Comparing PCIe with USB seems a bit off to me as PCIe is a bus expansion scheme for a distributed computing architecture while USB is intended as a peripheral connection scheme for data transfer purposes.Different applications have different tradeoffs and just because PCIe can be faster doesn't mean it's the right choice for all peripheral interconnects. For one thing, the USB can provide power to the endpoints, eliminating the need for them to have their own power source. PCIe external cabling makes power to the connector optional, so its presence is not guaranteed. The biggest factor, though, is installed base. USB is everywhere. PCIe has a lot of ground to make up before it can think about replacing USB, starting with solving the chicken-and-egg problem. Folks won't design it into host systems without there being enough peripherals available to make it worthwhile, and peripherals won't offer it until there are enough hosts to make the market size worth pursuing.
In this age, what applications can be benefited by USB3.0 while there are no other more effective ways that are also more user friendly. I am talking about wireless connectivity. Today, many applications can be served natively more effectively by WiFi or its variants. No-wire connectivity that is reliable and consistent is always more user friendly. For example, WirelessUSB was one of the worst technology ever pushed out and this came from pretty much the same folks who are now trying to brain-wash us with USB3 today. Would you trust them after the debacle of WirelessUSB and OTG (another waste of investment by the industry and consumers).
For the wired connectivity, in spite of the 5Gb/s "raw" bandwidth of USB3.0, the net bandwidth is far lesser especially when compared to PCI-e 3.0 (or 4.0). Popular applications often touted by those pushing USB3.0 include HDD or SSD. But, isn't the PCI-e External more efficient and cheaper by eliminating the myriads of protocol conversion for PCI-e all the way into the HDD and SSD? Cost, don't believe in what you've heard from the USB3 folks because they have their job security in stake!
Once again, USB has past its prime for many applications because USB3.0 has its limitation. Another limitation which not many people notice is: USB3.0 interface cannot be integrated natively into SoCs manufactured in the latest geometry like 20nm and 14nm. It is because of the 5V tolerance imposed by backward compatibility. In contrast, PCI-e I/Os employs the truly LVDS signaling and can therefore be integrated natively into SoCs manufactured in advanced geometry.
The world should move on, leave USB2.0 and USB1.0 to deal with the low-speed applications. I started working on USB since CY1994 and was dismayed by how this technology was bandaged to take on different applications which can be better served by other technologies. So, try to see through the hype and embrace the right ones.
I was excited about USB 3.0, too.... about 3 years ago. The problem with USB is that it seems to take so long to progress. It just seems stuck in time. We get chip refreshes every year, Moore's law carries on relentlessly... and yet it takes years and years for a standard to upgrade a wee bit? Having said that, looking forward to hearing about whether any progress is being made!
I thought the same thing (that USB was old and boring) until I had a meeting last week and learned about USB 3.0 and the potential for USB to be an enabling technology for many, many more years. If you think about it, it's already pretty much everywhere. And I'm told that the speed increase between USB 2.0 and 3.0 is massive.
Isn't USB already an ancient technology? I hope the ARM Tech Conf gives us a hint of what're going to coming down fast in the next year or so: like WHDI, WiGi, and the external PCI-Express. In particular, the external PCI-Express has the potential to replace USB for all high-speed usages in the future.
My Mom the Radio Star Max MaxfieldPost a comment I've said it before and I'll say it again -- it's a funny old world when you come to think about it. Last Friday lunchtime, for example, I received an email from Tim Levell, the editor for ...
A Book For All Reasons Bernard Cole2 comments Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...