Jump to content
Corsair Community

TWINX1024-4400C25PT not running at 275FSB stably.


Recommended Posts

Posted

I have the Corsair XMS 2x512MB DDR550 kit, part number TWINX1024-4400C25PT. I am running these sticks on an MSI K8N Neo2 Platinum motherboard with an AMD Athlon64 3000+. I have the CPU and RAM clocked at 275FSB. The RAM timings are set to "auto" in the BIOS, which works out to 1T 2.5 4-4-8 according to the POST screen.

 

When I have both sticks installed in dual channel mode, the highest stable speed seems to be right around 250FSB. I had to set the DIMM voltage to 2.8v to obtain stability at that speed. (I didn't feel comfortable pushing any higher over the 2.75v rating.) When I try to run 275FSB dual channel at 1T 2.5 4-4-8, the system will sometimes hard lock overnight. Other times, it will give Windows "memory could not be read" errors and crash whatever program (usually a game) that I am in at the time.

 

I can run each stick individually at 275FSB 1T 2.5 4-4-8 completely stable. I can also run both sticks in single channel mode stably with the CPU at 275FSB, but at that point, the it defaults to asynchronous DDR333, 228MHz, albeit at 2.5 3-7-3 timings. Forcing the RAM to run at DDR400 1T 2.5 4-4-8 results in a no-POST situation.

 

Please let me know if you have any suggestions, or if you think I may have a bad module. I'd really like to run this RAM at the rated 275FSB speeds. Thank you.

  • Replies 72
  • Created
  • Last Reply
Posted
What CPU do you have?

Click here to add your system specs to your profile and posts!

Are you OCing your CPU, or running the memory async from the CPU?

 

CPU is an Athlon64 3000+.

 

It is designed to run at 200FSB and I am running it synchronously with the memory at 275FSB. As far as I know, it is not possible to run the RAM on an A64 system asynchronously at speeds higher than the FSB. Or at least my MSI board doesn't have any such settings.

 

I am currently running the CPU at ~1.49 vcore. (In fact, it seems to be stable at 275FSB at around 1.42vcore--I am just running it higher since I am having stability problems and and to be positive the CPU is not at fault.) Cranking it up as high as 1.60vcore does not improve stability with the RAM in dual channel mode at 275FSB so I don't think the CPU's memory controller is at fault. Raising the RAM's voltage DOES improve stability at a given speed, which is why I suspect the RAM is at fault.

 

Also, I have lowered the CPU's multiplier to 6x (the default is 9x200FSB = 1.8GHz) and I still ran into the same 275FSB instability in dual channel mode. I've also played around with the HT setting (default is 5x at 200FSB) and running it at 3x or lower has no effect on relieving the instability at 275FSB either.

  • Corsair Employee
Posted

That sounds more like your CPU/MB is topping out. But I have no problem replacing your modules if you like.

Please follow the link in my signature “I think I have a bad part!” and we will be happy to replace them or it!

Posted
That sounds more like your CPU/MB is topping out. But I have no problem replacing your modules if you like.

Please follow the link in my signature “I think I have a bad part!” and we will be happy to replace them or it!

 

Just curious--what leads you to believe it is the CPU and/or MB that is the limiting factor?

 

I've got an Asus A7N8X with an Athlon XP on it at work--it'll let me clock the RAM at up to 200% of the FSB, so I'll test my sticks out on it and see if they work in dual channel at the rated 275FSB there before trying to RMA them.

  • Corsair Employee
Posted

That A7N8X will not run at 200% for the memory no matter what you use.

And I have not seen this MB and CPU run in our Lab at these settings, about 245-250 Mhz was the top end for this platform.

Posted
That A7N8X will not run at 200% for the memory no matter what you use.

And I have not seen this MB and CPU run in our Lab at these settings, about 245-250 Mhz was the top end for this platform.

 

Would the CPU or motherboard be responsible for the sticks being able to run stable at 275FSB by themselves but not in dual channel mode? Also, I've lowered the CPU's multiplier so that it runs at 275FSB while at or below its 1.8GHz rated speed--since I still get instability when the CPU is at stock speed, wouldn't that eliminate the CPU as a potential culprit?

 

I should have another 3000+ come across my desk at work in the next week or so--I'll give that a shot and see if that has any effect.

 

BTW, thanks for the quick replies!

  • Corsair Employee
Posted

The best to test the memory if you suspect it would be to test them in another system, preferably an Intel P4 with an I875 chipset and 2.4C CPU.

But if the system is stable running one up then its not the memory.

