• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Stress Testing

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
It takes 1.25 to get a stable 3.0ghz. It could be the last year of getting thrashed inside my 975X mobo trying to get stable @ 3.2. I have Dr. Moss turned off. I can run stable 3.2ghz @ 1.33 on my vcore with Prime95 for 24 hours.
 
If you've never updated your bios from 1.0, I suggest you do as there are substantial changes and fixes. The reviews i've read last night gave this mb fairly average score due to complication/bugs with bios 1.0.
http://global.msi.com.tw/index...ncat_no=1&prod_no=1479
I don't think your Q is degraded at all. You need some pretty crazy voltages and high temps to kill them. Even so, I put my money on the mainboard dying first.
 
I was browsing through coolaler's forum yesterday and looking at their results with your mb/quad and their results are normal. 400fsb-450fsb (some are hitting 500fsb) seems to be a cake walk with your mb. Although they aren't using 1.0 bios or any of the official, as they don't match the dates posted at MSI's site. I think you just have some settings that aren't tweaked right.

See if this works: 400x8
Clockgen Tuner
Cpu clock amplitutude: 800mv (you can try 900mv-1000mv)
Pci-e clock amplitude: 800mv
Cpu Skew: 200ps
MCH skew: 100ps
GTLRef0/1: 63%/67%
Perfromance Leve/Static Tread control: 8
Vcore: 1.28v
Vddr: 2.0 (Try 1.9v-2.1v) You have your ram currently underclocked with loose timings, it should run on its rated speed or it's defective if it won't.
MCH: 1.25v-1.35v (If Prime95 blend freezes within 1-2min, you have to adjust this voltage)
Vfsb/Vtt: 1.30v-1.35v (Set this voltage a notch higher or lower to your NB/MCH)
ICH: Default (Adjust it if you're having audio problems, devices that are hooked to your pc that isn't functioning normally as they should)
5-5-5-15-6-54-3-3-3-13@Sync with fsb 1:1

Goodluck
 
I tried to update my bios, but my old usb drive wont let me boot from it. I will have to pick up a new one. I will try it again after the weekend.
 
Originally posted by: james1701
I tried to update my bios, but my old usb drive wont let me boot from it. I will have to pick up a new one. I will try it again after the weekend.
Click to expand...

James, Here's a utility you can use to format your usb stick and make it bootable. The stick I use to store some data and mostly bios is formatted to fat16.
http://files.extremeoverclocking.com/file.php?f=197
Just read the guide at msi's site on how to flash your bios the old fashion way. Afudos + bios file from your usb drive.

Here's my finding so far with my quad...at 3.4ghz

Prime95 blend passed
377x9@3400
GTLref0/2=63%
GTLref1/3=67%
MCH GTLref=Auto
Vcore: 1.32v Load
CPU Clock Skew : 200ps
MCH Clock SKew: 100ps

Prime95 blend fail on core 0 @ 8K FFT Test
425x8@3400
GTLref0/2=63%
GTLref1/3=67%
MCH GLTref=Auto
Vcore: 1.32v Load
CPU Clock Skew : 200ps
MCH Clock SKew: 100ps

Prim95 blend passed
425x8@3400
GTLref0/2=67%
GTLref1/3=67%
MCH GTLref=64%
Vcore 1.32v
CPU Clock Skew : 200ps
MCH Clock SKew: 100ps
http://img386.imageshack.us/im...3/425x8settingsxw3.jpg

What I also noticed is that the MCH Clock skew makes a ton of difference when you're trying to achieve a higher fsb. If you've tweak everything and had not touched this setting. You might give it a try. 100ps-500ps makes a whole lot of difference from my testing. My fsb wall is somewhere at 490-495fsb with my quad. I say this becuase it needs a ton of vtt/vnb voltage to get everything to run stable. Hence I threw in the towel yesterday at 485fsb as it keeps on freezing hehehehe 🙂 It was very time consuming. I just ordered the same cooler as yours. What fan are you using on your TRUE?


Goodluck,
 
I am using the Antec 120mm tri cool. I am only running 1 fan and not 2. I could get a couple C cooler if I put the second one back, but I will see if I can get my Core voltage lower first.
 
