Intel forum is next front in x86 vs. ARM war

9/6/2012 4:15 PM EDT

Kandou bus is an interesting wild card, but if they claim to be relying on patent pending technology what is likelihood of getting this adopted by standards?...I would rather license it for free and get my money by selling IP blocks.

Here is some info (I am not paid by Kandou to disseminate this ;-)

Kandou Technologies has developed a new patent pending approach to serial link design that increases the bit rate for a given physical communications link. With this technology more bits can be sent per unit of energy, or less energy can be used to achieve a given bit rate. For common high speed links, speed increases on the order of 400% or bus power reductions to 25% are readily achievable and will yield meaningful net design and manufacturing cost reduction as well. These gains are complementary, thus additive, to known advanced serial link design techniques in use today. The magnitude and potential for tradeoffs of these gains dramatically improves the constraint space across numerous aspects of the design of high speed electronic devices.

Kandou Technologies has developed a new patent pending approach to serial link design that increases the bit rate for a given physical communications link. With this technology more bits can be sent per unit of energy, or less energy can be used to achieve a given bit rate. For common high speed links, speed increases on the order of 400% or bus power reductions to 25% are readily achievable and will yield meaningful net design and manufacturing cost reduction as well. These gains are complementary, thus additive, to known advanced serial link design techniques in use today. The magnitude and potential for tradeoffs of these gains dramatically improves the constraint space across numerous aspects of the design of high speed electronic devices.

9/7/2012 11:58 AM EDT

Many software vendors are attempting to price their applications based on Cores (Physical and Virtual). Not so much right now on mobile, but that could be coming later. When talking about Servers especially, this could have a big impact on the price of an application implementation, especially if the direction of a server with a lot of cheaper, more power efficient ARM processors. It will be interesting to see how this plays out.

9/7/2012 12:30 PM EDT

The title mentions mobile clients, but the article talks about high-speed interconnect, which is mostly irrelevant to this segment. If Intel wants to regain lost ground in client space, they need to get serious about inexpensive and energy efficient processing---almost all Intel chips require heatsinks and fans, while almost no ARM chips have one.

Another thing Intel needs to do is SoC systems aka microcontrollers. There's a plethora of ARM microcontrollers with integrated peripherals for communication (USB/ethernet/serial) and control (analog and digital)---but nothing I can think of from Intel.

9/7/2012 3:02 PM EDT

Rick, what was the conclusion of the HotInterconnect panel which speculated on Intel's acquisition of infiniband technology

9/7/2012 3:22 PM EDT

I think Intel are going about it the wrong way. They should ditch x86 in mobile, and loosen up with proprietary techs. Customers have choice now, we are not in the 80-90's anymore....

9/7/2012 6:47 PM EDT

There are a lot of interesting things shaping up in the server world. There is the ARM vs. x86 battle that is brewing (battle is an overstatement and a lot of folks are jumping the gun but it is surely going to come to a head in the next 5 years). Then there is the other factors like more efficient integration, interconnect like the appliedmicro folks. calxeda and seamicro are touting the scale-out fabric.

This space is definitely seeing a remorphing and fundamental assumptions that have stood the test of time for decades are being challenged. This leaves room for new players to get a toe hold in.

9/8/2012 8:18 AM EDT

I'm not so sure any of these electronic interconnects will be long lived, particularly in HPC. Without much fanfair, IBM, Intel, and several university groups have been making great strides in silicon photonic devices. An on-chip or integrated chip-to-chip optical interconnect would almost instantly render all of them obsolete both in performance and power consumption.

11/11/2012 12:09 AM EST

Exit X86-Windows 8 Prison Blues - http://www.youtube.com/watch?v=kqio-PxEjHc

The market for x86 based consumer client computers is watched intently by my Wall Street colleagues. The key indicator currently is Windows 8. If this new Microsoft facelift does not spark a popular revolution in consumer client then X86 will spend its last days in the IT closet hospice. X86 is known to our inner circles as X-cessive Energy and will equilibrate amongst the fixed platform gamer and enthusiast crowds and those TAMs are not exiting from a growth opportunity perspective.

The financial barrier to innovate around Windows 8 is an order of magnitude greater than ARM. Not to mention CISC power requirements. I know all the arguments about current ARM not being able to handle the popular client workloads. Those popular workloads are being abandoned and Windows 8 may serve as the prime motivator to help accelerate that exodus....as in goodbye Redmond.

Windows 8 is what we’re watching because it could help the soften X86 consumer client decline...bets are already placed.

Guaranteed, we're not going to let our money sit on energy obese tech in light of the cloud...no matter how many gamers overclock and buy multiple displays....those numbers don't work.