Windows 7 or vista..?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2007
you can reach far higher theoretical amounts of RAM, more then you'd ever need.


What was it that Bill Gates said about memory a few years back...:D

There are physical memory address limits with a 32 bit bus width that limit it to 4GB. This is a hardware limit. A 64 bit bus can address quite a bit more (double the amount for each extra bit width, for example: 33bit bus width can address 8GB, 34 - 16GB, etc. you get the picture).
 
What was it that Bill Gates said about memory a few years back...:D

There are physical memory address limits with a 32 bit bus width that limit it to 4GB. This is a hardware limit. A 64 bit bus can address quite a bit more (double the amount for each extra bit width, for example: 33bit bus width can address 8GB, 34 - 16GB, etc. you get the picture).

Wowzer that would be a lot of ram. Too bad the average memory slots non-server motherboards have is 4. That would pretty much mean a limit of 32GB.

That would only be roughly $800 worth of 4GB PC3-10666. :D
 
Last edited:
frugal-phile™
Joined 2001
Paid Member
There are physical memory address limits with a 32 bit bus width that limit it to 4GB. This is a hardware limit.

Are you sure it is a general hardware limit and not an OS limit? Many of the Intel Macs support >4 GB RAM even when running 32 bit OS X. The same machine booted into Windows using Boot Camp is limited to <4 GB.

dave
 
Disabled Account
Joined 2007
It's been quite a while since I thought about this stuff (high school, I was HEAVY into the hardware of the day and did a lot of research on the way my Color Computer worked. I did programming in assembly language as well, for what that's worth. Yeah, geek (but a cool one ;) )) but here goes:
To directly access ram, the memory controller (Northbridge for PCs or the intergated memory controller in the AM2 processors from AMD) needs to be able to address it, as in select a "bank" of ram that is at the physical address. To do this, it uses some of the bits in the bus width (I can't remember if they are the ones at the left or the right) to "pick" which bank to access. This is the physical limit, the bus width.
If there is a software work around, the amount of accessible memory would not be significantly higher and access time would longer (more clock ticks).
 
Last edited:
Why is the 64bit a better system than the 32bit?

Will the 64bit run on all systems?
64 bit offers higher floating point precision and support for larger numbers. As for its relevance to audio, higher floating point precision can possibly improve audio quality if floating points are used in processing, but I believe FLAC uses only integers so it's a moot point. 32 bit is already much more precision than any practical audio frequency delta sigma modulator so going up to 64 bits wouldn't make any difference.

In other applications, however, it can make for a big difference. I remember that the exact same version of D2X-XL ran at about 100FPS when compiled for 64 bit as opposed to only 60FPS in 32 bit.

Also note that business servers and workstations have been available in 64 bit for many years before the first consumer 64 bit PC. And there's a good reason for it. Going to 64 bit would roughly double the number of transistors, which would be acceptable to get more performance from a business machine but was cost prohibitive for home use. Now, advances in semiconductor technology made it very practical for home use.

Another way to think about it: Would you buy a car with a 7 speed transmission and only use 4 of the gears?

Note that a 64 bit CPU does not use less power running in 32 bit compatibility mode. The second half of the transistors apparently continue running but do no useful work.
 
Are you sure it is a general hardware limit and not an OS limit? Many of the Intel Macs support >4 GB RAM even when running 32 bit OS X. The same machine booted into Windows using Boot Camp is limited to <4 GB.

dave

It is both a hardware and software limit. Intel 32 bit CPU has 32 bit virtual adddress space, but it has 48 bit physical address space to address 64GB of physical memory. In order to access those memory, you will need to map them into the virtual address space in order to use them. This requires extra OS support -- Windows NT /PAE mode. Note that the paged memory is usually not first class citizen. You need to utilize special API in order to utilize them. SQL server is a good example of a program that utilize the PAE API -- we have them accessing 14GB RAM at work. I guess the OS could make the memory more available if it puts more efforts on it. I am not familiar with how MacOS implements/support PAE though.

Intel 64 bit CPU supports 64 bit virtual adddress space. If you run 32 bit OS on it, you do not utilize the extra address space. If you run 64 bit OS on it, the OS will utilze the whole address space. All 64 bit applications can utilize the 64 bit address space without special API as well.

Note that 64 bit OS is benefital even with just 4 GB of memory. The memory on your video card needs to be mapped to somewhere, and Windows NT 32 bit makes rooms for them in the 32 bit address space by not utilizing some of those 4 GB of memory. That's why on a 32 bit system you don't see the whole 4 GB available (usually somewhere between 3 - 3.75 GB available). However 64 bit executable is somewhat larger in size and that offsets some of the saving from video memory mapping.
 
