Frustrated with my TCCD, new memory ???

I have 2X512 PC3200 Patriot xblk. This should be TCCD chips. This is my second set, the first set was RMA'd for memtest errors at stock speeds. On this set, I can run stock timings and speed with 2.5v. That's DDR400 w/ 2-2-2-5. I can keep these timings up to about 210mhz, anything above that, while memtest stable, will crash windows. I can continue lowering timings and @ 2.5-4-3-10 and 2.8v 300mhz (htt multi x3) I can pass memtest86 with flying colors, but windows will not load. I have to back the fsb back down to about 210 to get it stable in windows, even with such loose timings. So far, after reading all the dfi mobo stickies, trying 3 different bios, and following every TCCD guide I can find, I can't get anyting over ddr420 stable in windows, although memtest runs fine all the way up to 300?? What gives ? Could this be a problem with my mobo ?
I'm loath to try another set of patriot after my experience with the last two sets. I've been considering some mushkin redline xp3500, I'm at my wits end with this patriot. I could try to RMA it again, but since it performs fine @ stock speeds/timings/voltage I really have no basis for RMA. I don't have another rig to test this stuff in. Seeing how well it does at stock speeds and voltage, I'm allmost inclined to think it's a problem with my mobo.

Oh, just to note, I can get into windows stable with the htt @ 300x3 if I run the memory on a divider keeping it around 200mhz , so I don't think the proc or memory controller is to blame.

You may need to attempt running at 2T.

I have definately tried running @ 2t. Does not seem to be any different than 1t in trying to get above 210 stable.

even 3-4-4-10 2t 2.9v is not stable in windows over 210. Yet 210 is stable @ 2-2-2-10 1t 2.6v ? I've been through all the "stable settings" threads I can find here and at DFI trying different settings. Nothing seems to make a difference.

Some more notes.. Memory is in the orange slots, I've tried swapping them. I've tried every voltage to 3.0, and all the general "stability" settings i've been able to find on the web for TCCD.

J.

Patriot had a good run but they are shady as of late, I had 5 sets of XBLK I played with, 2 clocked well 1 clocked decent and 2 sets were utter crap.

RMA then for some OCZ plats rev2 if you can, you'll be much happier

http://www.newegg.com/Product/Product.asp?Item=N82E16820146890

Like so ?

I just got these patriots back from RMA. The last set ran ok for a while, and then one of the sticks just went *poof*. After that it wouldn't even run underclocked with loose timings without errors in memtest.

I bumped the vdimm up to 3, and set cas @ 3.0, after that I was able to get to 275, but the bandwidth was shot, lol.

What's working me over, is how it can pass memtest overnight with 2.5,4,3,10 @ 300mhz but not even load windoze with the same timings and voltage @ 250mhz ? or even 220mhz... sheesh.

J.

yes those are the sticks.
memtest and windows stablity are 2 very different things, memtest all memory is free, its sort of step 1 in trying to find stabiltity, once in windows and the memory is filled and accessed its a different ballgame entirely.

Yes, I've experienced that. But I've never seen such a huge gap between what is memtest stable, and what is windows stable. 300mhz stable in memtest, but only 210 in windows ?? Now that's just crazy...

I wonder how to go about returning this stuff to newegg ? Since I just got this back from RMA, do you think they will let me exchange it ?

J.

juliendogg said:

Yes, I've experienced that. But I've never seen such a huge gap between what is memtest stable, and what is windows stable. 300mhz stable in memtest, but only 210 in windows ?? Now that's just crazy.

Click to expand...

I agree.

Thus makes me wonder if that's caused by the motherboard chipset?

The motherboard chipset possibly requires more voltage.

That's when a VDD? increase can help.

Have tried upping the chipset voltage as well. Alas, no help there. I considered that the chipset might be the issue, that's why I tested at 300+ with the memory on a divider to keep it below 200mhz. 333X3 was not a problem so long as the memory stayed below 200. Not sure if that tells me anything or not, but I would think if it was a chipset problem, then 333 would not be acheivable, memory divider or not.

