Jump to content

Dracula

Members
  • Content count

    215
  • Joined

  • Last visited

About Dracula

  • Rank
    Member

Contact Methods

  1. Dracula

    UNOFFICIAL tREF TABLE THREAD

    Thank you guys.
  2. Dracula

    UNOFFICIAL tREF TABLE THREAD

    ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ To answer your question simply, no. You stated, "If the 1:1 divider allows me to update 200 cells in a given amount of time, then the 5:6 divider should logically allow me to update only 166 cells in the same time. Therefore, I will have to request more frequent updates (less time between refreshes) to update the same number of cells in the same amount of time." AND "...calculations tell me that 166 MHz & 3.9 us, I will need a Tref of 0648 to achieve the same level of refreshment..." Regarding, "...tREF of 0648 to achieve the same level..." That is true, but why do that when you can do the same thing on a 1:1 divider? Especially when 1:1 is the overall best and most stable setting, as "dividers" go. The key word here, is of course, "divider". The DATA BANDWIDTH you see in the left column of the tREF TUTORIAL TABLE is just that, bandwidth, and directly means: amount of data potential per period of time. In other words, MAXIMUM DATA SPEED. The term "divider" means what it means and does not refer directly to "time". It literally means divider, as in part of an equation. "Divider", the way you're using it, is the wrong term to use here. The divider does not literally cause the updating of "200 cells" (1:1 divider) or "166 cells" (5:6 divider). The DATA BANDWIDTH represents the amount of potential data-throughput-cycles that can be executed in a set period of time (usually one second). 200MHz = 200 Million bits of data can be "put through" memory in one second. 166MHz = 166 Million bits of data can be "put through" memory in one second. The sole purpose of a "divider" is to serve us PC Enthusiasts as an additional tool in our quest for ever finely-tuned tweaking. It just puts another spin on the numbers so that we can force things to work in a combination that may otherwise not work. As far as a divider "updating cells", it has little or nothing to do with it, directly speaking. Also, dividers are not proven to REALLY improve or enhance your overall performance to any significant degree either. Regarding the equation you are using: "cycles/second * seconds = number of cycles, OR frequency * period = Tref" AND "(166 * 10^6 cycles/second) * (7.8 * 10^-6 seconds) ~= 1296 (the slight error is from the fact that we're not running at exactly 166 MHz)" AND "...multiplying the desired frequency (refresh data bandwidth) by the period (data bandwidth refresh time) should theoretically provide the correct Tref value..." How can I put this, NO! I have mentioned on several occasions, throughout this thread, that multiplying the DATA BANDWIDTH by the REFRESH TIME only gives you the per category REFRESHED DATA FREQUENCY as is circled in red (2589.6Hz) on the illustrated chart below. IT DOES NOT CALCULATE tREF. How do we know this? Simple, try using your equation for any part of the 1.95µ column and your equation will fall utterly apart. Not to mention that old "equation" you're using does not account for the other tREF setting in the same category. 200 x 3.9 might ape 0780 T but it completely fails to explain the 0016 T that resides in the same category. Now obviously this is not "your" equation, as I have seen it many times in the past. Years ago that equation applied nicely to old technology because it displayed data that generally appeared correct. Since you couldn't "tweak" settings nearly to the degree you can today, it was an acceptable shortcut to "correct-looking" calculations. That equation isn't truly correct; not then or now. It only apes accuracy. I had also explained this here. What it is REALLY calculating is the REFRESHED DATA FREQUENCY. What that means is: * Based on the amount "Refresh Data Bandwidth" in its given column (large vertical numbers on the left-hand side of the chart), which is how much potential data that can be refreshed and is expressed in Hertz (Hz), in a set period of time that is call the "DATA BANDWIDTH REFRESH TIME" (measured in microseconds). In other words, e.g.: 166Hz x 1.95µ = 323.7 Hertz can be refreshed in a 1.95 microseconds period of time. That's all! Very nice post though, LubenThrust. Too bad it seems that you only signed up at Diamond Flower Incorporated-Street for this one post that did nothing more than reiterate a fundamentally worthless interpretation of a grossly misguided algorithm. You demonstrated some excellent reasoning skills however, which would have been very valuable here ! If you have any other questions, please r-e-a-d through this thread thoroughly first then ask a relevant question. Thanks! ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ILLUSTRATION OF HOW THIS TABLE WORKS - DRACULA
  3. Dracula

    dfi venus finally out.