Last edited:
I have just moved up to Win 7.

Tried to upgrade from Vista and it failed with error messages.
So had to start with a clean hard disc which was a nuisance.
Since it installed had to go through the repair process three times !

Seems to be similar t oVista with the pain of upgrading but seems to have less bugs than Vista did, this is not surprising as WIn 7 is basically Vista with a few bells and whistles.

Got a bit upset when ebay wouldnt allow me to sell on my Vista !
Still sold it now privately so got there in the end.
 
I have just moved up to Win 7.

Tried to upgrade from Vista and it failed with error messages.
So had to start with a clean hard disc which was a nuisance.
Since it installed had to go through the repair process three times !

Seems to be similar t oVista with the pain of upgrading but seems to have less bugs than Vista did, this is not surprising as WIn 7 is basically Vista with a few bells and whistles.

Got a bit upset when ebay wouldnt allow me to sell on my Vista !
Still sold it now privately so got there in the end.

Yea,I've heard that "upgrading" is a major pain,and it's best to just nuke the drive,and do a fresh install. Upgrading also seems to take longer than a fresh install,sometimes alot longer.
 
Read and see where many of the 32 Bit Server versions address more than 4GB as phofman mentioned earlier:
Physical Address Extension - Wikipedia, the free encyclopedia

Basically no page can be larger than 4GB, but any serious operating system uses paging for memory protection and to support virtual memory. No single process can be larger than 4GB which is not a serious limitation.

I believe that you can hack XP and Win7 to support more than 4GB with a fairly simple registry change:
Make Windows 7 and Vista 32-bit (x86) Support More Than 4GB Memory » Raymond.CC Blog
 
Last edited:
frugal-phile™
Joined 2001
Paid Member
It's been quite a while since I thought about this stuff (high school)

In high school the closest thing to a PC i ever saw was the 1st HP calculators -- and that only because Gulf Oil figured the office needed one (which meant it was more or less my Dad's).

1st year uni i worked on an IBM 360. Algol W & assembler (4k memory segments)

dave
 
Re: Floating Point

Star, any general purpose computer supporting floating point is going to support single and double precision (64 bit). Even the early Intel 8087 that was used with 16 bit processors supported both formats. You don't need a 64 bit processor, indeed the 32 bit VAX super mini supported quad precision (128 bit) floating point.

The 8087 had a microcoded 80 bit ALU to do the math, and it had fairly graceful roundoff and underflow behavior as compared to brute force array processors that were much faster (usually single clock per operation) but less graceful:
http://www.cs.berkeley.edu/~wkahan/ieee754status/754story.html

and:
http://stillwaterscience.ning.com/forum/attachment/download?id=2015812:UploadedFi58:741

I've worked in CPU design, and array processor design for a good part of my career: http://www.linkedin.com/profile?viewProfile=&key=7589630&trk=tab_pro

64 bit machines were classified as Super Computers years ago, such as the Cray:
http://en.wikipedia.org/wiki/Supercomputer

From the above link:
"Supercomputers are used for highly calculation-intensive tasks such as problems involving quantum mechanical physics, weather forecasting, climate research, molecular modeling (computing the structures and properties of chemical compounds, biological macromolecules, polymers, and crystals), physical simulations (such as simulation of airplanes in wind tunnels, simulation of the detonation of nuclear weapons, and research into nuclear fusion). A particular class of problems, known as Grand Challenge problems, are problems whose full solution requires semi-infinite computing resources."

I don't see home or most business users having any real need for 64 bit systems. They are not necessarily faster unless you are doing 64 bit arithmetic which most software does not use. It probably will become mainstream due to marketing reasons - sell another OS = more money for MS.
 
Last edited:
Disabled Account
Joined 2007
In high school the closest thing to a PC i ever saw was the 1st HP calculators -- and that only because Gulf Oil figured the office needed one (which meant it was more or less my Dad's).

1st year uni i worked on an IBM 360. Algol W & assembler (4k memory segments)


We had something like ten TRS-80's in my HS (running on the Z-80 processor) and I had a Color Computer at home with a 6809E processor. I had the assembler cartridge that plugged into the "expansion" port and could do quite a number of things in machine language at the time. The opcode set was very basic and it took FOREVER to do the simplest thing but boy was it fast (at least faster than BASIC).
The good old days...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.