Jump to content

z3scott

Members
  • Content Count

    181
  • Joined

  • Last visited

Everything posted by z3scott

  1. On air (TR 948 and 80mmx32mm fan) I got mine to 2.5. With water it max's at 2.62 with my GSkill 4400LE 1GB (2x512Mb) kit and a heck of a lot of voltage p95 stable.
  2. I've been running 2Gb of OCZ PC4000 p/n OCZ5001024EL GXT w/o issues @260 3-4-8-3 1T no issues for a few months
  3. 9x bug affected some of us early users with certain cpu's possibly certain ram (TCCD based w/Brain Power boards was a conjecture) memtests fine at 10x and 8x multi but no go w/o errors on the 9x. Nv 6800's in different flavors seem to be problematic in some cases mine works just fine but many have had issues. Mine has been solid as can be except for the 9x issue.
  4. I've read a bit with peeps having some issues with Corsair ram. You will need to do some homework and tweaking to get them dialed in first. Likely having ram issues not cpu voltage issues. What bios are you on?
  5. I think you are a bit aggresive on the voltage starting out especially with air cooling albeit a decent solution (Zalman). You shouldn't need that kind of voltage to get a moderate overclock. May not even need to bump the volts to run 2300 even more. You should test how far you can go with very minimal volt increase first. Temps will rise along with the frequency even without raising the voltage. Have you checked out the ram fully using memtest per the guide? Or are you running a divider right now? Super Pi 32M will stress the system a little but Prime 95 Large FFT will generate max heat on the processor. I'd drop the voltage down and try that and see where the temps are.
  6. http://www.dfi-street.com/forum/showpost.p...431&postcount=2 This shows pics of the Genie Bios section you should be in to change voltages. The nominal CPU VID on the 148 opties is 1.3-1.4v. Your system may be in Auto. You will have to select it and the optional voltages will show. The CPU VID Special Control allows you to select a % additional to the base cpu voltage you select in the CPU VID ie 1.35v CPU VID with a 110% CPU VID Special would be 1.35v x 110% ergo 1.485v. It would be helpful to know what stepping your Opteron is. Those are the letters and numbers marking the processor IHS. You ought to try running nominal or with just a slight bit additional voltage to get your feet wet first and see what temps you get. These opties seem to do pretty good with default volts for a ways. Maybe 1.3v 110% or 1.325v 110% for 1.4575 and see how your temps are. My 148 ran 2750 at 1.4v w/o a problem. To change your HTT/FSB it is the 2nd item in the Genie Bios FSB Bus Frequency. If your going to test your proc speed as the OC Guide notes drop your ram speed by selecting a ratio in the DRAM section DRAM Frequncy Set
  7. Make sure you drop the LDT/FSB multi when you get up around 270 again probably 2.5 since 3x270 nets you 810 which may or may not be stable, (if you didn't this time). Glad to hear your back at it. These 250 Gb's can take a lot of punishment :nod:
  8. Running currently 2.7v + enabled: DRAM Frequency Set - 200=RAM/FSB:01/01 Command Per Clock (CPC) - Enable CAS Latency Control (Tcl) - 3.0 RAS# to CAS# delay (Trcd) - 03 Bus Clocks Min RAS# active time (Tras) - 08 Bus Clocks Row precharge time (Trp) - 02 Bus Clocks Row Cycle time (Trc) - 08 Bus Clocks Row refresh cyc time (Trfc) - 16 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 Write CAS Latency (Twcl) - Auto DRAM Bank Interleave - Enabled DQS Skew Control - Auto DQS Skew Value - 0 DRAM Drive Strength - Level 6 DRAM Data Drive Strength - Auto Max Async Latency - 8.0ns DRAM Response Time - Normal Read Preamble Time - 5.5ns IdleCycle Limit - 256 Cycles Dynamic Counter - Disable R/W Queue Bypass - 16x Bypass Max - 7x 32 Byte Granularity - Disable(4 burst) Memtested clean up to 277 with a low multi but no go at 11x past 265. Stable in all games and benchies.
  9. The bandwidth gains don't really buy you anything as you can see. Proc speed is paramount. Maybe some of the 3d bench jockeys will chime in. If you take a look at the OC Database postings you'll get a better idea of that. Not sure why the upper bandwidth levels seem to decrease your 3d scores though at the same proc speeds. :confused:
  10. Hey S_SS, sorry can't help you out on the ram direction. Not familiar with those sticks. Maybe you've done some searching here and in the Overclock Database for anyone using the Corsair? Another possibility is your psu is getting stretched thin at this point. What are the amperage ratings on it, especially the 12v? One other possibility is the cpu is a weak one. You have certainly pumped some volts to it so far though.
  11. If addressing my comments, you approved the first db entry at the time with no comment regarding the p95 shot. Technically the "Prime95 minimum 8 hours" rerquirement is met/exceeded. I have duly noted your requirement for stopping prime in any future db entries.
  12. "CPC" disabled will give you 2T good luck with 4 value sticks.
  13. I think ole ms got some rotterdam smoke in his eye's... lol. If he is counting my two post's #20 & #21 he just needs to look at the Task Manager showing them running past 12 hrs and 9 hrs. Took the shot's and left p95 running. Poor guy still fighting the prime battle eh?
  14. You'll likely need to tweak some timings beyond the basic's to get yourself running 1T. I'm not up on those ram sticks so you should search in this NF4 thread and the Overclocking forum using the OCZ description (part number). See if anyone is running them and any settings. You can also take a look see at the OCZ support forums Bleeding Edge: http://www.bleedinedge.com/forum/showthrea...70046#post70046
