Jump to content


  • Content Count

  • Joined

  • Last visited

About deeppow

  • Rank
    New Member

Profile Information

  • Location
    Los Alamos, NM
  1. Gentlefolk, Your images are not what they're shipping today; power connector setup is different. I can make the connections except I have no 8pin, it's a 4pin. Plus there isn't a Molex now. Can find some discussion on it but noboday bothers to say yes or no when the substitution of a 4pin is suggested. Since I checked here before purchasing, I'm a little irked that your images told me I was fine but now I have to ensure I'll not misconnect the mobo. Thanks for your assistance; greatly appreciate your time.
  2. Here is the basic math associated with the overclocking process, link. That page contains a link to even more equations. The memory timings come in separate from these overclocking equations but of course do have a reasonable effect on the end result. These a pages are within the advanced section of Motherboard, Memory, & CPU - Overclocking Guide. Predicting performance is an exercise in statistics, data collection and analysis off that work. One of the few links around is, link, by hitechjb1.
  3. The SD has been a great cpu for me thus far. You can see my sig below. I've not pushed it really hard yet but will probably run where I'm at most of the time. Good temps (less than 30C underload) with my cooling system. Enjoy and good luck!
  4. VGA shouldn't be an issue initially. Set it to default. What are the default speeds on your CPU and motherboard, 200 or 166? Sounds like your basic approach is right. - set memory ratio low, select 50% to be sure. - starting from the lower default speed of either the CPU or motherboard, increase the speed in ~5Mhz increments. For typical CPUs and motherboards, you should be able to go some way up in FSB speed before you need to increase the CPU voltage. If you can't go 5Mhz either your CPU or motherboard is poor. Anyway, those are some initial thoughts.
  5. I did a huge amount of testing awhile back for the AMD dual channel and single channel setups with various latency settings, 2-2-2-N and 2.5-3-3-M. For 2-2-2, N would typically be 5 or more. For 2.5-3-3, M would be 7 or more. Bottom line was that in dual channel the change in performance was ~3% if you changed from 2-2-2-N to 2.5-3-3-M. Thus for a 200 FSB (or HTT) using 2-2-2-N, you would require a 200*1.03=206 FSB for 2.5-3-3-M to perform at the same level. Systems running 300 HTT (which can be reached these days) would require an increase in HTT to 309 using the higher latencies. Thus a number of the comments above are right on the mark. - you'll see little difference in day to day use - speed can offset lower latencies, particularly when you realize memory that can run 2-2-2 often isn't able to reach the same levels of overclock (such as 300 HTT) as 2.5-3-3. - graphics differences can reach ~6% in performance but as you would guess the results are strongly dependent on your video card.
  6. Several ways to get to the same cpu voltage, i.e. different vid and multipliers can produce about the same voltage --- makes no difference the combo you use. A table of the various combos is here.
  7. You are correct but if he can get started then things get easier. Adding ram timings to the equation makes it a step more complex and you have to start evulating which ram settings, fsb and memory ratio, etc. work best and not just test stability. You want to eat the whole cake at once then here is a fuller link to the art of overclocking, linky.
  8. One way to test this setup (I assume the Barton is locked) is to go into your bios and set the memory ratio to 1/2 or 50%. This will mean the memory won't limit the testing you do and you can determine the max speed your cpu (and chipset) can deliver. Work with the cpu voltage and fsb to determine the fastest you can run the cpu. Test using Prime for stability to ensure your fsb is stable. End result is the max speed your cpu can run for the voltage you're willing to apply. Next you can reset the memory ratio back to 1:1. - If you can't boot, you'll need to lower the fsb or if worse comes to worse then reset the CMOS (reset all the bios setting, a pain but it happens). - If you can boot, run Prime95. If Prime crashes or fails the memory is limiting you and only solution is to raise the memory voltage or decrease you fsb. Keep testing till you find the "sweet spot" for fsb and voltages.
  9. Hope it helps a few overclockers before the topic scrolls off the bottom and falls into the bit bucket. Motherboard, Memory, and CPU Overclocking Guide. The guide is divided into 2 incarnations. One for newbies and the other being more advanced. The default is the newbie version but the more advanced version is easily accessed when the guide opens. New sections related to AMD64 processors and motherboards are under development and initial sections have been added. Automatic translation to a number of languages has been added as a capability. Automatic translation of technical English to other languages is generally poor but it has been added in hopes of providing a basic understanding to others. Enjoy!
  10. Angry, are those the ones under SiliconImage_SATAWinXP_64-RAIDSi3114_64bit_AMD-IA-64_rdvr-1000amd64 Never mind, I found them. Thanks for pointing me in the right direction.
  11. Trying to install Win64 and when I try to read the RAID drivers (downloaded the latest and greatest for the DFI site), I get the message "This disk cannot be read because it contains an unrecognized file system." The DFI download site says I should be able to write the floppy from the 32bit system. Any ideas how to solve the problem? I've tried a couple different floppies so the problem isn't a bad floppy. By the way, the problem is on new nF4 SLI-D mobo with San DIego cpu. WinXP is installed and running fine. Thanks.
  12. Can't get my nF3 250 board to recognize the gaming controller on the Audigy2. It says the controller isn't connected. Search was not successful or at least the words I used. Controller and Audigy2 previsously worked on the nF2 Infinity board so I believe they are OK. Help to fix the problem greatly apprecriated!!
  13. If linky doesn't work for you, go to http://www.mushkin.com/ and pick "Memory Test" near bottom of left hand page nav listing.
  14. Both work your CPU and memory. A failure in Memtest doesn't point particularly at the memory, just that there is some type of problem in the computational chain. Memtest will separate potential OS problems, see linky. EDIT: If linky doesn't work for you, go to http://www.mushkin.com/ and pick "Memory Test" near bottom of left hand page nav listing.
  15. Glad you enjoyed it demon and hopefully you have a better understanding of the issues associated with TIMs. The theory is really quite straight forward but the practice of making TIMs and highly OCed machines is an art form and will most likely remain that way for the forseeable future. A final note on the "synthetic benchmark" since it has been referred to a couple times plus appears in links. A "synthetic benchmark" as referred to here is special hardware built for testing. Why special hardware? To create a controlled environment where more accurate measurements can be made and experiments can be repeated with minimal systematic variation from one test to test. This control is necessary if you want scientific results that can be trusted. For example the fact my heater is twice as big as a CPU can be accounted for in my analysis. Unfortunately not everyone does it right and that creates a significant amount of confusion. Particularly hard on new folks as they begin working and reading in the area. Anyway, good luck and enjoy!! Taos, Los Alamos, Salt Lake City, Grand Targhee (was thin this year but had a few good days), among others.
  • Create New...