OCZ VX PC4000 - Burn In - OCZ rep please

[Tempest25]

I've looked around and haven't found an answer to my specific question. I've seen OCZ reps (Andy mainly) that recommend burning in at 3.2V and also it's been said at 200 with 2-2-2 settings, but I haven't seen a specific answer to:
Is it better to burn in at only 200 2-2-2 vs 245 (for instance) 2-2-2 when you can reach 245 at 3.2V with no problems?

I just received my PC4000 VX and so far have it up to 245 bus @ 3.2V with no problems. That seems to be the limit @ 3.2V. It would seem to me it would be better to burn in at the highest stable speed at the 3.2V.

What is best to burn in with? Should I use S&M, Prime95, or MemTest? If it's MemTest, should I just let all tests run continuously or loop test #5 only?

If you've found differently in your testing (i.e. burning in at 200 ultimately results in higher final bus speeds), please let me know.

Also, when you're verifying the 250 bus speed in your testing, is there a standard voltage your verifying that at?

[justwOo]

The answer I got from the OCZ reps when I was having problems was... 2-2-2 at 200mhz and 3.0-3.2v for 24 hours straight, then go up as far as you can go at each voltage, running 1 complete loop of memtest *The entire loop, not just test *
Then up voltage and continue going, running only 1 loop to test each new speed

[andyOCZ]

The answer I got from the OCZ reps when I was having problems was... 2-2-2 at 200mhz and 3.0-3.2v for 24 hours straight, then go up as far as you can go at each voltage, running 1 complete loop of memtest *The entire loop, not just test *
Then up voltage and continue going, running only 1 loop to test each new speed

As justWOo says. This is what i recommend. You can also increase in steps after the first 24 hours. BTW nice speeds at 3.2v. :)

BTW nice ferret justWOo

[nOx34]

As justWOo says. This is what i recommend. You can aslo increase in steps after the rirst 24 hours. BTW nive speeds at 3.2v. :)

BTW nice ferret justWOo


lol , someone's typing in a hurry :p:

What's your recommendation on burn-in procedure for the gold vx pc3200 ? :banana:

[andyOCZ]

lol , someone's typing in a hurry :p:

What's your recommendation on burn-in procedure for the gold vx pc3200 ? :banana:

Who me?

Burn in is the same as above. What justwOo says. Same procedure.

[mdzcpa]

I just started testing my new VX last night. great stuff. Doing 2-2-2-8 at 265 just fine at 3.3v. :banana:

If burn in can help, then this is indeed some awesome memory.

[Torin]

I got mine yesterday, burned in for 12 hours at 200 2-2-2-10 @ 3.2V. Early this morning I kicked it up to 250 2-2-2-10 @ 3.2V, so far no issues. When I get home for lunch, I'm gonna bump it as high as it can go at 3.2V, let that run till tonight then start moving it up to 3.3V and beyond.

[kimandsally]

Andy OCZ, could you explain what burning in does? I have some OCZ 1GB (2 x 512MB) PC3200 Dual Channel Platinum Series EL-DDR REVISION 2 CAS2
Would this also benefit from burning in?

I must say I am happy with it I can get 260 stable @ 3-4-4-8 2.7v

The voltage is strange on this ram it does not improve with more voltage it does run the same @ 2.85v but is there any benefit running 2.85v as opposed to 2.7v if I'm only getting the same speed?

Thanks :)

[Torin]

No benefit in using more voltage if it doesn't make the OC more stable, or offer a higher OC. Sometimes more voltage, specifically on memory with TCCD (like your's) makes it unstable.

And no, burning in does not help TCCD chips... but it does help Winbond UTT chips, which is what the VX series uses.

[Tempest25]

Thanks for the replies. I set it to burn-in at 3.0V at 200 2-2-2-7-1T and have it going now while at work. I used it for about 5 hrs last night at 3.2V @ 245 though. Hopefully that won't adversely affect my end results. I'll start cranking it up this evening and let everyone know what I end up with.

[xsky]

does anybody else have a problem with memtest errors in test 1-3 when voltage is set to high? (~3.45V)?

[kimandsally]

No benefit in using more voltage if it doesn't make the OC more stable, or offer a higher OC. Sometimes more voltage, specifically on memory with TCCD (like your's) makes it unstable.

