This is something of a pipe dream thread. I know GDDR5 exists in graphics cards and that memory type is not in modules for desktop systems or even on servers. I have studied tech a bit, and I thought the original DDR was really cool when it first came out, performing 2 memory operations per clock, one on the up stroke and one on the down stroke. As far as I know, this has been the general approach for since then.
What intrigues me is the claim that GDDR5 does four operations per clock cycle, two up and two down. So it's more like DDDR 😛
I know GRAM performs a read,write and update all in one clock cycle and desktop memory doesn't have this constraint.
My last major system purchase was a 90nm dual core system. I am patient having all I "need" for a computer. I made the bold statement (a few years back) that when I can have an 8-core, 22nm CPU, (4 times as many cores in 1/4 the process) USB3.0, and DDR5 memory, I will upgrade. USB3.0 is hitting it's stride, and Ivy Bridge 8-cores are due out this year.
Will we see DDR5? if so, will the current memory controllers adapt easily being on the cpu?
What intrigues me is the claim that GDDR5 does four operations per clock cycle, two up and two down. So it's more like DDDR 😛
I know GRAM performs a read,write and update all in one clock cycle and desktop memory doesn't have this constraint.
My last major system purchase was a 90nm dual core system. I am patient having all I "need" for a computer. I made the bold statement (a few years back) that when I can have an 8-core, 22nm CPU, (4 times as many cores in 1/4 the process) USB3.0, and DDR5 memory, I will upgrade. USB3.0 is hitting it's stride, and Ivy Bridge 8-cores are due out this year.
Will we see DDR5? if so, will the current memory controllers adapt easily being on the cpu?
