Tell me more ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Why is RAM module development seemingly stuck at the same size for a while now (a couple of years)? I bought 2x 2 GB modules 2 years ago, and now it's all the same size, with prices even bigger. I want more memory, because I work a lot on my computer and I just need it. What is going on?

Hardware/memory progress was being made constantly until these couple of years, and I'm a big computer user for over 15 years. Why aren't there 4 GB/8 GB modules yet? I would gladly replace my DDR2 motherboard for a DDRX one if it had at least 4 GB DDRX modules for a reasonable price.

Now we have a situation with very cheap USB drives reaching 64 GB size, and RAM modules with a pathetic 2 GB size. Sounds like some sort of conspiracy.

share|improve this question

closed as too localized by studiohack Jan 20 '12 at 18:23

This question is unlikely to help any future visitors; it is only relevant to a small geographic area, a specific moment in time, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet. For help making this question more broadly applicable, see the FAQ.

4 Answers

Probably because there's little consumer demand for larger modules. The average user is still probably not going to push the 4GB barrier (except gamers), and PC makers are still installing 32 bit operating systems on new machines. Also, most new machines come with at least 4 and up to 6 memory slots. With 6x4GB (24GB) of RAM, you've got way more than anyone in their right mind is going to be using, even for games.

The one exception is for servers, things like SQL databases, and they simply have many more memory slots and use FB-DIMMs instead of getting larger modules; several servers have up to 8 DIMM slots per CPU socket, which would allow some 64 GB of ram in a typical 1U enclosure --- and even massive SQL servers don't use that much most of the time.

More importantly, yes, flash memory has drastically reduced in size and increased in capacity, but it has several attributes that RAM does not:

  • Large breakthroughs (such as MLC flash) have happened in the flash memory arena relatively recently. Flash is still feeling a boost from things like this; RAM technology has remained relatively constant since SDRAM came out in the 1990s
  • There is large consumer demand for large flash memories, because of increasing competition in the SSD market.
  • RAM is several orders of magnitude faster than flash, and does not have a write count limit, nor does it have to erase and read in large chunks. RAM is truly random access; flash is not. (Which is why, after all, flash is called flash)
share|improve this answer
3  
Surprisingly, I've come close to 20GB of memory usage just with PhotoShop (editing huge .RAW collages of my kids with many layers/effects). Luckily my home server has 20GB installed. – Jess Sep 21 '10 at 5:30
Note that server memory consumption is likely to accelerate given the massive increase in virtualization workloads, because the hosts of multiple virtual servers will need lots of RAM. – Justin Grant Sep 21 '10 at 19:38
@Justin: Even with virtualized workloads I think 64 GB of RAM in a typical 1U enclosure is plenty. When you start getting more than that, you start running into other bottlenecks (i.e. CPU time) And I believe there are systems out there with more than 8 DIMMs/CPU Socket, but I've just not had the luxury of playing with such systems. Finally, virtualizing actually decreases memory load over multiple single servers actually uses more RAM because the different VMs can share pages -- which can't be done with separate servers. – Billy ONeal Sep 21 '10 at 21:15

First thing first: There are 4GiB modules for DDR3 which aren't extremely expensive. The biggest DDR3 modules I've seen are 16GiB. There are also 4GiB modules for DDR2, but they are extremely expensive.

Another problem are transistor sizes. Chips with large capacities require smaller process in order to make large yields possible and we are slowly reaching technological limit (different parts of transistor are measured in memory chips, so the tend to have larger numbers than processor chips even though they are smaller) so it's getting more and more difficult to design smaller transistors (if anyone needs a source, I can look it up).

So basically (small demand for big modules) + (small yields for big modules) = (high price for big modules)

In my opinion others gave good non-technical reasons large modules aren't in use, so I'm not going to go into that area.

share|improve this answer
1  
+1 for funny equation. It should be noted that the linked to modules are fully buffered, have ECC, and are registered; any of the three essentially kills their utility for consumer systems (but is good for servers...) – Billy ONeal Sep 21 '10 at 2:41
@Billy ONeal Yes, perhaps I should have noted that. It's quite logical in my opinion. Would anyone with a server, who is a potential customer, even consider non-ECC chips? Combine that with home PC motherboard support for large modules and it everything should become clear. – AndrejaKo Sep 21 '10 at 2:44
Because lots of chipsets have the buffering builtin to the chipset for machines that support that many sticks of RAM. At least, the PowerEdge 2950s I work with over the summer work that way. EDIT: I just realized you meant EEC not buffering -- doh! I agree with you completely then :P – Billy ONeal Sep 21 '10 at 3:08
1  
well indeed there are 4gb modules at a fair price now. it was about time. still no 8gb. capacity growth is definitely very slow in the last few years – user49951 Sep 22 '10 at 18:16
@user49951 I agree! I remember that compute I bought few years ago started with 128 MiB of RAM and ended up with 2 GiB of ram (and that's because of bugs in chipset or it would have been even more). – AndrejaKo Sep 22 '10 at 18:27

I think it's more of a matter of demand.

Demand drives development (actually profit drives development, profit is driven by demand) and if there isn't a demand or need for something, it likely won't get developed.

For most people's average computer usage, 1 or 2 GB is enough, more avid users will need up to 4 GB, anything past that is usually unnecessary as most mainstream programs don't use that much memory even if it's available.

Only when you get more into the high level professional or scientific programs do you need more than 4 GB of RAM. Most workstation computers come equipped with at least 4 memory slots and if you use 2 GB sticks, thats 8 GB, plenty for pretty much any current practical computer application.

The only systems to ever need more than this are special dedicated systems such as servers and computing clusters. The thing is though, when you deal with large amounts of memory, its usually not the amount of memory that becomes the issue, its more on the speed of access of that memory that becomes a bottle neck (latency and bandwidth). Thats why you are seeing more development on the speed side as opposed to capacity.

Basically put, people don't need more memory. Even specialized applications that demand more memory run into the problem of bottlenecks in latency and bandwidth. Development is progressing towards increasing bandwidth and decreasing latency instead.

share|improve this answer

For all of you who don't need more RAM - try using ramdisk. It's VERY good when you've got a SSD. It reduces SSD usage, and also speeds up things. Put there browser cache, temporary files ... etc. Ramdisk also speeds up my Visual Studio building events. I've got a ramdisk of 3 GB and 5 GB for Windows. It's very tight infact.

share|improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.