GEIL Ultra - wich chips ?

[Vassili Zaitsev]

http://www.tigerdirect.com/applications/Se...&Tab=2&NoMapp=0

 

Hi there, just bought that GEIL at tigerdirect.

 

2-3-3-6 at SPD (everything at auto).

Ran 2 Hours and 8 minuts of memtest, "5 Pass"

 

 

Wich chips have those babys??? recommendations?

 

another link:

http://www.geil.com.tw/portal/product_400_t_dc.php

 

 

I dont wanna use extreme voltages or something like that, im thinking in a Async OC, dont wanna use more than arround 2.8-2.85vdimm

 

 

 

Thanks!

[cjoe7]

http://www.tigerdirect.com/applications/Se...&Tab=2&NoMapp=0

 

Hi there, just bought that GEIL at tigerdirect.

 

2-3-3-6 at SPD (everything at auto).

Ran 2 Hours and 8 minuts of memtest, "5 Pass"

 

Wich chips have those babys??? recommendations?

 

I dont wanna use extreme voltages or something like that, im thinking in a Async OC, dont wanna use more than arround 2.8-2.85vdimm

 

 

Thanks!

 

I have a set of those. Those use UTT or new BH5 chips. In general they perform very well at default voltages.

 

Why don't you want to use higher Vdimm and achieve 1:1 CPU/Memory performance? You purchased a DFI mobo? Its main purpose is to OC and do extreme things.

 

If you're willing to pump in some more volts and use active cooling, you can probably do 2/2/2/5 1T 1:1 timings @ 245mhz+ with about 3.4-3.6V. My set has achieved 2/2/2/5 @ 250Mhz 1T 1:1 with 3.4V. Just slowly burn them in with memtest and use active cooling.

 

good luck

 

cjoe

[Vassili Zaitsev]

Thanks for your help

 

But i dont wanna get extreme OCs, cause with them comes extreme risks, and aint got too much money to replace computer parts.

 

 

Well, 2-3-3-6, memtest for 8 Hours 37 Minuts, at DDR400 SPD, they seems to be fine, and working without trouble, i'll see later their max OC 1:1 using 2.85v as much.

 

 

Cya!

[cjoe7]

Thanks for your help

 

But i dont wanna get extreme OCs, cause with them comes extreme risks, and aint got too much money to replace computer parts.

 

 

Well, 2-3-3-6, memtest for 8 Hours 37 Minuts, at DDR400 SPD, they seems to be fine, and working without trouble, i'll see later their max OC 1:1 using 2.85v as much.

 

 

Cya!

 

No problem. Anytime.

 

You have some good parts. Your PSU is very nice. With good parts, your risks are mitigated to some measure. Do you have a decent UPS? If you do, just place a 120mm fan over the memory and you're good to go up to 3.3V without using the risky 4V jumper.

 

I haven't encountered any issues using the 4V select jumper. I used it up to 3.7V for short durations and usually keep it around 3.33V to maintain 250HTT 1:1 1T 2/2/2/5 speeds (even though my memory will do 259HTT). Many of the issues people have encountered are due to using crappy PSUs without good line conditioning.

 

Unfortunately, most UTT memories need 3.5V initially to burn them in before they work at lower voltages. I burned in a set of Mushkins 2x512 UTT for around 2weeks. After a month of usage, they could run 250Mhz 1:1 1T 2/2/2/5 @ 3.4V or 200Mhz 1:1 1T 2/2/2/5 @ 2.7V.

 

cjoe

[Vassili Zaitsev]

Thanks.

 

 

Well, I want my memory to lives at least 3-4 years...i just have lots of computer that would apprecciate this RAM before it dies for electronmigration...

 

 

But im making a minireview, i found a very rare thing, even with the RAM running at memtest86 for 8 hours, prime for two hours, UT 99 with full bots and me spectating arround 1-2 hours, and 0 problems at DDR400 2-3-3-6-12 2.66v, it was enought to freeze, hang or random reboot run a PC Mark 2005 Memory Test suit. 3 min 20 seconds, usually hangs or reboots at the write memory test, with 16 or 8 MB.

 

And even i get this error:

http://img469.imageshack.us/img469/3838/error0fl.jpg

 

 

Did somebody with this same RAM or a similar experienced this?

I have increased the voltage to a measured 2.88v and it still giving me trouble in this test, it started standing more time testing at DDR400 with 2.77v, at 2-3-3-6-12 and 3-3-3-6-12 2.77v. No crashes with both test (im gonna try tomorrow run those test at 2.77v 2-3-3-6-12 DDR400, everything at SPD and lets see if it passes at least 1 hour under PC Mark 2005 memory suit test), at 2.66v it seems that DDR400 at SPD wont run troubleless. I increased the vid control even to 1.46v at stock speed (1.4v +104% special VID Control) to see what happens with 166 MHz 2-3-3-6-12 and it still crashing, but runs more time the test.