And no, burning in does not help TCCD chips... but it does help Winbond UTT chips, which is what the VX series uses.

Cheers mate thanks for that, there is so much to read and learn it's hard to keep up but it's nice if you can do it right :toast:

[OCZEric]

Andy OCZ, could you explain what burning in does? I have some OCZ 1GB (2 x 512MB) PC3200 Dual Channel Platinum Series EL-DDR REVISION 2 CAS2
Would this also benefit from burning in?

I must say I am happy with it I can get 260 stable @ 3-4-4-8 2.7v

The voltage is strange on this ram it does not improve with more voltage it does run the same @ 2.85v but is there any benefit running 2.85v as opposed to 2.7v if I'm only getting the same speed?

Thanks :)


Hi Guys,

Please don't "burn in" any TCCD based product. In general, TCCD's already get very very hot, and if you go through the burn in procedure that Andy mentioned for VX for the TCCDs, you may permanently damage your module. At high voltage, under burn in conditions, I've seen TCCDs get as high as 90C and suffer thermal damage resulting in them performing worse than before.

Thanks :)

-Eric

[HiJon89]

Hi Guys,

Please don't "burn in" any TCCD based product. In general, TCCD's already get very very hot, and if you go through the burn in procedure that Andy mentioned for VX for the TCCDs, you may permanently damage your module. At high voltage, under burn in conditions, I've seen TCCDs get as high as 90C and suffer thermal damage resulting in them performing worse than before.

Thanks :)

-Eric
What about MemTesting them at just 2.8V for 24 hours to break them in?

[OCZEric]

It should be ok, just be sure to keep them relatively cool. You have to understand that normal operation never gets to the full load 24/7 that memtest will stress the modules at.

Under normal operation, there will be lags in the memory access that give the core of the dram chips time to cool down. But if you are stressing the memory constantly, the cores will not have much time time to cool off and so residual heat will build up unless you are careful to keep the memory cool.

Also, I personally have not seen real increases in performance due to burning in TCCDs, even at normal voltages. Your mileage may vary of course :)

-Eric

[HiJon89]

I'm getting 2 80mm fans to put directly over them, mainly for my BH-5 but I'll use them when I use the TCCD also, together there should be about 50CFM going over the RAM at all times.

[spytech]

I just started testing my new VX last night. great stuff. Doing 2-2-2-8 at 265 just fine at 3.3v. :banana:

If burn in can help, then this is indeed some awesome memory.

damn mine (vx4000) are in the mail. but still dont have a system to put it in
:mad: waitin for venice and the new ati board (or dfi whichever turns out better)

mdzcpa = how far do you have it so far? need an update, since i cannot test mine yet i can be happy for others

[charlie]

I just started testing my new VX last night. great stuff. Doing 2-2-2-8 at 265 just fine at 3.3v. :banana:

If burn in can help, then this is indeed some awesome memory.

Mike,
VX4000 or 3200?

C

[Torin]

His is 4000.

[kimandsally]

Hi Guys,

Please don't "burn in" any TCCD based product. In general, TCCD's already get very very hot, and if you go through the burn in procedure that Andy mentioned for VX for the TCCDs, you may permanently damage your module. At high voltage, under burn in conditions, I've seen TCCDs get as high as 90C and suffer thermal damage resulting in them performing worse than before.

Thanks :)

-Eric

Thanks I'm glad then that I haven't tried it and I never thought about the ram never getting a rest I feel much better that I haven't cooked them :)

What does burning in actually do then to the vx to improve it?

[Torin]

No one really knows, we just know that it works.

[mdzcpa]

Mike,
VX4000 or 3200?

C

VX4000.

Although 265 2-2-2-8 is nice, I'll be pushing for more tonight. Right now my Gskill LA at 295mhz 2.5-4-3-7 is still outperforming the the VX. I'll need to get the VX into the 275 range to beat the LA.

[kimandsally]

No one really knows, we just know that it works.

Any idea how it started? While on the subject do you think the cpu's i;e amd winchesters benefit in any way from a burn in as I read on here that people got a better clock after stressing it?

Then someone said they go off after a while of being overclocked and putting the cpu in the fridge takes it back???

