Worst ISA evar

[jhu]

This thread probably won't go over too well either.

I hope it does. We won't hesitate to lock it if it ends up being a flamefest
-ViRGE

 

[ALIVE]

well i do not think everyone will come on the defense of the x86 isa
it really sucksssssssssss

 

[_Rick_]

The thing with x86 was, that it's based on an iterative process of "a good idea at the time" and "we've got to support ALL of it".
On the other hand, IA64's VLIW ISA didn't exactly get compiler developers excited either.

Soon we might be at the point, where x86 software emulation might be fast enough to re-arrange the instruction set, and standardize it, so we don't get any more 3Dnow vs SSE and FMA3 vs AVX vs SSE5 vs FMA4 stupidness.

Of course, in the end, in real life, it doesn't matter, as Java is slow on all platforms.

 

[gevorg]

end users might consider x86 as the worst/bad, but x86 is Intel's best weapon to maintain their near-monopoly on the desktop arena

 

[Homeles]

end users might consider x86 as the worst/bad, but x86 is Intel's best weapon to maintain their near-monopoly on the desktop arena

And here the issue lies...

 

[TuxDave]

well i do not think everyone will come on the defense of the x86 isa
it really sucksssssssssss

Call me crazy.... but I don't think you know what you're talking about.

 

[alyarb]

The existence of this thread says much more than any post inside it

 

[fierydemise]

Having worked with the MIPS ISA a fair amount I despise the branch delay slot but most ISAs have their own "fun" quirks.

 

[Idontcare]

This is one poll that I won't be voting in because I know absolutely jack-diddly about the nuances of these different ISA's.

I know of them, but I don't know them, certainly not enough to rank one as better or worse than another.

Everything that I ever coded for was on x86, save for one project on PA-RISC that was truly the blind leading the blind (that was me, leading myself).

 

[Ferzerp]

This poll is for armchair processor architects. x86 will win because of the geek urban legend that it is awful, has too much cruft, etc


I gave arm a spite vote

I doubt more than one or two of the votes even has a clue how to vote (and I'm not one of them).

 

[Cerb]

IA-64, due to being the computer ISA equivalent of supply-side economics.

Every other computer/ISA in the list, and most others not on there, was fundamentally good, coming about as a response to business needs and technological innovations from its time of inception.

 

[dmens]

There's something other than x86? IMPOSSIBRU.

 

[Ben90]

I doubt more than one or two of the votes even has a clue how to vote (and I'm not one of them).

I'm amongst those who have no idea about ISAs. I can safely say it would probably suck pretty hard to program for the ENIAC though.

 

[palladium]

Probably need people like ruby or pm to chime in...

 

[jhu]

The correct answer is ENIAC unless you don't mind rewiring your computer just to program it.

 

[Homeles]

The ENIAC hasn't been holding back the computing industry for however many years.

 

[denev2004]

The correct answer is ENIAC unless you don't mind rewiring your computer just to program it.

......Calling it a computer NOW just seems to be so...strange

Although I know It does has some basic structure of modern computer.

 

[ll350]

This thread probably won't go over too well either.

I hope it does. We won't hesitate to lock it if it ends up being a flamefest
-ViRGE

Now I'm interested. How exactly are individuals supposed to express their extreme dislike for a particular idea, without those who believe the idea has merit reacting negatively? And then each side incrementally ratcheting up the heat? Perhaps if the poll was "Best ISA EVA?" then it MIGHT be possible. I believe what you are expecting here is the Internet equivalent of a chupacabra sighting. I don't think it's impossible, just highly improbable.

On a related note, a hypothetical question: Does this post count as flaming?

 

[Cerb]

The correct answer is ENIAC unless you don't mind rewiring your computer just to program it.

For the technology and general understanding of the time, ENIAC was genius. It was a fully programmable (like FPGAs, but with people to handle the FP part ), 'cycle'-level debugable, computer system, back when the best logic devices available were punch cards and overcomplicated light bulbs.

Sure, I could make something more powerful that runs on lava and water in Dwarf Fortress (if only I could keep my dwarves and cats in check!), but hey, that's what over half a century of technological development gets you.

Perhaps if the poll was "Best ISA EVA?" then it MIGHT be possible.

On that note: RX, IMO. Combining the institutional knowledge from SuperH and 68k. In 10-15 years, it'll need to be something different, of course. 15+ years ago, it probably would have been ARM.

 

[sandorski]

As an End-User for almost 30 years, x86 is my favourite. I don't care if Software Developers hate it, that's not my problem.

 

[nenforcer]

Considering the amount of money spent on developing it and how short its actually been around (1 decade), the IA-64 must take the cake.

The massive subsidies paid to the likes of HP, Microsoft and Oracle just to continue supporting it is bewildering.

Even every Linux distribution has pretty much abandoned the architecture in favor of SPARC64, MIPS64, ARM and/or X86-64.

Also the fact that AMD first designed and built X86-64 and Intel had to copy as the rest of the industry followed Opteron instead of Itanium.

