Friday, October 3rd 2008

Theory of DDR3 Voltage Limitations for Bloomfield Gains Ground

Nehalem does promise to be a processor to look out for, it would be Intel's next installment, post the successful Core 2 series processors. This time however, Intel made a core modification with the way the system handles memory. The Bloomfield processors house a massive 192-bit wide memory controller for supporting tri-channel DDR3 memory. It however was found that the controller could bring in limitations to the DIMM voltages that the system could support.

The retail version of ASUS P6T Deluxe OC Palm Edition motherboard was unboxed by XFastest. Being the retail product, as usually, it comes with precautionary labels attached to parts of the motherboard. The one that covers the 6 DDR3 DIMM slots reads:
According to Intel CPU SPEC, DIMMs with voltage setting over 1.65V may damage the CPU permanently. We recommend you to install DIMMs with voltage setting below 1.65V.
It could have implications on the current DDR3 memory market as well as you, if you happen to have DDR3 modules, which you plan to retain for use in the future platform, that operate above the said voltage. It also means that in the near future, we could be seeing memory sticks that facilitate overclocking at much lower voltages. From a technology standpoint, companies such as Samsung, Elpida, Micron, etc., are working on releasing DRAM chips based on newer silicon fabrication technologies, that operate at lower voltages.
Add your own comment

89 Comments on Theory of DDR3 Voltage Limitations for Bloomfield Gains Ground

#1
InnocentCriminal
Resident Grammar Amender
\m/

Triple w00!

:rockout:

That is all... for now.
Posted on Reply
#2
DaMulta
My stars went supernova
So for chips is dual then? And 5 sticks will be still dual channel maybe.

I'm just wonder I can't find any links yet about it.

Also been looking for what happens when you stick 3 sticks in on current motherboards. But I think that he is right that one stick will not be in dual channel. AMD XP chips had motherboards with 3 slots and it would run in dual channel mode.
Posted on Reply
#3
DaMulta
My stars went supernova
While doing my digging I found this about the old nforce2 and it using 3 chips for dual channel. I'm guessing 5 could do the same thing.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=1719&p=4

This is the new Intel
http://www.sysopt.com/features/cpu/article.php/3775041
The onboard memory controller of the Core i7 platform supports the de facto single and dual-channel DDR3 architecture, but adds a new wrinkle: triple-channel memory. This announcement lit up some eyes in the enthusiast community, and although Intel has been quiet in terms of performance results, demo sessions have yielded some very impressive visual scores.

The first Nehalem iteration is rumored to implement a triple-channel DDR3 architecture, and now that the CPU determines the memory type, speed, capacity, and number of modules, this is going to put some limitations on both motherboard vendors and end users. For example, existing DDR3 memory kits are sold in a dual-channel format, with two modules. The Intel X58 requires three modules for top performance, and many potential upgraders are waiting for memory manufacturers to introduce their triple-channel kits, featuring a trio of fully matched and guaranteed DDR3 sticks.

The presence of an integrated memory controller also means that the Northbridge components of the X58 chipset will be greatly simplified. Intel will even release a Core i7 processor with an integrated graphic core, further limiting the role of the chipset in this new architecture. The Nehalem also marks the debut of Intel's QuickPath Interconnect, with each bidirectional link supplying up to 12.8GB/sec of bandwidth each way for a total bandwidth of 25.6GB/sec per link -- that's over 50GB/sec for the top-end models that will have two QuickPath links.
cool, you never know they could of taken out dual, but that would be a crazy thing to do.




Edited:thought about what it said.
I feel like an idiot wondering about these things, lol So it does have dual channel support . I was talking to someone else that wondered the same thing. So if you put 4 chips in that triple channel motherboard what happens. I wonder if you put 4 in will it still be in triple mode like it was dual mode for 3 chips on the older nforce2.

I think it's rambus ram that made me think it. Because in that system you had to have 2 chips. If you didn't it didn't work, but of course DDR is different.
Posted on Reply
#4
DaMulta
My stars went supernova
Am I going to far with this lol.

I could test the 3 sticks in my machine tonight sometime.
Posted on Reply
#5
Morgoth
i can also test it on my i got 4 gigs in my p4
if you like to
Posted on Reply
#6
Wile E
Power User
Here ya go D. 3 sticks in an X38 board.

Posted on Reply
#7
DaMulta
My stars went supernova
BTW if anyone was wondering his SS is of DDR2 not DDR3.



Cool, so with the new Intel it should be triple channel if you just bought two sets 2 DDR3 chips.
Posted on Reply
#8
spud107
could have one stick as backup lol.
Posted on Reply
#9
DaMulta
My stars went supernova
spud107 said:
could have one stick as backup lol.
What would be the point if it's still running in triple channel?
Posted on Reply
#10
kid41212003
Wile E said:
Here ya go D. 3 sticks in an X38 board.


Is that your board and memory? I think It might be 2x1GB and 2x512MB...
Posted on Reply
#11
Wile E
Power User
kid41212003 said:
Is that your board and memory? I think It might be 2x1GB and 2x512MB...
Yes, it's my board and memory. It's 2x1GB Transcend aXeRam 1200 and 1x1GB Crucial Ballistix 800, for a total of 3x1GB sticks.
Posted on Reply
#12
Fitseries3
Eleet Hardware Junkie
remember that garbage that the core voltage is directly linked to the ram voltage?

myth dispelled....





give me a min to explain.....

QPI/dram voltage is related to the path between the memory controller and the actual ram itself. this needs to be within .5v of the vcore to prevent damage to the memory controller.
Dram bus voltage is the actual voltage that the ram is running at.... IE: Vdimm, Vram, Vmem, etc. this should be as close as possible to within .5v from the qpi/dram voltage.

bit-tech
While Asus and Intel (rightly) scare everyone (read: uneducated) into thinking that 1.65V on the DRAM voltage should be the absolute limit before you reach for the fire-blanket, all that's really needed it to obey this: keep the CPU uncore voltage within 0.5V difference of the DRAM voltage and there's no problem. Over this potential difference and you’ll greatly increase the chance of CPU death, but it certainly won't happen instantly in a big ball of fail fire if you make a mistake.
source.... http://www.bit-tech.net/hardware/2008/11/06/overclocking-intel-core-i7-920/3

and if you think 2000mhz isnt possible on x58....

here ya go.........

2XXXmhz ram on x58 in tripple channel.

Posted on Reply
#13
Scrizz
fitseries3 said:
remember that garbage that the core voltage is directly linked to the ram voltage?

myth dispelled....





give me a min to explain.....

QPI/dram voltage is related to the path between the memory controller and the actual ram itself. this needs to be within .5v of the vcore to prevent damage to the memory controller.
Dram bus voltage is the actual voltage that the ram is running at.... IE: Vdimm, Vram, Vmem, etc. this should be as close as possible to within .5v from the qpi/dram voltage.



source.... http://www.bit-tech.net/hardware/2008/11/06/overclocking-intel-core-i7-920/3

and if you think 2000mhz isnt possible on x58....

here ya go.........

2XXXmhz ram on x58 in tripple channel.


thx :toast:
Posted on Reply
#14
Sasqui
^ ditto - real screenies excite me, specially over 4GHZ!!!! :toast:
Posted on Reply
Add your own comment