Posted

I thought the whole reason behind buying a dual channel "kit" was that some single sticks of RAM might not perform at their rated specification in dual channel mode. Buying a kit rated at a certain speed is supposed to guarantee that both sticks will indeed work together at their rated speed.

 

I'll see if I can come across a P4 system like you described to test out my RAM as well.

  • Corsair Employee
Posted
I thought the whole reason behind buying a dual channel "kit" was that some single sticks of RAM might not perform at their rated specification in dual channel mode. Buying a kit rated at a certain speed is supposed to guarantee that both sticks will indeed work together at their rated speed.

 

Yes, but that is not the issue/question to ask ATM. Now if you said One modules runs at the tested spec but one does not that might be relevant.

XMS4400C25 modules are very aggressive and not many MB/CPU's can run at these settings. In fact we have to sort the 2.4C CPU we have to get 10 out of about 60 that will run reliably at the tested settings for these modules and we have to sort the MB's as well.

Posted
Also, regarding the A7N8X and memory clocking, I run the CPU on that MB at 200FSB. If I ran the RAM at something like 135% (if that's an available setting) that would put the RAM at 270MHz. Now I obviously don't expect to go running the RAM at 200% of the FSB, but is there any reason I should run into problems attempting to run 135%?
  • Corsair Employee
Posted

No I am sorry, that is a limitation of the CPU and chipset. The settings that are there are for older CPU's, but with a CPU speed of 166 MHz or faster the memory frequency would need to be set to 100%, sorry!

This is a chipset limitation.

If the FSB of your CPU is 400 For Intel and 100 MHz for AMD, the memory speed supports only DDR 266.

If the FSB of your CPU is 533 For Intel and 133/166 MHz for AMD, the memory speed supports DDR 266/333

If the FSB of your CPU is 800 MHz for Intel and 200 MHz for AMD, the memory speed supports DDR 333/400

Thank You!

Posted

I'm pretty sure that the motherboard isn't the limiting factor as far as RAM speed goes. The reason being that the memory controller for the Athlon64 is on the CPU, not on the MB, and my MB seems to have no problems with high FSBs. I've currently got my 3000+ clocked at 290FSB, which is 2.61GHz out of a 1.8GHz-rated CPU, with the RAM at DDR333 / 241MHz 1T 2.5 4-7-4 in dual channel mode. Everything appears to be stable at that speed. (I run Folding to keep the CPU at 100%, run 3dMark and SiSoft benchmarks, play HL2, and just do some wordprocessing/internet browsing to decide if I think the system is stable.)

 

So I am assuming that the weak link, when attempting to run the CPU and the RAM synchronously at 275FSB, could be the memory controller on the CPU, or the RAM itself.

 

250FSB seems to be about the stable maximum synchronous FSB. I was able to run synchronously at lower FSBs with the RAM voltage at or below 2.7v, but it took 2.8V to ensure stability at 250FSB. Now if the CPU's memory controller was at fault, would increasing the RAM voltage (but not making any changes to the CPU settings at all) have any effect on increasing stability?

 

What I'm getting at is that if the CPU's memory controller was causing instability running 250FSB or higher synchronously with the RAM, then raising the RAM voltage shouldn't help improve stability. Especially considering that the RAM is rated to require only 2.75v at 275FSB and I had to use 2.80v just to be stable at 250FSB.

 

Please let me know if my deductions here make sense, or if you think I am incorrect. I hope that I'm not being a pain in the *** and I appreciate the RMA offer, but I'd like to be reasonably certain of a culprit before I go sending back any parts, especially if the RAM might not be at fault. Thank you!

  • Corsair Employee
Posted
As I said, when we tested this MB and others based on the same chipset they would top out about 240-250 MHz when running the memory 1-1. Just seems to be the limit of this chipset. And the memory controller is in the CPU and will share the same power source as the memory so upping the memory voltage may increase stability and can be because the memory or memory controller need a bit more power. I would maybe see if a friend might have another system you can test the modules in before you go thru with an RMA, we have a lifetime warranty on our memory so we will be happy to replace them at any time.
Posted

I just got ahold of another A64 3000+, which I installed for these tests. In the BIOS, I noticed that there is a setting to clock the RAM at DDR500 speeds. I first tried clocking the CPU at the stock 200FSB with the RAM at 250MHz. The system appeared to run stably. The CPU was at 1.42vcore and the RAM was at 2.75v with 2.5-4-8-4 1T timings.

