Sign in with
Sign up | Sign in
Your question

96/128 bit cpus

Tags:
Last response: in CPUs
Share
February 8, 2013 4:36:21 AM

Do you think we will see those become mainstream in the next 5~10 years? (Do they even exist as of now? O.o) or will we just keep on adding more cores and speed?

More about : 128 bit cpus

a b à CPUs
February 8, 2013 5:32:57 AM

Anything above 64 bit CPUs is more than we can possibly use ATM. Most people can get by on ~4 GB of memory which can saturate a 32 bit processor system but doesn't even come close to the limits of 64 bit memory-CPU. Somebody correct me if I'm wrong but I believe the max theoretical limit of memory that can be addressed with 64 bit processors is 16 exabytes, which you can imagine is a long way off from becoming the norm.

IMO in the next 5 years, chip design will have a heavy emphasis on lowering power requirements.
a b à CPUs
February 8, 2013 6:46:44 AM

Memory addressing is only one aspect. 64-bit XP was not sold on the basis of memory addresses... at that time 4GB was considered huge and not needed.

Rather, the idea was that some programs needed to work with numbers that could only be expressed in 64-bit. Some programs were supposed to see improvements in speed because of this. I think there ARE some scientific programs that benefit, but mostly for the bulk of computing it made no difference. So these days we just talk about memory. And 64-bit will be enough for at least the next 20 years.
Related resources
a b à CPUs
February 8, 2013 8:27:37 AM

Clearly you haven't seen the numbers 64 bits can handle. Just 32 bit alone handles 0 to 4,294,967,296 and each bit you add doubles the capacity.
a b à CPUs
February 8, 2013 9:43:03 AM

2^64 = 1.844674407*10^19 so 16 Exabits
February 8, 2013 10:12:37 AM

The first mainstream 16-bit CPU started production in 1978 (Intel 8086).
The first mainstream 32-bit CPU started production in 1985 (Intel 80386).
The first mainstream 64-bit CPU started production in 2003 (AMD Opteron). (note: Intel had the 64-bit Itanium from 2001, but that was not mainstream).
So 7 years of 16-bit, 18 years of 32-bit....so 20 more years we are safe. I would rather say the mobile chips will take over before we increase the bits.
About RAM size, only a single process is limited to 4GB, but ever since 1995, the 32-bit x86 Pentium Pro CPUs could use up to 64GB RAM (36-bits). Thus you could use 16 4GB-limited applications all in RAM (for windows, it was actually 32 maxed applications, as windows only gives 2GB for its processes. PS: to prevent trolls, you can increase the app limit to 3GB).

What I want to say is that the "bit size" is given by internal architecture (like the size of the instruction pointer, registers and integer arithmetic units) not by external bus size.
a b à CPUs
February 8, 2013 8:56:29 PM

FinneousPJ said:
2^64 = 1.844674407*10^19 so 16 Exabits

What?
a b à CPUs
February 9, 2013 8:46:52 AM

What what :lol: 
a b à CPUs
February 9, 2013 8:14:02 PM

mathew7 said:
The first mainstream 16-bit CPU started production in 1978 (Intel 8086).
The first mainstream 32-bit CPU started production in 1985 (Intel 80386).
The first mainstream 64-bit CPU started production in 2003 (AMD Opteron). (note: Intel had the 64-bit Itanium from 2001, but that was not mainstream).
So 7 years of 16-bit, 18 years of 32-bit....so 20 more years we are safe. I would rather say the mobile chips will take over before we increase the bits.
About RAM size, only a single process is limited to 4GB, but ever since 1995, the 32-bit x86 Pentium Pro CPUs could use up to 64GB RAM (36-bits). Thus you could use 16 4GB-limited applications all in RAM (for windows, it was actually 32 maxed applications, as windows only gives 2GB for its processes. PS: to prevent trolls, you can increase the app limit to 3GB).

What I want to say is that the "bit size" is given by internal architecture (like the size of the instruction pointer, registers and integer arithmetic units) not by external bus size.


The part about the Opteron being the first 64-bit CPU is partially wrong; mainstream 64-bit CPUs were available in the 90s but they where all RISC and non x86 designs.
a b à CPUs
February 9, 2013 8:25:54 PM

Here is an example of what we are talking about:

http://www.stata.com/products/64-bit-intro/

In this case, the company claims better performance of their software under 64-bit. How much of this is due to memory and how much is due to such things as word size? Or could it be that they simply re-wrote the core software for 64-bit and so got better efficiency? We don't really know.
I just remember the claims about how much better 64-bit was going to be for the general desktop user.
a c 100 à CPUs
February 9, 2013 11:09:50 PM

mathew7 said:
The first mainstream 16-bit CPU started production in 1978 (Intel 8086).
The first mainstream 32-bit CPU started production in 1985 (Intel 80386).
The first mainstream 64-bit CPU started production in 2003 (AMD Opteron). (note: Intel had the 64-bit Itanium from 2001, but that was not mainstream).
So 7 years of 16-bit, 18 years of 32-bit....so 20 more years we are safe. I would rather say the mobile chips will take over before we increase the bits.
About RAM size, only a single process is limited to 4GB, but ever since 1995, the 32-bit x86 Pentium Pro CPUs could use up to 64GB RAM (36-bits). Thus you could use 16 4GB-limited applications all in RAM (for windows, it was actually 32 maxed applications, as windows only gives 2GB for its processes. PS: to prevent trolls, you can increase the app limit to 3GB).

What I want to say is that the "bit size" is given by internal architecture (like the size of the instruction pointer, registers and integer arithmetic units) not by external bus size.


You can use up to 64 GB of RAM (36 bits) using the Physical Address Extension with most newer CPUs in 32-bit mode. However Windows generally works very poorly with PAE (broken drivers) and only a couple of the 32-bit Windows Server variants even allow it to be turned on. Ironically PAE not working is probably the reason we have much in the way of 64-bit Windows.

Proximon said:
Here is an example of what we are talking about:

http://www.stata.com/products/64-bit-intro/

In this case, the company claims better performance of their software under 64-bit. How much of this is due to memory and how much is due to such things as word size? Or could it be that they simply re-wrote the core software for 64-bit and so got better efficiency? We don't really know.
I just remember the claims about how much better 64-bit was going to be for the general desktop user.


It is likely a combination of being able to use more real and virtual memory, larger word size, twice the number of registers, and also knowing that any 64-bit CPU supports at least SSE2 and using SSE2 rather than legacy x87 code for FP work.
a b à CPUs
February 10, 2013 6:11:25 AM

I had forgotten about that. I knew there was some relationship to instruction sets... the AMD64 CPUs had SSE2 and that was why people were interested in both at that time.

Thanks for clearing that up!
a c 186 à CPUs
February 10, 2013 6:29:24 AM

Do da math. :p 
a b à CPUs
February 10, 2013 6:35:59 AM

:sol: 
!