Any thoughts on that?

[iboomalot]

think 285 can be hit at 2-2-2 ???

[isp]

VX4000.

Although 265 2-2-2-8 is nice, I'll be pushing for more tonight. Right now my Gskill LA at 295mhz 2.5-4-3-7 is still outperforming the the VX. I'll need to get the VX into the 275 range to beat the LA.

How close is it?

[chinkgai]

yeeeeehaw! mine came in this morning! burning in right now at 200 @ 3.0v

one question...what timings beyond 2-2-8-2 are we supposed to use?

im using the same timings as my value vx set which was at 250Mhz:

2-2-8-2-7-15-2-2-2-2, rest auto

i guess i wont have "real results" till this weekend after burn in!

[Tempest25]

I'll post my results tonight. I was going to last night, but I had no luck increasing my speed beyond 247 Mhz. I hadn't tried beyond 245Mhz prior and switched back to 200 2-2-2 3.0V for burning in the prior night, so when I got home last night after nearly 24hrs at that speed (and about 5 hrs at 3.2V @ 245Mhz), I couldn't get it beyond 247Mhz MemTest stable no matter what voltages I tried. (wow, that's a long sentence... lol)

I tried increasing all voltages (LDT, VCORE, DRAM) independently and together and could not improve my results. I'd always error out in the first loop of all tests. It was easy to determine by using test #5. It'd error out at the end every time, even at 250Mhz @ 3.3V or 3.4V.

I have it looping at 3.2V @ 247Mhz and have had no errors in 16 hrs so far, so I'm going to see what I can do with it tonight after I get home from work.

I may need to change around my timings a bit more and I'm also going to put a high-speed house fan blowing in the case to make sure it's not a heat issue, though I've felt the heatspreaders and they haven't been very hot to the touch. I have a ThermalTake Tsunami case and the fan mounted in the plexiglass side blow directly over the ram.

[Torin]

From what I've found so far, it doesn't seem to benefit by adding more voltage immediately. It seems best to step up the burning by voltage... you won't get much out of 3.4V until you burn in for a while at 3.3V, which you won't get much out of until you burn in for a while at 3.2V.

[chinkgai]

tempest, its very most likely your timings rather than heat

[Tempest25]

Well, I was able to get it a bit higher with no errors for 20 passes of loop #5 (I know, in no way conclusive). I got it to 250 using 3.3V with the fan (one of those windtunnel small household fans) blowing on it. I'm wondering how much timings are restricting me. This board, as Andy has said, can be a nightmare. I'm running the following timings:
2-2-8-2-8-15-2-2-2-2- 3072 (I've tried 1560 here also) - Auto - Enabled
Auto - 0 - Auto - Level 4 - 7.0ns - 5.0ns - 16 - Enabled - 16X - 7X - Disabled

I'm back to running at 3.2V @ 247 tonight so I can do something other than test, but will see if I can put some high speed fans in the case tomorrow to see if that'll help, though it doesn't seem like it will very much.

Anyone have any suggestions on changes to my timings? As best I can tell from looking over posts, these are optimal.

Also, if an OCZ rep looks over this again, how do you verify the 250 setting? What voltage do you use? And what timings do you guys use in your setups?

[ryanpgroovy]

PM OCZeric and get him to shoot you the test bios settings

[Torin]

Alright, I've got a kind of weird issue.

Last night and today I was running memtest #5 @ 268 2-2-2-10 w/ 3.4V, it ran for about 13 hours with 0 errors.

When I got home from work, I kicked it up to 269 and starting memtest #5 again. After 1 loop, I got about 10 errors. I then changed the voltage to 3.5V and ran test #5 again, this time I got about 30,000 errors. I dropped it down to 3.3V and got about 300 errors. So to summarize:

268 2-2-2-10 @ 3.4V, test #5 for 600+ loops: 0 errors

269 2-2-2-10 @ 3.3V, test #5 for 1 loop: 300 errors
269 2-2-2-10 @ 3.4V, test #5 for 1 loop: 10 errors
269 2-2-2-10 @ 3.5V, test #5 for 1 loop: 40,000 errors

What's going on? Is this a timings issue? All I specify in the BIOS are 2-2-2-10, the rest are Auto, which gives me this: 2, 15, 17, 2, 2, 10, 2, 3, 2, 4, 200 Mhz 3.9us, 8x, 64 bit, Disable, Disable, 4x, 7 ns, 5.0 ns, 16 clks, Enable, 200

Drive strength is auto, which I believe is defaulting to 2, but not sure.

