Comments on: Is hardware customization obsolete? http://www.embeddedinsights.com/channels/2011/09/21/is-hardware-customization-obsolete/ Shedding Light on the Hidden World of Embedded Systems Mon, 28 Jul 2014 16:18:37 -0400 hourly 1 http://wordpress.org/?v=3.0 By: OAZ @ TI http://www.embeddedinsights.com/channels/2011/09/21/is-hardware-customization-obsolete/#comment-8379 OAZ @ TI Thu, 29 Sep 2011 00:43:59 +0000 http://www.embeddedinsights.com/channels/?p=634#comment-8379 The PC succeeded because it was an open architecture. And I believe in the mobile world this will continue to be true. Not in terms of adding new HW to a device rather than inter-connecting devices using wireless technologies e.g. like M2M or utilizing the cloud. Customization will happen via SW - and the SW needs to be open ..... The PC succeeded because it was an open architecture. And I believe in the mobile world this will continue to be true. Not in terms of adding new HW to a device rather than inter-connecting devices using wireless technologies e.g. like M2M or utilizing the cloud. Customization will happen via SW – and the SW needs to be open …..

]]> By: L.R. @ LI http://www.embeddedinsights.com/channels/2011/09/21/is-hardware-customization-obsolete/#comment-8318 L.R. @ LI Fri, 23 Sep 2011 05:48:27 +0000 http://www.embeddedinsights.com/channels/?p=634#comment-8318 Robert, you pose a question relating to several trends in the industry, and it is not clear if you need each trend to be individually addressed, or perhaps you are trying to hint that these trends are inter-related? With regards to the fist observation, with respect to the addition of application-specific peripherals, I completely agree with S. - the trend is still here except that the additions are happening outside of the box, instead of inside. Looking to the future, I think that while USB should satisfy most consumer requirements,external PCIe (including the Thunderbolt variant) is fully capable of accommodating high-end professional peripherals of any conceivable function, including custom-built (e.g. on an FPGA) hardware blocks. Explaining the reasons for this trend may actually touch on the other trend you mention - the underlying trend is to improve the user experience and reduce the total cost of ownership. With respect to the extension of system capabilities - in-box customization is prone to mistakes like overloading of the power supply or the overwhelming of the ventilation system, which are very difficult to diagnose in a telephone support call, and thus moving towards outside of the box addition greatly reduces the chance of system failure as a result of such additions (e.g. USB tight power consumption rules and over-current protections), and reduces the difficulty of troubleshooting (and hence reducing TCO). The trend of consumers moving away from the traditional desktop form factor IMO can be traced to the same motivation - why buy a cow for just a glass of milk ? Digging even deeper it seems all these trends are rooted in the mass-market adoption of computing devices and is feeding from it. Due to the vastness of the mass-market volume it has become practicable to build devices in a variety of form factors and configurations to fit that particular market segment, instead of selling a base system plus a repertoire of add-on components, essentially as a kit. Going back a century, to when the radio receiver was gaining acceptance, during the first wave DIY kits were sold to hobbyists, as they were few, but when market adoption has ramped up and demand increased, the manufacturer had to address a much less competent consumer with a more cost-effective product, at a vastly bigger volume. So this may be all the result of the scaling economies. Robert, you pose a question relating to several trends in the industry, and it is not clear if you need each trend to be individually addressed, or perhaps you are trying to hint that these trends are inter-related?
With regards to the fist observation, with respect to the addition of application-specific peripherals, I completely agree with S. – the trend is still here except that the additions are happening outside of the box, instead of inside. Looking to the future, I think that while USB should satisfy most consumer requirements,external PCIe (including the Thunderbolt variant) is fully capable of accommodating high-end professional peripherals of any conceivable function, including custom-built (e.g. on an FPGA) hardware blocks.
Explaining the reasons for this trend may actually touch on the other trend you mention – the underlying trend is to improve the user experience and reduce the total cost of ownership.
With respect to the extension of system capabilities – in-box customization is prone to mistakes like overloading of the power supply or the overwhelming of the ventilation system, which are very difficult to diagnose in a telephone support call, and thus moving towards outside of the box addition greatly reduces the chance of system failure as a result of such additions (e.g. USB tight power consumption rules and over-current protections), and reduces the difficulty of troubleshooting (and hence reducing TCO).
The trend of consumers moving away from the traditional desktop form factor IMO can be traced to the same motivation – why buy a cow for just a glass of milk ?
Digging even deeper it seems all these trends are rooted in the mass-market adoption of computing devices and is feeding from it. Due to the vastness of the mass-market volume it has become practicable to build devices in a variety of form factors and configurations to fit that particular market segment, instead of selling a base system plus a repertoire of add-on components, essentially as a kit.
Going back a century, to when the radio receiver was gaining acceptance, during the first wave DIY kits were sold to hobbyists, as they were few, but when market adoption has ramped up and demand increased, the manufacturer had to address a much less competent consumer with a more cost-effective product, at a vastly bigger volume.
So this may be all the result of the scaling economies.

]]> By: S.A. @ LI http://www.embeddedinsights.com/channels/2011/09/21/is-hardware-customization-obsolete/#comment-8312 S.A. @ LI Thu, 22 Sep 2011 16:08:40 +0000 http://www.embeddedinsights.com/channels/?p=634#comment-8312 The availability of robust external interfaces, particularly USB but also including Firewire, eSATA and so on, has been one factor in this move away from internally-bussed hardware devices. Also, there are now higher levels of integration of devices on the motherboard - for example, a one-chip solution that integrates 802.11 a/b/g/n and Bluetooth is easily integrated. Depending on the application, the wireless capabilities of the device may be all you need to connect to peripherals. In those examples, the driver architecture is often greatly simplified. For example, if you have a low-speed I/O device, it may be possible to simply use the USB HID interface and avoid needing a device-specific driver altogether. But for high bandwidth or low-latency I/O, plugging into a board-level bus (like one of the PCI flavors) seems like it won't be going away any time soon? The availability of robust external interfaces, particularly USB but also including Firewire, eSATA and so on, has been one factor in this move away from internally-bussed hardware devices. Also, there are now higher levels of integration of devices on the motherboard – for example, a one-chip solution that integrates 802.11 a/b/g/n and Bluetooth is easily integrated. Depending on the application, the wireless capabilities of the device may be all you need to connect to peripherals.

In those examples, the driver architecture is often greatly simplified. For example, if you have a low-speed I/O device, it may be possible to simply use the USB HID interface and avoid needing a device-specific driver altogether.

But for high bandwidth or low-latency I/O, plugging into a board-level bus (like one of the PCI flavors) seems like it won’t be going away any time soon?

]]>