Dual Channel vs. Quad Channel – no difference

[cogeanumarius]

Salve dear community,
this is my first post on this forum and I am not sure if I am posting it in the right place, for the administrators, please feel free to move it to the appropriate section.
I have just built a system based on 2011 socket, using a 3820 CPU and 4 modules of memory 4 GB at 933 MHz
I have used AIDA64 to create the attached reports, first report is using just 2 memory modules (Dual Channel setup – 128-bit) and the second report is using all 4 memory modules (Quad channel setup 256-bit) I see no difference in the memory read/write/copy results.
Thank you for your suggestions.
*PS: sorry for running a full report and not only attaching a hardware related report, I was just thinking that someone would be interested and could use those details as well.

 

[cogeanumarius]

I forgot to add the links to the report files:
http://cogeanu.com/test/128.htm.zip
http://cogeanu.com/test/256.htm.zip

 

[Uoppi]

Have you tried Sandra 2012?

 

[yumri]

i am wondering were all 4 memory modules in when you did the test for dual channel then put in another 2 modules for quad channel?
Then what latency did you use on each? As the speed of the modules is so low that the latency on them will matter more.
Also was it 4GB in all or 4GB per module?
did you have the same amount of RAM in each test also?
Lots of factors go into what makes up the performance of the computer and your post didnt specify most of what should have been.
also how did you test them in AIDA64 as the testing method matters also?
Main thing about a Quad channel setup though is for when you have to access big chucks of memory so a test of using a 4KB random read, write, copy method will not produce any difference but if you do a 1GB or more read, write, and copy method you will see a difference thus quad channel is more for workstations and severs atm then for consumer computers as we do not need to move so much data at a single time.

 

[cogeanumarius]

Salve yumri,
Just to answer all the questions:
- I have just attached 2 memory modules when I did the dual channel test, and I than added 2 more modules when I have tested the quad channel.
- I have not modified any parameters for the memory modules; I have had the same settings, BIOS defaults for both tests.
- As I have the reports attached, I have tested using a 16GB kit (4x4GB)
- For the dual channel test I have 2x4GB installed and for the second test I have just added the other 2 modules
- As I did not knew what factors I should mention I have just added the full reports from AIDA64.
- If you say that this is the reason, it may be that only for big amount of data to be transferred the quad channel could be visible, but I would be really disappointed by the technology if it will only show benefits for servers and workstations, and not for consumer computers like mine.

 

[constepatdyak]

latin reference? Salve? sorry couldn't resist 😀

 

[cogeanumarius]

old Latin, yes, it is the mother of my native language (Romanian).
Thank you for the remark, not too may that could realize that.

 

[constepatdyak]

haha nice, i did latin 5years in high school 😀
Sorry don't know much about this topic, hopefully you get it resolved!

 

[cogeanumarius]

I am still waiting for somebody that could give me some hints about what have I done wrong in my configuration so I can not see any improvement when enabling quad channel.

 

[Uoppi]

I am still waiting for somebody that could give me some hints about what have I done wrong in my configuration so I can not see any improvement when enabling quad channel.

All the benchmarks I've seen with AIDA vs. SiSoft Sandra 2012 suggest that Sandra will give you much more truthful results. In reality there is a bandwidth difference.