Why don't we use SRAM

[Accessless]

In virtually all computers nowadays we use Dynamic Random Access Memory (DRAM) for our main volatile storage and Static Random Access Memory (SRAM) as a cache memory for higher speed memory reading/writing built into the CPU.

So why don't we use SRAM in place of DRAM and free up space in our processors?

I already know that SRAM is more expensive but by how much? Surely enthusiasts would pay many dollars for even faster RAM and CPU designers would be happy to have more space to work with.

[BC_Programmer]

As you said, it's more expensive. It's about 6 times the price for SRAM of the same capacity as DRAM. Of course, that only applies where SRAM can actually be manufactured in the same size. Die size for SRAM also grows quite large at higher capacities. It also consumes far more power.

If SRAM were used as the main memory of a PC, aside from the higher cost, the RAM that get's installed would be large enough that cases would need to be designed specifically to hold each SRAM Shelf that get's installed. Each SRAM shelf would need an effective cooling solution, equivalent to perhaps the air-channel-cowling used on graphics cards to keep the SRAM cool. It would also require a much more powerful power supply than is typically found in a PC.

it simply isn't worth the requirement to re-engineer a standard to accomodate RAM shelves rather than RAM sticks, and the added price of the power usage and cooling to use the higher speed SRAM as main memory.

Also much of the performance advantage from SRAM on the CPU die itself is a result of literally being physically close to the processor core. If the CPU had to reach across the bus for every memory access, it wouldn't matter how fast the memory is because the distance would impair performance below what the on-chip cache's currently have, and since a large majority of CPU memory access is able to access cached data acquired from main memory via the prefetcher, it would likely result in a net slowdown in performance.

[Geek-9pm]

I like the reply by BC.

The OP may have an erroneous idea. The CPU will  generate a refresh for the RAM at regular intervals. This has been a built-in feature from the early days** . It would do nothing to make the CPU  smaller, faster, sexy or anything.

Perhaps the OP received his training in the no shoelace school. At that place  teachers taught students not to have shoelaces because it was a waste of time. 

** The Intel 8080 did not have refresh, but the Zilog Z-80  did.

[Accessless]

Geek-9pm, I have no formal computer training (aside from a laughable ICT GCSE from my school days). Wikipedia is my friend.

[camerongray]

The OP may have an erroneous idea. The CPU will  generate a refresh for the RAM at regular intervals. This has been a built-in feature from the early days** . It would do nothing to make the CPU  smaller, faster, sexy or anything.
Perhaps the OP received his training in the no shoelace school. At that place  teachers taught students not to have shoelaces because it was a waste of time. 

There was nothing wrong with the original question - It was a valid and interesting question.  It's good to ask questions like that rather than restricting yourself to the by-the-book thinking such as "All PC RAM is DRAM and that's just the way it is" that so many people seem to get stuck with.  If people don't ask questions or think about how things can be done differently, technology will never advance!

BC's response hits the nail on the head though.  While SRAM is technically faster, the cost would be ridiculous and the distance from the CPU would mean that it still would be nowhere near the speed of on-die cache - With the microscopic distances involved in modern machines, the sheer distance between the CPU and RAM makes up a large amount of the latency involved.  Also, IIRC SRAM has higher power consumption than DRAM which is a critical factor in modern devices.

[Geek-9pm]

Sorry if I offended anybody. 
Well, the question has brought some interesting points.
Now in favor of static RAM, it can kept the memory state for a long time using very low power. This may sound like a paradox. There are very low power SRAM chips that are made for applications where the machine has to rest for a longtime. Such as some kind of telemetry device that only communicates for a few minuets every day. Reference:
[ur=http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=4260999&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D4260999]
A Low-Power Embedded SRAM for Wireless Applications [/url]

This paper introduces a novel ultra-low-power SRAM. A large power reduction is obtained by the use of four new techniques that allow for a wider and better trade-off between area, delay and active and passive energy consumption for low-power embedded SRAMs. The design targets wireless applications that require a moderate performance at an ultra-low-power consumption. The implemented design techniques consist of a more efficient memory databus, the exploitation of the dynamic read stability of SRAM cells, a new low-swing write technique and a distributed decoder. An 8-KB 5T SRAM was fabricated in a 0.18-mum technology. The measurement results confirm the feasibility and the usefulness of the proposed techniques. A reduction of active power consumption with a factor of 2 is reported as compared to the current state of the art. The results are generalized towards a 32-KB SRAM.

For some remote afflictions, 32K of RAM could hold a significant amount of data. But it would not do for a video system. Instead, a slow system that reports some moderate collection of data. Perhaps weather consolidations, water flow, sunlight level and so on.

[BC_Programmer]

It would be more expensive, need a bigger PSU, and better cooling solutions.

But it would be worth it to be able to say "Yep, I was thinking of adding another Shelf of RAM"

[Geek-9pm]

Bad link. Fixed.

A Low-Power Embedded SRAM for Wireless Applications