Get Firefox! "my blog doesn't just deal with my life, it deals with some important stuff too"

9th of October 2004

Flash RAM

I just had a huge argument over MSN with a friend who seems to think that within the next 2-3 years RAM will be a thing of the past. Or, looking at it another way, HDDs will be a thing of the past. He seems to think we will have one big lump of Flash RAM in our computers which acts as storage. We will partition it so that part of it is used like RAM and the rest is used as plain simple storage like a HDD.

Personally I can't see it. There are reasons why we have multiple tiers of storage. While Flash RAM is getting faster and faster, so too are other RAM technologies. There will always be something that's fast but expensive, and there will always be something that is slow but cheap. This is how it's been since the dawn of the computer and I don't see it changing within the next 2-3 years.

While I can't deny I do believe that Flash will soon be used as some sort of non-volatile fast storage internally, I don't see it becoming one unified storage method. If it reaches DDR RAM speeds then I can see it replacing existing memory as the standard meaning we can have lovely non-volatile computers (rebooting would be a thing of the past!) and that would certainly be a great advance in personal computing. However, speed and large volumes of storage do not go hand in hand. Because we all want large amounts of storage we need cheaper technologies. Plus, large amounts of storage usually needs larger amounts of space to live in, and a massive wad of flash on the motherboard strikes me as a bad idea.

Look at it another way, latency is very important in memory access. If the latency is high the computer will be sluggish. If the latency is low it will be far nicer to use. In mass-storage higer latencies are normal, and it's not a problem because generally we are just dumping large amounts of data to them. Especially now that RAM modules are getting so big.

Anyway I can't be bothered discussing it any more. We agreed that in 2006 and 2007 we will see who is right and who is wrong. However, considering his previous track record with these kinds of arguments, I don't think he should get his hopes up. ;)

Blog #311, posted at 13:15 (GMT)