Legit Reviews

Intel DX58SO 'Smackover' Motherboard Does Triple-Channel Memory

Details on the Intel DX58SO 'Smackover' motherboard are becoming crystal clear here at the Intel Developer Forum in San Francisco! For those that haven't been keeping up on the bleeding edge of technology the new chipset that Intel will be releasing to support Core i7 processors (Nehalem) is the X58 Express (Tylersburg).

In a meeting between Legit Reviews and Intel it was discovered that three slots should be used if you want the best performance. If you pop in a fourth module and occupy all the slots the DX58SO motherboard won't be running in quad-channel. It will actually default back down to single channel mode, which will obviously drastically impact memory bandwidth. The official support memory clock speed is 1066MHz, but Intel was quick to say that 1333MHz, 1600MHz and even 2000MHz memory kits will work on the platform.

Article Title: Intel DX58SO 'Smackover' Motherboard Does Triple-Channel Memory
Article URL: http://legitreviews.com/article/775/1/

hmm, ive never seen RAM above the CPU, i wonder how RAM temps are

its too bad the VDIMM is limiting the board, i'm sure this is for the superhigh end, so they could've went the extra 1/2 mile and made it go higher since current boards can

skierkid450 wrote:hmm, ive never seen RAM above the CPU, i wonder how RAM temps are

its too bad the VDIMM is limiting the board, i'm sure this is for the superhigh end, so they could've went the extra 1/2 mile and made it go higher since current boards can

Actually, this is a very good thing... They are hitting 2GHz at 1.65V... That is lower power consumption and less heat that current platforms... Newer IC's like Samsung's are doing great on these platforms. I'll have to give this a thumbs up!

Apoptosis wrote:

skierkid450 wrote:hmm, ive never seen RAM above the CPU, i wonder how RAM temps are

its too bad the VDIMM is limiting the board, i'm sure this is for the superhigh end, so they could've went the extra 1/2 mile and made it go higher since current boards can

Actually, this is a very good thing... They are hitting 2GHz at 1.65V... That is lower power consumption and less heat that current platforms... Newer IC's like Samsung's are doing great on these platforms. I'll have to give this a thumbs up!

okay, wasnt sure if they were making headway in the lowering power consumption there, i only read it as "more volts=higher speed" basically, oh well.

i'm also curious why triple channel is the max, you think a BIOS update might allow quad in the future? or is triple the way intel is choosing to go, and quad later

I don't think quad is possible, as it isn't implemented into the chipset/architecture (for all that I know). Why even include the 4th RAM slot then? What a waste, especially since it looks like the 4th one that disabled triple-channel memory is closest to the CPU, which means it would overclock better than the furthest one. Doh!

skierkid450 wrote:okay, wasnt sure if they were making headway in the lowering power consumption there, i only read it as "more volts=higher speed" basically, oh well.

i'm also curious why triple channel is the max, you think a BIOS update might allow quad in the future? or is triple the way intel is choosing to go, and quad later

I'm with you here, low power consumption is good but... this is an enthusiast board, let it get to 2Vdimm atleast!

And channels are hardware, you can't just from dual to triple or triple to quad with a software update. There are cost issues too, sure they can probably make octo-channel or whatever, but the number of pins required for the memory controller would sky rocket

The reason why we can't push 2v through the memory is that the memory controller is now on the die. The VDIMM is the signal used to communicate between the memory controller and the modules. Now I myself have no qualms about pushing 2v through a 45nm processor albeit under liquid nitrogen. Now.. how many of you even push 1.65v through your 45nm processors on water cooling? The fact that Intel is even pushing 1.65v is surprising as that's a bit high for 45nm for 24/7 usage.

Gomeler wrote:The reason why we can't push 2v through the memory is that the memory controller is now on the die. The VDIMM is the signal used to communicate between the memory controller and the modules. Now I myself have no qualms about pushing 2v through a 45nm processor albeit under liquid nitrogen. Now.. how many of you even push 1.65v through your 45nm processors on water cooling? The fact that Intel is even pushing 1.65v is surprising as that's a bit high for 45nm for 24/7 usage.

I don't think that's right... where did you find that DDR1/2 that could run on 1.35V for AMD processors in the last few years?

Memory controller voltage may be the same (or not) as CPU voltage but Vdimm is somehting else

Guess I should start saving up for all this awesome kit that is coming out

Dan

Zertz wrote:

Gomeler wrote:The reason why we can't push 2v through the memory is that the memory controller is now on the die. The VDIMM is the signal used to communicate between the memory controller and the modules. Now I myself have no qualms about pushing 2v through a 45nm processor albeit under liquid nitrogen. Now.. how many of you even push 1.65v through your 45nm processors on water cooling? The fact that Intel is even pushing 1.65v is surprising as that's a bit high for 45nm for 24/7 usage.

I don't think that's right... where did you find that DDR1/2 that could run on 1.35V for AMD processors in the last few years?

Memory controller voltage may be the same (or not) as CPU voltage but Vdimm is somehting else

For the same reason that 2.0v was blowing up 65nm Phenoms. High vdimm and low vcore on Phenom was causing the magic dust to escape. I was trying to find the link I had read before describing the muxs and demuxs working in Phenom and the high voltage deltas causing the chips to blow up but I can't find it for the life of me. Anywho, what a lot of us guys that ram stupid volts into stuff are afraid of is this attribute will extend itself to Core i7, thereby ruining our fun in tweaking memory on air cooling.



note to self: don't hit the edit button, hit the quote button

Gomeler wrote:note to self: don't hit the edit button, hit the quote button

I think all mods have done that at least once before

Dan

I knew the new socket was going to be bigger but now I have seen one its going to be real interesting to see how coolers for it are going to look due to the RAM being above the socket.