• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Help a newb with memory overclock (Crucial Ballistix PC3200 on an Asus A8V Deluxe).

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

KillrBuckeye

Senior Member
Joined
Jan 9, 2005
Location
Livonia, MI
Help a newb with memory overclock (Crucial Ballistix PC3200 on an Asus A8V Deluxe).

I have never messed with memory timings before, and this is the first time that I have purchased some nice memory that is "supposed" to be good for overclocking. I am running 2x512MB of Crucial Ballistix PC3200 on an Asus A8V Deluxe motherboard. This memory is rated at timings of 2-2-2-6 with a stock voltage of 2.8V. Everything works great at stock settings, as I'd hope, but I'm a bit discouraged by my initial testing at overclocked speeds.

Specifically, I am getting some memtest errors while running 220 MHz with timings of 2.5-3-3-8. Voltage is 2.8V, which is the max that the Asus board can supply. I have read through some of the other threads about these Ballistix modules, and my results seem far below what others have been able to achieve, even at the same voltage.

I have a question about these memtest errors: does ANY error in memtest, regardless of how often it occurs, indicate system instability? The reason I ask is that the errors I am getting at the settings mentioned above are VERY infrequent. I had 3 failures, all during Test 7, after making 34 passes (memtest was running for about 10 hours). Does this mean that I would not want to run my system at these settings, or does the infrequency of the errors indicate that it would probably run fine?

What suggestions would you give me other than to try the other dual channel slots in the motherboard? I have left all memory timing settings on AUTO except for the ones I listed above. Are there other settings that might be giving me problems? Thanks in advance for your help.
 
Right now I am getting a 12% overclock on this ram. You are getting 10%. I am far from being any kind of ram expert so I really couldn't tell you any variables with the timings. I can run 1:1 with my cpu and can't really test to see how much further they will go because my cpu's HTT won't go any higher. I really don't believe that the ram you have can run 1:1 with your cpu at like HTT of 230/240. That would be a 15%/20% increase.

Hey, just noticed you are from Detroit. I'm from Adrian. It is a small town like 30 min. SW of Detroit:).
 
avalanche83 said:
Right now I am getting a 12% overclock on this ram. You are getting 10%. I am far from being any kind of ram expert so I really couldn't tell you any variables with the timings. I can run 1:1 with my cpu and can't really test to see how much further they will go because my cpu's HTT won't go any higher. I really don't believe that the ram you have can run 1:1 with your cpu at like HTT of 230/240. That would be a 15%/20% increase.

Hey, just noticed you are from Detroit. I'm from Adrian. It is a small town like 30 min. SW of Detroit:).
Hmm, I was expecting to be able to run 1:1 with an HTT of 230/240 with loose timings on this memory. Searching through other threads, it seems that other people have had success doing this. However, I think my motherboard may be the limiting factor. Even people who have tried otherwise fast TCCD memory on this board didn't seem to have much success. I was hoping that these Micron chips would be a different story, but so far it doesn't appear that way.

I'm surprised you didn't give any crap about my user name. You must not be a scUM fan. :) I'm from Columbus, Ohio, originally.
 
If I'm not mistaken, Micron chips are generally to be avoided.

3 errors over 10 passes could be a mere satistical feature. See if you can boot into Windows and run SuperPi 32M followed by loops of 3DMark01. If you encounter a BSOD, make sure its not because of the CPU (I doubt it could be the CPU with your Venice :) ).
 
Try putting the NB to 1.65v and the HTv to 1.35v.

I will have to put my Corsair VS back in and see if I can get the same kind of % increase.

lol. No, not a scUM fan. I am from Florida originally. Gators fan:).
 
Super Nade said:
If I'm not mistaken, Micron chips are generally to be avoided.

3 errors over 10 passes could be a mere satistical feature. See if you can boot into Windows and run SuperPi 32M followed by loops of 3DMark01. If you encounter a BSOD, make sure its not because of the CPU (I doubt it could be the CPU with your Venice :) ).
Really? Every review of the Ballistix modules that I read was very positive. In almost all of the tests, it beat out comparable TCCD-based modules. So 3 memtest errors in 10 hours might be stable enough? Good, I will try your suggestion of running SuperPi 32M and 3DMark01. That is, after I finish setting up my installation for the 2nd time in 4 days!!!

avalanceh83 said:
Try putting the NB to 1.65v and the HTv to 1.35v.

I will have to put my Corsair VS back in and see if I can get the same kind of % increase.

lol. No, not a scUM fan. I am from Florida originally. Gators fan.
Okay, I'll try that once I resume my testing. Florida, eh? I guess that explains the orange and blue lettering in your sig. :p
 
Ok. I did the Corsair VS test and the speed for it was better than your Ballistix for some reason.

