Opteron 170 Overclock Help

[JMWarren]

FSB 250
HTT 4x250=1000
Multiplier 10x
Memory on 166 Divider for 208 Memtest stable

VCore is set to 1.5v in the bios yeilding 1.456v in windows.

I can't even get to 2.55 Prime95 stable without more volts. I'm at 49c under load but am wary of adding more voltage.

Have I missed something or is this just a poorer OCing chip than average?

 

[Furen]

Your voltage is way too high for that clock in my opinion. I'd try lowering the HT multiplier to 3x and increasing the memory divider (to 200, I guess) in order to increase the HTT a bit more. If you are able to do so then your problem is either the RAM or the HT speed.

 

[Markfw]

I only get 2500 on mine.....

 

[Heckler 5th]

did u increase the chipset voltage a tic or two?

 

[JMWarren]

Yeah, I increased the chipset voltage just to be safe although I don't have issues if I lower my multiplier. I know this board can do 300+ easy.

I've tried 260 @ 3x but have stability issues that I don't believe are ram related unless the ram is only having problems in windows and not in memtest.

I had similar issues with my 3200+ build hitting a wall at 2.4, but a new bios and some memory settings (even though it was memtest stable) allowed me to hit 2.6. I just don't know where to ask now that DFI-Street is down.

 

[Noubourne]

What are you cooling that with, chilled water? TEC? That's the only way to be sub-50C under load with that much volts. Don't trust any program in windows to tell you what you are pushing. You are pushing 1.5!! The only way to think you are doing less than that is by testing with a multimeter, which I seriously doubt you have done. Trust what you set and ignore those programs, they lie.

Most good Optys don't like volts dude. Go to stock voltage. Overclock as far as you can with STOCK VOLTAGE, and only then start adding volts.

I think the temps you are reading are wrong, just like the volts you are reading. I'm on air and I break 50C at stock speeds and volts under load with a Scythe Infinity. Use CoreTemp to measure your temperatures. It reads the temps right off the die.

Also, what is the stepping on your chip?

 

[JMWarren]

Core Temp reports 52-53C after hours in Prime set to Max Heat, 50c on Small FFTs. I'm using an SI-120, for reference my 3200+ @ 1.55v ran 37c under Core Temp.

This chip won't even make it into windows at 2.5 on stock volts. I'll post the stepping when I'm back at home.

When I checked voltages with a DMM on the 3200+ the windows voltages were almost dead on. My board seems to undervolt. I tested with a DMM since installing the 170.

 

[Noubourne]

Originally posted by: JMWarren
Core Temp reports 52-53C after hours in Prime set to Max Heat, 50c on Small FFTs. I'm using an SI-120, for reference my 3200+ @ 1.55v ran 37c under Core Temp.

This chip won't even make it into windows at 2.5 on stock volts. I'll post the stepping when I'm back at home.

When I checked voltages with a DMM on the 3200+ the windows voltages were almost dead on. My board seems to undervolt. I tested with a DMM since installing the 170.

Well that settles it. I'm buying one of those SI-120s!!! I knew they were good, but WOW!

I guess the stepping is going to determine how you treat it. I've got CCBBE and it hates voltage. There are certainly steppings that act different though.

I have not heard of the DFIs undervolting like that. I haven't tested mine, because temps are clearly my issue, not volts.

 

[Noubourne]

Have you checked your rail stability with that PSU? I checked my NEO 480 and it was ROCK solid on all three rails, at bootup, idle in windows, and full load too. Never budged.

Since you have a DMM...

 

[JMWarren]

Yeah, last time i checked the PSU was solid and the voltage reported in windows was within 1.5% of reality. The newer versions of the SI-120 are reported to cool even better. I'll see how far I can get on stock voltage when I get home.

 

[Noubourne]

Originally posted by: JMWarren
Yeah, last time i checked the PSU was solid and the voltage reported in windows was within 1.5% of reality. The newer versions of the SI-120 are reported to cool even better. I'll see how far I can get on stock voltage when I get home.

Oops. I meant the Ultra 120 is what I was looking at.

I've got a pile of HSFs anyway. One more can't hurt heh.

 

[JMWarren]

Stepping is CCBWE 0551UPMW

Just trying a new bios now. I'll post the results as soon as I have them.

 

[Big Lar]

I had a 170 with that stepping, and couldn't get it much over 2800 with chilled water. However, it could be your ram timings, what is your MAL at, as well as the Read Preamble? May want to try and loosen them up a bit.

 

[JMWarren]

Flashing to the new BIOS doesn't seem to have done much, but my temps have drop by 2c in MBM, Core Temp and RMClock. I though Bios made no difference to Core Temp or RM Clock.

The best I can do on stock voltage is 2300. 2400 needs 1.45 in the Bios (1.42 reported in windows). 2500 needs 1.5 in the Bios (1.456 in windows).

I will try adjusting the MAL and Read Preamble tomorrow but since I'm passing Memtest I'm doubtful that it will help but it's worth a shot.

I'm running a heavy divider so that my ram is underclocked to keep it out of the equation so I'm not too hopful.