And all this time, my +3.3V rail is at 3.71V

Edit: Oh, forgot I temporarily removed my sig..... this is with a DFI LP UT 250gb, 3400+ Mobile Clawhammer, chipset @ 1.8V and OCZ PowerStream 520W. And cooling the memory is definitely not an issue... got at least 150CFM of 65F air on it.

[B5I8]

PM OCZeric and get him to shoot you the test bios settings
Why not just post it for all to view.

[chinkgai]

what timings beyond 2-2-8-2 are we supposed to use?

also, during burn in time, what if you get errors? are you supposed to continue?

i've had two errors after 200something passes

[Tempest25]

What passes did the errors occur in? From my understanding, if it's only a few errors, they should disappear after burning in long enough. Also, 200 passes of what? Are you not looping the whole series? There's no way you have 200 passes of the whole series ... I got about 50 in 16 hrs. When burning in, you should be doing the full series.

[xsky]

Alright, I've got a kind of weird issue.

Last night and today I was running memtest #5 @ 268 2-2-2-10 w/ 3.4V, it ran for about 13 hours with 0 errors.

When I got home from work, I kicked it up to 269 and starting memtest #5 again. After 1 loop, I got about 10 errors. I then changed the voltage to 3.5V and ran test #5 again, this time I got about 30,000 errors. I dropped it down to 3.3V and got about 300 errors. So to summarize:

268 2-2-2-10 @ 3.4V, test #5 for 600+ loops: 0 errors

269 2-2-2-10 @ 3.3V, test #5 for 1 loop: 300 errors
269 2-2-2-10 @ 3.4V, test #5 for 1 loop: 10 errors
269 2-2-2-10 @ 3.5V, test #5 for 1 loop: 40,000 errors

What's going on? Is this a timings issue? All I specify in the BIOS are 2-2-2-10, the rest are Auto, which gives me this: 2, 15, 17, 2, 2, 10, 2, 3, 2, 4, 200 Mhz 3.9us, 8x, 64 bit, Disable, Disable, 4x, 7 ns, 5.0 ns, 16 clks, Enable, 200

Drive strength is auto, which I believe is defaulting to 2, but not sure.

And all this time, my +3.3V rail is at 3.71V

Edit: Oh, forgot I temporarily removed my sig..... this is with a DFI LP UT 250gb, 3400+ Mobile Clawhammer, chipset @ 1.8V and OCZ PowerStream 520W. And cooling the memory is definitely not an issue... got at least 150CFM of 65F air on it.

That might be the same I described some posts ago in this thread.
Could you check if you get errors in tests 1-3 when using 3.5V?

[andyOCZ]

Alright, I've got a kind of weird issue.

Last night and today I was running memtest #5 @ 268 2-2-2-10 w/ 3.4V, it ran for about 13 hours with 0 errors.

When I got home from work, I kicked it up to 269 and starting memtest #5 again. After 1 loop, I got about 10 errors. I then changed the voltage to 3.5V and ran test #5 again, this time I got about 30,000 errors. I dropped it down to 3.3V and got about 300 errors. So to summarize:

268 2-2-2-10 @ 3.4V, test #5 for 600+ loops: 0 errors

269 2-2-2-10 @ 3.3V, test #5 for 1 loop: 300 errors
269 2-2-2-10 @ 3.4V, test #5 for 1 loop: 10 errors
269 2-2-2-10 @ 3.5V, test #5 for 1 loop: 40,000 errors

What's going on? Is this a timings issue? All I specify in the BIOS are 2-2-2-10, the rest are Auto, which gives me this: 2, 15, 17, 2, 2, 10, 2, 3, 2, 4, 200 Mhz 3.9us, 8x, 64 bit, Disable, Disable, 4x, 7 ns, 5.0 ns, 16 clks, Enable, 200

Drive strength is auto, which I believe is defaulting to 2, but not sure.