HTT229 was at 2.5-3-3-8 (229MHz) with 12.75% gain in MHz
HTT232 was at 2.5 4-4-10 (232MHz) with 13.8% gain in MHz
 

Attachments

  • HTT229.JPG
    HTT229.JPG
    64.7 KB · Views: 48
  • HTT323.JPG
    HTT323.JPG
    67.3 KB · Views: 60
Last edited:
avalanche83 said:
Ok. I did the Corsair VS test and the speed for it was better than your Ballistix for some reason.

HTT229 was at 2.5-3-3-8 (229MHz) with 12.75% gain in MHz
HTT232 was at 2.5 4-4-10 (232MHz) with 13.8% gain in MHz
LOL, $80 RAM >> $180 RAM??? Are you using memtest to check for stability? What voltage are you running?

Does the voltage supplied to the CPU affect the memory overclock??? The reason I ask is that I have only nudged up the Vcore a tad bit. Is it possible that this is causing the instability?
 
No. I haven't done any memtest on it but it is windows stable at HTT229. It gets a little flakey at HTT232. I run it on 2.8v. I don't really know about the Vcore part.

The Ballistix that I currently have will do 278MHz windows stable at 2.8v 2.5-4-4-10 but it likes 3v doing 278MHz 2.5-3-3-8. I haven't really tried everything out with this ram yet so don't know too much about what it can really do. I think my cpu is holding me back. It is so damn unpredictable at HTT280.
 
That seems low. I have had my ballistix (2x512 3200) running at 250Mhz with timings of 2.5-2-3-10 with 2.8v.

Although, I cant get my memory to run at 255Mhz even with 3-4-4-10 with 2.85v.

Try putting your multiplier on the cpu down just to rule it out.

Let me know if you want to to run any tests, Ill be happy to run them.
 
KillrBuckeye said:
Really? Every review of the Ballistix modules that I read was very positive. In almost all of the tests, it beat out comparable TCCD-based modules.

:)

TCCD PC3200 @ DDR600 here. Could you point me to some links discussing this? I may have something new to learn. Sounds interesting though, after all the complaining I've heard about Micron, this is a bit surprising.
 
avalanche83 said:
I am from Florida originally. Gators fan:).

G8TRS_R_NUM_1

Well, at least in my way of thinking...Go Gators! I'm also a big U.M. fan...not as in University of Miami, but as in "Go Get'm Urbar (Meyer).


I just picked up some Ballistix to play with. My impression was also that Micron chips were decent with an A64 rig, although all in all, I would have to say that Samsung TCCD/5 are probably the best of what is currently available.

I have personally jumped off the UTT bandwagon (edit: although as many here might already now...I'm still a big fan of original bh-5/bh-6).

Continuing with my edit:

I have always understood Micron based stuff, like the EB series from OCZ example, to really be good in the 245-255 or so range, giving you tighter timings than most other comporable products, but not being a high HTT part like the TCCD. It was very well suited for 3.0GHz - 3.4GHz P4s which operated in this range.

Keep in mind that when TCCD first came out it was not tweaked for S939 A64 rigs and usually not found using BrainPower PCB, except on the OCZ PC3200 which was usually neck-to-neck with the Ballistix on A64 rigs (many of which were s754 at that time, too). Since the introduction of the NF4 and particularly the DFI NF4 boards, nothing touches TCCD when it comes to high HTT. Everyone tweaks the Samsung based stuff to run of a64/NF4 rigs these days.
 
Last edited:
Super Nade said:
:)

TCCD PC3200 @ DDR600 here. Could you point me to some links discussing this? I may have something new to learn. Sounds interesting though, after all the complaining I've heard about Micron, this is a bit surprising.
Sorry, I have to revise my previous statement. I went back and reviewed the articles that offered glowing reviews of the Ballistix. It turns out that they conclude exactly what Reefa_Madness described: the memory does great and tends to be on par with or beat out comparable TCCD memory in the ~250 MHz range. However, the articles do not compare the memory with TCCD-based modules at much higher speeds. So, I guess if you want really high HTT, TCCD is the way to go. However, as you're all too familiar with, TCCD doesn't do well in all board/chipsets, so it may not always be the best choice. That's why I decided to give the Ballistix a try with the A8V. I would be really happy if I could get an HTT of 240 or above with this memory.

I think the CPU may have been the cause for some of instability I was experiencing. Before I left this morning, I set my memory to 230 MHz with the same timings (2.5-3-3-8), and reduced the CPU multiplier to 8. There were no errors during the first couple of minutes, but we'll see the results when I get home this evening. I'm hoping for the best.
 
Super Nade said:
:)

TCCD PC3200 @ DDR600 here. Could you point me to some links discussing this? I may have something new to learn. Sounds interesting though, after all the complaining I've heard about Micron, this is a bit surprising.

my friends pc3200 ballistix does 230 2-2-2 at 2.9v, 260 2.5-2-2 with 2.8v and 290 3-3-3 with 2.9v. its basically a freaking unbeatable ram, because of its versatility. and EB used -5b C while ballistix of today uses -5b G.
 