I have tried using the HP program already, with no luck. My usb stick is about 5 years old. So I think its time for an update. I was playing around and was able to get 3.5ghz stable at 1.45 on my core. It ran my temps up to 70C, and 56C on the cpu case temp. I think I will stick with 3.4 for now and try to get my bios updated. It looks like they got a few more features working, like speed step, in the new bios. I tried changing all the gtlref and scew settings one at a time, and sometimes couple at a time with no luck. It just takes a lot of power for my cpu not to drop the 3rd core. I even unversed the gtlref voltages, and it had no effect either way. I wonder if it might be malfunctioning in my bios and not giving me the effect its designed to do.

Just a couple of months ago, my thoughts would have been to run the 3.5ghz and not worry about the temps too much, because when gaming, I only hit 56C with the core set to 1.45. Now that new games are being released that use a quad core better, it might run the cpu up to prime 95 temps. So I think I had better not take the chance.

Did you get the 120 Extreme? You will like it, but its big. I have mine pointed up toward the but 220mm fan at the top of my case and all fans back on high again. I just wish I had longer power supply cables, so I could do a good wiring job. I really need a new power supply. This unit is going on 4 years old.

Its amazing what a good OC will do for just general responsiveness to you system. Before I got my Q6600, I was running a P4@4.0ghz. That desktop was screaming. You can really tell a difference when you slow down you processor in just surfing and email operations.
 
Well, thanks to my ignorance of Coretemp, I have been running with a Tjmax of 100C instead of 90C. Now my temps are much better. I am only hitting 60C @ 1.45 volts. That leaves me plenty of headroom till the 70C mark Intel posted. I just wonder, I have also been using Hardware monitor from CPUZ for information. It looks like they are using Tjmax at 100C instead of 90C. Will this affect Tcase temperature readings?
 
Originally posted by: james1701
I have tried using the HP program already, with no luck. My usb stick is about 5 years old. So I think its time for an update. I was playing around and was able to get 3.5ghz stable at 1.45 on my core. It ran my temps up to 70C, and 56C on the cpu case temp. I think I will stick with 3.4 for now and try to get my bios updated. It looks like they got a few more features working, like speed step, in the new bios. I tried changing all the gtlref and scew settings one at a time, and sometimes couple at a time with no luck. It just takes a lot of power for my cpu not to drop the 3rd core. I even unversed the gtlref voltages, and it had no effect either way. I wonder if it might be malfunctioning in my bios and not giving me the effect its designed to do.

Just a couple of months ago, my thoughts would have been to run the 3.5ghz and not worry about the temps too much, because when gaming, I only hit 56C with the core set to 1.45. Now that new games are being released that use a quad core better, it might run the cpu up to prime 95 temps. So I think I had better not take the chance.

Did you get the 120 Extreme? You will like it, but its big. I have mine pointed up toward the but 220mm fan at the top of my case and all fans back on high again. I just wish I had longer power supply cables, so I could do a good wiring job. I really need a new power supply. This unit is going on 4 years old.

Its amazing what a good OC will do for just general responsiveness to you system. Before I got my Q6600, I was running a P4@4.0ghz. That desktop was screaming. You can really tell a difference when you slow down you processor in just surfing and email operations.
Click to expand...

For a P45 chipset, I can safely say that it's not behaving as it should. It should be easy to find the max fsb for the quad compared to the previous mb you had which is 975. That chipset had difficutlies oc'ing a quad to 400fsb due to lack of GTL ref options via bios. Hopefully the newly released bios at msi site would ironed your issues out. You should be able to clock your quad to 3.6Ghz it's do-able but not impossible.

Ordered the cooler out yesterday, I had a good deal on it. Did your true have a concave base? I have a 120mm fan here that came with my case. Just by looking at it, I will have to agree with your statement that it's big hehehe 🙂 I hope to not be able to lap mine, because I only have 900 grit sandpaper here on hand. It's going to be 2-3hr journey just trying to fit this cooler on my midtower case. My wire management solution for now is to stuff all the wires under my dvd drive. It looks tidy ti'll you tilt your head on the left and see a big ball of wires hiding under my dvd drive 🙂



 
Originally posted by: james1701
Well, thanks to my ignorance of Coretemp, I have been running with a Tjmax of 100C instead of 90C. Now my temps are much better. I am only hitting 60C @ 1.45 volts. That leaves me plenty of headroom till the 70C mark Intel posted. I just wonder, I have also been using Hardware monitor from CPUZ for information. It looks like they are using Tjmax at 100C instead of 90C. Will this affect Tcase temperature readings?
Click to expand...

