First overclock on Opteron 180 query.

mikeo

mikeo

mikeo

Soldato
Joined
17 Aug 2005
Posts
2,828
Location
SW Scotland
Hi folks

Thought I'd have a try to see what I could get on this. First "real" attempt at overclocking.

CPU is an Opteron 180 (2.4ghz) and mobo is an MSI K8N SLI. Memory is Crucial "value" 2 GB (dual channel).

So...

- Set FSB to 210mhz
- HT multi set to 4
- Set PCI clock to 101mhz. Don't know why, but this locks the PCI bus to 33mhz (found this out on the MSI forums).
- Set memory to 180 mhz

Now this boots into windows fine and CPU showing as 2.5ghz.

Now CPUz usually (on BIOS default) shows memory as: 3,3,3,8,11,2T
But now shows: 2.5,3,3,8,12,2T

Try increasing FSB to 215 mhz and gets as far as trying to boot into Windows and Blue screens with a stop message (not like I can tell much from it).

Reset BIOS to default and all is fine.

Question. Is this likely to be my "value" memory? Or is there something else that I can try?

Thanks in advance.
 
Your ram timings are set to SPD, change them to manual, and change the timings to what it says in the SPD table for 400mhz. 3-3-3-8 etc.

You will definatley want to run 1T though if your chipset allows it
 
Defcon5

Thanks for that.

Now I've been able to identify he following:

Case Latency = TCL = 3
RAS to CAS delay = TRCD =3
RAS precharge = TRP = 3
Cycle Time = TRAS = 8
Bank Cycle Time = TRC =11
Command Rate = 2T (1T if it can cope)

But... this leaves a number of other values for the memory that I don't know what they should be set to:

Row to Row delay = TRRD
Row Resfresh Cyc Time = TRFC
Read-Write time = TRWT
Write Recovery Delay = TWR
Write to Read delay = TWTR
Refresh Period = TREF

and what's the DRAM idle timer? CPUz says it's currently 16

Any help would be appreciated.

Thanks again
 
It will probably have an 'auto' option, use this for the ones you dont know, they dont have much of an effect anyway.

Once you have found the limit of the cpu, then you can start messing with the ram, but definatley sort the cpu first.
 
Decided to try again.

Set memory timings as above and still get a STOP screen.

Also tried a memory divider at 166, still no luck.

Mind you, I do remember reading somewhere that these boards don't overclock very well.

Any other suggestions?
 
Row to Row delay = TRRD
Row Resfresh Cyc Time = TRFC
Read-Write time = TRWT
Write Recovery Delay = TWR
Write to Read delay = TWTR
Refresh Period = TREF

i run these at:

2
14
2
2
1
1560 cycles (i cant alter in bios, this is how A64 info reports)

that is with
3-3-3-8 1T @2.7v

for me i believe TRWT is the most vital, 1 = no post, 3 = dire performance

this is running 5b-f based PC4000 balistix however.

my first guess would be your hitting cpu limit and need more v-core.
most of the "standard" crucial ive ever seen has been phenominal (Pc133 @ 166 2-2-2, pc2100/2700 @ pc3200 3-3-3-8 etcetc)
 
Thanks for that.

Now I can get into the forum (seemed to be down yesterday afternoon for me).

Question. From looking at CPUz, the CPU voltage (Vcore I'm assuming here) seems to vary from 1.28 - 1.33. Is this variance normal? This is while the machine is idle.

I had sort of assumed that for a fairly mild overclock (IE. 2.4 ghz to 2.6, which is really all I was trying for) that I would not need to play around with the voltages. Or am I being the naive newbie here?

If I decided to have a go at increasing the vcore, how much should I initially try? Set it to maybe 1.4?
 
Last edited:
mikeo said:
Question. From looking at CPUz, the CPU voltage (Vcore I'm assuming here) seems to vary from 1.28 - 1.33. Is this variance normal? This is while the machine is idle.
Click to expand...

I don't know your mobo, but i think you may have "Cool and Quiet" enabled in the bios. If you have, dissable it, as well as "thermal Throttling" if that's in your bios.
 
I have always had a vcore flutter, nothing to worry about.

What cooling are you using? Id just put the voltage to 1.5, you can bring it down later.
 
mikeo said:
Thanks for that.

Now I can get into the forum (seemed to be down yesterday afternoon for me).

Question. From looking at CPUz, the CPU voltage (Vcore I'm assuming here) seems to vary from 1.28 - 1.33. Is this variance normal? This is while the machine is idle.

I had sort of assumed that for a fairly mild overclock (IE. 2.4 ghz to 2.6, which is really all I was trying for) that I would not need to play around with the voltages. Or am I being the naive newbie here?

If I decided to have a go at increasing the vcore, how much should I initially try? Set it to maybe 1.4?
Click to expand...

Its normal the v core fluctuating a bit. V core 1.33v is low if you want to overclock; I need 1.44v to reach 2850 with my cpu (fluctuates between 1.41v and 1.44 at load. Idle can go up to 1.46v). I wouldn't try anything above 1.5v unless you have water cooling (always watch your temps with speedfan or something similar).

For more help we need the stepping of the Opteron, temps and volts at full load. I remember I couldn't go above 2700 with memory 1:1 at the beginning with my opteron until I bought better cooling and start playing with the memory timmings and dividers.
 
Last edited:
Defcon5 said:
I have always had a Vcore flutter, nothing to worry about.

What cooling are you using? Id just put the voltage to 1.5, you can bring it down later.
Click to expand...

Cooling is "air". But case pretty well ventilated. 92mm blowing in at bottom front of case and 120mm extract at top/back of case. Only last week I added an 80mm fan blowing in just opposite the CPU. Obviously the 8800 GTS and the PSU add to extracting warm air as well. CPU / GPU temps at both idle / under load (at stock settings) are pretty good compared to some threads I've looked at. CPU 28/30C at rest and hit mid 40's under load on ORTHOS.

Phoenix BIOS does it by adding a percentage (rather than setting the actual value). I'll give it another shot later today.
 
Last edited:
Well...

Vcore would appear to have been the "problem".

added 3.3% to Vcore and booted into Windows and started ORTHOS

After a couple of minutes get:

Test 2, 4000 Lucas-Lehmer iterations of M19922943 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 4 minutes 52 seconds - 1 errors, 0 warnings.
Execution halted.

More Vcore required maybe?
 
You must log in or register to reply here.
Back
Top Bottom