Bought hardware from friend, doesn't OC nearly as well for me

[Micand]

Howdy! I recently purchased a used AMD Opteron 165 processor, DFI LanParty UT NF4 Ultra-D motherboard, and two 1 GB sticks of Geil PC4000 RAM from a friend. He had been running the CPU at 2.6 GHz with a 1.4 v vcore (stock vcore being 1.35 v) ever since buying the hardware new a year-and-a-half ago. Limited experimentation on his part suggested he could run as high as 3.0 GHz on that voltage, albeit with compromised stability.

 

Sadly, since installing the processor, motherboard, and RAM myself, I haven't been able to get anywhere near 2.5 GHz. With my vcore at 1.45 v, HTT at 280, and CPU multiplier at 9 for a clockspeed of 2520 MHz, Prime95 fails after one minute. At a vcore of 1.5 v (the highest I dare test at), a HTT of 290 and CPU multiplier of 9, yielding a clockspeed of 2610 MHz, Prime95 crashed immediately upon launching it. Throughout my overclocking adventures, I have kept the memory/HTT divider at 1:2, so my RAM has never been running above 145 MHz, with relaxed timings. Upping DRAM voltage from the default 2.6 v to 2.7 v or 2.8 v yields no improvement. The PCI express bus has remained locked at 100 MHz. I've tried the LDT/FSB frequency ratio both at "auto" (which I understand is a good setting for this board, regardless of HTT) and to numbers yielding the closest approximation to 1000 without exceeding it.

 

The only major change from my friend's hardware configuration is my power supply, a 550 W Enermax EG565V-PE – I'm using exactly the same CPU, motherboard, and RAM that he was. My power supply, as well as all my other hardware, was previously in a system using the same model of motherboard and an AMD64 3200+ 1.8 GHz Venice, which I had been running perfectly stably at 2.78 GHz for the preceding two-and-a-half years.

 

Idle temperatures are around 24 C, only a couple degrees above ambient; load temperatures, even at a vcore of 1.475, never exceed 45 C. I'm using a Thermalright SI-120 heatsink, which I took great pains to lap to a mirror finish; for thermal interface material I'm using Arctic Silver 5, which I applied exactly in accordance with the official instructions. Mounted on the heatsink is a Delta 120 mm fan pushing 130 CFM. I also lapped the chipset heatsink and then applied Arctic Silver 5 to it; the chipset's temperature hovers around 35 C. Cooling throughout the rest of the case is quite ample, particularly since I have my side panel off.

 

The one hazy area is with BIOS revisions – before I began overclocking, I flashed to the latest BIOS provided by DFI, 2006/04/06. After receiving poor results with the official 2006/04/06, I flashed to the official 2005/11/14, but saw no improvement. Both flashes had been performed with WinFlash, and I had neglected to clear the CMOS after flashing, though I had used the “load optimized defaults” function both before and after flashing. In talking to my friend, I learned he had installed a version modified by OCZ Tony using Tmod's bootable BIOS flash CD, but could not remember what version he used. I, of course, overwrote it with my ill-thought-out initial flash. I tried flashing to OCZ Tony's 704-2bta, which I assumed to be very similar to the version my friend had been using, if not exactly the same. To do so, I used Tmod's bootable BIOS flash CD, not WinFlash. Afterwards, I was careful to shut down the system, remove the power cord, remove the CMOS battery, and activate the clear CMOS jumper for ten minutes. Still, upon loading optimized defaults, I saw no improvement in overclocking results.

 

What could possibly account for the discrepancy in overclocking results between my friend and I? Using almost the same hardware, save for a different power supply (which ought to be more than up to the task), why can I not even reach 2.6 GHz stable with a 1.5 v vcore, while my friend achieved it all-but effortlessly with only a 1.4 v vcore? Anything above 2.6 GHz, such as the 3 GHz at which he was able to boot Windows and run Prime95 at, is but a pipe dream. Temperatures seem well within the acceptable range, and while BIOS version discrepancies are a cause for some concern, I saw no improvement moving between two official revisions, as well as a modified one by OCZ Tony. Could the hardware possibly have been damaged during shipping? Given that it was shipped from North Carolina to Alberta, Canada, could some border official have handled it outside of an anti-static bag and caused electrostatic discharge-related damage (even though it sounds quite ludicrous that the damage would manifest itself in so subtle a manner)? Is there anything I might try to obtain the results that my friend did with the very hardware he sold to me?

