Jump to content
Sign in to follow this  
Antoneo

Signs of instability? Overclocking my Sempron 3100+!

Recommended Posts

Hey guys, I'm on my second Sempron computer now and am experiencing different "quirks" between the two while overclocking and testing for stability.

 

Behavior on Sempron 2800+ before reaching a stable overclock:

1. Give me the "nvraid detect" loop while attempting to boot up.

2. Fail to boot into windows.

3. Prime95 would tell me that there was a hardware/rounding error and would consequently fail.

 

Behavior on Sempron 3100+ before reaching a stable overclock:

1. Fail to boot into windows.

2. Simply restart without attempting to boot into windows.

3. I believe while running Prime95, the computer would freeze and I would have to press reset.

 

When I am nearing a stable overclock, step 3 for both is where I am adjusting the voltages and then Prime95 testing. CnQ is disabled for both.

 

On my 3100+, I haven't been able to get Prime95 to error out with a hardware/rounding error. I have only been getting freezing inside of windows where I am forced to press the reset button. I am currently running [email protected] Prime95 stable for about 5-6 hours now.

 

I will also note that I am able to boot into Windows at 300x9 with 1.450 vcore but I get freezes.

 

The troubling thing is that at 290 FSB, with 1.475 vcore I would get a freeze and at 1.5 vcore it seems prefectly stable. I feel that I am missing the Prime95 software errors as I have never seen any yet, it has been either stable or simply freeze up. I was wondering if I should be concerned at all or if anyone else have found it to be their experience as well. Is this a sign that my PSU is inadequate?

Share this post


Link to post
Share on other sites
Guest Blooz1

Are those voltages you're giving "total" (sum of VID Control + VID Special), or just "VID Control"?

 

1.5 "total" seems low to me for what you're trying to achieve. Your rigs are so similar, it might be worth the time to swap CPU's to test your PSU question!

Share this post


Link to post
Share on other sites

those are insane speeds for 1.5v and a .13u process (it is .13u right?) on air

 

took me over 1.7v to get even close to those speeds (see sig)

 

i agree with blooz the voltage is funky.

Share this post


Link to post
Share on other sites

The voltages I am stating are the "total," I never applied any VID special voltage in the BIOS so it is only the VID control. They are both revision E6 .90u process chips.

 

I have swapped the Sempron 2800+ between both motherboards and have achieved the exact same stable overclock so in that case, the CPU is the limiting factor. I haven't swapped the Sempron 3100+ cpu because it's just such a hassle to remove both of the XP-90/120 and then reinstall. I have to take out the motherboard and I've gotten tired of the whole process hehe. If I had access to a more powerful PSU then, a PSU swap would be doable but unfortunately I've none at the moment.

 

Another peculiarity I've noticed is that CPU-Z does not report the Frequency or FSB:DRAM in the memory tab while the 3100+ is on this overclock. At stock, the Frequency is correctly 200 MHz and FSB:DRAM is CPU/9. The two labels are grayed out along with the text boxes next to them. At other overclocks, the program does display information I think. My 2800+ shows the correct 200 MHz and CPU/12 information. What gives?

Share this post


Link to post
Share on other sites

I am running the memory at 203Mhz (7/10 divider) with 2.7 vdimm. You have a good point, I have not run memtest on it yet, I've partially forgotten about it because the RAM was running close to 200MHz. But I will try it tonight.

 

I am running 2005/05/04 BIOS on both my machines without testing for which BIOS version the best for either one. I didn't want to fiddle in and out with flashes so I updated to the latest one and wanted to leave it at that.

 

I've updated my sig with more details!

Share this post


Link to post
Share on other sites
Guest Blooz1

Well, I still think your low voltage is the main "culprit" on the 3100, although that "value" ram isn't helping your cause with such an extreme divider. I'm curious to see the results of your Memtesting.

 

 

My 2800 Palermo will run fine at 2400+/1.55V, but it has to be bumped up to the 1.6V range to pass heavy stability tests such as OCCT or SP2004.

