DRAM Command Rate Explained (not really)

Toggle sidebar Toggle sidebar
N

Nafryti

Distinguished
Sep 26, 2008
414
6
18,865
I am trying my best to figure out and understand DRAM Command Rate, not just the mainstream quartet of timings, i mean just the command rates, and explain them all in lamens terms (1 2 and 3... 4?).

If anyone has the capacity, or a resource to put it out clear as day without stupid equations thought up by Albert Einstein and Steven King.

Was getting close to that here: IN THIS THREAD when it was abruptly ended by a moderator. Don't care how old something is, i would still like to see it answered! I watched that thread for ages! and besides, isn't TH about giving answers? Maybe i missed the memo amending that.

Anyways, i am making this thread to continue that one, and the last thing i said before it was shut down: "regardless of what the TomsHardware tutorial says, it has left me with the same amount of questions i started with."

I am only interested mainly on the command rate, EVERY google result and even the tutorial here on TH, says there is just 1 and 2 command rates... where as, i discover only a few people have posted about 3N. MY DRAM IS A 3, CAN IT BE ANSWERED!? ASUS Crosshair V Formula-Z only has selections for 1 or 2 not 3N!!! What does 3N even mean!?

Thank you all in advance.

BONUS:
pulling my hair out over this!
 
Solution
P
Sort by date Sort by votes
P

Pinhedd

Champion
Aug 7, 2007
10,342
20
56,965
Hi, I'm currently in the process of rewriting the DRAM FAQ. It's much more clearly explained in my draft, which I will copy from:

Command Rate: The command rate is the number of bus cycles for which the signals for a command is held on the DRAM IO bus before it is considered valid. A command rate of 1T means that a command is held for a single cycle, whereas a command rate of 2T means that a command is held for two cycles. The longer command rate is used to ensure that the signals are stable at the input pins of all ICs on the active rank at the time of the clock. 2T is common for configurations that use either high speed memory or more than 2 ranks per channel.
Click to expand...

A command rate of 3T/3N simply holds it for 3 cycles. This is common on standard high-speed memory (above DDR3-2000) but since high-speed memory on the PC is the domain of enthusiasts most vendors pick high-quality modules or apply higher voltages to keep the command rate at 2 or below.

If you would like me to explain the technical reasons behind it, I'll be happy to do so
 
Upvote 1 Downvote
N

Nafryti

Distinguished
Sep 26, 2008
414
6
18,865
delighted to hear from you sir, so this being said, my RAM being a 3N and my Motherboard not having 3N in it's list, am i not going to have the full rated DDR3-2133 of my AMD Radeon G-Series memory?

I'm fairly concerned now, i went with the ASUS gaming boards for the high end performance, now it looks as though i will not get that.

seeing as how the voltage is 1.65v and the CR is 3N, at DDR3-2133 what can i do to stabilize my RAM? i've left it at the AMP settings but it keeps defaulting to DDR3-1866 if i reset the bios and just apply the AMP setting without specifying the speed. I've over clocked and volt bumped just about everything else in my PC, have not touched the reference clock, the 200mhz thing, i leave that at auto cuz if i put it to anything other than 200 everything else is messed with and the computer crashes.
 
Upvote 0 Downvote
P

Pinhedd

Champion
Aug 7, 2007
10,342
20
56,965


Hi,

I find it difficult to believe that a 1.65 volt DDR3-2133 module would require a command rate of 3T. This is something that I would expect from a 1.5 volt DDR3-2133 module. The smart thing to do would be to field an email to Asus to see if they have tested your particular memory module with your particular motherboard. It may be that 1.5 volts requires 3T, but 1.65 volts requires only 2T. Go ahead and test it, it won't hurt.
 
Upvote 0 Downvote
N

Nafryti

Distinguished
Sep 26, 2008
414
6
18,865
my RAM: http://www.newegg.com/Product/Product.aspx?Item=N82E16820403002
my Motherboard: http://www.newegg.com/Product/Product.aspx?Item=N82E16813131876

The SPD details state 3T (or N, not sure) as the command rate for this ram, its the fact that the motherboard doesn't have 3T listed in the DRAM setup, just 1 or 2. I have had it running in Auto for as long as i've had the RAM, but the RAM states it is a 1.65v and the RAM has a 3T.

Seeing no option for 3T i thought maybe my Motherboard wasn't compatible with 3T/3N ram, i've hardly seen any OC'ers ever mention anything over 2, always saying 2 is best for stability over 1, but 1 is fastest. Seeing the majority of overclockers stating that 1 is faster than 2, i'm hard pressed to believe any "Quality" claims by the Radeon Memory with a CR of 3. However, the motherboard does not have an option for 3, even though the motherboard supports up to DDR3-2400.

D3vBA6D.png

Image is to my memory, currently, AMP and XMP settings are identical and is what is in use.
BtQ99u8.png


I don't fully understand what is going on here... but.. in BIOS the settings between AMP and XMP are the same, but here they are different, prolly a translation error generously provided by Windows.
 
Upvote 0 Downvote
P

Pinhedd

Champion
Aug 7, 2007
10,342
20
56,965


There is no noticeable performance difference between 1T, 2T, and 3T. The hold is only necessary whenever a rank is switched. The Chip Select assertion takes time to complete as the physical address needs to be translated into a tuple of [channel, rank, bank, row, column]. If the rank needs to be changed, then the CS for the desired rank needs to be asserted. This translation and assertion period takes time not encountered on the multi-drop aspects of the bus and the time taken increases with the complexity of the installed memory set and the speed at which the target rank responds to the CS assertion. Once the assertion is complete, the chip can begin processing commands. On a 1T the first Row Active command can be sent on the cycle immediately following the CS assertion, but on 2T it must be delayed by one cycle while it waits for the CS assertion to complete (I'm not sure if a NOOP has to be inserted in its place, or if the same command can be present on the bus). 3T simply necessitates 3 cycle latency between CS assertion and Row Active. If the command and accompanying data are sent before the rank has activated, data corruption will occur.

In the grand scheme of things though, one or two cycles on every rank switch are negligible compared to the number of useful cycles performed between rank switches.

My suggestion is to simply leave it on auto or 2T with the rest of the settings entered manually, if it isn't broken don't fix it.
 
Upvote 0 Downvote
Solution
N

Nafryti

Distinguished
Sep 26, 2008
414
6
18,865


Thank you, i've been uneasy about this setting mismatch, but it doesn't seem to BSOD on me so i'll take your advice, and not fix it, as it isn't broken... hopefully that doesn't jinx me.
 
Upvote 0 Downvote
P

Pinhedd

Champion
Aug 7, 2007
10,342
20
56,965


As you stated earlier, 3T is incredibly rare. There have been DDR3-2133 2T modules around for quite a long time.

Keep an eye out for crashes, and run Memtest86 overnight for piece of mind if you wish. The memory access patterns generated by Memtest86 should stress rank switching.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom