Intel planning to get rid of DRAM on PCs | Hacker News Reader

Intel planning to get rid of DRAM on PCs

semiaccurate.com

25 pointsalecco22 comments

And anyway, nowadays the usual advice to clients is to max out the RAM on purchase since it's a mess to replace (be it a non-technical user or a datacenter cluster with critical uptime) and the specific configuration of RAM will become pricer per GB in the mid term (because the market moves constantly forward to different versions).

And also note one of the main reasons some algorithms run faster on GPUs is the memory bandwidth bottleneck. If Intel gets rid of the BW problem it's a level game againsta Nvidia.

This move would allow Intel to have some custom performance tricks. For example, memset multiple GB in a fraction of the time.

Open articleView on HN

That article has no technical information at all, it seems to be pure click-bait.

Does anyone know more? Are they moving the RAM or replacing with some flash derivative? Something else?

I think they want to move the CPU into the hard drive so Oracle is faster (IA64)

I assume they mean memory will be integrated into the CPU package itself? If that provides for faster RAM, smaller computers, and greater power efficiency I would find it to be an acceptable trade-off. I've found over the years my RAM needs are tied pretty closely to my CPU needs. Today I can get by fine with 8GB but by the time I need 16 or 32GB I will probably want a newer CPU as well. So doing this all in one package wouldn't bother me. I'm guessing on desktops/servers there would be some ability to add slower modular RAM to compliment the faster RAM integrated into the CPU?

Wouldn't the slower RAM mean either inconsistent performance (yuck), or having to slow down the CPU memory? (like today's user RAM with a slow and fast card)

Performance is already so inconsistent today you'd never notice.

It would depend on how the slower RAM was treated. Boosting the memory to support memcache for example would seem to make sense.

You could use the on-chip memory as "L4 Cache".

So what are they replacing it with?!

EDIT: from the comments I get the impression that they want to move the memory closer to the CPU.

Stacked DRAM, if I understand it correctly.

I hadn't heard of stacked DRAM. For anyone else looking for a reference, I liked http://blogs.intel.com/research/2011/09/15/hmc/

The description of "getting rid of DRAM" sounds misleading if this is what they are deploying.

Only if you have showdead on.

In other HN curiosities, I lost the ability to flag a few weeks ago.

From that article: "This groundbreaking prototype has 10 times the bandwidth and 7 times the energy efficiency than even the most advanced DDR3 memory module available."

(argh)

1 point by alecco 2 hours ago | link | parent [dead]

Initially I didn't like the idea. But thinking it better it might be actually good. The big PC bottleneck nowadays is DRAM. A big on-die/stacked memory at many times faster bandwidth and lower latency could be a game changer. And anyway, nowadays the usual advice to clients is to max out the RAM on purchase since it's a mess to replace (be it a non-technical user or a datacenter cluster with critical uptime) and the specific configuration of RAM will become pricer per GB in the mid term (because the market moves constantly forward to different versions).

And also note one of the main reasons some algorithms run faster on GPUs is the memory bandwidth bottleneck. If Intel gets rid of the BW problem it's a level game againsta Nvidia.

This move would allow Intel to have some custom performance tricks. For example, memset multiple GB in a fraction of the time.

I think they may be referring to non volatile (phase change) memory.