    Well yeah but you're forgetting about the newly discovered 10th planet designated as UB313 , which has yet to receive a colloquial name. The DFI LANParty UT nForce4 SLI-DR "UB313" will be "100% Japan-made, solid Copper electrolytic capacitors with 24 karat, gold plating". It will capitalize on the newer Silicon Image Sil 3132 chip, while being capable of utilizing the latest in XDR RDRAM memory technology; circumventing SDRAM's DDR2 and DDR3 flavors altogether. It will OverClock your processor to 400MHz FSB x 10 CPU multiplier and do much, much more all for a staggering $700 price-tag! Expensive yes , but for those of us who comprise the edgiest of the Xtreme Performers , it'll be "well worth the penny for the performance!" - DRACULA
  4. Dracula

    UNOFFICIAL tREF TABLE THREAD

    Good question MikalCarbine257. Truly, how can someone tell the good tREF settings from the bad, or even the worst from the best? The answer is three fold, generally speaking: TEST EACH SETTING IN YOUR BIOS Using the MemTest86+ v1.65 program in a pre-boot environment, you can observe the amount of data-throughput for each setting. It will be located in the upper left-hand quandrant of MemTest86+'s blue testing screen. By using the settings that mirror each other in having the highest data-throughputs, you can narrow things down a bit by eliminating the poor performers and taking note of the best ones. Now you can move on to point #2. OVERCLOCKING Over Clocking is not only a great means by which to measure the overall performance and stability of your computer but it is also a great way by which to gauge each of your tREF settings. You will need to test ALL of the remaining settings leftover from point #1, individually, if you wish to yield the best results . By Over Clocking your computer as far as it can go and then testing those remaining settings in your OS with software such as ( click on any of the links to download the corresponding program ) : BOINC Manager v5.2.13. CPU Burn-in v1.01 EVEREST Home Edition v2.20 Build 405 [email protected] v5.03 OCCT v0.91 PassMark Performance Test v6.0 PCMark05 PC WIZARD 2006 v1.661 Prime95 v24.14.1.0 SiSoftware Sandra Lite 2005 SR3 Stress Prime 2004 Super PI v1.4 mod World Community Grid Agent v3.0 Build 2844 ( [email protected] ) etc... you should be able to determine which settings are the most stable and/or give the best performance with your particular setup. You can also use the COMPLEX tREF BANDWIDTH Chart I created, as a guide for additional assistance with this endeavor. [*]GENERAL USAGE AND "REAL WORLD" PERFORMANCE Believe it or not, this is the overall most intuitive process, but also the least precise and longest approach. Since you are not gathering data from testing, benching or "folding", you won’t have an exactly metered method by which to determine how one tREF setting is better than another, other than observations made by your applicable "real world" usage. For example, a lot of the members here at DFI-Street are gamers. Over the course of 1 to 3 months, simply pay attention to how well one setting performs while you’re "fragging" your opponents, or whatever else it is you do for your gaming experience. Please note however, that this technique could take a very long time, therefore I really don’t recommend relying on this phase of the process until you’ve at least tried the other two steps thoroughly. ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ By doing the above three steps, you should be able to easily discover which settings are the best. Also, and contrary to popular belief here, these tREF settings in your BIOS have a much larger impact on your computer’s overall performance and “OverClockablility” than some other threads may lead you to think. For example, if you use any of the 4 WORST tREF Settings, generally speaking, you will notice a massive performance drop. That is why I cannot recommend implementing them, for any reason. For your convenience, here again is a list of the 4 BEST and the 4 WORST tREF BIOS settings I have personally found; your results may vary: THE 5 BEST BIOS tREF SETTINGS 0780 1560 3072 3120 4708 (stability) THE 4 WORST BIOS tREF SETTINGS (ones you should definitely avoid) 0016 0032 0064 0128 ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ In the face of all this testing, though it may not always be the best performer, the ‘AUTO’ setting may be the easiest path to take for the majority of you, as it works quit well in most instances and for most applications, yielding solid results while requiring no effort to generate those results. For the slavishly hardcore, ‘Xtreme’ enthusiast however, the absolute cutting-edge of performance is always the end goal; it is for them that this thread was created . Again, and as always, I truly hope this information was helpful in your quest for better gaming, OverClocking and performance. - DRACULA
  5. Dracula

