Featured Articles

AMD sheds light on stacked DRAM APUs

AMD sheds light on stacked DRAM APUs

AMD is fast tracking stacked DRAM deployment and a new presentation leaked by the company  points to APUs with stacked DRAM,…

More...
Nvidia officially launches the 8-inch Shield Tablet

Nvidia officially launches the 8-inch Shield Tablet

As expected and reported earlier, Nvidia has now officially announced its newest Shield device, the new 8-inch Shield Tablet. While the…

More...
Intel launches new mobile Haswell and Bay Trail parts

Intel launches new mobile Haswell and Bay Trail parts

Intel has introduced seven new Haswell mobile parts and four Bay Trail SoC chips, but most of them are merely clock…

More...
Aerocool Dead Silence reviewed

Aerocool Dead Silence reviewed

Aerocool is well known for its gamer cases with aggressive styling. However, the Dead Silence chassis offers consumers a new choice,…

More...
AMD A8-7600 Kaveri APU reviewed

AMD A8-7600 Kaveri APU reviewed

Today we'll take a closer look at AMD's A8-7600 APU Kaveri APU, more specifically we'll examine the GPU performance you can…

More...
Frontpage Slideshow | Copyright © 2006-2010 orks, a business unit of Nuevvo Webware Ltd.
Thursday, 26 July 2007 11:14

AMD G3MX memory socket for Opteron

Written by test
Image

Ready in 2009

AMD has revealed that it's working on a new socket for memory expansion for its Opteron processors called G3MX or G3 Memory Extender. AMD is working together with IDT and Inphi to develop this new technology which is based on DDR3 memory.

The G3MX technology allows AMD to use 16 memory modules per CPU socket instead of the current limitation of eight. This might not seem like a big deal right now, but with larger databaes and virtulization technology, this could be a crucial feature for AMD's Opteron servers.

G3MX is only expected to appear in quad socket systems and beyond, so don't expect to see this appear on workstation motherboards or AMD's FX platform, at least not initially.

The trick here is that G3MX works similarly to the buffer on a FB-DIMM, but it's a single buffer instead of separate buffers on each memory module, which should allow for much lower latency.

By using standard DDR3 memory modules, the cost of adding 16 memory modules would be far cheaper than using FB-DIMM's as well.

You can read the AMD press release here
blog comments powered by Disqus

 

Facebook activity

Latest Commented Articles

Recent Comments