Adjusting ram timings

[Ballz2TheWallz]

well i really never botherd to seriously tweak my system to the full extent that i could, but today i have some free time. I wanna know when im upping the FSB and the system starts to become unstable because of my ram what timing settings to i loosen first? do i change 2-2-2-5 1T to 2-3-2-5, or 2-2-2-8 or what? whats the best order to go about this. My ram is pc3200 TCCD and becomes unstable under torcher tests at about 220mhz(440mhz)

 

[tornadotsunamilife]

I have a 3200+ running at 2.2GHz using TCCD ram at 2-3-3-5 1T (stock timings 2-2-2-5 1T), ran prime95 and OCCT fine with 2-3-2-5 1T but failed memtest test #5 after about 20 passes.

Maybe try those although I know, no 2 chips are the same

 

[Ballz2TheWallz]

im priming atm sitting at 2405mhz on a 267mhz HTT, ram is on a 133mhz divider at 218mhz with slightly raised timings(2-3-3-6) for prime stabilty

edit nvm just got an error, damn wonder if its the ram, it was a rounding error .5 expected less then .4

 

[BallerX]

i have a dfi ultra infinity with a mobile barton.. my ram is corsair xms 2-3-3-6. mine is actually very strange. i have two different kinds of ram in the machine.. i have 1x1024 of tccd in 1 channel and 2x512 of some other type from their earlier generation.. normally i would not exptect to get good fsb oc with this mismatch.. however i am able to get a very respectable fsb oc of 225mhz at 1t-2-3-3-6. with a 10x multiplier.. i would start with some overly lax timmings and see what the absolute max stable memory oc is 1:1 and then start making the timings tighter and go from there.. with an xp 1to1 is where it is at.. my highest max fsb is around 235 fsb or so but with timmings of 3-4-4-8. with real world testing i discovered that 2250 mhz with the tighter timmings was faster overall. i also have my memory at 2.9 volts..
ymmv also tccd likes voltage so dont be afraid to give it some. stay below 3 volts though unless your cooling it directly with a fan.

 

[(cf)Eclipse]

quoted from my a64 oc'ing guide.. though i plan to make a better ram specific oc'ing guide soon


There are two basic options for making your ram go faster:
More voltage
Higher timings

here's what I do. it takes a while, but it works well.
set the following in the bios:
HT link multi = 2x (edit: some boards have weird quirks, try 2.5x or 3x instead if you feel this is an issue)
CPU multi = 5x
RAM divider = "200mhz 1:1"
and set the stock timings for your ram in the memory page

reboot with a memtest floppy in (or with memtest enabled if you have a dfi board)
in memtest, change to test #5, do one pass. if there are no errors, go back into the bios, and bump up the HTT (FSB) 3mhz, and boot back into memtest. repeat until you get errors in test #5. once this happens, you have two options:
1.) raise voltage. this won't always fix the problem, but it can help
2.) relax timings. kinda specific to the ram type you have, but you can go through and try to raise the minimal amount of settings to get it stable.
I would advise you write down the max speeds of each set of timings before errors occur. go back and fine tune the speed until it can do >5 whole passes of memtest without errors.
then write down these max stable speeds at each set of timings.



one thing i can tell you for tccd... the mhz/timing usually goes sorta like this, in order of slowest to fastest

2-2-2 < 2-3-2 < 2-3-3 < 2.5-3-3 < 2.5-4-3 < 3-4-4

i would personally stop at 2.5-3-3 or 2.5-4-3, depending on how many mhz that 4 gets you.
also, don't give tccd too much voltage... and definitly put a fan over any ram you overclock to keep it cool

 

[Ice Czar]

2-3-2-5, or 2-2-2-8

and generally speaking you leave the tRAS alone
or at least dont shorten it, the consequences arent pretty
loosening it might help depending

tRAS violations

tRAS is the time necessary to warrant data integrity, not only within the memory array but also within the entire system. If incorrect data are written back to the hard disk, drive corruption is the logical consequence, that is, violating tRAS can, among other things, result in corruption of not only some data on the HDD but also cause bad sectors on the drive.

Let's take a look at how the other candidates perform. As it turns out, only the double sided (64 Mbit chips) are able to function at a bank cycle time of 7 cycles (tRC = tRAS + tRP; 4+3 = 7 x 7.5 ns = 52.5 ns and even at the 140 MHz setting: tRC= 49 ns). All other DIMMs, including the single-sided Micron samples failed at this setting. Keep in mind that the bank cycle time is somewhat related to the depth of a memory chip. That is, the smaller the chip, the shorter can the bank cycle time be. Therefore, it is not surprising that the DIMMs with the lowest depth chips outperformed all other DIMMs with regard to tRAS valid.

Eight Ways To Kill Your HDD

TRAS Violation: The Creeping Corruption of a HDD

One of the most common reasons for HDD failure is what is called tRAS violation. tRAS is the minimum bank open time of the DRAM, that is, we are talking about system memory here. Many mainboard manufacturer still include Ultra and Turbo settings in their CMOS setup options that are only workable at 100 MHz memory bus settings, a.k.a PC1600 mode. One setting that has absolutely no impact on performance is the minimum bank open time or tRAS, while the same setting can have catastrophic consequences for data integrity including HDD addressing schemes if the latency is set too short. In theory, tRAS can be as short as tRCD + CAS delay, however, in reality, the minimum bank open time is dictated by the RAS Pulse Width, that is the time required to reach a voltage differential between memory bitlines and reference lines to safely identify a 0 or 1 logical state.


The main reason why tRAS violation does commonly lead to HDD corruption may relate to the translation of the physical memory space into virtual memory sub-spaces by the operating system and finally writing the data back to the storage media but it is not entirely clear what is going on there. A fact is, though, that a tRAS value of 5 is adequate for PC1600 or 100 MHz operation. At 133 Mz or PC2100, tRAS should never undercut 6T, likewise, at PC2700, the value should be increased to 7T where applicable. In terms of performance, tRAS settings hardly make any difference. We challenged some performance gurus at AMD on this matter and they reported a drop in Quake frame rates from 792 fps to 790 fps when increasing tRAS from 5T to 6T.

they are somewhat dated at this point in particulars of course

 

[tornadotsunamilife]

From what I understand (from your quotes) your talking about lowering the tRAS yes? I don't think anyone here is trying to lower it past the stock timings

 

[Ballz2TheWallz]

hm, thanks guys...just if anyone wants to know my ram tops out at 317fsb with so loose ass timings lol, my processor a 3000+winchester tops out at 2.4ghz so ill just be playing with my ram for awhile(i ended up with a shitty stepping a long time ago because no one had my MSI motherboard in stock)

 

[plywood99]

Another thing to keep in mind, when clocking ram try to set it so tras + trp = trc.

 

[Ice Czar]

From what I understand (from your quotes) your talking about lowering the tRAS yes? I don't think anyone here is trying to lower it past the stock timings

true, however I always post for posterity

 

[(cf)Eclipse]

they are somewhat dated at this point in particulars of course

that applies to any mis-set memory timing though.. is you don't get memory errors in memtest and you can do a couple round of superPI 32m, i'd say your data is pretty safe. if you're really worried, find a big file, run a checksum on it, copy it a few times and run checksums again