    UNOFFICIAL tREF TABLE THREAD

    Good question MikalCarbine257. Truly, how can someone tell the good tREF settings from the bad, or even the worst from the best? The answer is three fold, generally speaking: TEST EACH SETTING IN YOUR BIOS Using the MemTest86+ v1.65 program in a pre-boot environment, you can observe the amount of data-throughput for each setting. It will be located in the upper left-hand quandrant of MemTest86+'s blue testing screen. By using the settings that mirror each other in having the highest data-throughputs, you can narrow things down a bit by eliminating the poor performers and taking note of the best ones. Now you can move on to point #2. OVERCLOCKING Over Clocking is not only a great means by which to measure the overall performance and stability of your computer but it is also a great way by which to gauge each of your tREF settings. You will need to test ALL of the remaining settings leftover from point #1, individually, if you wish to yield the best results . By Over Clocking your computer as far as it can go and then testing those remaining settings in your OS with software such as ( click on any of the links to download the corresponding program ) : BOINC Manager v5.2.13. CPU Burn-in v1.01 [email protected] v5.03 OCCT v0.91 PassMark Performance Test v6.0 PCMark05 Prime95 v24.14.1.0 Stress Prime 2004 Super PI v1.4 mod World Community Grid Agent v3.0 Build 2844 ( [email protected] ) etc... you should be able to determine which settings are the most stable and/or give the best performance with your particular setup. You can also use the COMPLEX tREF BANDWIDTH Chart I created, as a guide for additional assistance with this endeavor. [*]GENERAL USAGE AND "REAL WORLD" PERFORMANCE Believe it or not, this is the overall most intuitive process, but also the least precise and longest approach. Since you are not gathering data from testing, benching or "folding", you won’t have an exactly metered method by which to determine how one tREF setting is better than another, other than observations made by your applicable "real world" usage. For example, a lot of the members here at DFI-Street are gamers. Over the course of 1 to 3 months, simply pay attention to how well one setting performs while you’re "fragging" your opponents, or whatever else it is you do for your gaming experience. Please note however, that this technique could take a very long time, therefore I really don’t recommend relying on this phase of the process until you’ve at least tried the other two steps thoroughly. ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ By doing the above three steps, you should be able to easily discover which settings are the best. Also, and contrary to popular belief here, these tREF settings in your BIOS have a much larger impact on your computer’s overall performance and “OverClockablility” than some other threads may lead you to think. For example, if you use any of the 4 WORST tREF Settings, generally speaking, you will notice a massive performance drop. That is why I cannot recommend implementing them, for any reason. For your convenience, here again is a list of the 4 BEST and the 4 WORST tREF BIOS settings I have personally found; your results may vary: THE 4 BEST BIOS tREF SETTINGS 1560 3072 3120 4708 THE 4 WORST BIOS tREF SETTINGS (ones you should definitely avoid) 0016 0032 0064 0128 ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ In the face of all this testing, though it may not always be the best performer, the ‘AUTO’ setting may be the easiest path to take for the majority of you, as it works quit well in most instances and for most applications, yielding solid results while requiring no effort to generate those results. For the slavishly hardcore, ‘Xtreme’ enthusiast however, the absolute cutting-edge of performance is always the end goal; it is for them that this thread was created . Again, and as always, I truly hope this information was helpful in your quest for better gaming, OverClocking and performance. - DRACULA
  6. Dracula