 

Thanks,

 

Jeff

[bldegle2]

answered this over at sharkies for you....

 

i am leaning towards the PS, or your friend fudged big time with the supposed stable clockage.

 

Ps linkee:

 

http://www.diy-street.com/forum/showthread...ed+power+supply

 

Enermax has a Love/hate going on w/the 600 watt version, maybe the 550 is falling into the same gap?????

 

also, in all this, did you ever talk to the seller???? maybe if he gave you the settings he ran.......since this is a complete package purchase, it would seem the prudent thing to do...i am assuming it prolly shipped with the settings he was running, unless he cleared the CMOS before shipping....

 

laterzzzz.....

[Sharp]

What settings are you using?

EG.

GENIE BIOS Setting

>DRAM Configuration..................Press Enter = New Menu

FSB BUS Frequency....................300

LDT/FSB Frequency Ratio..............X 3.0

LDT BUS Transfer Width...............16 16

CPU/FSB Frequency Ratio..............X 10

PCI eXpress Frequency................100Mhz

K8 Cool 'n' Quiet Support............Disable

Cool 'n' Quiet MAX FID...............AUTO

CPU VID StartUp Value................1.350 V

 

CPU VID Control......................AUTO

CPU VID Special Control..............AUTO

LDT Voltage Control..................1.20 V

Chip Set Voltage Control.............1.50 V

DRAM Voltage Control.................2.70 V

+0.3v if not 3.2v....................Disabled

 

Run MemTest86+.......................Enabled

 

GENIE BIOS Setting >DRAM Configuration

DRAM Frequency Set (MHZ).............166 (DRAM/FSB:5/6)

Command per clock (CPC)..............Enabled

Cas latency (tCL)....................3

RAS to CAS delay (tRCD)..............3

Min RAS active time (tRAS)...........5

Row precharge time (tRP).............2

Row cycle time (tRC).................07

Row refresh cycle time(tRFC).........12

Row to Row delay (tRRD)..............2

Write recovery time (tWR)............2

Write to read delay (tWTR)...........1

Read to write delay (tRWT)...........2

Refresh period (tREF)................2064

Write CAS latency (tWCL).............N/A

Odd Divisor Select...................Disabled

DRAM Bank Interleave.................Enabled

 

DQS Skew Control.....................Decrease

DQS Skew Value.......................0

DRAM Drive Strength..................Level 8

DRAM Data Drive Strength.............Level 3

Max Async Latency....................8ns

Dram Response........................Fastest

Read Preamble Time...................6ns

Idle Cycle Limit.....................x16

Dynamic Counter......................Enabled

R/W Queue Bypass.....................X4

Bypass Max...........................X4

32 Byte Granularity..................Disabled (4 bursts)

[Micand]

answered this over at sharkies for you....

 

i am leaning towards the PS, or your friend fudged big time with the supposed stable clockage. ...

 

Thank you for both responses. I'll respond over here as well, for the benefit of other potential saviors. It seems that the warning in the thread you linked to is specific to the Enermax Liberty 620W -- several other Enermax PSUs are on the recommended list, including the 600W EG701AX-VE, whose model number suggests it is similar to my EG565P-VE. A search for EG565P-VE on these forums yielded a number of people who've been running it in similar setups quite well. Additionally, I've been using the same PSU to power my 3200+-at-2.78 GHz since May 2005. Unfortunately, I don't have any other PSUs readily available to test with.

 

I asked my friend regarding the settings he used, but he could only remember CPU clockspeed (2.6 GHz) and vcore (1.4 v), as he had not used the hardware for three months prior to selling it to me, and had not played around with the BIOS for some time prior to that. Between the three months it spent inactive and the month it took for me to receive it after he shipped it, by the time I installed it the BIOS settings had reset to defaults.