 

I dont know what can be, i havent try with higher LDT than 1.3v nor higher chipset voltage.

Havent change nothing at DRAM setup, everything seems to be only at auto and optimal, any advise using this RAM?

 

 

*Im trying to do a Integrated Memory Controller in A64 and Performance Review, im gonna give you the link after i get this freaky thing at least stable enought to run all test without trouble.

 

 

Thanks!

[cjoe7]

Thanks.

 

Well, I want my memory to lives at least 3-4 years...i just have lots of computer that would apprecciate this RAM before it dies for electronmigration...

 

But im making a minireview, i found a very rare thing, even with the RAM running at memtest86 for 8 hours, prime for two hours, UT 99 with full bots and me spectating arround 1-2 hours, and 0 problems at DDR400 2-3-3-6-12 2.66v, it was enought to freeze, hang or random reboot run a PC Mark 2005 Memory Test suit. 3 min 20 seconds, usually hangs or reboots at the write memory test, with 16 or 8 MB.

 

...

 

Did somebody with this same RAM or a similar experienced this?

I have increased the voltage to a measured 2.88v and it still giving me trouble in this test, it started standing more time testing at DDR400 with 2.77v, at 2-3-3-6-12 and 3-3-3-6-12 2.77v. No crashes with both test (im gonna try tomorrow run those test at 2.77v 2-3-3-6-12 DDR400, everything at SPD and lets see if it passes at least 1 hour under PC Mark 2005 memory suit test), at 2.66v it seems that DDR400 at SPD wont run troubleless. I increased the vid control even to 1.46v at stock speed (1.4v +104% special VID Control) to see what happens with 166 MHz 2-3-3-6-12 and it still crashing, but runs more time the test.

 

I dont know what can be, i havent try with higher LDT than 1.3v nor higher chipset voltage.

Havent change nothing at DRAM setup, everything seems to be only at auto and optimal, any advise using this RAM?

 

 

*Im trying to do a Integrated Memory Controller in A64 and Performance Review, im gonna give you the link after i get this freaky thing at least stable enought to run all test without trouble.

 

Thanks!

 

Ever wonder why that memory is relatively inexpensive? Its called UTT because its UnTesTed. I wouldn't trust UTT SPD values being stable. I would use it as a guideline to follow. In general, many of those memory sticks need a little extra (0.1-0.2) voltage to achieve benchmarking stability. You need to memtest with tests #5 and #8 exclusively because it really heats up the memory. When you run all the tests, some tests are not as stressful - letting the memory cool down.

 

So when you run PCMark, it uncovers errors or instability in your OC. Also, memtest stability does not equate to Windows stability. You still need to verify your OC. Try OCCT, Prime95, SP2004, S&M, CPU Burn, ...etc to verify system stability. You'll be very surprised when you run OCCT. Many rigs that seem to run perfectly fine for gaming and Prime95 can still fail this test.

 

Don't worry about dividers and extra settings for now because you're only running standard DDR400. My friend needed to increase his vdimm to 3.0V before he achieved full stability. Search the NF4 thread for BH5 / UTT memory settings. If you can't find any, I'll provide you with mine.

 

cjoe

 

BTW: I wouldn't worry about electromigration. If you stress your memory for a short amount of time and then back down, it should be okay. I'm not telling you to run your memory at 100% of their max potential. Just find the max and then run 85% of that max at moderate volts. The secret is keeping the memory cool. If you allow the memory to cook, your memory WILL die an early death.

[Vassili Zaitsev]

LOL untested dont seems like a nice name, that's maybe why tell call them UTT

 

http://www.dfi-street.com/forum/showthread.php?t=24719

Readed that.

 

Im planing to get some 2 GB RAM Kit in future, I think I'll take those babys, make a review and sell them to a extremer OCer in some months. Maybe somebody gets interested on buying a reviewed RAM that overclocks nice @_@ not my kind of RAM and volts.

 

 

I'll try to run just #5 and #8 test at memtest, then some pc mark memory test.

 

Havent heard about OCCT Neighter SP2004

 

http://sp2004.fre3.com/download.htm Is that the download?

 

 

But i wonder something, do really i need to test in all those test?

If a rig can run lots of games for hours totally stable, win, prime95 and memtest86 full, do will the rig really crash or be unstable under gaming or something, if the OCCT or SP2004 crashes?

 

How many time should i run those test to say: this thing is fully stable! :tooth: ??

[cjoe7]

LOL untested dont seems like a nice name, that's maybe why tell call them UTT

 

http://www.dfi-street.com/forum/showthread.php?t=24719

Readed that.

 

Im planing to get some 2 GB RAM Kit in future, I think I'll take those babys, make a review and sell them to a extremer OCer in some months. Maybe somebody gets interested on buying a reviewed RAM that overclocks nice @_@ not my kind of RAM and volts.

 

I'll try to run just #5 and #8 test at memtest, then some pc mark memory test.

