Linked by Thom Holwerda on Mon 16th Mar 2009 17:04 UTC
AMD Recently, AMD spun off its manufacturing business in a partnership with the Abu Dhabi government into Global Foundries. Apparently, Intel isn't very happy about this, and has said in correspondence to AMD that the patent cross-license agreement from 2001 has now been broken by AMD.
Thread beginning with comment 353335
To view parent comment, click here.
To read all comments associated with this story, please click here.
poundsmack
Member since:
2005-07-13

While there are many good architectures they all have limitations:

SPARC: currently only developed by Fujitsu and Sun. While an open standard it hasn't gained much headway.

ARM: ARM has only recently started a push on preformance. While it woul be fine if it offloaded virtually everything to a GPU it isn't suited well for multi aplication work load. this however has improved greatly over the last year and the upcomming products are very promising.

Itanium: I like this chip I do. it's not that hard to code for, and is faily straight forward. downsides is that it is really expensive due to limited quantities and a few other things. the good news is that windows is suported on it and the new Itaniums comming out are smaller faster and cheaper.

Cell: a nightmear to code for, enough said.

PPC: not to hard to code for, loads of linux suport, and freescale has some chips that make great desktop processors.

MIPS: no

x86-64 is good. is it the best archetecture out there? no, but it does what is needed and its a lot better than it used ot be. I don't mind it as the standard and think that this will go away soon as neither side can survive the loss of the other.

Edited 2009-03-16 23:15 UTC

Reply Parent Score: 2

lemur2 Member since:
2007-02-17

While there are many good architectures they all have limitations: SPARC: currently only developed by Fujitsu and Sun. While an open standard it hasn't gained much headway. ARM: ARM has only recently started a push on preformance. While it woul be fine if it offloaded virtually everything to a GPU it isn't suited well for multi aplication work load. this however has improved greatly over the last year and the upcomming products are very promising. Itanium: I like this chip I do. it's not that hard to code for, and is faily straight forward. downsides is that it is really expensive due to limited quantities and a few other things. the good news is that windows is suported on it and the new Itaniums comming out are smaller faster and cheaper. Cell: a nightmear to code for, enough said. PPC: not to hard to code for, loads of linux suport, and freescale has some chips that make great desktop processors. MIPS: no x86-64 is good. is it the best archetecture out there? no, but it does what is needed and its a lot better than it used ot be. I don't mind it as the standard and think that this will go away soon as neither side can survive the loss of the other.


People don't code for chips any more ... compilers do that.

Did you know that GCC has just gained support for "automatic parallelization"?

Obligatory supporting link:
http://www.phoronix.com/scan.php?page=news_item&px=NzEzNA

That sort of thing could make the multi-RISC-core approach (SPARC, ARM and Cell) a lot more interesting, don't you think?

There is a lot of source code widely available (with no-risk-of-being-sued license terms) that is readily compiled by GCC.

Reply Parent Score: 3

bannor99 Member since:
2005-09-15

While there are many good architectures they all have limitations:

SPARC: currently only developed by Fujitsu and Sun. While an open standard it hasn't gained much headway.



MIPS: no



What kind of arguments are these? Sparc being an open standard would be a big win for consumers as any one could take it to a bigger market - or design an enhanced SPARC chip, if they have the know-how.

And, MIPS? It has been long stated by many engineers that the MIPS is one of the best designed processors ever. I believe it's for that reason it has been the real platform of choice for computer architecture classes.
Of course, all of those courses eventually focus on x86
( and PPC ) due to their ubiquity.

The greatest obstacle any alternative platform that tries to make headway on the desktop faces is to get Ms to port their OSes and apps.

Reply Parent Score: 1

poundsmack Member since:
2005-07-13

@lemur2
yes I know about automatic parallelization and setting a compiler to go for multiple targets and letting it write the machine code. I write cross platform apps across multiple architectures, including every one i listed ;)

@bannor99

i never said SPARC being an open standard was a bad thing, its a GREAT thing! SPARC is my favorite platform of them all.

Also I agree with you about MIPS its a great architecture, but since SGI no longer makes desktop and server grade MIPS chips, it has turned into the next "likely to die off" architecture ;)

Reply Parent Score: 2

Treza Member since:
2006-01-11

SPARCs :
Developped also by Gaisler Research, open source GPL version available (LEON3), in some of the satellites flying over your head.

MIPS :
The traditional (32bits) version is expected to be free by now, the awful unaligned accesses patent is no relevant anymore.
There is also the Chineese Loongson, MIPS 64bits, 4 way superscalar (officially licenced from MIPS Tech. by ST for uses outside of China, AFAIK)

Itanium :
Not hard to code ? Really ? Grouping the instructions by 3, using 41bits opcodes, and taking care of dependancies ? I've seen many DSPs more easily manageable than that.

If the many core architecture goes mainstream, with software able to use it, then the x86 may have some contenders, beyond the GPU with many parallel pipes.

Reply Parent Score: 1