I'm using 95C tjmax as my settings with everest. Because 90C doesn't make sense, My cpu at idle are 2-3c lower than my mb temps (I used mb temp to measure ambient temp inside my case). If your true barely gets hot, I think this why the 100C tjmax made everyone cpu temp reading too high. My rule with P4 was to not hit 60C max with any oc. With my Dothan, max was 50C and my Quad at 65C max. Heat degrades everything very quickly, could be the cpu or the mb components onboard. Due to some to of these tower coolers. Some of these components aren't getting sufficient cooling when they are pushed out of spec.
 
"The digital thermal sensor (DTS) accuracy is in the order of -5°C ~ +10°C around
90°C; it deteriorates to ±10°C at 50°C. The DTS temperature reading saturates at
some temperature below 50°C. Any DTS reading below 50°C should be considered to
indicate only a temperature below 50°C and not a specific temperature. External
thermal sensor with ?BJT? model is required to read thermal diode temperature."

Thats is from Intel, I guess everyone needs to take it with a grain of salt.
 
Wow, i'm surprised what the difference tweaking the skews and gtl does to a quad. Check this out James. It was completely impossible for me to run 485fsb, no matter how much vtt/nb voltage I shoved to the system without lockups. So decided to tweak the skews/gtl refs to their most optimal settings yesterday (Lots of prime95 blend and memtest). I was about to throw in the towel once again 😛 One thing I noticed that at 470fsb-480fsb, the system no longer needs a ton of juice to the nb every 1-2fsb step and instead of locking up/freezes. It would produce a rounding error instead (multiple error in memtest) So I gathered all my data from yesterday and fine tune my voltage settings. Now 485fsb is a peice of cake for my quad 🙂 I don't think I can optimize the voltage setting used for my cpu oc.
http://img247.imageshack.us/im.../485x7settings2jj0.jpg

After:
GTLref0/2=67
GTLref1/3=67
NB GTLref=64
Cpu Skew=200ps
NB Skew=400ps

Before: (freezing and lots of it)
GTLref0/2=63
GTLref1/3=67
NB GTLref=Auto
Cpu Skew=200ps
NB Skew=100ps
 
Have you ran any benchmarks? Other than memory, why run 485 with multi of 7 to get 3.4ghz. I really don't see much of an advantage to running a fsb that high when you don't need too unless that is what it takes to get your memory to a specific mhz. Have you tried raising your multiplier to get 3.6ghz, using the GTLref adjustments. That would get you more gains than anything.
 
Originally posted by: james1701
Have you ran any benchmarks? Other than memory, why run 485 with multi of 7 to get 3.4ghz. I really don't see much of an advantage to running a fsb that high when you don't need too unless that is what it takes to get your memory to a specific mhz. Have you tried raising your multiplier to get 3.6ghz, using the GTLref adjustments. That would get you more gains than anything.
Click to expand...

Looks like you're running your quad now at 3.5Ghz nice...What multi are you using? The reason why I'm testing high fsb is that I can't get a good ram bandwidth on my setup as the ram max out at 515Mhz. 3.4Ghz imo is a good setting as with games at high resolution, i'm always gpu limited.

With everest ram bandwidth: Ram maximum limit is 515Mhz@2.3v 5-5-5-15
485x7@3400Mhz 1:1 5-5-5-15@2.2v
Read- 9200mb/s
Aquamark Cpu Score: 20.100
Vtt voltage@1.376v
Vnb voltage@1.368

425x8@3400Mhz 5:6 5-5-5-15@2.32v
Read- 8800mb/s
Aquamark Cpu Score: 19,500
Vtt voltage@1.30v
Vnb voltage@1.276v

This setting took me all the way to 488fsb without too much voltage adjustments
488x7@3.4Ghz+
Cpu Skew - 200ps
NB Skew - 400ps (vital to stability at high fsb)
GTLref0/2 - 67*vtt
GTLref1/3 - 67*vtt
NB GTLref - 64*vtt (never leave this on auto when trying to achieve high fsb)
Vcore: 1.320 (fully loaded)
Vddr: 2.2V
Vtt: 1.376v
Vnb: 1.368




 
You must log in or register to reply here.

TRENDING THREADS

Back
Top