Sign in with
Sign up | Sign in
Your question

Instability using different multiplier

Last response: in Memory
Share
July 7, 2005 5:12:31 AM

When testing my single DDR Ram Corsair 3200 ValueRam I came into this issue:
Ram will overclock fine running sp2004 ram test to 230mhz. But when the multiplier is either on 9.5 or cpu is underclocked the ram crashes the system. Currently FSB=241 CPU mult is 10.5 = 2532mhz = no ram crashes. Is this a memory controller issue?
Also I'm assuming since I can get 214 to 215 overclock in DDR with 2 sticks then this must be equal to 230 in single I'm getting now?

Lastly, all 3 sticks I have show the same specs in cpuz/everest as 2.5/3/3/8. But 1 stick is a model vs512mb400 and the other 2 are vs512mb400c3. The 400 is advertised as 3.0 latency for some reason. If I mix the two the system is unstable despite the same software readings. What's with this? I assumed the advertised 3.0 was an error and the all three sticks were the same.

Amd 3500 (Clawhammer)
Abit AV8
6800gt<P ID="edit"><FONT SIZE=-1><EM>Edited by scwam on 07/07/05 01:16 AM.</EM></FONT></P>

More about : instability multiplier

July 7, 2005 6:25:39 PM

Possibly, when you are underclocking the RAM your system is running the memory at a different latency. Sometimes the frequency the module is running determines the latency.

Quote:
Also I'm assuming since I can get 214 to 215 overclock in DDR with 2 sticks then this must be equal to 230 in single I'm getting now?

The assumption that two sticks with a 215 overclock is equal to one stick with a 230 overclock is false. Two sticks running at 215 are equal in speed to one stick running at 215. Also, equality cannot be determined by speed alone.

Quote:
Lastly, all 3 sticks I have show the same specs in cpuz/everest as 2.5/3/3/8. But 1 stick is a model vs512mb400 and the other 2 are vs512mb400c3. The 400 is advertised as 3.0 latency for some reason. If I mix the two the system is unstable despite the same software readings. What's with this?

CPU-z will only read the latency of one module. Any modules in tandem will all run at the same speed and latency. Also, you can manually set the latency in the BIOS, completely ignoring the values in the SPD.

Furthermore, a module with a CL3 specification being run at CL2.5 can cause stability issues for your system. This may be what is happening by combining one CL2.5 module with one CL3 module and running both at CL2.5.

<b>correction</b>
CPU-Z will read the SPD of more than one module. However, the system will run all memory modules at the same speed and latency.

<font color=green>******
"Believe nothing, no matter where you read it, or who said it, unless it agrees with your own reason and common sence." -The Buddha
AIM BrentUnitedMem</font color=green><P ID="edit"><FONT SIZE=-1><EM>Edited by BrentUnitedMem on 07/07/05 01:41 PM.</EM></FONT></P>
July 8, 2005 6:50:17 AM

Thanks so much for your time in your detailed post. I realize that I need to go through each one of these sticks individually and check the latency, and come to think of it, the Default timing (1T vs 2T). I will then note any descrepencies.
Also, regarding the crashing with mixing the two models types was wrong for me to say. I am now using the same identical sticks in ddr and it still crashes. It is only in Slots 3 and 4 that this occurs, never slots 1,2. Apparently it is an issue with those slots on the motherboard itself or it must be a memory controller problem from the cpu on those slots. Although I'm still testing stability now. It seems so far that the stability is better when I lower my Dram in Bios to 266. Both 333 & 400 setting creates much instability (overclocked, default or underclockedw, slots 3,4 only). I'm hoping that this is not the same problem the Winchester cores had with their memory controllers using all 4 slots. I have a Clawhammer core 3500. Does anyone know if there is a memory controller issue with this model? Oddly enough I played battlefield 2 last night using the 2 identical sticks in slots 3,4 underclocked to 158 and the game seems to play noticably smoother than when it ran at stock in slots 1 and 2. Why? I got another 512mb stick comming to fill all four slots to play this game but I hope I'm not forced to take a speed reduction or accept instability issues using all 4 slots.