 

What settings are you using?

 

Thank you for your response. After loading optimized defaults, I left all overclocking-related settings at default, save for FSB Bus Frequency, DRAM Frequency Set (MHz), LDT/FSB Frequency Ratio, CPU/FSB Frequency Ratio, and CPU VID Control. Currently, my (quite unstable) settings are as follows:

 

GENIE BIOS Setting

>DRAM Configuration..................Press Enter = New Menu

FSB BUS Frequency....................280

LDT/FSB Frequency Ratio..............X 3.0 (tried X 2.0, auto)

LDT BUS Transfer Width...............16 16

CPU/FSB Frequency Ratio..............X 9

PCI eXpress Frequency................100Mhz

K8 Cool 'n' Quiet Support............Disable

Cool 'n' Quiet MAX FID...............AUTO

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

 

CPU VID Control......................1.475 V (began at 1.35 V [stock], worked way up in 0.025 V increments -- still unstable at 280*9 MHz, 1.475 V)

CPU VID Special Control..............AUTO

LDT Voltage Control..................1.20 V

Chip Set Voltage Control.............1.50 V

DRAM Voltage Control.................2.60 V (tried 2.70 V and 2.80 V)

+0.3v if not 3.2v....................Disabled

 

Run MemTest86+.......................Enabled

 

GENIE BIOS Setting >DRAM Configuration

DRAM Frequency Set (MHZ).............100 (DRAM/FSB:1/2)

Command per clock (CPC)..............Enabled

Cas latency (tCL)....................AUTO

RAS to CAS delay (tRCD)..............AUTO

Min RAS active time (tRAS)...........AUTO

Row precharge time (tRP).............AUTO

Row cycle time (tRC).................12

Row refresh cycle time(tRFC).........24

Row to Row delay (tRRD)..............3

Write recovery time (tWR)............3

Write to read delay (tWTR)...........2

Read to write delay (tRWT)...........3

Refresh period (tREF)................3120

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........................Normal

Read Preamble Time...................Auto

Idle Cycle Limit.....................256 Cycles

Dynamic Counter......................Disabled

R/W Queue Bypass.....................16 X

Bypass Max...........................04 X

32 Byte Granularity..................Disable(4 bursts)

 

Any additional assistance will be much appreciated.

[Sharp]

Any idea what chips are on your memory modules?

 

A starting point, what happens with these settings?

 

GENIE BIOS Setting

>DRAM Configuration..................Press Enter = New Menu

FSB BUS Frequency....................280

LDT/FSB Frequency Ratio..............X 3.0

LDT BUS Transfer Width...............16 16

CPU/FSB Frequency Ratio..............X 9

PCI eXpress Frequency................100Mhz

K8 Cool 'n' Quiet Support............Disable

Cool 'n' Quiet MAX FID...............AUTO

CPU VID StartUp Value................1.4V

 

CPU VID Control......................1.4 V

CPU VID Special Control..............AUTO

LDT Voltage Control..................1.20 V

Chip Set Voltage Control.............1.50 V

DRAM Voltage Control.................2.70 V

+0.3v if not 3.2v....................Disabled

 

Run MemTest86+.......................Enabled

 

GENIE BIOS Setting >DRAM Configuration

DRAM Frequency Set (MHZ).............166 (DRAM/FSB:5/6)

Command per clock (CPC)..............Enabled

Cas latency (tCL)....................AUTO

RAS to CAS delay (tRCD)..............AUTO

Min RAS active time (tRAS)...........AUTO

Row precharge time (tRP).............AUTO

Row cycle time (tRC).................11

Row refresh cycle time(tRFC).........14

Row to Row delay (tRRD)..............AUTO

Write recovery time (tWR)............AUTO

Write to read delay (tWTR)...........AUTO

Read to write delay (tRWT)...........AUTO

Refresh period (tREF)................1560

Write CAS latency (tWCL).............N/A

Odd Divisor Select...................Disabled

DRAM Bank Interleave.................Enabled

 

DQS Skew Control.....................Decrease

DQS Skew Value.......................0

DRAM Drive Strength..................AUTO

DRAM Data Drive Strength.............AUTO

Max Async Latency....................8ns

Dram Response........................Normal

Read Preamble Time...................5ns

Idle Cycle Limit.....................AUTO

Dynamic Counter......................Enabled

R/W Queue Bypass.....................AUTO

Bypass Max...........................AUTO

32 Byte Granularity..................Disabled (4 bursts)

[Micand]

Any idea what chips are on your memory modules?

