Walk before Run

[wolfwatcher51]

Just getting my system together and want to make sure I have it "right", so I do not burn anything up when trying a little oc.

I have coretemp 0.96.0.0
cpu-z 1.4.2.0
speed fan 433

The temps in coretemp are about 15C higher than speedfan, so I am guessing this version, latest?, is still 15C low?

I read the testing thread and just a little too noob. It says to run memtest and keep your temps below75C, or something like that. When I make the memtest floppy and reboot to it to run, it is the only screen there, looks dos like. So, how do I monitor my temps while running memtest?

On the screen durring memtest, towards the right side there is an entry for errors. So far mine says 0. If there were errors, they would show up here, right? Or are they logged someplace and how do I get to them.

I have not tried Stress Prime 2004-dual core, so I do not know what is visible when it is running. What can I use, and how to measure/monitor temps durring SP 2004?

Thanks in advance, Chris.

 

[Idontcare]

Originally posted by: wolfwatcher51
The temps in coretemp are about 15C higher than speedfan, so I am guessing this version, latest?, is still 15C low?

See section 5 in Graysky's 1st post in the following forum thread: http://forums.anandtech.com/me...=2057083&enterthread=y

This will fix your speedfan temperature issue.

Originally posted by: wolfwatcher51
I read the testing thread and just a little too noob. It says to run memtest and keep your temps below75C, or something like that. When I make the memtest floppy and reboot to it to run, it is the only screen there, looks dos like. So, how do I monitor my temps while running memtest?

That is correct, you should see a blue background screen verymuch like a dos application. This is normal. You can't monitor temps thru memtest. If you want temps then you have to hookup some kind of external temp monitor. Usually no one monitors temps during memtest as it is supposed to test your RAM and not so much your CPU...although you should be aware that a hot/overclocked CPU can cause memtest errors as it does test some of the L1/L2 caches.

Originally posted by: wolfwatcher51
On the screen durring memtest, towards the right side there is an entry for errors. So far mine says 0. If there were errors, they would show up here, right? Or are they logged someplace and how do I get to them.

Correct, if you have errors they will show up there as well as on the bottom half of the screen in very noticable red background text.

Originally posted by: wolfwatcher51
I have not tried Stress Prime 2004-dual core, so I do not know what is visible when it is running. What can I use, and how to measure/monitor temps durring SP 2004?

You use Prime95 in Windows. Once you are logged into windows and have installed the program then run it. You will have "Torture Test" choices - small FFT is for really testing out your CPU, Blend is more for checking out RAM and FSB stability.

To monitor your temps while Prime95 (also called Orthos ans Stress Prime, all same thing) is running you just simply open Coretemp and/or Speedfan and watch.

Coretemp logs the temps by default so it can be handy if you step away and your computer reboots in the meantime, you can go back and see what temp the system was before Coretemp was killed by OS reboot or hang.

Also conider CPU-Z in addition to having Coretemp and Speedfan open, it will show you your CPU's core voltage in effectively realtime so you can see how many volts are actually making it into the CPU during load.

 

[wolfwatcher51]

Idontcare,

Thank you for your very specific answers. That is no doubt the best reply of any of my posts to date.

Again, thanks, Chris.

 

[Idontcare]

Happy to have helped. It hasn't been so long ago that I too was learning to walk before running...I remember all the uncertainty and now I hope to reduce some of it.

John's and Graysky's threads in the CPU forum are pure e-treasure. Use them well and they will serve you well.

When you get to running remember to pay it forward.

 

[wolfwatcher51]

OK,

Used 2004 for orthos testing. I left it at the default of blend. Also opened cpu-z, speed fan, and core temp. Before starting the test the readings were:

Sys fan 518 sys temp 28
cpu 0 fan 1259 core 0 temp 36
cpu 1 fan 1223 core 1 temp 36

Started test and almost immediately got this:
core 0 temp 49
core 1 temp 50

As test continued to run the core temps stayed 48-51 each. First time I ran it I only went 13 minutes before I had to go. Results displayed no errors. Later, set up for the test again, this time I have more time, so will see what happens. So far, it has run 30 minutes, passed the 1024 and 8K tests and the temps are 50-51 ech.

This is a stock E2180, IP35 mobo, 2X2gb ddr2-800, enzotech ultra-x in a Rosewill case with 2X120mm case fans.

Is this the expected behavior? I guess the temps could jump that much since they are going from 0 load and 6x multiplier to 100 percent load with 10x multiplier. What are your experiences?

One more thing, I am alittle confused about. The cpu fan control. I have mine set for target temp 45, temp tol 5,start control 80, and stop control 50%

I believe this means the fan should run at 50% until the cpu temp gets to 50 (45+5) at which time it should go to 80% and ramp up as the temps go up. Is that right?

The sys fan is set to ref temp CPU, target temp 45, temp tolerance 5, start control 70 and stop control 50%. This works just like the cpu setup, except at different speeds, right?

While the test is running and the core temps are 51, should't the cpu and sys fans have gone to 80 and 70% ?. According to speed fan they seem to be at the same rpms that they were when the test started. Is it because the cpu temp it is referring to is the cpu temp in speed fan, rather than the core temps? Speed fan shows cpu temp 38 and the cores 51/51. Does the cpu temp in speed fan need the 15 offset like I put in the core temps, and if so, what do I select to do it? Or do we set the cpu and sys temps down to the cpu temps?

Thanks in advance, Chris.