    UNOFFICIAL tREF TABLE THREAD

    AND Thanks for the compliment! Try not to think of those tREF numbers you see in your BIOS in terms of "high" and "low", numerically speaking. Those tREF numbers you see in your BIOS represent the duration of the delay between the actual refreshes. High data-throughput, high data-bandwidth and good stability are what you should seek . To date, the best tREF settings I have found are: THE 5 BEST BIOS tREF SETTINGS 0780 1560 3072 3120 4708 (stability) As always, you should experiment with all the different tREF settings in your BIOS to determine precisely which one will work best for your unique setup. I hope this helped ! - DRACULA
  7. Thanks EdLSmith, for the outstanding post ! The only thing, however, is that 1.425 volts VID x 113% Special VID = 1.61025 voltage yield (which is a little low-sided too). The voltage readings displayed in the BIOS always low-ball the true voltage readings. You do have awesome temps for that level of vCORE on air-cooling though. So far my best OverClocked setup that is reasonably stable has been at 3051MHz. Here are those settings again: HIGHEST OverClocked Genie BIOS Settings (15 hours, Prime95 stable with very cold ambient room temperatures) FSB Bus Frequency...................................... - 218 LDT/FSB Frequency Ratio............................. - x 5 CPU/FSB Frequency Ratio............................. - x 14 PCI eXpress Frequency................................ - 116Mhz CPU VID StartUp Value................................ - 1.500v CPU VID Control......................................... - 1.500v CPU VID Special Control.............................. - 104% LDT Voltage Control.................................... - 1.50v Chip Set Voltage Control............................. - 1.50v DRAM Voltage Control................................. - 2.70v DRAM+ .03V If it’s not 3.2V......................... - Enable DRAM Configuration Settings: DRAM Frequency Set................................... - 200=RAM/FSB:01/01 Command Per Clock (CPC)........................... - Enable CAS Latency Control (Tcl)............................ - 2.0 RAS# to CAS# delay (Trcd).......................... - 03 Bus Clocks Min RAS# active time (Tras)........................ - 05 Bus Clocks Row precharge time (Trp)............................ - 02 Bus Clocks Row Cycle time (Trc)................................... - 07 Bus Clocks Row refresh cyc time (Trfc).......................... - 14 Bus Clocks Row to Row delay (Trrd).............................. - 02 Bus Clocks Write recovery time (Twr)............................ - 02 Bus Clocks Write to Read delay (Twtr).......................... - 02 Bus Clocks Read to Write delay (Trwt).......................... - 03 Bus Clocks Refresh Period (Tref).................................. - 1560 Cycles DRAM Bank Interleave................................ - Enabled DQS Skew Control...................................... - Decrease Skew DQS Skew Value........................................ - 0 DRAM Drive Strength.................................. - Normal 4 DRAM Data Drive Strength........................... - Level 3 Max Async Latency..................................... - 07.0 Nanoseconds DRAM Response Time................................. - Fast Read Preamble Time................................... - 06.0 Nanoseconds IdleCycle Limit........................................... - 64 Cycles Dynamic Counter........................................ - Disable R/W Queue Bypass..................................... - 04 x Bypass Max............................................... - 02 x 32 Byte Granularity..................................... - Enable (2 Bursts) I am still endeavoring to break through the 3100MHz ceiling on plain air-cooling, but I've had no success with it as yet. Any tips on what I could do to achieve this goal and remain at least 5 hours Prime95 stable on air-cooling, while keeping the vCORE at or below 1.6 volts? Let me know . " ILLEGITIMIS NIL CARBORUNDUM " = " Don't let the bastards grind you down ." Cool!
  8. Dracula