 

[customcoms]

You say you are running a heavy divider? That may be your problem. Many Opty's don't work below the 150 divider, and others in-between don't work sometimes either! You need to find ram limits and processor limits individually. Also, secondary ram timings are crucial to Opteron overclocking. Unfortunatly, the known good settings are mainly at dfi-street, which as everyone knows is down. these might help:

http://www.bleedinedge.com/forum/showthread.php?t=11998

http://www.bleedinedge.com/forum/showthread.php?t=9833

http://www.bleedinedgesupport.com/ocz/forum/index.php

I'm sure there are street members here besides myself that have their bios settings available. here's mine:

http://www.newegg.com/Product/Product.asp?Item=N82E16813136152]
DFI LANPARTY nF4 Ultra-D SLI Mod - 7/04-2 BigToe BIOS[/URL]

AMD Opteron 165 CCBBE 0616XPMW

2x1GB Kingston Value Ram

Leadtek "Extreme" 6600GT 128 mb, 550/1120 93.71 Driver

Seagate 120gb SATA HD (8MB cache)

Maxtor 11gb IDE HD

Samsung 8x DVD-Rom IDE drive
http://www.newegg.com/Product/Product.asp?Item=N82E16827152059]
NEC 3550A@4551A DVD-RAM IDE drive[/URL]
http://www.newegg.com/Product/Product.asp?Item=N82E16817101516]
E-Power Tagan 480 Watt PSU (Single rail version of the newegg link, 12v rail 28A)[/URL]



==========

[b]Genie BIOS Settings:[/b]
[color=red]
FSB Bus Frequency.............................  -  300
LDT/FSB Frequency Ratio.......................  -  x3
CPU/FSB Frequency Ratio.......................  -  x9
PCI eXpress Frequency.........................  -  101Mhz

CPU VID StartUp Value.........................  -  StartUp

CPU VID Control...............................  -  Auto
CPU VID Special Control.......................  -  Auto
LDT Voltage Control...........................  -  1.40v 
Chip Set Voltage Control......................  -  1.60v
DRAM Voltage Control..........................  -  2.90v[/color]

[b]DRAM Configuration Settings:[/b]
[color=red]
DRAM Frequency Set............................  -  150=RAM/FSB:03/04
Command Per Clock (CPC).......................  -  Enable 
CAS Latency Control (Tcl).....................  -  03
RAS# to CAS# delay (Trcd).....................  -  03
Min RAS# active time (Tras)...................  -  07
Row precharge time (Trp)......................  -  03
Row Cycle time (Trc)..........................  -  12 Bus Clocks
Row refresh cyc time (Trfc)...................  -  15 Bus Clocks
Row to Row delay (Trrd).......................  -  03 Bus Clocks
Write recovery time (Twr).....................  -  03 Bus Clocks
Write to Read delay (Twtr)....................  -  02 Bus Clocks
Read to Write delay (Trwt)....................  -  03 Bus Clocks
Refresh Period (Tref).........................  -  3120 Cycles
DRAM Bank Interleave..........................  -  Enabled

DQS Skew Control..............................  -  Auto
DQS Skew Value................................  -  0
DRAM Drive Strength...........................  -  Auto
DRAM Data Drive Strength......................  -  Auto
Max Async Latency.............................  -  Auto
DRAM Response Time............................  -  Normal
Read Preamble Time............................  -  Auto
IdleCycle Limit...............................  -  256 Cycles
Dynamic Counter...............................  -  Disable
R/W Queue Bypass..............................  -  16 x
Bypass Max....................................  -  07 x
32 Byte Granularity...........................  -  Disable(4 Bursts)
[/color]

==========


Dual Prime95 + 3dMark2001SE + CPU-Z 1.37:

http://img169.imageshack.us/my.php?image=dualprimesn8.jpg]

[/URL]

3dMark2003:
http://img135.imageshack.us/my.php?image=3dmark03postrd6.jpg]

[/URL]

3dMark2005:

http://img149.imageshack.us/my.php?image=3dmark05postdg3.jpg]

[/URL]

3dMark2006:

http://img138.imageshack.us/my.php?image=3dmark06postmf2.jpg]

[/URL]

==========



short description:
300x9 @ 2700Mhz, 1.3V
3-3-7-3 @ 225mhz, 2.9v vdimm

 

[JMWarren]

Thats for the recommendations. I'll give them a go tomorrow.

How would I know if a divider isn't working correctly? Simply instabilit or would CPUz report it?

Could it be the case that I'm Memtest stable yet my memory settings cause instability in windows? In my case this instability can be cured by extra voltage...probably has to do with the on die memory controller?

 

[JMWarren]

Memory Tweaks haven't helped at all. This chip just seems to be a really poor overclocker. Orthos will fail at 2.6 almost instantly unless the voltage is higher than I'd like it to be.
The jump from 2.3 to 2.4 moves the voltage from 1.35 to 1.45 and from 2.4 to 2.5 brings it to 1.5.

Looks like I'll be running it at 2.4 for now and building a new rig sooner rather than later.

 

[Noubourne]

I googled that stepping and I am not seeing good results. Well, not spectacular anyway. Many are reporting sub-2.6Ghz on stock volts. CCBBE 0616 chips almost all do very, very well.

My chip would boot up at 2.8+ on stock volts. These reports are topping out at 2.8 with high volts like 1.5v, and you can bet they're mostly extreme cooling.

0551 appears to be not a very good week overall (not that there aren't good ones, just not a lot).

 

[Makaveli]

ya that is a bad stepping

 

[customcoms]

Thats why I tend to recommend the Opteron 165's: cheaper, better steppings, and all the Opteron's tend to hit the same speeds anyhow; only reason for a faster stock speed is if your board or your memory can't handle higher FSB speeds.