<P ID="edit"><FONT SIZE=-1><EM>Edited by scwam on 07/08/05 03:24 AM.</EM></FONT></P>
Related resources
July 8, 2005 7:20:48 AM

Well, I installed each of the 3 sticks on their own, before doing so I reset the cmos each time. The readings of all 3 sticks are exactly the same 2.5/11/14/3/2/8/3/3/2/4 with Command rate set to "auto" and burst rate at 4. Everything was identical despite one model# not having "C3" on the end of it. Newegg also advertises that this should have a 3.0 latency with customer reviews showing its reflecting 2.5 in their apps too. So I must exclude that there are no differences in the ram and that it must be the a MB problem with Dimms 3&4 or the CPU memory controller on those two Dimms. I'll find out what my results will be on Monday when I'm able to install all 4 sticks.

<P ID="edit"><FONT SIZE=-1><EM>Edited by scwam on 07/08/05 03:29 AM.</EM></FONT></P>
July 8, 2005 7:28:47 PM

Quote:
It is only in Slots 3 and 4 that this occurs, never slots 1,2

Always populate the first two slots first.

Quote:
It seems so far that the stability is better when I lower my Dram in Bios to 266

Memory run at lower speeds tend to be more stable. Many registered modules are sold at DDR266 because of the increase in stability.

DDR 400 with a specification of CL3, may run at 333 or 266 if you change the latency to CL2.5. This is normal.

Quote:
Oddly enough I played battlefield 2 last night using the 2 identical sticks in slots 3,4 underclocked to 158 and the game seems to play noticably smoother than when it ran at stock in slots 1 and 2. Why?

Maybe because you underclocked the modules? I'm not sure how to further comment on this.

Quote:
The readings of all 3 sticks are exactly the same 2.5/11/14/3/2/8/3/3/2/4

Not sure what this means. Are you reading the modules one stick at a time, or all together? Any case, it still seems like you are running all the modules at CL2.5. If one of the modules is rated for CL3, this may cause problems for your system.

If you continue to have problems, you will need to relax the timings. Try running all the modules at CL3.

I don't think VALUE RAM is good for gaming. No doubt, you will put additional strain on your system running 4 modules of VALUE RAM together. It is never a good idea to use 4 modules on a desktop board.

<font color=blue>******
<font color=green>"Believe nothing, no matter where you read it, or who said it, unless it agrees with your own reason and common sence." -The Buddha
<font color=blue>AIM BrentUnitedMem
July 9, 2005 6:50:03 AM

OK, all these sticks are identical, despite the difference in model numbers they are the same stick same specs. 2.5/3/3/8. Bios, cpuz, everetst all show the same. I ran each stick by itself withough the other two and all the specs were the same. Later this afternoon, I got the fourth stick from chiefvalue.com. Last night I could only run the idential model numbers in slots 3,4 at 266 and 170max clock rate without crashing. Now, get this. ALL 4 sticks runn at 212, overclocked cpu 2530.
Yes, the 4 stickd defaulted at 333 on my amd 3500 Clawhammer. However, I was still permitted to overclock the ram to 212 with no crashing. It would not do this with the 2 stick in slots 3,4. Apparently for those that have a 3500 Clawhammer it will work when you finnaly install all 4 sticks. I was concerned after placing my order that if 2 sticks didn't work in 3,4 then why would all four sticks work. In addition I was told performace would be reduced. Now Command Rate "Auto" now shows as 2T. Who cares.

Naturally I though twice about accepting the package. I'm glad I did. The performance is superb compared to 1g. Battlefield 2 has absolutly no lag anymore. Its nearly a nearly flawless playing experince at 2540/212 6800GT 410/1.11. They say you reduce performace, but honestly, I feel windows boot has shaved off 4-6 seconds. Everest did report a memory write speed reduction of 1800 to 1450. But the read speed increase from 5100 to 5800. Latency went from ~60-70 to 59-62. So you figure. I notice no difference in application speed. Seems to be fast, but that may be subjective. I'm happy with my decision to insteal the 4th stick. Everything runs stable, downclock, overclocks, etc.


Addition: I just ran the sytem overclocked with a different CPU multiplier and it seems it only boots with certain intergers.<P ID="edit"><FONT SIZE=-1><EM>Edited by scwam on 07/09/05 04:19 PM.</EM></FONT></P>
!