The Intel Core2 was around then and was getting power levels down in that range. The newer portable processors with the x64/x86 instruction set are even lower.
The Intel Atom was around in 2008 and it runs x86 code and is very low power, though the versions available then probably didn't stack up against the Tegras available then.
I just think it was a mistake to go with a processor designed for handheld devices. As I'm sure you know, everything in engineering is a trade off and even if the processor families are upgradable the handheld processors are always going to be more limited than desktop/laptop processors. Parasitic power drain is a concern in an electric car, but the parasitic losses from all the other systems monitoring the rest of the car's systems, an extra watt or two on the central display unit is going to be minimal when the car is idle. When the car is driving, it is down in the noise compared to the power the drive train is pulling.
The Tegra, like other cell phone CPUs is designed to pack as much as possible into the smallest possible space. This is important in something that needs to fit in your pocket, but it's completely unnecessary in something that fits in a 4500 lb car. The car has more room in front of the dash board than 20 of the largest desk top computers you can buy.
If it was my design, I would have chosen the processor family that is likely going to allow the easiest hardware upgrades in the future (x86/x64) and a standard form factor single board computer that has been around a while. There are a number of standardized backplanes used in industry. PC/104 is a common standard with lots of available options:
PC/104 Consortium - Supporting legacy technology while developing new solutions for the future
PC/104 is just one standard, there are others. PC/104 would probably work best in this application from the standards I can remember, but I haven't worked with a current industry standard bus in a few years. What I"m working with on my current project is a modified ISA bus.
If the car's systems all have standardized interfaces and the OS and apps are designed to run on any computer hardware, production could upgrade the computer in the car once or twice a year and it wouldn't disrupt production at all. If it's a standard board or boards, Tesla could always by pushing suppliers to lower prices because there would be competition.
It would also make expansion of things like Autopilot very easy. If there is some new system that needs to interface, design a custom board to handle it and integrate it into the existing design.
I'm a big fan of flexible design. If power is not critical and space is not at a premium, trade off and go with a little bigger and maybe a little more power hungry solution and have a very wide upgrade path for the future. If they had gone that route, the processors in use today would probably be more efficient than the current processors in use by Tesla and almost certainly faster.
The Intel Atom was around in 2008 and it runs x86 code and is very low power, though the versions available then probably didn't stack up against the Tegras available then.
I just think it was a mistake to go with a processor designed for handheld devices. As I'm sure you know, everything in engineering is a trade off and even if the processor families are upgradable the handheld processors are always going to be more limited than desktop/laptop processors. Parasitic power drain is a concern in an electric car, but the parasitic losses from all the other systems monitoring the rest of the car's systems, an extra watt or two on the central display unit is going to be minimal when the car is idle. When the car is driving, it is down in the noise compared to the power the drive train is pulling.
The Tegra, like other cell phone CPUs is designed to pack as much as possible into the smallest possible space. This is important in something that needs to fit in your pocket, but it's completely unnecessary in something that fits in a 4500 lb car. The car has more room in front of the dash board than 20 of the largest desk top computers you can buy.
If it was my design, I would have chosen the processor family that is likely going to allow the easiest hardware upgrades in the future (x86/x64) and a standard form factor single board computer that has been around a while. There are a number of standardized backplanes used in industry. PC/104 is a common standard with lots of available options:
PC/104 Consortium - Supporting legacy technology while developing new solutions for the future
PC/104 is just one standard, there are others. PC/104 would probably work best in this application from the standards I can remember, but I haven't worked with a current industry standard bus in a few years. What I"m working with on my current project is a modified ISA bus.
If the car's systems all have standardized interfaces and the OS and apps are designed to run on any computer hardware, production could upgrade the computer in the car once or twice a year and it wouldn't disrupt production at all. If it's a standard board or boards, Tesla could always by pushing suppliers to lower prices because there would be competition.
It would also make expansion of things like Autopilot very easy. If there is some new system that needs to interface, design a custom board to handle it and integrate it into the existing design.
I'm a big fan of flexible design. If power is not critical and space is not at a premium, trade off and go with a little bigger and maybe a little more power hungry solution and have a very wide upgrade path for the future. If they had gone that route, the processors in use today would probably be more efficient than the current processors in use by Tesla and almost certainly faster.