    UNOFFICIAL tREF TABLE THREAD

    Yes, there clearly is a discrepancy here that still has no solution as yet. As far as if I “worked this out”, for the tREF delay/refresh/frequency categories, by means of some complex mathematical equation, the answer is no. I used programs, as you did, to display the results of settings made from within the BIOS. The COMPLEX tREF BANDWIDTH CHART, on the other hand, is pure math. The one I have currently displayed in Post #1, is based on the A64 Tweaker program. I will make another COMPLEX tREF BANDWIDTH CHART based on MemTest86+, but I will have to recalculate each and every setting, which will be a major time-consuming drag... I am also aware that there have been equations used in the past to calculate which refresh/frequency category the tREF Delay settings should be assigned to, and it is entirely possible that such equations exist for today’s motherboards. To wit, I have yet to find any that are current and applicable to the “Modern MotherBoard”. The equations I have found are outdated and are not based accurately on current technology, which is why I have not displayed those equations here in this Thread. If you have any tREF or other memory equations (other than the already stated 1/F) you’d like to share with us Sharp, or if anyone else has some for that matter, please feel free to post yours here along with the proof. We will all be very happy to see NEW data :nod: ! I am still waiting for a response on that. And I am sorry it has taken so long for a response, but I was doing research and making inquiries. Back on 11-13-2005, 10:49 AM - CST, I wrote a Post (#395) at XtremeSystems Forums addressing this “discrepancy” issue. >>>>> ( http://www.xtremesystems.org/forums/showth...t=37345&page=16) <<<<< In that Post, I address the discrepancy between CodeRed’s A64 Tweaker program and Memtest86+. I asked CodeRed, the author of A64 Tweaker, directly for help on this issue. I even shot CodeRed some Private Messages, but the guy never responded to me. I really don’t know what his deal is and I really don’t care. The fact of the matter is that this disagreement between these programs is obviously more complicated than anyone had initially thought. I will, however, try to respond to you as best I can Sharp! Regarding this inquiry, “You are implying that for a 7.8us refresh to choose 0016 or 0064 but memtest shows a decrease in performance. 0016 = 1605MB/s in memtest 0064 = 2354MB/s in memtest” The answer is true and not true. The reason it IS true is because those tREF Delay settings have the highest data-bandwidth rate, in mathematical terms only. Think of it like a car, Sharp. Even if you know what the “fastest”, “most aggressive” or “broadest” timings are, does not mean they will give you the “best” overall performance. Whenever I have tried the settings you mentioned in your two examples, I get horrible bandwidth readings in MemTest86+ too. The 0016 setting is the worst! Sometimes my rig won’t even post with that setting. Ergo, you should avoid using any of the lowest tREF Delay Settings in each of the four given “refresh categories” (refresh categories = 1.95µs, 3.9µs, 7.8µs and 15.6µs, respectively). THE 4 WORST BIOS tREF SETTINGS (ones you should definitely avoid) 0016 0032 0064 0128 Now I know that some of you here have used one of these four settings with a certain degree of success. Be that as it may, and I would still like to see rock-solid proof of that success, I still recommend avoiding them (a.k.a. – any tREF Delay setting below 200T), because they ARE generally bad! Getting back to the “Car Analogy”, even though it is true that 0016, 0032 and 0064 will yield the highest amounts of data-bandwidth, your computer most likely will not like those timings because they are too aggressive and thereby too incompatible. This incompatibility leads to the poorest data-throughput, thereby leading us to the overall poorest performance. In other words Sharp, “OverKill” can actually, and very logically, windup “killing” real-world performance. Keep that in mind when OverClocking your FSB as well. I certainly hope this information has been useful . “#1. The tREF setting in the BIOS = number of cycles...” Being more specific, it is the number of Clock-Cycles/Ticks (5 nanoseconds, based on a 200MHz FSB) of DELAY. AND “#2. cycles = 1/f...” A FACT, I might add, that I have never disputed. If you read the contents in the uppermost