Rumors abound that Intel is designing 64bit extensions to it’s Pentium line, in case Itanium turns out to be a flop: “Intel’s decision to back the novel Itanium architecture had upset a small group of Intel engineers in Oregon, who preferred to build on the x86 legacy. When AMD released the specifications of its upcoming 64-bit chips in the summer of 2000, these “cowboy” engineers decided that Intel needed to match its rival. They began developing their own 64-bit extensions to the Pentium line, making sure the code was compatible with AMD’s design.” Update: MercuryCenter has an article about this too.
How ironic: Intel playing catch-up with AMD.
I found it especially interesting that they are keeping compatibility with *AMD*’s instruction extensions. How about that? ;^)
Yamhill is an Indian tribe hear in Oregon, Thats pronounced OR-EE-GUN.
There is a county, river, and town named after this tribe, In fact a large percentage of the names of places in OR-EE-GUN are named after indian tribes.
Now I have heard rumors that Itanium will be a flop, seems that it just might be.
It is raining right now. It has been raining for the past several days.
I expect it will be raining for the next 4 months(on and off).
Yes I like rain.
OOps I used the wrong “hear” I should have said
Yamhill is an Indian tribe here in Oregon.
i live in oregon too. just thought i would let ya know….and i too like the rain…ppl in oregon either like rain or move….i met a family once that did move cause they hated the rain….but anyway.
that is very ironic the intel is playing catchup but it will likely not matter as the right amoumt of advertising and the consumer won’t know who came up with it. i would like to see ppl start goin the way of the powerpc…they are 128 bit suckers…
Hey there mike,a fellow oregonian.
I have been reading about the Amiga SDK and DE enviroment, and I think this might be a good solution to our problems.
If this idea works then the underlying system is no longer of any relevance.
You write the code once and it runs on almost any system.
Being able to disconnect the hardware from the software will stop all this backwards compatability nonsense we have now. Someone like Intel would be free to develope any kind of hardware they wanted to all that would be need is the translation layer and all your apps run.
Games would be really helped by this because they would need to only target the translation layer instead of the whole os/harware. This would greatly increase their sales because anyone who could run the DE enviroment could run their games. If games would work then any software would because games tax systems the most.
Am I understanding this correctly?
Like microsoft will ever let that happen, hey a guy can dream can’t he?
things were beautfiful in the land of processors, back in the early days when things were new and (very) simple. we only had 8 bits then, and that was all we wanted. everyone was happy when 8086 was crowned by eye-bee-emm, one of the most powerful gods.
8086 reigned for a few years, then his son and his son and his son and eye-bee-emm began to lose power to another god: Mike Rowsopht. he didnt love all the processors and only bestowed power on 80386, with this new power ‘386 built vast new cities, glittering like the stars,
unknown to most, ‘386 had build them on top of the old cities, still built in the days of horse and carriage. with each passing generation the new city became more convoluted and polluted as the shiny new buildings were squeezed round and over the old city.
the god who created all the processors decided that pentium, a weak and hobbled king, would be the last of his line. he teamed up with another god, HP, to create a super-processor worthy of being king and able to clear away the stench and decay of the x86 line.
unfortunately Mike Rowsoft’s legacy of power seems to be winnign again… we can only hope that the kingdom of D’Esktop will one day be invaded by the many tribes calling themselves the Embedded Alliance, and they all have better kings, even ARM…
NOTE: sorry for the horrible symbolism, i dont know what came over me
In fact, it’s only 32bit right now. The registers for Altivec are 128 bits in width, but that doesn’t mean it can address 128 bits worth of memory. I believe that its general registers are 64bits, but this is only to accomodate easy implementation of double floats and long ints.
Just because a processor can handle datatypes of a certain bit length doesn’t mean that they ARE that size. Hell, under that rationale you could say that the x86 CPU’s after the Pentium are 80bit machines since the FPU uses 80 bits of precision. Or, with segmentation you can address up to 36bits worth of memory, so maybe it’s a 36 bit processor?
Don’t be fooled by FUD. If you don’t know, read a technical reference or don’t speak. Consumer PPC’s up to at least the G4 are all 32bit.
> Don’t be fooled by FUD. If you don’t know, read a technical reference or don’t speak. Consumer PPC’s up to at least the G4 are all 32bit.
I think the point here is that FUD rules supream. Apple has tried to avoid fud, and learned their lesson–if you are the only manufacturer advertising monitors by visable size, not tube size, you lose.