 

Havent heard about OCCT Neighter SP2004

 

http://sp2004.fre3.com/download.htm Is that the download?

 

But i wonder something, do really i need to test in all those test?

If a rig can run lots of games for hours totally stable, win, prime95 and memtest86 full, do will the rig really crash or be unstable under gaming or something, if the OCCT or SP2004 crashes?

 

How many time should i run those test to say: this thing is fully stable! :tooth: ??

 

I wasn't kidding - UTT actually means untested.

 

Sharp did a great job summarizing all the memory types and their characteristics.

 

Stability is something that you and only you can decide for your application.

 

OCCT, Prime95, and Sp2004 are just good stability testing programs. If its OCCT and Prime95 stable, then you can be pretty confident that it won't crash in PCMark or any other application. I like OCCT because its only a 30min test and it will usually find OC stability issues within 15min. Prime95 and SP2004 may take as long as 7hrs before any issues are found. Typically, it requires 10 errorless hours of Prime and SP2004 before its considered stable.

 

OCCT and other Utilities

 

You don't need to be this anal about stability but when you encounter unpredictability in your apps such as PCMark, you need to address the issue of stability. The worst case scenario would be you running an application and it generates one error out of millions of computations. You would never be aware of it until it was too late and it would be almost impossible to find. Basically, these testing programs are doing millions of computations and checking the results. Programs crash because they encountered something unexpected. Will the programs that you use check their results for errors? Probably not.

 

However, do you stress your CPU upto 100% utilization for more than 2hrs continuously? If you don't, I wouldn't worry about it so much. If you play games or do benchmarking, I would seriously consider OCCT and SP2004 stability.

 

When you run memtest #5 and #8, you need to run them for 5hrs+ continuously. You need to insure that the errors generated per pass are less than 10; otherwise, memtest is logging errors instead of burning in your memory. When you get no errors for several hours, bump up the memory speed 5-10mhz. You should run memtest while you sleep and let the memory slowly burn in. There are many memory burn-in strategies on this forum.

 

If you still can't run stable with your default memory settings, you may need to increase your Vcore slightly.

 

good luck

 

cjoe

[Vassili Zaitsev]

I like OCCT because its only a 30min test and it will usually find OC stability issues within 15min. Prime95 and SP2004 may take as long as 7hrs before any issues are found. Typically, it requires 10 errorless hours of Prime and SP2004 before its considered stable

30 min of OCCT is enoguth to see that there is a problem with your OC?

And if you think you reach the "top" speed of the rig, you will use more than 30 min or what?

 

This thread become a chat xD

 

When you run memtest #5 and #8, you need to run them for 5hrs+ continuously. You need to insure that the errors generated per pass are less than 10; otherwise, memtest is logging errors instead of burning in your memory

Less errors than 10 for more than five hours? i dont know if i missunderstood you or something, im not english native speaker =$, but i tought that the memtest test should NOT give errors (if everything's fine).

 

Well, when i'll use memtest86 i should just have to use #5 and #8 to make sure the RAM doesnt overheats and is a stable OC? What are the other test useful for? testing the RAM Integrity? i mean, if it's not damanged or something?

 

Do you have a guide or tips to see the most common cause of memtest error in a specific test? It would be useful.

 

**im gonna try to run the OCCT at 2.77v at DDR400 2-3-3-6-12 CPU at 1.36v chipset 1.5v LDT 1.28v. Lets see what happens =) *im trying to get a rock solid stock configuration.

[grease monky]

HaHaHa Your funny cpu at 1.36 whats the use of having the board

[Vassili Zaitsev]

The use is for a part of the computer. (Sarcasm)

 

I just need to get the computer stable at the stock voltages, dont you think that may sound more funny to start overclocking without even a stock speeds stable rig? that would be very funny to me

 

**passed 30 min test of OCCT, it just did it and go to the main screen, didnt told me nothing about an error or something...i guess it was ok =)

Now im gonna test the RAM with memtest86 #5 and #8 while i see starwars

[cjoe7]

I was sure that you would get your system stock stable pretty quickly. OCCT is a good quick check to test for stability. Once you get your system stock stable, its good to really stress it which further burns it in so it can run the same speed with less volts eventually.

 

You probably only needed to increase your Vcore and Vdimm slightly to get that stability.

 

Like I said earlier, running benchmark stable usually requires slightly more voltage.

 

Memtest is a good memory stresser. Memory can be analogous to fabric because if you stretch it enough times it will conform to the new shape. As you increase your FSB, your memory will give lots of errors. How much do you increase your FSB? I use 10 errors as a baseline. I increase my FSB to a point where it only generates 10 errors / pass. Then I start my burn-in. In the beginning, it will quickly adapt and generate no errors. Later it will be more difficult to get zero error runs.

 

Trust me. By OC'ing your memory and CPU, your machine will be more stable and more efficient at stock speeds by requiring less voltage.

 

cjoe