J.

juliendogg said:

Yes, I've experienced that. But I've never seen such a huge gap between what is memtest stable, and what is windows stable. 300mhz stable in memtest, but only 210 in windows ?? Now that's just crazy...

I wonder how to go about returning this stuff to newegg ? Since I just got this back from RMA, do you think they will let me exchange it ?

J.

Click to expand...

PM Jason from newegg he will get you sorted out.

juliendogg said:

I have 2X512 PC3200 Patriot xblk. This should be TCCD chips.

Click to expand...

Yep. They are TCCD and good ones to boot!

This is my second set, the first set was RMA'd for memtest errors at stock speeds. On this set, I can run stock timings and speed with 2.5v. That's DDR400 w/ 2-2-2-5. I can keep these timings up to about 210mhz, anything above that, while memtest stable, will crash windows. I can continue lowering timings and @ 2.5-4-3-10 and 2.8v 300mhz (htt multi x3) I can pass memtest86 with flying colors, but windows will not load.

Click to expand...

Memtest stability doesn't equate to Windows stability (I think you know that, but this is for anybody who doesn't). Did you cut the LDT to 2.5? These sticks fly at 2.9V. You need active cooling though. Having tested 3Gigs myself (testing the fourth set right now), I can tell you that 2.9V is almost a prerequisite to stability. Drive strengths I used were 8/7 and 1.

I have to back the fsb back down to about 210 to get it stable in windows, even with such loose timings. So far, after reading all the dfi mobo stickies, trying 3 different bios, and following every TCCD guide I can find, I can't get anyting over ddr420 stable in windows, although memtest runs fine all the way up to 300?? What gives ? Could this be a problem with my mobo ?

Click to expand...

Did you refer to my timings in the Memory section and the DFI section? I don't think its a problem with the Mobo. Usually, there is one weak stick in the set.That should be placed close to the CPU for stability.Use the Orange slots.

I'm loath to try another set of patriot after my experience with the last two sets. I've been considering some mushkin redline xp3500, I'm at my wits end with this patriot. I could try to RMA it again, but since it performs fine @ stock speeds/timings/voltage I really have no basis for RMA. I don't have another rig to test this stuff in. Seeing how well it does at stock speeds and voltage, I'm allmost inclined to think it's a problem with my mobo.

Oh, just to note, I can get into windows stable with the htt @ 300x3 if I run the memory on a divider keeping it around 200mhz , so I don't think the proc or memory controller is to blame.

Click to expand...

If you wish, I can test them for you. I am in the process of testing a set for somebody else on the forums, so I have a large set of timings I could choose from. Regardless of what anybody else says, high HTT will take its toll eventually. So, if you see a progressive decline in stability over a period of several months, it isn't something to be surprised at.

Good luck mate!

S-N

Super Nade said:

Yep. They are TCCD and good ones to boot!


Memtest stability doesn't equate to Windows stability (I think you know that, but this is for anybody who doesn't). Did you cut the LDT to 2.5? These sticks fly at 2.9V. You need active cooling though. Having tested 3Gigs myself (testing the fourth set right now), I can tell you that 2.9V is almost a prerequisite to stability. Drive strengths I used were 8/7 and 1.


Did you refer to my timings in the Memory section and the DFI section? I don't think its a problem with the Mobo. Usually, there is one weak stick in the set.That should be placed close to the CPU for stability.Use the Orange slots.


If you wish, I can test them for you. I am in the process of testing a set for somebody else on the forums, so I have a large set of timings I could choose from. Regardless of what anybody else says, high HTT will take its toll eventually. So, if you see a progressive decline in stability over a period of several months, it isn't something to be surprised at.

Good luck mate!

S-N

Click to expand...

Thanks man,

I did indeed refer to the timings you have posted, yours and at least a dozen others, combined with tweaking based on my own knowledge of memory and timings. I find that one stick will give errors just slightly before the second stick, testing individually they are VERY close to each other, but I do have the weaker stick closest the cpu, and using the orange slots.

