Micro-OPs are not RISC because have no relative roots between them, and because there are not instructions. Actually today’s CISC instructions are very complex and they don't executed at once but in frames in space or time/hz. These frames are decided by the MicroCode of the processor and called Micro-OPs. This is the only way to produce a modern CISC, by lowering the different execution units in a number between 3 and 5 instead of 10-20(impossible), while RISC needs only one (vectors). There are more clever ways like MISC and OISC but these are not for today’s intelligence. Anything I said above is 100% correct, some of you must learn to study first.
You can actually find studies on the Internet about: MicroCode vs RISC vs VLIW.
x86 has more risk of going dead than ARM. Dead = mainstream rarity.
Over time there has been many attempts to introduce a new CPU for Windows users. More and more stories have been added to the x86 house and it is getting tall and weak at the foundations. The makers would love a new set of registers, op codes and features. For it to die though would need a concerted effort from Intel, AMD, Microsoft, Apple, and other users+makers of the CPU structure.
IIRC, Linus was laughed at for creating a *nix kernel that would run on his 386.
Originally Posted by maligor
ARM: Fedora 17 ARM Release Candidate VFAD results
Sat Jun 16
"Paul Whalen redhat.com Good day all,
Good day all,
Thanks to everyone who was able to participate in our VFAD this afternoon. Below is a summary of the results:
The following RC1 images (http://scotland.proximity.on.ca/arm-...o-mirrors/RC1/) worked as expected and all tests were successful:
Pandaboard w/serial console & SD
Trimslice Bare/Value Pro/H/H250 w/serial console & SD
Trimslice Pro/H/H250 w/serial console & SD/SATA
Highbank w/serial console & SATA
Kirkwood(Guruplug) w/serial console & MicroSD
The following images had issues that should be resolved in RC2:
Versatile Express(Qemu) - Did not boot, kernel oops. Resolved in RC2.
Versatile Express+XFCE (Qemu) - Did not boot, kernel oops. Resolved in RC2.
Pandaboard+XFCE w/SD - The RC1 image incorrectly booted to multi-user mode. Fixed in RC2. Using the GUI, options to reboot/shutdown, automatic mounting and updating failed with an authentication error. Bug to be filed.
The following images had issues that require further investigation/action:
Beagleboard XM w/SD - Booted successfully on a Rev B board, subsequent boots produce a kernel oops. Failed to boot using a Rev A2 board.
Efika SmartTop iMX w/SD - Booted successfully, subsequent boots produce a kernel oops.
The unofficial nightly images for the Raspberry Pi (http://scotland.proximity.on.ca/arm-nightlies/) were also tested:
Raspberry Pi w/SD - Works as expected.
Raspberry Pi+XFCE - X not installed. To be resolved in future nightly images.
The full results and tests performed can be found here - http://fedoraproject.org/wiki/Archit...ra_17_Test_Day
New images for RC2 are being created now!
Due to the overwhelming success of today, we are holding another VFAD next week:
Fedora 17 - RC2 image validation VFAD
Monday June 18th - 12pm EDT
#fedora-arm on Freenode
Special thanks to the participants today, and we hope you can join us next week.
On behalf of the Fedora ARM Team,
Lately, Phoronix seems to like to mislead their reader with the title, Or use 'strong words' to add more clicks. Like this one. Fedora Discussion: "ARM Is A Dead End. It's as if majority in the discussion agree that ARM is a dead end.
It's unfortunate. I hope for better jounalism on phoronix with each day passing, but looks like it went the opposite.
The reason ARM is going to take over the market isn't because of performance or power consumption, it's because of price. Intel has the X86 architecture by the balls, but anyone can go out and buy an ARM license. Looks at companies like Texas Instruments, Samsung, Qualcomm, and even Apple. These are not companies known for making processors, but they're giving Intel a run for their money.
All we need is someone to come along and make a desktop ARM chip. One that is tweaked to hell and clocked at 3-4 Ghz. I could easily see AMD or Nvidia making a motherboard chipset that works for both X86 and ARM CPUs to be inserted.
Much like AMD, the purpose of ARM is to help remind Intel that they aren't a monopoly.
I'm highly skeptical of any attempt to make an ARM chip competitive with Intel on the desktop. Intel's manufacturing capabilities give it a large advantage over everyone else, and the ARM companies have no experience building those kinds of chips.
Originally Posted by Dukenukemx
Where they can succeed is in the server space, where efficiency and power usage are important, and they can easily scale to many cores to provide performance.
I'd love a 32-core ARM laptop with two days' battery life.
You can already buy a tablet and attach an external keyboard.
Originally Posted by curaga
The problem with a 32-core ARM laptop is that nothing on the desktop uses 32-cores, and each one individually is way too slow.
Last edited by smitty3268; 06-17-2012 at 07:50 AM.