Linked by Thom Holwerda on Tue 4th Jun 2013 22:08 UTC
PDAs, Cellphones, Wireless "Several Android tablets running on Intel Clover Trail+ Atom processors broke cover at Computex Taiwan. Intel's dual-core, 1.6GHz Atom Z5260 is fueling a Samsung Galaxy Tab 3 10.1 tablet, as well as Asus's 6-inch Fonepad Note and 10-inch MemoPad FHD10 tablets, while Asus also unveiled a hybrid 11.6-inch Transformer Book Trio, combining an Android slate based on a 2GHz dual-core Atom Z2580 with a keyboard dock running Windows 8 on an Intel Haswell processor."
Thread beginning with comment 563891
To view parent comment, click here.
To read all comments associated with this story, please click here.
Kochise
Member since:
2006-03-03

You should show me who prefer x86 and its limited non orthogonal register allocation (AX, BX, ...) vs. full orthogonal processing on ARM Rx registers, and its exceptionally good conditional execution. Its 16 bits Thumb ISA. Etc... OK, it lacks of MUL/DIV in early versions, no FPU in many incarnations, but it works pretty well.

Coding/debugging the x86 in assembly, on the other hand...

Kochise

Reply Parent Score: 2

tylerdurden Member since:
2009-03-17

At least two of them work on compilers, actually.

ARM's ISA can be as easily criticized. E.g. Even though it's RISC some implementations break many loads and jumps into multiple instructions (due to truly half assed encodings). Some criticism can also be targeted towards predication issues. Bit-twiddling approaches in ARM can also get monumentally "ugly." Etc.


As I said, it depends on the culture of the developer. In my experience, people who have no issue grasping stack/accumulator machines, can deal with x86 just fine. Whereas others raised to equate microprocessors with register-to-register machines have a hard time wrapping their head around it.

Orthogonality is another of those things. Some people have a quasy religious expectation of a flat architectural register file just because, whereas other programmers deal with banked register groups as being easier for explicit scheduling optimizations.

Furthermore, the OO and AMD64 extensions have made a lot of the issues moot for a long while. Alas...

Edited 2013-06-06 21:05 UTC

Reply Parent Score: 2

Kochise Member since:
2006-03-03

The ARM "bit-twiddling" is yet effective, especially if it can be done in the same instruction instead to explode this into multiple instructions with a branch to select the path. More compact, no prediction.

The OO and AMD64 are nice, they are 21st century technologies though. And you want to put this into a tablet ? I know times will come, but right now, I think it's a little too early.

Kochise

Reply Parent Score: 2