 

A starting point, what happens with these settings? ...

 

Thanks for your response. Alas, I do not know what chips are on my Geil PC4000 memory modules, for heatspreaders are applied to it. With the settings you suggested, my results are the same as with the following, which were suggested on the HardForums by Sta!nless:

 

DRAM Frequency Set (MHZ).............133 (DRAM/FSB:2/3)

Cas latency (tCL)....................3

RAS to CAS delay (tRCD)..............3

Min RAS active time (tRAS)..........8

Row precharge time (tRP).............3

Command per clock (CPC)..............Disabled

 

Using either those settings or the ones you posted, with a CPU vcore of 1.4 v, two instances of Prime95 run concurrently fail instantly at 276*9=2484 MHz, within 8 minutes at 272*9=2448 MHz, and are stable for 12 hours at 270*9=2430 MHz. With a vcore of 1.5 v, at 280*9=2520 MHz, two instances of Prime run for 19 hours without erroring, but fail within one minute at 285*9=2565 MHz.

 

Any additional assistance will be much appreciated.

[Sharp]

Read Preamble Time...................increase this to 6ns.

 

What are the temps when at load for the CPU, Chipset and PWM?

[Micand]

Read Preamble Time...................increase this to 6ns.

 

What are the temps when at load for the CPU, Chipset and PWM?

 

Thanks for your suggestion. No improvement was evident after increasing Read Preamble Time from 5 ns to 6 ns. With a 1.4 v vcore, the CPU is still limited to 268*9 = 2412 MHz; with a 1.5 v vcore, it is only able to reach 280*9 = 2520 MHz.

 

While running an instance of Prime95's In-place large FFTs stress test on each CPU core, at 268*9 = 2412 MHz on a 1.4 v vcore, the CPU's temperature is 37 C, the chipset's is 34 C, and the PWM's is 40 C. While idling at the same clockspeed and vcore, the CPU's temperature is 27 C, the chipset's is 33 C, and the PWM's is 37 C.

[Sharp]

Temps are fine.

 

You could try using just one memory module in the top orange slot.

This will decrease the CPU's work load.

 

If that doesn't help I am out of ideas.

 

Do you have all 4 power connectors plugged into the board?

Now would be a good time to try another power supply

[Micand]

Temps are fine. ...

 

Thanks for your continued help. I tried using the 2x 512 MB Samsung RAM modules (TCCC chips) that had allowed my Venice-based Ultra-D system to reach 2.78 GHz, with both in the orange RAM slots on the motherboard; afterwards, I tried with only one of the Samsung RAM sticks, in the top orange slot as you suggested. In neither case was any improvement evident.

 

All four power connectors are indeed plugged in. I've tried taking the system completely apart and then reassembling it, including reseating the CPU and reapplying thermal compound, but to no effect.

 

As a last-ditch effort, within the next week or two I shall try using a 400 W Sparkle power supply in place of my 535 W Enermax, and report back here once I have.

[bldegle2]

if you got the complete setup your friend had, CPU, MB, ram, and you can't get the clockage, then you have two possibilites that i can see....

 

one, the PS is not up to the task....personally, I do not like Enermax PS's....

 

or two, your friend was puffing up the performance figures.....

 

laterzzzz......

[red930]

Hi

 

Can you make a sig for us with all of your hardware. If it's a power issue, it might be easier to say "oh, it's definitely the PSU" if we know what else you've got in there.

 

PS. Don't try the sparkle...that's not going to get you anywhere.

PPS. Which power supply are you running? 550 or 535W? How old is it?