The only way you are going to be able to take advantage of the additional RAM is by running a bit OS as others have stated. Having dealt with ADP as the Director of IT for the largest auto dealer group in Wyoming it will be at least another 10 years before they figure out what bit computing is as they just barely upgraded all of their code to 32 bit 2 years ago when I was running them for our auto dealer group. My condolences to you but keep up the good fight. I doubt it.
Most decently new processors support bit. All processes running on that system get a virtual memory address space of 4 GB, 2GB for private memory , 2GB for operating system stuff; regardless of how much RAM is available. If you have 3GB installed on the systems in question, then that should be sufficient for your requirement.
I was just thinking the same, that my only option without having to switch to 64bit XP is to go up to 3GB RAM and hope that it'll be sufficient to run all processes and applications.
If you move to a 64bit OS don't move to XP 64bit as it is very badly supported plus XP itself is being phased out , move to Windows 7 64bit instead, it is a much better OS and works perfectly in 64bit plus has a lot more support in terms of drivers for hardware. Intel Core 2 Duo should be a x64 bit processor. If you purchase a copy of Windows 7 Professional, you can use XP mode and should be able to run your software that is only designed for x86 bit.
This will allow you to take full advantage of both the additional memory and the 32 currently unused lanes in your hardware. Does this mean that Win 7 Pro is 64bit by default or is it also 32bit? You should be fine with your processor. I would definitely recommend moving to Windows 6 Professional or Ultimate if you need encryption bit instead of XP.
XPbit is horrible, out of support, and generally a Very Bad Idea. If you get Professional or Ultimate, it comes with XP-mode for any legacy apps you may have.
I want to play Crysis though have a GTS , so I am curious, which processor would you guys recommend for this? A core2quad or core2duo? My main thing is gaming, I don't really do anything else on this comp. Or must I get 4GB to play crysis fine? Including memory CPU and motherboard Thanks.
Joined Apr 1, Messages 4, 0. Gamax said:. Wile E Power User. Joined Oct 1, Messages 24, 4. HTC said:. Joined Nov 8, Messages 5, 0. With the price of DDR2 so low now I see no reason not to get 4Gb even if you only use bit, because when you go bit in the future you'll be able to use all the memory. Also, for gaming, I would recommend either the Q, or E depending on your other uses.
If you do other things like video encoding, or don't intend to upgrade in a few years, than the Q Otherwise get the E, which is a dual core, not a quad. Currently, there are very few, if any, games that actually take advantage of more than 2 cores. Wile E said:. No, it won't recognize all 4GB, but Darknova said:. Last edited: May 12, Faced the same problem myself. Not sure. May be something that you did. I have tried everything I could think of, but still it is stuck at 3 GB.
Gave up trying long ago. Arie ,. If you are in the mood, go for WinXP 64 bit. It can recognise anything you can throw at it for a couple of years at least. Now, most perhperials only support a bit physical address. By convention, the latter is used. This means that all hardware[2] uses a physical address below the 4 GiB, or bit, boundary. Also by convention, RAM is generally mapped to the bottom of this space[3] and the hardware goes at the top of this space. How much space is needed for the hardware is dependent on how much hardware and what hardware you have.
Doesn't it use the entire space and leave no room for the devices? So this is the first problem: you have to have hardware that supports remapping the RAM.
Most modern workstation, and server boards support this, as well as many desktop boards. Santa Rosa and future laptop boards do as well, but not previous ones. If you hardware doesn't support this, then nothing else matters. So, now the operating system comes into the picture.
The operating system has to support bit physical addressing, as it's really an extension to x86 processor architecture. Thankfully, all modern operating systems, including Windows, support this and have for quite some time. However, you sometimes have to do some trickery to enable PAE support, for example, on Linux you normally have to install a "highmem" kernel, as default kernels frequently have PAE support disabled.
Desktop Windows adds one more wrinkle. Tasks applications don't use physical address directly. They use virtual addresses which are always bits wide hence a bit processor. Virtual addressing is used to allow applications to ask for more memory than you have available, and to provide isolation between applications so one task crashing doesn't crash the other.
It works by mapping the virtual address to a physical address in blocks called a page; these mappings are arbitrary and can be changed at any time. The kernel, where all the drivers are located, also uses virtual addressing it's an all or nothing deal Now, drivers have to play a role in this translation process. The kernel provides a set of routines and methods that must be followed to do this.
However, some drivers don't follow the rules correctly when a physical address is above bits. This causes everything to break and results in a system crash. MS' solution to this problem is to never use a physical address above bits on desktop Windows XP, Vista. They assume on server Windows the drivers will be better written, I guess.
Hopefully that's understandable. If it's not, please ask questions. TL;DR version -- Windows has the capability, but it is intentionally restricted to workaround broken drivers. There's nothing that can be done beyond switching operating systems. How much RAM you lose is dependent on your hardware. You have 36, 40, or 48 bits of physical address space depending on your CPU. However, chipset limitations always reduce that to a smaller number.
Sometimes, that's bits, but it isn't always. Some space below 4GiB is reserved for hardware. Space above 4 GiB is not reserved in any special fashion. Virtual addressing has nothing to do with reserved addresses for hardware.
Virtual addresses are just that. Any 4-kiB page of virtual addresses can be made to point to any page of physical addresses. There's no tradeoff. And, as I said eariler, the only OS that can't do this is desktop Windows, due to a delibrate limitation. However, they work just fine with bit addresses too, and since the system has to support those, they're rarely used or seen. There's some special stuff at the very bottom of the address space, normally.
But again, it's not worth worrying about. TLB thrash. Which has nothing to do with the amount of RAM the operating system can use. I'm going to say this explictly, because I should have done so sooner: the restrictions on virtual addressing have nothing to do with the restrictions on physical addressing. They're totally unrelated. What you're talking about is a restriction on virtual addressing.
This is why everyone else does not do this. Err, nothing put together like that. As I said earlier, tasks on a modern operating system use virtual addresses. These addresses have to be mapped to physical addresses the processor can assert to talk to RAM and perpherials. Now, different tasks may map different virtual addresses to different physical addresses. The best way to think of this is that every task has its own unique virtual address space.
However, the TLB doesn't store what task a mapping belongs to. This means when the CPU switches tasks and virtual address spaces , it has to throw out all the cached mappings and look up the mappings for the new task. Now, the most common task to switch to is the kernel. Also, anytime the hardware interrupts the machine, its is the kernel that responds to the interrupt. The point is that the kernel is switched to very often.
To mitigate the cost of throwing away the TLB mappings every time the kernel runs, most bit x86 operating systems split the virtual address space into two pieces. One piece is always used by the kernel, and one piece is used by your user-space applications. This is why your userspace applications can only allocate 2 GiB or 3 GiB, under the right conditions of virtual memory. Half of the address space has been walled off.
But by walling the space off, the address space for the kernel is always present, no matter what task is running on the CPU. So when your task switches to the kernel, a flush of the TLB mappings is no longer required.
This speeds up access to the kernel. However, OS X doesn't do this.
0コメント