Oh please just let the x86 instruction set die. Give me a reasonable number of general purpose registers, real floating point registers (not a stack), push the intelligence into the compiler, and make it big endian for the love of all that is good and does not suck.
has been on the AMD extensions for a while. I felt that Intel ignored that at their own peril. Good to see that they have some sense.
Having said that,I ‘ve always felt that the x86 architecture sucked and there have been plenty of other cleaner architectures that on paper are much cleaner.
I can’t agree with the big endian comment. little endian is logically more sensible for any m/c architecture as byte, word, longword etc can all share the same address location. The only argument that ever held any sense for big endian addresses/data is that they are easier to read on a hex listing. Fine if you like programming with switches – ah I remember the days entering the bootstrap loader for an old PDP8 with the front panel switches.
P
Since the 64 bit x86 is binary compatible with 32 bit x86, this *extremely* good news for asm programmers/applications/os’s.
x86 is outdated, period.
AL, AH ;what remains of an 8bit world
AX ;god awful 16bit world
EAX ;the hack attempt at making things 32bit
I applaud anyone who bashes little-endian, im no fan of it at all. I always end up cursing against it while coding in x86 assembly. I like the alpha instruction set where almost ever instruction has 3 operands. You can mov a 64bit value from one register into another and do a logical (AND|OR|XOR) operation to it in a single instruction and cycle. 6 general use registers? come on! oh wait, there is ebp, that makes 7, still 9 shy of the Alpha. The alpha has registers that are designed to hold parameters for functions. Imagine a compiler for an alpha that could use those registers instead of pushing all of the arguments onto the stack like on the x86.
I know this isn’t the place to play endian wars. I’m just curious, why exactly do you like big endian? Also, refresh my memory as to what the Alpha is (big or little?) Found this reference http://www.bitbanksoftware.com/code5.htm which pretty much sums it up for me.
P
Subject tells all. Alpha can be configured to be either endian. Yet another thing to love about Alpha: it leaves endian flame wars to the masses/implementors.
Gimme low power consumption and no fans. The latest chips are consuming power like crazy.
How cool would it be to assemble your computer, turn the power on and hear…nothing. No whine, the silent HD the only hint that it’s breathing. Ahh yes, that be the puter for me! (Macs are out of the question here, unfortunately, before someone jumps in 🙂
Why is x86 so much more inefficient with power? 50-60W for a chip (I don’t care how fast it is) is ridiculous.
They already exist! You can get x86 clone chips from VIA (the division that used to be Cyrix) that are intended to only run with a heatsink. I believe that they run up to 1 GHz, too!
Of course, you take a performance hit. Then again, if you want all that power, but don’t want to accept having a fan, then you’re just being unreasonable.
What I like about big endian, among other things, is that you bit-centric operators (shift, and, or, etc.) are actually doing to your data what you think they are doing. In other words, C and C++ are already big-endian. It is also much easier to read memory when debugging, or to read file formats on disk, etc.
Although I understand the argument that you can cast little-endian data to different sizes without changing the pointer, I don’t buy it. This is not genuinely useful to software engineers – it only makes a difference to chip designers. If you compare the number of people who program for x86 chips to the number of people who architected them, this should be a no-brainer. Endianess should come down to the convenience of the largest number of people who have to deal with it, which is overwhelmingly in the software camp.
Not to mention big-endian is good, right, and has been ratified as the standard by almost every human language (Arabic and Hebrew being notable exceptions).
Of course, you take a performance hit. Then again, if you want all that power, but don’t want to accept having a fan, then you’re just being unreasonable.
—-
I have a habit of doing that
I really, really hate fans and their incessant whine, though. (power supply fans included). I think it should be possible to make a silent high perfomrance x86 computer, but for some reason, it just hasn’t caught on. Hmmm, they can send man on the moon but can’t make Joe User’s computer silent. *scratches head* It ought to be possible & cheap, I reckon, with a little thought.
OpinionBoy thus wrote:
I really, really hate fans and their incessant whine, though. (power supply fans included). I think it should be possible to make a silent high perfomrance x86 computer, but for some reason, it just hasn’t caught on. Hmmm, they can send man on the moon but can’t make Joe User’s computer silent. *scratches head* It ought to be possible & cheap, I reckon, with a little thought.
I follow up with:
For a completely silient PC just take a look at the humble i486. Open up any old 286/386/486 and you’ll find no CPU fan. So I suggest that in order to create a fanless CPU, you need to control the heat, so in order to control the heat you need to use less power, and to use less power you make it slower and smaller, (aka the common embedded CPU).
As for the 64bit x86, please don’t. While legacy support is a good thing, one must realise when it is time to drop that support (and the 32->64 bit move would be a good time, as Apple dropped the 68K for the PPC).