1009 product and some of it actually made us better understand why we so far only have a limited amount of USB 3.0 host controllers in the market.

It turns out that the FL1009 is using the third design that Fresco Logic worked on while developing its host controllers, although we weren’t given any precise reasons as to why the previous solutions weren’t chosen apart from that they didn’t deliver what Fresco Logic believed would be a good solution. Fresco Logic uses a very different design internally compared to NEC/Renesas, as the company didn’t have any USB 2.0 PHY experience to build upon. As such the company not only had to develop a USB 3.0 PHY and transceiver, but also a USB 2.0 PHY to make a backwards compatible solution.

NEC/Renesas uses a different design with a dedicated USB roothub for the USB 2.0 and USB 3.0 part of the host controller, whereas Fresco Logic designed a combined solution for each of the ports. Fresco Logic also uses what they call an xHCI accelerator engine, which goes under the name of GoXtream. This allows not only for increased performance, but also lower power utilization by processing as many requests as possible as quickly as possible, rather than processing the requests one at a time.  Fresco Logic also designed a firmware-less host controller.  Although this took some extra time in the development process, Fresco Logic claims that this is not only easier and more affordable to manufacture, but it’s also easier to integrate into retail products.

The significance of xHCI 1.0 pretty much boils down to two things, the potential of a universal USB 3.0 driver supplied by Microsoft and  other OS platforms, as well as vastly improved support for virtualization, something USB 2.0 doesn’t do nearly as well thanks to several different standards being used. From our understanding, Microsoft is pushing for a unified USB 3.0 driver stack, but this isn’t going to happen overnight, so don’t expect this to be part of the next service pack for Windows 7. While on the topic of drivers, to our knowledge, Fresco Logic is the first company to officially support Linux which is good news for those of you that are running Linux and want some USB 3.0 action. We can also expect to see UASP support in future drivers from Fresco Logic.

Fresco Logic provided a demo of what the FL1009 is capable of, although sadly we didn’t see it compared to the NEC/Renesas controller on the Gigabyte X58 motherboards used for the demo. A SuperTalent USB 3.0 RAIDDrive was used together with CrystalDiskMark 3.0 to give us an idea of the performance of the host controller. In the sequential read test the RAIDDrive hit a not insignificant 373.7MB/s and even SuperTalent issued a press release just over a week ago bragging about how fast the RAIDDrive is when paired up with Fresco Logic’s controller. The performance difference is said to be about 55MB/s to Fresco Logic’s favor. The write performance ended up just a smidgen under 200MB/s, again, some 40MB/s faster than previously possible.

The FL1009 is also the preferred host controller of Blackmagic Design, a company that specializes in HD video capture devices and its Intensity Shuttle allows for HDMI video capture in real time. However, to capture uncompressed 10-bit 4:2:2 1080p video with the Intensity Shuttle, you have to use Fresco Logic’s controller according to Blackmagic Design, as the NEC/Renesas solution just isn’t fast enough to cope with the bandwidth. Now this might be a niche application, but it’s still interesting to see that there’s that much difference between different host controllers.

So what does the future hold for Fresco Logic? Well, Bob McVay hinted on solutions for mobile devices smaller than notebooks, various video related solutions, as well as non-USB 3.0 related products, but wouldn’t go into any more details at today’s event. He also said that Fresco Logic is considering a four port solution, if there’s enough demand for such a product. However, the company is expecting to ship over one million USB 3.0 host controllers by the end of this year, but considering that it’s only about one third of Renesas claimed monthly output, it might not be that significant a number.

According to a slide during the presentation, Fresco Logic is expecting about a year and a half to two years’ worth of profitable USB 3.0 host controller shipments until USB 3.0 becomes an integrated part of the motherboard chipset. Considering that we have yet to see a single roadmap from Intel indicating any kind of support for integrated USB 3.0 support and the fact that as far as what is known, AMD is so far only planning on offering USB 3.0 support in one of its upcoming chipsets for this year, there might be more life in the USB 3.0 host controller market than expected. For next year, some 3.5 billion USB devices are expected to ship, although USB 3.0 is expected to account for less than 250 million of those devices and isn’t expected to reach the billion mark until about 2012.

Fresco Logic’s CTO Bob McVay and MD Jeryaun Yen

When we asked about Intel’s Light Peak technology, Bob McVay said that he isn’t overly concerned about Light Peak at the moment, as for one it’ll be much more costly to implement than USB 3.0 and on top of that, it’s questionable if it’s even needed at this point in time, as USB 3.0 is already offering more than enough bandwidth for at least the next couple of years. That is, unless there’s some kind of dramatic change in performance of storage devices that we’re not aware of. That said, Light Peak will come sooner or later and it might well end up replacing USB 3.0 in the long run.S|A

The following two tabs change content below.

Lars-Göran Nilsson