And all this time, my +3.3V rail is at 3.71V

Edit: Oh, forgot I temporarily removed my sig..... this is with a DFI LP UT 250gb, 3400+ Mobile Clawhammer, chipset @ 1.8V and OCZ PowerStream 520W. And cooling the memory is definitely not an issue... got at least 150CFM of 65F air on it.

Heat is causing your errors. Actually if you are at 268 you are darn lucky. You have the fastest set of VX I know of.

VX needs voltage to run, but voltage creates heat. I experienced the same thing 2 night. 3.4v no errors, 3.3 volts many errors and 3.5v it just dumped errors.

If you could cool the DIMMS better they would handle more volts. As it is you are doing great with 268mhz with 3.4v IMHO.

I suppose you have some healthy fans blowing on the memory for cooling. If not add or upgrade them and see what happens.

[xsky]

268 is really great...
I get "only" 256MHz 2-2-2-8 but that's on Intel.

Heat might be a problem but I don't know what I could do more than placing two 80*80 fans at full speed right in front of the vx ^^

ocz should think about a watercooled edition...maybe with integrated TEC's :D

[gundamit]

Right now my Gskill LA at 295mhz 2.5-4-3-7 is still outperforming the the VX. I'll need to get the VX into the 275 range to beat the LA. When you say outperforming do you mean bandwidth tests, 3Dmark tests, SuperPi and others? My OCZ TCCD at 300 (at those same timings) seem about the same as my BH-5 at 250 2-2-2-6.

[chinkgai]

What passes did the errors occur in? From my understanding, if it's only a few errors, they should disappear after burning in long enough. Also, 200 passes of what? Are you not looping the whole series? There's no way you have 200 passes of the whole series ... I got about 50 in 16 hrs. When burning in, you should be doing the full series.

where did you hear that it was the whole series?

Hi, you can run at 200mhz, aith 2-2-2 timings using 3.2v for about 48 hours. 24 hours minimum. Run memtest and loop test 5. Memtest can be turned on in your BIOS.

This procedure seems too help when you crank up the volts and speeds later.

I will be starting a thread tomorrow with some suggested VX timings here on XS. Check it out then.

[Torin]

Heat is causing your errors. Actually if you are at 268 you are darn lucky. You have the fastest set of VX I know of.

VX needs voltage to run, but voltage creates heat. I experienced the same thing 2 night. 3.4v no errors, 3.3 volts many errors and 3.5v it just dumped errors.

If you could cool the DIMMS better they would handle more volts. As it is you are doing great with 268mhz with 3.4v IMHO.

I suppose you have some healthy fans blowing on the memory for cooling. If not add or upgrade them and see what happens.
If heat is the problem, I dont know what I can do to fix it. I've got 2 strong 60mm fans practically sitting on the ram, blowing about 50 CFM, then behind that I've got a house fan that easily blows 150-200CFM, and I've got the temp in the house down to 65F. If that doesn't cool them down, nothing short of watercooling them will. :)

But 268 is fine with me, I really only need 260 before my CPU tops out at 11x 1:1. Couldn't even use the extra speed if I wanted to. :p:

[formyfaith]

Guys make sure you tighten your specific timings.


If you're upto 268 with somthing like a 2-2-2-2 in the specific timings area
then it means that it's not that good.


Tightening the specific timings to something like
0-2-1-2 with 262 HTT is a lot better than 2-2-2-2 with HTT 268.


Actually it's kind of like you can't say that you've reached 268
with those loosened specific timings.


I'm running my PC3200 VX at tightened timings on HTT 262 as shown below.


200-1T-2-2-3-2-7-15-00-02-01-02-3120-1-Enabled
Increase Skew-43-1-4-7-5-16 cycles-enabled-16X-7X-disabled 8 burst.


I too can push up the HTT to 268 when I sacrifice the specific timings
as shown below.


200-1T-2-2-3-2-7-15-02-02-02-02-3120-1-Enabled.
Auto-00-Auto-4-7-Auto-256 cycles-Disabled-16X-7X-Disabled 8 Burst.


But the bandwidth and the performance is a lot lower
than 262 with tightened timings.


Me and a few of my friends have the very first batches of PC3200 VXs,
and all of them are very good.