Just so there is no confusion...I referenced the EB not because is uses the same exact chip as the Ballistix, but because it has the same general characteristics, allthough theMicron "G" die is clearly superior to the "C" (tighter timings and higher clocks).

I've been playing around with a 2x256 set of PC3700 EB and for the life of me, can not get it clean at 250, no matter what timings, CPC or voltage I use. This is on an MSI NF3/winney 3200 combo. I'm going to pull them out and throw in some Ballistix and see what happens. I'll throw the EB into an Intel 865PE and see what they do there.

I am of the opinion that the EB will work better on an Intel rig, but possibly the Ballistix will do better on the A64. I'll post back in a day or two and share how it went.



KillrBuckeye,

The good thing about the A64 is that with the dividers at the upper end very little performance is lost when they are used so even if the ram is not doing 250, the cpu can be. With respect to your question about vcore. If the cpu is not stable, you are going to get errors all over the place. Are you doing full loops of memtest or just some of the selected tests (like #5). Run full loops and you'll get an indication of whether your cpu is stable or not.
 
Well it turns it that it WAS my CPU that was causing the errors in memtest :). I decreased the multiplier to eliminate the possibility of CPU error, and it was perfectly stable at 230 Mhz for 10 hours of memtest. I'm trying 240 MHz right now.
 
Sorry for the double-post, but I wanted to make you aware of my progress and ask for your suggestions. The memory was perfectly stable at 240 MHz (2.5-3-3-8), but had some errors with those same timings at 250 MHz (13 errors over ~8 hours of memtest). I'm testing 245 MHz right now.

Some questions I have are as follows:
- Which timings would you suggest that I loosen first to try to push my HTT?

- Is it possible that the motherboard doesn't like HTT above 250? To test this, would you suggest employing a memory divider to run memory slower than the HTT?

- Generally speaking, do people tend to get better results with tighter timings at lower memory speeds (this would require a memory divider to push the CPU to its limit), or loose timings with as high of an HTT as possible (to strive for max overclock with memory at 1:1)? I know it's difficult to answer, but in your experiences what has worked the best?
 
KillrBuckeye said:
I have never messed with memory timings before, and this is the first time that I have purchased some nice memory that is "supposed" to be good for overclocking. I am running 2x512MB of Crucial Ballistix PC3200 on an Asus A8V Deluxe motherboard. This memory is rated at timings of 2-2-2-6 with a stock voltage of 2.8V. Everything works great at stock settings, as I'd hope, but I'm a bit discouraged by my initial testing at overclocked speeds.

Specifically, I am getting some memtest errors while running 220 MHz with timings of 2.5-3-3-8. Voltage is 2.8V, which is the max that the Asus board can supply. I have read through some of the other threads about these Ballistix modules, and my results seem far below what others have been able to achieve, even at the same voltage.

I have a question about these memtest errors: does ANY error in memtest, regardless of how often it occurs, indicate system instability? The reason I ask is that the errors I am getting at the settings mentioned above are VERY infrequent. I had 3 failures, all during Test 7, after making 34 passes (memtest was running for about 10 hours). Does this mean that I would not want to run my system at these settings, or does the infrequency of the errors indicate that it would probably run fine?

What suggestions would you give me other than to try the other dual channel slots in the motherboard? I have left all memory timing settings on AUTO except for the ones I listed above. Are there other settings that might be giving me problems? Thanks in advance for your help.

Any error is bad!!! Don't bother trying to use Windows if Memtest86 fails, or else, Windows probably is going to end up getting corrupted!!!

If it took at least approximately 10 hours for an error to appear, then the RAM likely is overheating!!! I thus recommend a fan setup change.

If it was caused by the processor, then the processor OC is definitely unstable!!! Memtest86 don't tax the processor as much as Prime95, thus if Memtest86 found a problem, then it's unstable with the stability problem being caught early.
 
Last edited:
RJARRRPCGP said:
Any error is bad!!! Don't bother trying to use Windows if Memtest86 fails, or else, Windows probably is going to end up getting corrupted!!!

If it took at least approximately 10 hours for an error to appear, then the RAM likely is overheating!!! I thus recommend a fan setup change.

If it was caused by the processor, then the processor OC is definitely unstable!!! Memtest86 don't tax the processor as much as Prime95, thus if Memtest86 found a problem, then it's unstable with the stability problem being caught early.
It's not the CPU, as I can run 255x10 @ 1.50VCore with a 5:6 memory divider, and it's stable. The memory runs at 212 MHz at this speed. However, I can't seem to get the memory running at ~217 MHz, even by loosening the timings (I tried running 1:1 with the HTT, so it definitely wasn't the CPU). I still haven't found time to try the other two DIMM slots to see if it helps. That may be my next step. I think I made a mistake by getting memory that runs 2.8V by default, as it gives me no room for overvolting with this board. I didn't pay close enough attention to the specs when I purchased.
 
Back