Replies to post #10722 on Advanced Micro Devices Inc (AMD)

[yourbankruptcy]

wbmw, I will side with you. The rate memory requirements grow will not change in the next few years. There is no need of 4 Gb at desktop until 2005 or 2006. But I beleive that software developers must get the hardware in volume at least 1 year before that. Which means Microsoft must get the hardware at least 1.5 years before this 1 year. So Intel must give first hardware to Microsoft this year. And I don't mean 5 manually assembled prototypes. I mean at least a hundred of functional developer's computers. Obviously, this is not going to happen.

Look at AMD, they gave first Opteron servers to MS this year. The first release will come in 2005, not counting beta patches.

[jhalada]

wbmw,

what are some of those "radical" things that will change the pace of memory that has been constant over the last 3 decades?

I don't see anything really radical, but I don't think it will take anything radical to get to 4 GB.

The next stop for the memory makers is apparently going to be 1 Gbit chip. I have not heard much about companies going to 512 Gb (so I guess they are skipping it). This means that the same way a power user can but today 1 GB for a dual channel mobo, which consists of 2 DIMMs, each DIMM with 16 256 Gb chips.

You don't have to go out of your way to get this. The price now is less than $200 for this configuration.

When the 256 Mb chip is replaced with 1 Gb, these 2 DIMMs become 4 GB of memory. At that point, it becomes inefficient to have an OS that doesn't let you use 1/2 or 1/4 of it in your app.

When will the 1 Gbit chips become mainstream? I don't know. Has anybody seen industry predictions / roadmaps?

Samsung will have them next hear. Initially, they will of course be expensive, but my guess is that they will become a standard within 2 years.

Server guys will of course always be dealing with 2x the amount of memory, since the registered DIMMs can have 32/36 chips. And, to make things even easier, servers typically have more DIMM slots.

Joe

[j3pflynn]

wbmw(CJ, chime in!), as I said, no one knows exactly when they'll get to market, but just look at Coatue's acquisition by AMD. If this technology pans out, it has the power to cause an earthquake in the memory and storage industries. Cheap, non-volatile, fast, easy to manufacture. Just think about the instant-on possibilities, even for complete and complex Windows installations. Hard disks would only be used for rarely used, very large files.
Now, as always, the technology has to be translated into a viable product, but the progress they've made is encouraging. Then there's MRAM and various others, but somehow the polymers seem most promising.
Paul