  15. CZ, the guide is one of the best and most thorough breakdown's of how to go about the job at hand. The OC database here is also a great reference to check out settings others have had success with. You really need to go over them many times to get a handle on things. Getting 2.8 means you need to get your FSB/HTT up to 255 from it's default of 200 ([email protected]). Post #3 & #4 in the guide explain how to go about it by lowering your ram speed to eliminate it (or at least minimize it) from the testing for max cpu speed. All those settings are outlined and in your Genie bios. I've got a new rig on my bench now with the 148 and it seems to be a good one. I don't have windows loaded yet so no true stability testing but mine hit 3102 MHz running memtest with CPU vid at 1.425 Special 104% with the stock retail HSF. Ran 11x250 with default voltage. I'd try staying at default volts until you get a better handle on the settings but follow the guide and you can get quite far with not too much tweaking. There is no magic bullet "couple of settings". Takes some time and playing.
  16. The mechanism of destruction is electromigration primarily over gradual periods of time unless you "blow" it out in one fell swoop with a major volt shock such as linked to earlier. Google electromigration and you'll get info on the phenomena. http://www.semiconfareast.com/emig.htm http://www.csl.mete.metu.edu.tr/Electromigration/emig.htm http://en.wikipedia.org/wiki/Electromigration Failures can be a result of a void (open circuit) or when enough conductive atoms migrate a great enough distance they can short two normally separate but close circuits. The substrate materials used also play a part in their ability to minimize this impact. SOI materials were needed to make the move to the smaller dimensional circuit topologies we are now enjoying the benefits of partially because of this. Controlling your temps does help but doesn't as Drac eloquently points out eliminate the problem. Part of the temp increase is due to the atoms themselves getting "excited" and "jostling" themselves creating additional heat not just from the pure voltage/current applied to the circuit. The folks at AMD aren't going to release data even if they have some available and that data wouldn't be able to reflect the longer term cycles we as users and overvolters would truely need to determine a benefit analysis of x voltage/x temp/x time cycle. They would be doing some sort of accelerated thermal/voltage testing which must incorporate projections not real life plug it in and run it for 3 years at 30% over nominal voltage. The EOL (end of life) cycle and next gen proc cycles are too short. Rest assured they have headroom or we would all have dead proc's on our hands wouldn't we? lol I run mine 24/7 to eliminate thermal cycling stress from contributing to the problem as well. Sharing info on succesful oc results helps but doesn't mean your proc will run x volts like jb's proc. Bottom line is anything beyond stock decreases a chips chance of making past it's warranty period. If you can't afford the pain of blowing one up don't push it. :shake:
  17. Hey folks! I'm joining the NF4 crowd here with #2 rig in sig (details to follow.. will update when components come). Building for my son as an X-mas/birthday combo. Started my link file for Ultra-D threads and stickies but if any of you care to toss some in here to gander at would be appreciated. I have plenty of time to read up before tackling things. Will be looking at the database etc but curious what you folks think of the option of building with the GSkill vs OCZ ram I have to work with, (already own both sets). Thanks in advance
  18. Running a single Raptor Sata and a single Pata drive here with the original drivers off the DFI disk and no Nvidia ide driver.
  19. If your temp isn't over about 50 c and things are stable don't worry about it. you will drive yourself nuts watching volts in MBM5, Smart or any other windws based proggie. if you are stressing it and it doesn't crash on you it s/b fine. My 12v rail according to MBM5 runs from 12.14v to12.47v (this isn't even priming it) w/o probs. Does it BSOD, lockup, drop to desktop on you? If not don't get workred up about it.
  20. Betting moneys on the psu... maybe the Nvidia Ide driver but likely psu issue.
  21. Way to go Nikki. Hope it stays clean for you. Sounds good so far :cool:
  22. Bodhaguru, thank you. There alot of good folks here. Have some fun with the system for some time. Play some games, video, whatever floats your boat Then you have some experience to reference performance when you do overclock. You likely can easily get your proc speeds up to 2400 and higher without breaking a sweat eventually. OCCT takes abot 30 minutes to run. 3dMark depends on the version, '01, '03 & '05. You can set it loop (repeat) say 5-10 times for a quick stress (5 minutes per loop approx) and they are nice because it will work the vid card also. P95 you can set up to run overnight try "Blend" and "Large FFT". It won't stop unless it errors or crashes. Blend works the ram & cpu. L FFT will generate maximum heat and stresses the cpu. You can use the system while running Prime albeit it will be sluggish. EDIT: http://www.dfi-street.com/forum/showthread.php?t=28049 This is a downloadable pdf version of the DFI Overclocking Sticky. You should download and read it a few times. Keep it as a reference. It will help you on the next leg of the journey
  23. Not sure how you got the load temp in the bios? I don't trust any readings to be exact anyway. It's common to see a difference from bios to windows temps with the windows being lower (due to power mangagement absent when in the bios) as you are seeing; although the difference normally isn't quite that large from what I've seen. Mine is 30-31 C in the bios currently and was 27-28 C in windows when I rebooted to the bios (water cooling in a cool room). General rule of thumb is 50 C (windows) thereabouts when stressing the system. You will want to watch your other temps as well (PWMic & System). Good airflow through the case is important.
×
×
  • Create New...