Yellow-Portion of any one of my “enhanced” tREF Tables, you will see that I display this formula along with several included examples. Truly, "there is no conflict" with this particular part of this subject. Thanks for the screen shots Sharp, but in all reality and in this case, a screen shot of a program truly proves nothing at all and is not a good foundation on which to base an argument. Now I am not saying that you are right or that you are wrong about the accuracy of A64 Tweaker’s displayed tREF values because I now feel that there is a strong possibility that it is exhibiting tREF values incorrectly. I have no proof of this, however, other than the discrepancy displayed among these differing programs. That is why I am currently showing two different tREF Tables along with a tREF Discrepancy Table here: What we need is a Master Programmer "extraordinaire" who is familiar with these types of programs, and their codes, to look into this problem for all of us. Just in case anyone was wondering, this "tREF Discrepancy" problem with CodeRed’s A64 Tweaker program is not confined merely to DFI motherboards; it is a problem that spans across and includes all 64bit based boards out there. So this truly is a BIG problem because EVERYONE on the internet is basing their tREF values on CodeRed’s, most likely erroneous, A64 Tweaker program. This is unfortunate not only because it is creating more confusion with RAM Memory timings, which is already a very complex and confusing subject, but it is also vastly spreading MISinformation on a now GLOBAL LEVEL. I don’t now if Code’s program is just reading something else or if all the other programs out there besides A64 Tweaker are wrong, but I find this explanation to be highly UNlikely, however. In short, your pictures aren’t “evidence” of anything right or wrong other than the fact that it shows the now obvious and gnawing discrepancy that exists between A64 Tweaker and all of the rest of the other similar programs. If you have “real” evidence, as in the form of programming code along with a lengthy and detailed summary as to what the problem is/was, please enlighten us Sharp. Otherwise, you and I could sit here for the rest of time displaying pretty pictures of programs that disagree with each other, while still resolving nothing. Excellent! I am a very big advocate of questioning EVERYTHING. It is only through research and questioning that any of us truly learn anything at all. You have very good questions and concerns here Sharp, I just wish that I had an entirely complete answer for you regarding the discrepancy. At the end of the day I feel like I’ve been left holding a bag of questions with no answers in sight. Maybe someone sometime can answer our questions. Until then, keep asking good questions and keep searching for the COMPLETE answer!- DRACULA
  9. Thanks sQuid for the post. It’s good to see others who are interested in experimenting with their computers in outside-the-box ways, yet remain within the realm of reasonability .
  10. These are some interesting Posts from another good Thread on OverClocking that contain some very helpful information, links and programs. I have coalesced these discussions into one Post in order to keep the data congruous and easy to locate. Hopefully you will find some useful details in them. ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ Simple answer, use the Central Brain Identifier (CBI) program. You can download it by clicking on the picture below. If you set the LDT Multiplier to 'AUTO' when OverClocking, it will automatically default to whatever Multiplier is bootable or suitable, whether it is above or below the standard 1000MHz :nod: . ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ If your X2's are having an issue running CBI, there is one other program that I can think of that can also display your LDT as "Hypertransport Link" in real-time; it's called nTUNE. Now before you freak out, I propose that you use it for OBSERVATION ONLY. DO NOT use it to 'tweak' anything. If you use the "NVIDIA Monitor" portion of it, you should be able to see your LDT even when set to 'AUTO'. I would have displayed a picture of it for you guys, but it turns out that NVIDIA Monitor is "camera shy", LOL... If that doesn't work, you can just use the "Info" section of nTUNE to see the LDT as well. I have a picture of my nTUNE's main interface. You can click on the picture to go to the nVIDIA download site/window to obtain this program. Again, as a WARNING, use nTUNE to monitor you system's information only! Here is the info I promised: Go to the bottom of the screen and click "Accept" to download. Click on any of the photos to download the notorious nTUNE. Hopefully THIS will work for you guys. Good luck ! ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ That's because you're thinking in terms of Double Data Rate (DDR). The 945MHz that CBI displays for your rig is in terms of Single Data Rate (SDR). However, when expressed in terms of DDR, your machine is running at 1890MHz Mega-Transfers per Second (MTS), which is 110MHz below the stock DDR 2000MHz MTS. Mine is shown as 1125MHz, that is in SDR. When in terms of DDR my 1125MHz SDR is equivalent to DDR 2250MHz MTS, which is 250MHz above the stock DDR 2000MHz MTS, to which you are referring. While mine is actually OverClocked, yours is really slightly UnderClocked. I hope this information was useful . ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ Programmers who create statistical and measuring software, almost always make the GUI's of their programs display system data in terms of SDR because it is the raw and "real" speed, which is separate from "effective" speed. It's all about accuracy and actuality. As far as your LDT being slightly UnderClocked, that's not really important at all, and just so you all know, your LDT can be safely set between 1400MHz and 2200MHz DDR. Personally, I can run my LDT up to 2280MHz DDR without issue. Once I hit the 2300MHz DDR and above range (1150MHz SDR and above), I begin getting lock-ups and BSoD's. You should be able to run it up to 1100MHz SDR and as low as 700MHz SDR without too many concerns. I would avoid dropping below the 700MHz SDR mark though, as it will become more a hindrance than a help . A possible reason as to why my LDT goes as high as it does, when the multiplier is set to AUTO, may have just a little to do with the BIOS version (704-2 BTA) I am running . Everything else you have said here is very good and reasonable advice, however, you should be able to easily set your LDT to 1100MHz SDR (2200MHZ DDR) without it factoring into any potential stability issues. But, then again, having it slightly UnderClocked really won't hurt anything either. Yeah, I know I have read all the same stuff here at Diamond Flower Incorporated-Street about keeping the LDT below 2000MHz DDR at "all costs", but that is simply not true. As previously stated, I have had mine as high as 2280MHz DDR for months without any problems. I have also tried UnderClocking it when experimenting with new and unstable settings to see if putting it below 1100MHZ SDR or even below 1000MHz SDR would give me more stability and I have never seen it demonstrate any such improvement. Your personal experience may vary however, due to unique equipment and settings. ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ Please share with us your personal experiences with LDT settings and OverClocking. If you have some useful personal data, links or programs that would be beneficial here, as well as for the rest of the DFI-OverClocking-Community, please drop us a post .
  11. This is a re-posting of an original and colossally enormous post. While it still contains all the content from the original, it will not "appear" as it did in its original format. All the areas you see in "code" still contain important data and can still be read (albeit, not too easily); this is the how and the why they are still being included. The purpose for using the "code" on these specific sections of this post is to put less emphasis on them, thereby fostering a "friendlier" environment in this thread. As much as I hate censorship, especially when I’m censoring my own work, it is better than the alternative (deleting the entire thread, again). Hopefully, this version of the original post will suite everyone’s interests more appropriately and garner more valuable input from the entire DFI-Street Community. - DRACULA ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ [INDENT][FONT=Georgia][SIZE=2][COLOR=SandyBrown][FONT=Verdana][SIZE=3][B]THIS [U]NEEDS[/U] TO BE SAID... I want to reaffirm the intent of this thread here. Firstly, I was the one who authored this thread for my purposes and my purposes alone. When was the last time anyone in this thread, other than myself of course, actually took the time to ask the CREATOR of THIS thread what he was after? Not once, is the answer. Well then, let me clarify it for everyone:[/B][/SIZE][/FONT][/COLOR] ****************************** [B][U]”HAD YOU BOTHERED TO ‘R-E-A-D’ MY SIGNATURE...”