 

[ShintaiDK]

Considering the amount of money spent on developing it and how short its actually been around (1 decade), the IA-64 must take the cake.

The massive subsidies paid to the likes of HP, Microsoft and Oracle just to continue supporting it is bewildering.

Even every Linux distribution has pretty much abandoned the architecture in favor of SPARC64, MIPS64, ARM and/or X86-64.

Also the fact that AMD first designed and built X86-64 and Intel had to copy as the rest of the industry followed Opteron instead of Itanium.

IA64 is actually still very popular and outsells Sparc. The bigtin market is between POWER and IA64 atm. Revenue of IA64 is also higher than AMD for example.

AMD64/EMT64 is a huge disaster longterm. But it was the easy way out. The market and selection of ISA is unfortuntaely not about whats best. But instead what contains the smallest risk and yields the best profit in a short term view. And why we will most likely sit in the year 2040 with a CPU that can still execute 8bit code with the entire legacy package back to 1972.

 

[ALIVE]

IA64 is actually still very popular and outsells Sparc. The bigtin market is between POWER and IA64 atm. Revenue of IA64 is also higher than AMD for example.

AMD64/EMT64 is a huge disaster longterm. But it was the easy way out. The market and selection of ISA is unfortuntaely not about whats best. But instead what contains the smallest risk and yields the best profit in a short term view. And why we will most likely sit in the year 2040 with a CPU that can still execute 8bit code with the entire legacy package back to 1972.

well i do not see anythign bad if the 20435008 cpus can still run 8 bit code
that would be great
well if the cpu were only 64bit support
i do not think the adoption of them would have been so fast
even now there is still business that still run 16bit programms at their work

amd64 was better than ia64 cause it kept compatibility with the past
if intel was not so crazy about pricing and throw the ia64 a lot cheaper at the begining then the discusion here might have been completly diferent

intel had the reputation
intel had the money
intel lacked the right pricing

 

[ShintaiDK]

Problem with AMD64/EMT64 is it was like peeing your pants to stay warm. It works fine in the beginning, but it quickly goes cold.

Only good thing is that its alot easier to port code from x64 to IA64 and others, unlike from x86. Its also why the adoption of x64 is still so terrible slow 10 years later. Just look on IPv6, its epicly obsolete and actually worse than IPv4 today. But its been in "deployment" now 1996. But we rather jump over the cliff than take a risky move. We already ran out of adresses.

Intel didnt lack the right pricing. They could have sold IA64 chips for 10$ and x64 would still win because MS feared it would lose Windows monopoly. And because taking high risk in a competitive environment is something nobody wants to do. Its all about money and profit today, not tomorrow.

Maybe you dont have a problem with running 8bit code in the future. But chip designers do. It adds development time and cost 10s if not 100s of millions for validation, implementation etc for an uarch. Not to mention all the restriction it adds to improving the overall design.

Or look at BIOSes. We are slowly moving to UEFI tho. But we still need 16bit realmode BIOS programmers today. Problem is, they are rare and I dont think new are coming in.

 

[ALIVE]

Problem with AMD64/EMT64 is it was like peeing your pants to stay warm. It works fine in the beginning, but it quickly goes cold.

Only good thing is that its alot easier to port code from x64 to IA64 and others, unlike from x86. Its also why the adoption of x64 is still so terrible slow 10 years later. Just look on IPv6, its epicly obsolete and actually worse than IPv4 today. But its been in "deployment" now 1996. But we rather jump over the cliff than take a risky move. We already ran out of adresses.

Intel didnt lack the right pricing. They could have sold IA64 chips for 10$ and x64 would still win because MS feared it would lose Windows monopoly. And because taking high risk in a competitive environment is something nobody wants to do. Its all about money and profit today, not tomorrow.

Maybe you dont have a problem with running 8bit code in the future. But chip designers do. It adds development time and cost 10s if not 100s of millions for validation, implementation etc for an uarch.

Or look at BIOSes. We are slowly moving to UEFI tho. But we still need 16bit realmode BIOS programmers today. Problem is, they are rare and I dont think new are coming in.

yes there is the
cpu maker
the software developers
and the consumer

these 3 forces in the market do not want exackly the same product
and thus comptete with each other
what is best for one is worse for the other
the market spins around these 3 forces
the consumers wantes x64 and the consumers made x64 to take the share of the market
is x64 worse than ia64?!?!?
not from the consumers part
if intel sold ia64 for 10$ then right now there would be only ia64 in the market
after all everyone upgrades windows every 2-3 years so it would not be that strange to buy the ia64 and a win version for that
but what was the introdusing price for he ia64?!!??! aiming the server market a very nicht product and even windows were more expensive to buy a license for them
intel become giant because it sold cheap cpu
intel failed to marketia64 cause it was expensive
as it failed with p4 and rambus
pc market was born and still run by one factor
dead cheap product
other you have it or you get out of the game
thats how amd survived some periods
person goes to buy computer person buys cheaper computer
cause persons do not know anything about computers