 

[Idontcare]

Originally posted by: wolfwatcher51
Used 2004 for orthos testing. I left it at the default of blend. Also opened cpu-z, speed fan, and core temp. Before starting the test the readings were:

Sys fan 518 sys temp 28
cpu 0 fan 1259 core 0 temp 36
cpu 1 fan 1223 core 1 temp 36

Started test and almost immediately got this:
core 0 temp 49
core 1 temp 50

As test continued to run the core temps stayed 48-51 each. First time I ran it I only went 13 minutes before I had to go. Results displayed no errors. Later, set up for the test again, this time I have more time, so will see what happens. So far, it has run 30 minutes, passed the 1024 and 8K tests and the temps are 50-51 ech.

Looks good so far. Once you know blend does not cause your computer to have issues (as you have done) then usually you jump right into testing small FFT as that will raise your CPU temps even higher (to pretty much the max any program could raise them to at the current hardware setup) and this is the quickest way to find out when your CPU heat or voltage has become the weakest link.

Also, it is OK to leave Orthos running while you walk away unless you are particularly nervous about something catching fire or something. I leave my systems running orthos overnight so I get some good 6hr runs in.

Originally posted by: wolfwatcher51
This is a stock E2180, IP35 mobo, 2X2gb ddr2-800, enzotech ultra-x in a Rosewill case with 2X120mm case fans.

Is this the expected behavior? I guess the temps could jump that much since they are going from 0 load and 6x multiplier to 100 percent load with 10x multiplier. What are your experiences?

Yep absolutely normal behaviour. Orthos will raise temperature nearly as fast as you release the mouse button after clicking to start the run. They will continue to rise ever slowly though for minutes. Usually you won't see your peak CPU temps for at least 15 minutes.

Originally posted by: wolfwatcher51
One more thing, I am alittle confused about. The cpu fan control. I have mine set for target temp 45, temp tol 5,start control 80, and stop control 50%

I believe this means the fan should run at 50% until the cpu temp gets to 50 (45+5) at which time it should go to 80% and ramp up as the temps go up. Is that right?

The sys fan is set to ref temp CPU, target temp 45, temp tolerance 5, start control 70 and stop control 50%. This works just like the cpu setup, except at different speeds, right?

While the test is running and the core temps are 51, should't the cpu and sys fans have gone to 80 and 70% ?. According to speed fan they seem to be at the same rpms that they were when the test started. Is it because the cpu temp it is referring to is the cpu temp in speed fan, rather than the core temps? Speed fan shows cpu temp 38 and the cores 51/51. Does the cpu temp in speed fan need the 15 offset like I put in the core temps, and if so, what do I select to do it? Or do we set the cpu and sys temps down to the cpu temps?

I am of no help to you here as I don't use speedfan (or any software control) for controlling my fan speeds. I do use speedfan for monitoring them though. (edited to clarify my speedfan usage)

I let the BIOS control the CPU heatsink fan speed (motherboard = Gigabyte DS3L, HSF = Tuniq Tower 120) and I use Thermaltake A2029 120mm case fans which have a remote thermocouple to enable fanspeed control by the fan itself.

I will say this though - on my DS3L my CPU's fan barely increase rpms if it heats up to only 50C...rpms really ramp between 50 and 60C. This could purely be a product of my hardware configuration and not explain your observations at all. But seeing your temps are so low it is my first suspicion. Wait till you start running Orthos with small FFT (your temps will increase another 6-10C over the orthos blend temps) or until you crank up the overclock before getting to concerned over whether the fanspeed control is not behaving to expection.

Good luck!

 

[wolfwatcher51]

Moving right along,

I also just use speed fan for monitoring and let the bios control the cpu and sys fan
headers. Just wondering which of the cpu temps the abit ip35 bios v12 uses to control
by?

Lets see what variables get the highest priority in the o.c. stratgey.

There is the multiplier, the fsb controlled to some extent by Vcore, and the ram speed
controlled by the divider ratio, and the strap for 1066 or 1333 gets in here also. Timings?

I want to use this computer for daily use, very little to no gaming, but embroidery
software, photoshop, corel draw, some video incoding etc. So I would like a good fast
oc, but not all you can get at expense of life. I am thinking 3 years or so, depending on
the next round of chips and prices. Not going for record in oc thread

Rig is E2180,IP35,GSkill 2X2gb ddr2-800 at 1.8v,Enzotech ultra-x,Antec earthwatts
500, and Rosewill case; two 120mm fans plus a side vent and grill with internal cone.

Since I think that C1E and EIST are good to use, I will stick with the stock 10 X. Next
I think 3.2G for the 2180 should do, so fsb becomes 320.

Here is where it gets more convuluted, or at least for me and where I need your help.
Choosing the divider. Synchronous or asynchrous? Divider for 1333 strap or 1066. And
the "tighter" timings for slower than rated speed, right? What gets first priority?

Could someone with the ip35/ip35-e please tell me what dividers are available for this
board? I am using bios v12 if that matters. And, what strap each is associated with?

The 320 fsb would give me 640 for the ram, assuming 1:1. Does running 800 ram at 640
make you able to run 4-4-4-12 instead of 5-5-5-15 and is that better? Or ?? Help!

Thanks in advance, Chris.
P.S. The vcore reading in core temp does not agree with cpu-z. I have seen this in
several screen shot pictures in the oc thread and believe it was that way when I ran
orthos last night. Does it show default vcore, or ?

Thanks in advance, Chris.