InterviewSolution
Saved Bookmarks
InterviewSolution
| 1. |
Solve : AMD makes pin compatible ARM or x86 chip designs.? |
|
Answer» Here is one report: AMD unveils plans for ARM, x86 chips that are pin-compatible Another report says about the same. AMD Announces ARM-based "K12" Custom Core, Pin-Compatible x86/ARM Chips Coming In 2015 Quote ... Earlier AMD say they were going to do this kind of thing. What this does NOT mean. It does NOT mean an ARM chip will execute a x86 binary. NO! It does NOT mean a hobbyist can simply swap out ARM and replace it with a the x86 chip. For one thing, the BIOS must be changed. Not easy. The advantage is for OEMs who get orders from resellers who want custom low-cost motherboards. This new design motherboard can be tailored to a specific application. Thus reducing the need for a diverse inventory. But what could this mean for user ordinary users? I think that is a good idea. It gives companies a choice without having to a board for each cpu.I like the idea, but it would be nice is they added a bios feature to have a dual-bios so that is its an ARM or x86 it will auto detect the CPU and run with it. All they would need is say an extra 2 pins to the CPU for a HIGH or LOW when the CPU is installed. If the pins that act as a enable/disable of ARM are used and the BIOS at the lowest level detects to use one or the other then there is no having to worry about reflashing a board to switch from ARM to x86 CPU. It would make a switch between ARM and x86 as QUICK as a normal CPU swap, but then there is the OS that needs to be installed for either ARM or x86. If I was AMD this is what I would do to make it handle both and autodetect via an electrical contact pin key. Both CPU's could appear the same at the bottom, but one CPU is open between the pins and the other is SHORTED between the 2 pins to act as an identity for the CPU.The problem is that ARM and x86 have conflicting requirements- eg certain pieces of information (interrupt tables and whatnot) which of course have their own specific format and arrangement with both, and the addresses required for each overlap, so it's impossible to have both mapped at the same time. The only way would be to make it solid state and handled via circuitry- eg an extra pin as mentioned for one or the other that can be used to bankswitch to a different ROM. The 'switch' would need to occur before any booting takes PLACE at all- because booting obviously would run Machine code and they aren't cross-compatible in any way. Thing is, this would of course require motherboard support- you would never be able to drop in an ARM CPU in a motherboard designed for an x86 CPU and vice versa; which raises further interesting questions particularly since Video BIOS ROM and the ROM code found in other expansion cards is going to typically be X86, and as I recall ARM does not have similar allowances for ROM Shadow by add-on cards (I'm not even sure what Bus it would use, I'm assuming PCI-E would work for ARM).The other problem is is that this is 2 year old news...Quote from: patio on May 08, 2014, 08:19:51 PM The other problem is is that this is 2 year old news...Google it... Quote AMD Unveils Project Skybridge, a Single Chip for Both x86 ...EDIT. Kinks repaired. I saw a WW2 documentary yesterday, That means it happened last year, right?Quote from: BC_Programmer on May 08, 2014, 11:28:05 PM I saw a WW2 documentary yesterday, That means it happened last year, right?Facebook started the Open Computer Project** in 2011. And in 2012 AMD announced their intent to make new chips with 64 bit designs This year, 2044, AMD announced the idea of a physical design for a socket that could work on both chips. They claim the product will be ready in 2015. Recently other companies have shown serious interest in the Facebook Open Computer Project. So it was not a documentary of WW II. ** http://en.wikipedia.org/wiki/Open_Compute_Project Quote ...The above quote shows this is a real project for improving data center. The stories I linked are not stories from two years ago. The AMD thing is this year. For one think that the author is quite right. AMD can't win Intel at its own game (they have tried, they even had once a better product, but we all know that at the end it didn't matter), so they need to take a huge risk and try to change the game.Quote from: Geek-9pm on May 09, 2014, 12:11:19 AM Facebook started the Open Computer Project** in 2011. And in 2012 AMD announced their intent to make new chips with 64 bit designs This year, 2044, AMD announced the idea of a physical design for a socket that could work on both chips. They claim the product will be ready in 2015. None of this is even tangentially related to what I said. Patio said that it was 2 years old. Partly true, considering AMD announced ARM-based Opteron's in 2012; AMD working with ARM is not news- it is at least 2 years old. That is what you refer to there. My analogy was that your "citations" that this is not news is based on the age of articles posted. Google results are not citations- they are Google results. If somebody said WW2 was old news and I posted a link to a google result to a page written 2 days ago- does that make it news? Of course not. That said, the actual Pin compatible stuff IS news. I was mostly put off by your inability to actually defend that yourself. Quote from: jackj on May 09, 2014, 12:28:24 AM For one think that the author is quite right. AMD can't win Intel at its own game (they have tried, they even had once a better product, but we all know that at the end it didn't matter), so they need to take a huge risk and try to change the game. It's not really a case of "Winning", I don't think. It's a case of whether the market is profitable for them. Intel and AMD have been leapfrogging each other since 2000 or so, and both companies have done quite well in that- and consumers have gained because of the competition- winning teams all around. AMD's huge risk was the Slot A Athlon which was pretty much when it came down to two choices for CPUs. I don't think it is fair to AMD to say they cannot compete, when they haven't just competed but actually defined standards that Intel now follows (IA-64). |
|