 

I then tried 210FSB which put the RAM at 262.5MHz. With the memory timings set to "auto" in the BIOS, it defaulted to 2.5-5-10-5 1T. Vcore and Vram were unchanged. I ran a couple of benchmarks successfully.

 

I then bumped it up to 220FSB, with the RAM at 275MHz. Timings stayed at 2.5-5-10-5 1T according to the POST screen. The system made it to the desktop and the gave a BSOD and spontaneously rebooted. I upped the RAM to 2.8v but the system gave another BSOD and reboot at the same point. I'm now back at 210FSB/262.5MHz RAM at 2.75vram.

 

I haven't tried clocking the RAM and the new CPU synchronously at high speeds yet.

 

So with the new CPU, clocked at only 220FSB, I am unable to run the RAM at 275MHz stably. I will switch back to my original CPU (after attemping high FSB synchronously clockings with the new one) and see if it runs into the same wall at 220/275.

 

What is your opinion on these most recent results? Thanks!

Posted

Well, the new CPU was something of an overclocking dud. I could get it up to around250FSB, but I had to lower the multiplier from 9x to 8x in order for it to be stable. Running at a 9x multiplier, the highest stable FSB I could hit was around 230. (The next step up I tried was 240, and that crashed.)

 

I switched back to my previous CPU. I checked the RAM settings in the BIOS, and the DDR500 setting was gone. I'm guessing that the new CPU was one of the "Venice" cores, and apparently that's why I had the option to clock the RAM faster than the CPU. So it looks like any further asynchronous testing is a moot point with my original CPU.

Posted
Please let us know how you make out!

 

Have you guys had any experience with the new Venice core Athlon64s? (IE will their memory controllers typically handle RAM speeds in excess of 275MHz?)

 

Since I was not able to run the new test CPU I borrowed at 220FSB/275MHz RAM stably (although it did appear to be stable at 210/262.5), does that help at all determining whether the RAM is at fault?

 

(So far I have not come across a local Intel setup which will run at 275FSB or better to be able to test the RAM, but I'm still looking. I've found one that's 5 hours away, but that's not exactly practical to test with. :p: )

 

Thank you!

  • Corsair Employee
Posted
There seems to be many of them out now, but I have not seen them able to hit 275 Mhz, I think as the MB bios's mature you will start to see better over clocks with them. We are still using the older cores or FX55 CPU's to test AMD systems.
  • 3 months later...
Posted

Time to resurrect an old thread with some updates.

 

Over a month ago (early October-ish) I was able to get my system stable--running with only one stick of RAM with the CPU and RAM clocked at 200FSB. Two sticks would result in instability. At that point, I still wasn't positive that I should point the finger at the RAM.

 

Then I borrowed a pair of XXXX (well I guess you don't get to know the brand name :laughing: ) 256MB DDR533 CL3-4-8-4 sticks. I popped them in, set the RAM voltage to 2.7v, and cranked the CPU and RAM up to 250FSB at CL2.5-4-8-4 timings. I have been running stably for over a week now in dual channel mode. Benchmarks, games, internet, etc have all run fine with no crashes whatsoever.

 

Today I swapped back in both of my Corsair 512MB DDR550 sticks. I made no changes to the BIOS settings. After booting up Windows, I got a BSOD crash that said "windows has detected an error and is shutting down to prevent damage". I swapped the XXXX sticks back in, and here I am typing this post.

 

In short, I can't think of anywhere to place the blame for the crashes besides on at least one of the Corsair sticks being bad. If you think there is something else I should try or that another part might be at fault, please let me know. Otherwise, please give me an RMA number so I can send my Corsair RAM back for replacement. Thanks!

 

<edit> Oh yeah, I forgot about the online RMA request form. I'll wait until you reply before I fill it out, though.

  • Corsair Employee
Posted
I have no problem replacing your modules but you are comparing apples to oranges when you run a 256 meg against a 512 meg module. Most 256 meg modules are single sided and most 512 Meg modules are double sided and would be more loading on the memory controller. But if you just want to get them replaced. Please follow the link in my signature “I think I have a bad part!” and we will be happy to replace them or it!
Posted

Well crap. After almost two weeks of apparently being stable with the 256MB sticks, I got a BSOD error yesterday. At this point, I guess it's pretty safe to say that since I've gotten crashes with two completely different pairs of sticks of RAM, the problem lies elsewhere.

 

I understand what you're saying about the single-sided RAM (as the 256MB sticks indeed are) placing less of a load on the memory controller.

 

Anyway, thanks for your help. :):

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...