 

Have you read AG's thread on testing settings at 1T vs. 2T? Very interesting findings! Running at 2T is known to help with stability issues:

 

Angry's Testing On 1T/2T

Share this post


Link to post
Share on other sites

I haven't been able to run memtest with the current settings as I've been testing at 1.475 vcore one more time and have had the machine freeze after 8 hours of Prime95. So I placed the machine back on the 1.500 vcore for about 6 hours now.

 

Blooz1: Does running a divider cause "stress" to the CPU's memory controller? I'm intrigued because you described the 7/10 divider I am using as extreme. I have 2.7 vdimm on the RAM but I'll have to run memtest to be sure that the ram isn't the problem.

 

If you copy and paste the following address in a new browser window, you'll see the CPU-Z quirk I tried to describe: http://pics.bbzzdd.com/users/antoneo/pic1.JPG

Share this post


Link to post
Share on other sites
Guest Blooz1

Antoneo -

 

Maybe I used a poor choice of words when I said "extreme".

 

Running a divider actually takes stress off of a memory controller. Most people, when running dividers, try to keep the divider as low as possible (I.E., try to run a 9:10/180mhz vs. a 7:10/140mhz divider) in order to run the ram as fast as possible to get as much bandwidth as they can.

 

Although there's very little difference between running memory 1:1 and running with, say, a 9:10 or 5:6 divider, once you run a divider of 3:4/150 or higher, there is a slight performance penalty.

 

Here's a link to a chart which can be useful when deciding what to use for a divider and the resulting memory speed;

http://www.zen21587.zen.co.uk/overclockers...em%20ratios.JPG

 

(I have a printed copy of this chart, and find it handy!)

 

With your "value" ram, I can see where you don't have much of a choice but to run a higher divider. I typically run a 5:6 divider, but I'm always "tweaking" to see if I can run a 9:10. I know I could do it by slowing my HTT, but that would defeat the purpose!

Share this post


Link to post
Share on other sites

I thought I might update what was going on with my computers since my last post.

 

I've changed out the power supply in the Sempron 3100+ computer from a Antec TruPower 330 to a Fortron 450.

 

I also decided to run Prime95 on both machines just for fun and it turns out the the Sempron 2800+ machine got a rounding error and failed within several hours. That is very strange since I am pretty sure it for 24 hours+ before without problems. So I turned down the overclock from 300x8 @1.55vcore to [email protected] and am Prime testing. It's been stable for 24 hours+ again. Anyone know if there is a way for Prime95 to tell me how long it has been running without having to stop the program?

 

What puzzles me is that the systems are much more stable when I dial in RAM timings as opposed to leaving everything on AUTO. When I am trying to overclock with the settings for RAM are left on AUTO, I would have trouble booting into Windows or would get a system reset loop. Anyone else experience this? Does this mean anything in terms of stability for other settings that have been left on AUTO?

 

The LDT/FSB multiplier for this machine and the other is 2.5x. Vdimm for both is 2.6v and they both were memtested for 12 hours without errors.

 

The Sempron 3100+ machine began to show me software/rounding errors in Prime95 after installation of the new PSU. I should have measured the rails with the old PSU to be sure heh. So, I began Prime95 testing again.

290x9 @1.475 vcore failed as expected

290x9 @1.500 vcore failed after 44 hours and 10 minutes. Prime95 got 100 "Illegal Sumout" warnings and stopped. Grr.... looks like Prime95 stable for 24 hours doesn't cut it either. I'll try 1.525 next. :(

Share this post


Link to post
Share on other sites
44 hours? That's ages. I'd be happy with 44 hours. But i'm happy with just Game Stable

 

Yeah, well I just left the computer on all day and did some work (saving very frequently lol :O) on it. When I wasn't doing work, I was playing World of Warcraft and to my surprise, it ran fine with Prime95 in the background.

 

EDIT: I am seeing temps of 45-49C on load for the Sempron 3100+ CPU (I assume Temp1 in speedfan is the CPU temp as it fluctuates the most during Prime95 and has the largest difference from idle to load). Is this too much for a 90nm CPU?

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...