Memory timing help needed

S

SavageThrash

Limp Gawd
Joined
May 16, 2007
Messages
249
Alright guys I come to the only forums that ever seem to provide useful advice in a timely matter with an issue that I am totally lost on. It involves reprogramming the spd on some ddr 2 so-dimm, heres where Im at.

I have 2 tester sticks of unlocked eeprom ram and i want to keep the sticks at the looser 333mhz (pc2-5300) timings whilst running them at 266mhz (pc2-4000). The issue im having is that the software Im using (thaiphoon burner) allows me to adjust timings via ns. Now when i set the memory to 266 the timings in the software automatically jump down to the proper 266 timings.

I want to set them back to the 333mhz timings however the timings in what i call standard format (5-5-5-15 for example) have MANY sub settings within the ns bracket that im using to adjust. Let me use tRCD for an example. It is set to 15.00ns and that gives me a timing of '4t'. The 333 spd is a 5t for tRCD however I have the option of anywhere from 15.25ns to 18.75 ns under the 5t timing. The software doesn't give me the ns timing of the 5t 333mhz setting so I'm lost as to what I can/should set it to. I thought the best route would be to just set them all at the very loose end of the spectrum (18.75 ns) but I have a feeling that if i do that with all the timings that maybe something wont match up and cause the stick to function improperly/very poorly.

If you guys have read all this I thank you very much and appreciate any insight you can give. If theres something i haven't explained well enough let me know and i can get some screenshots of the software in here as well but for now I need to sleep this nightmare off and hope for some help from you guys in the morning :)
 
Alright maybe you guys can answer this question then.

How come when i rewrite my spd to 266 the memory still boots at 333 speeds? Should i be clearing my cmos as well or is this just a bios thing?
 
How come when i rewrite my spd to 266 the memory still boots at 333 speeds? Should i be clearing my cmos as well or is this just a bios thing?
Click to expand...

I would have thought on a cold boot it would have re-read the SPD. On a warm boot /shrug Since you did not make clear if you checked/reset the bios settings or shutdown completely after the SPD write. /shrug.

Typically everthing is in "tics" of the clock. 1/fx (inverse of clock speed will give time.

So 333MHz has a tic of 1/333,000,000 = 3.003 x e^-9 = .0000000003003= 3 nS. = one tic or T

266 is 1/266e^6 = 3.759 x e^-9 = 3.76 nS = one tic or T

when multiplying the above tics x some number and coming with with a final timing value in NS. round up all time is the general rule. So if math comes out 15.2 nS and you can only set whole numbers, 16nS would be the first choice to try.

The 333 spd is a 5t for tRCD however I have the option of anywhere from 15.25ns to 18.75 ns under the 5t timing. The software doesn't give me the ns timing of the 5t 333mhz setting so I'm lost as to what I can/should set it to.
Click to expand...

tRCD - Row Address to Column Address Delay:
tRCD is the number of clock cycles taken between the issuing of the active command and the read/write command. In this time the internal row signal settles enough for the charge sensor to amplify it.


@333 5T(ics) would be 5 x 3nS = 15nS. propagation delay etc cannot be calculated etc. clock jitter etc.
so the 15.25 would be the tightest for 5T and may or may not work. Each and every one of those settings depend on the memory chips, slot installed, and every other setting you see in the SPD profile. So you might need to go 16.25 all other things being the same. /shrug Thats the beauty and the pain of being able to dial in the settings yourself. The main timings are not a big issue nor are some of the others that are derived from the main timings but as you say, there is a bunch of stuff that unless you have the specs for the memory chips and are a memory design engineer, I hold little hope of you being able to optimize them.

Going with the upper end of the allowed range on stuff you cannot find info on would be a good idea to start.


---------
In short, if you did not have a record or some other source of the complete settings for 333MHz operation of those ram sticks, I see little hope of actually finding by testing or trail and error all the settings for absolute best performance or to even match the stock 333 configuration unless you can get the values from another stick etc.

One stragety is to look/find/determine exactly what memory chips are used by manuf and exact part number and then find a 333 stick and read the SPD values and use them.

Another option ( from the way you are posting I bet this is not an option as the bios settings are limited/non-existant) is to put the SPD setting on manual, input the main settings and let the hidden tables in the BIOS set up the memory. You could then use something like Memset to read the values, convert to nS and use those values for a lot of your "blanks".

Finding someone with a 333 stick using the same exact memory chips would be the easiest way out.

Thats my best shot. Likely I am completely wrong about everything, its been a while since I messed with it.
 
I did fully shut down the pc after flashing the new spd, I guess ill try clearing cmos as well.

I have another stick of 333mhz of the same ram that I haven't messed with but like I said the software won't tell me what the exact NS timings are. Im just going to take your advice and start at the loose end of the spectrum and work from their - That's if I can even get it booting at 266.

And yes your correct in the lack of bios support. Im trying a theory overclock on the rig in my sig. If im correct the RAM is whats killing my overclocks so I wanted to run one step down (pc2-6400 to pc2-5300) and keep the pc-6400 timings In hoping that when I bump my FSB up my memory wont crash. So far I've only been able to find pc-5300 sticks with a writable eeprom.

Thanks for the lesson on the tics, helps me wrap my head around all of this :)
 
You must log in or register to reply here.
Back
Top