Old school OCZ Platinum OCZ4002048ELDCPE-K suggested settings

[caps_buster]

For my old computer to get some interesting overclock and also (mainly) more memory I got a 2G of these DDR (1) OCZ Platinum 2-3-2-5 sticks labeled as OCZ4002048ELDCPE-K.


(image only illustrational from eBay, not mine, no idea why one of these sticks have v1 and other v2 anyway)

So these should have the nice Samsung TCCD chips, so I was trying to power the machine on with 229MHz (250 FSB/183 divider that gives 229MHz on the rams) 2.5-3-3-7 settings and it - failed

WTF!

So I reset the bios and started going step-by-step. 200MHz divider (so, 1:1 to FSB) and I see the default settings are TCL - 2, TRCD - 3, TRP - 2, TRAS - 5, TRRD - 2, TRC - 11, TRFC - 24, TRWT - 3, Read Preamble 5.5ns and Async latency 7nS. Since I did not plan on getting FSB over 280, then these two last settings I leaved untouched.

However I do relax the timings to TCL - 2.5, TRCD - 3, TRP - 3, TRAS - 7, TRRD - 3, TRC - 11, TRFC - 24, TRWT - 3 and tried at witch clock the memory fail to post/run Memtest with voltage at auto to the rams.
220 was fine, 225 too, 230 no post. Backed down to 227MHz and that seems to work.
Then I relaxed the timings even more to TCL - 2.5, TRCD - 4, TRP - 3, TRAS - 7, TRRD - 3, TRC - 14, TRFC - 24, TRWT - 4 and tried the 230MHz clock and again it failed...

Previous lame Kingston 2x 512MB sticks can do 229MHz at these settings...

So I set the divider 183 and started raising the FSB while I know, that the CPU can do about 2850MHz with 1.425Vcore (and x11 multi), so up to about 255-260MHz I'm fine.

This went this way - I went back to these settings: TCL - 2.5, TRCD - 3, TRP - 3, TRAS - 7, TRRD - 3, TRC - 11, TRFC - 24, TRWT - 3 ... and started pushing - 235MHz FSB gives 198MHz, that was fine. 240 gives 203MHz on the rams and also it seems to be fine in DOS/Memtest. 245 gives 207MHz and it also looks fine... but 250 (as I run before) failed to post. So I set 248 with gives 209MHz on the rams and that worked well, but on WinXP boot it give BSOD. Damn. So backed down to 246 and that boot WinXP, but then it fail on BSOD
WTF... I changed the voltage from auto (IIRC 2.5V it should be) to 2.70V and... reboot instead of BSOD. So voltage are no-no, therefore I tried voltage back at auto + 245MHz FSB and that somewhat works, but very unstable.

So I had to back down the divider to 166, so I get 250 - 196.4MHz for the rams. Then 255MHz FSB gives out 200.4MHz on the rams, so I set back their tightest default settings, as at 200MHz, there is really no need to relax timings at all.
That gives 2805MHz for the Opty 148 using Asus A8R32-MVP Deluxe (Radeon Xpress 3200)...

But I'm somewhat disapointed. I used to run OCZ P4001G rams (also 2x 1G) at 231MHz with 2.5-3-3-7 settings at 2.64V ...

So, what I doing wrong? At 1:1 divider these rams managed to 227MHz, witch is under the 231 wanted, but at least they can be overclocked somewhat. Now when I push the divider, then even very slight overclock (209MHz) is impossible. WTF! Old OCZ P4001G rams run their tightest 2-3-2-7 settings up to 216MHz.

Do I need to play a bit with the skew or relaxing the drive in order to get higher? I refuse to believe that I cannot get these rams clock faster. So I asking for suggestions. That was just my first quick attempt to get the machine working by quick & dirty way

[cbjaust]

at 200MHz 2-3-2-5 2.8v. Also raising TRAS might give better bandwidth as measured in Memtest86+. At higher frequencies try 2.5-3-3-11 up to 3.0v. The IC on these modules are probably not TCCD, more likely they are ProMOS.

[caps_buster]

http://www.ocztechnologyforum.com/fo...002048eldcpe-k

...according to this, that was pretty much confirm what I managed to discover - 225MHz require extreme voltages and very relaxed secondary timings to get 2.5-3-3 to work... much less to get it to work stable. Hence they are really not worth the trouble, so they go to the P4 box, where is basicaly no overclocking happening and these old OCZP4001G sticks seee more action.

Looks like that the only way to get something out of the OCZ4002048ELDCPE-K is to get the Rev 2 of them and nothing else.

...


So in the end, I was disapointed. Also I tested how far one can go with the default 2-3-2 timings and stock 2.50V on these OCZ4002048ELDCPE-K. Not very far. 205 seems to be stable, but at just 210 it started reporting bios corruption and so on... so somewhere between 205 and 210 is the limit of the current timings with no bumped up voltage. That is not much headroom, as my OCZP4001G sticks 2x1G do 216MHz at 2-3-2 timings (2.60V tough, might be worth to try with 2.60V these ones too) ...

In the end, I use these OCZ4002048ELDCPE-K sticks at P4 mobo, where is no overclocking with very tight timings at 2.50V they worked just great. I was able to get them stable at 200MHz and 2-3-2-5 with TRFC 9, TRRD 2, TRTP 2, TWTR 1 and TWR 2 timings. That is very very nice.

But for overclocking I use the OCZP4001G sticks, that can do much, much better - at 233.7MHz they go with 2.5-3-3-7 TRC 9, TRFC 12, TRRD 2, TWR 3, TWTR 2, TRTW 2:
http://hwbot.org/submission/2496577_...am_392.4_marks
...in same mainboard. So I declare OCZ4002048ELDCPE-K sticks useless for overclocking, unless they are v2 with the Samsung chips. This was a bad purchase. So much for "Platinum"...

[Reefa_Madness]

So these should have the nice Samsung TCCD chips, so I was trying to power the machine on with 229MHz (250 FSB/183 divider that gives 229MHz on the rams) 2.5-3-3-7 settings and it - failed

WTF!

Looks like that the only way to get something out of the OCZ4002048ELDCPE-K is to get the Rev 2 of them and nothing else.

...in same mainboard. So I declare OCZ4002048ELDCPE-K sticks useless for overclocking, unless they are v2 with the Samsung chips. This was a bad purchase. So much for "Platinum"...

Part of your disappointment with the results appears to come from your comparing them to Samsung TCCD based modules. The kits you tested were most likely using Infineon chips.

Based on the parts number, the OCZ Plats that you're posting about are 2x1GB kit and the coveted Plats with TCCD were only available in 2x512MB modules, so you aren't comparing apples to apples. The TCCD modules weren't "v2", but instead had "Revision 2" in the name, and again, were only available in 2x512MB kits. Not to beat a dead horse, but I'll say it again...you will not find TCCD in ANY 2x1GB kit.

This is a link to the product page on Newegg which shows the name and parts number.

http://www.newegg.com/Product/Produc...82E16820146890

These ended up being made in three different "versions", the first, original release did not have a "verx.x" number on them, subsequently OCZ released them with not TCCD chips (which were binned by Samsung), but instead with TCC5, which were the cheaper un-binned chips and OCZ did the binning. These have a tiny, tiny "ver1.1" just to the right of the timings. After an uproar from customers about them using TCC5, a "ver1.0" was released which returned to using TCCD. The problem with the TCC5 was that while the chips were just as good on the highly binned PC4800 stuff, the headroom on the lower PC3200 was no where as good as the original release.

If you're looking for some crazy clocks on DDR1, then go find a kit of either the modules with no version number on them, or those with "ver1.0".