With tightened specific timings, all of them can go upto 262 to 267.


But my experiences with the new PC4000 VXs weren't so nice.
They(a total of 4 set) do go upto HTT 260-265 with loosened timings,


but just can't go up more than HTT 255 with tightened timings.


That's why I called the guy who bought my PC3200 VX and
persuaded him to sell them back to me. :rolleyes:


Make sure that you test your rams with tightend timings,
unless you can't really say that you have reached what HTT :)


(If you can reach HTT 268 with tightened timings then you are
really lucky. :D )

[mdzcpa]

When you say outperforming do you mean bandwidth tests, 3Dmark tests, SuperPi and others? My OCZ TCCD at 300 (at those same timings) seem about the same as my BH-5 at 250 2-2-2-6.

Sandra benches.

[gundamit]

Sandra benches. Yeah. Everest read is about 1K difference but it doesn't seem to translate to much difference (if any) on SuperPi or 3DMarks. I'm still hopeful I'll get my hands on some low latency (BH-5, CH-5, UTT whatever) memory that'll help me out on 3DMark. For every day gaming though, TCCD all the way. :)

[Torin]

(snip long post about tight timings)
For all my testing, I am using these timings:
2.0-7-15-2-0-8-2-3-1-2-200 3.9-8x-64bit-disable-disable-4x-7ns-5.0ns-16clks-disable-200

[arctic-k20]

my new set of vx 3200 is able to do 260 @ 3.2v using these timings:

1.5-2-5-2-7-15-1-2-1-2-auto-1-en-icr-0-6-4-auto-auto-auto-8-7-en

currently burning 3.4@265 with the same timings

[Tempest25]

where did you hear that it was the whole series?

I saw that post as well, but there was another where someone talked to OCZ on the phone and they were told to use the whole test. As proof this is needed, I looped again last night and for some reason, even though I had 16hrs of straight looping the night prior with no errors, I had 10 errors total in about 12 hrs of testing and none of them were in test #5. They were in tests 4, 6, & 8.

These tests were again ran at 3.2V @ 247. I'm just not having good luck with this memory.

[chinkgai]

i only use 5 and 8

yeah i've seen 4 get me errors before...but 8 almost always gets errors for sure if 5 doesnt

only downside is 4 and 8 take much longer than 5

i never loop the entire series, waste of time to me

[andyOCZ]

I saw that post as well, but there was another where someone talked to OCZ on the phone and they were told to use the whole test. As proof this is needed, I looped again last night and for some reason, even though I had 16hrs of straight looping the night prior with no errors, I had 10 errors total in about 12 hrs of testing and none of them were in test #5. They were in tests 4, 6, & 8.

These tests were again ran at 3.2V @ 247. I'm just not having good luck with this memory.

What revision is your board. Look on the bottom of the last PCI slot at the sticker. You'll see R.AA0 or something like that.

[B5I8]

Thanks to this thread I now have the correct timings set in my BIOS. Now all I have to do is a 24 hour burn-in before I start cranking it up.

[Tempest25]

I'll check this when I get home this evening and post the revision.

As an update, I ran the tests again for a while (not 16 hrs) with no errors and ran Prime overnight with SuperPI running along side it for about 32 mins (to complete the 32M test) and had no problems there either.

I haven't received a response from Eric on the timings yet either, though it was the weekend. :)

I found another thread (I don't have the link with me at work) on another set of forums where an OCZ rep (sorry, I forget the name) posted some timings to try that was updated 3/19. Those timings were looser than what I was trying and they still didn't help me obtain 250 error free.

What revision is your board. Look on the bottom of the last PCI slot at the sticker. You'll see R.AA0 or something like that.

[Tempest25]

I've made some progress. It's currently running at 3.3V @ 252 with no problems in various tests and an hour of memtest looping #5. Apparently, the high speed tornado fan wasn't enough. I modified an old card cooler bracket to work with my case, so I have 2 80MM fans blowing over the ran now and that seems to have done the trick.

It won't run much higher without errors though, so I'll do some burning in at this point and report my results at the end. I'm at the rated speed (with very tight timings) though at least, so OCZ is in the clear. :toast:

Thanks for the help!