Windows is not stable at your recommended timings, nor anyone elses. I do realize that memtest stable does not = windows stable, but as i posted above, I have never seen THAT large a gap between the two. 300mhz memtest stable, 210 windows stable with same timings/voltage ?

If I had any other memory to use while you test these, I might take you up on that offer, but I don't at present have any other sticks.

J.

The discrepancy you speak of is highly unusual. It is quite possible that your XP install is dying. Did you run tests on a fresh install?

yhpm regarding my offer.

Super Nade said:

The discrepancy you speak of is highly unusual. It is quite possible that your XP install is dying. Did you run tests on a fresh install?

yhpm regarding my offer.

Click to expand...

It's a brand new box with a fresh install and all new hardware. I was considering formatting and reinstalling as I did not do that when I switched from a 6800nu to an x800gto2. Games and Apps seem to be stable at stock speeds though, and I can hit 2.7ghz stable on my venice. I'm thinking 2.8 will be doable with my chip, but I wanted to get this memory issue sorted before finding out...

J.

SN,

I know I had tried your exact timings before, but could not remember the exact result. So I came home and plugged them in again.

Here's what I set. I am underclocking the CPU to take him out of the equation...

DRAM Timings:
Ram Multi 200 1:1
CPC enabled
Tcl 2.5
Trcd 04
Tras 10
Trp 03
Trc 10
Trfc 20
Trrd 02
Twr 03
Twtr 01
Trwt 08
Tref 3.9us, 166mhz (i think that equates to 2560)
Bank Inter Enabled
Dynamic Skew Control Auto (off)
Skew Value 0
DRAM Drive Strength 07
DRAM Data Drive Strength 01
Max Async 8ns
Read Preamble 5ns
Idle Cycle 256
Dynamic Counter AUTO
R/W Queue Bypass 16x
Bypass MAX 7x
32 Bit Gran DIS

Voltages,

HTT 300 (also tried 280)
LDT X3
CPU Multi x6
PCI-E 100
Vcore Startup
Vcore 1.4
Vcore Special AUTO

LDT V 1.3
Chipset V 1.6 (tried 1.7 also)
Vdimm 2.9 (tried 3.0 also)
Dram +0.03 Disabled


With these settings I got 2 errors in 10 runs of memtest test #5. Dropped the HTT to 280 and ran 10 loops of memtest #5 with no errors. As a quick test, ran a full loop of memtest with no errors. Booted into windows ok.

I opened up Clockgen, and Superpi. Using superpi 512K as a quick test, it failed out immediately with the "not round" error. I had to lower the HTT with clockgen to 220 to complete superpi 512k.

What do you think ? I'm going to try to reinstall windows and drivers to see if that helps. I'll work on that tonight.

J.

Same thing with a fresh install, minimal software loaded.

On one of my sets Drive strength of 08 and 01 worked flawlessly. Try that.
Use the 6-23-3 BIOS. I see thatyou are using an older BIOS, as the "DRAM Response Time" option is missing. This is very important and useful

I re-installed windows and had the same results. I decided to do some more testing on each stick alone. Very interesting results,

Bios 6-23-3
HTT 280
DDT X3
CPU Multi X6
PCI-E 100
Vcore Startup
Vcore 1.4
Vcore Special -

LDTv 1.3
Chipset V 1.7
Vdimm 2.9
Dram + 0.03 disabled

Using the exact timings posted above ^^

Module A
SuperPI stable @ 315mhz

Module B
SuperPI stable @ 310mhz

It would seem that the memory clocks just fine, and the problem lies with dual channel on my mobo.

This is with 1 module in DIMM #2 (the upper orange slot)

Put it in dual channel and it fails the same tests at 210mhz.

Ideas ???

I'm going to spend the next few minutes testing the yellow slots. I keep hearing not to use them with TCCD. Worth a shot i reckon .


J.

yeah they say the yellows are better for BH5/UTT but you never know. You may have a bad orange slot.