[/U][/B][/CENTER] [COLOR=Red][B]THERMALS[/B][/COLOR], is the word for the day! [COLOR=RoyalBlue][B]40C[/B][/COLOR] degrees idle and [COLOR=Red][B]50C[/B][/COLOR] degrees load is the rule of thumb for “[B][U]thermal[/U] [U]limits[/U][/B]” when [COLOR=Red][B]Over[/B][/COLOR]Clocking and [COLOR=DarkOrange][B]Over[/B][/COLOR]Volting. Most people out there build their computers with between 3 to 5 total internal drives. Maybe 2 or 3 SATA drives 1 or 2 opticals and a Floppy drive. As thus, they can supply a fairly generous amount of volts to their components without the worry of “[COLOR=Red][B]thermals[/B][/COLOR]” factoring into their little [COLOR=Red][B]Over[/B][/COLOR]Clocking/[COLOR=DarkOrange][B]Over[/B][/COLOR]Volting equation. Ergo, they can run from [B]1.65[/B] to [B]1.80 volts[/B] through their CPU on plain [COLOR=Yellow][B]AIR-Cooling[/B][/COLOR] without it getting too [COLOR=Red][B]HOT[/B][/COLOR]. However, in conjunction with what my signature [I]plainly[/I] says (hence why I created it sooo "[COLOR=Sienna][B][I]anally[/I][/B][/COLOR]" from the very beginning...), I am running a total of [B]11[/B] [U]internal[/U] drives with 2 more opticals on the way in the next month or three (8 SATA’s, 2 Opticals and 1 FDD currently). That’ll be [B]13 INTERNAL DRIVES[/B] [U]TOTAL[/U], when all is said and done! Honestly, do you think that internal drives have NO impact on how [COLOR=Red][B]hot[/B][/COLOR] my case temperatures can get? It [U]GREATLY[/U] effects how high the ambient temperatures are in my case, which in turn effects how [COLOR=Red][B]hot[/B][/COLOR] my CPU will become, especially under load. Where you might run your air-cooled San Diego core at [B]1.7 volts[/B] with 3 to 5 internal drive and walk away with a cool [COLOR=Green][B]47C[/B][/COLOR] to [COLOR=Green][B]49C[/B][/COLOR] degrees under load, I easily hit [COLOR=Red][B]50C[/B][/COLOR] at just [B]1.55 volts[/B] under load. It is because of all the hardware I’ve got in my tight rig. I will NEVER get cool temps like you, under load, with just plain [COLOR=Yellow][B]air-cooling[/B][/COLOR]; this why I’ve been very seriously considering moving over to [COLOR=DeepSkyBlue][B]liquid[/B][/COLOR] or [COLOR=RoyalBlue][B]refrigerant[/B][/COLOR] (yes, I know, “Phase-Change-Cooling”) cooling. The [COLOR=Red][B]Thermals[/B][/COLOR] ARE the killer here. So please, NO ONE “[B][I]order[/I][/B]” me to throw voltages at my CPU that I KNOW WILL NOT WORK WITH MY SETUP! I know my rig better than anyone else ever will. ****************************** THE TRUTH ABOUT “Burn-In” When people use “burn-in” they do it under the assumption that they are increasing the ceiling of performance on their hardware permanently. Although this can be true in most cases, the fact of the matter is that the performance increase you normally obtain in doing so is marginal at best and in many cases, there is NO performance boost at all. In a few instances, there can even be a LOSS in performance delta. Diminished performance upon “burn-in” completion is atypical, however. I’m not telling you this to ‘scare’ you away from burning-in your hardware, because I do it as well, I am telling you this so that you are well informed on the topic. When most people do the “burn-in” on their equipment, they do so immediately after they’ve finished assembling all the components of their system and have loaded and installed all essential and appurtenant software. Having completed the initial setup, they begin the 72 to 96 hour process of “burning-in”. They crank-up their hardware with the highest “postable” OverClock and OverVolt that does not result in a BSoD, and begin running a program like ‘[email protected]’, Prime95’s PrimeNet, OCCT, etc... When I “burn-in” my components, I use a similar method but I do it in 3 phases. First I start with a modest OverClock and OverVolt, 100MHz, burning-in for 24 hours. I then increase the OverClocking and OverVolting another 100MHz, and burn-in for 48 hours. Finally I push it as high as it will go without pulling the BSoD card and “burn” it for 96 hours. Even though it’s longer, I do it this way because I have seen slightly better results over the standard way; at least this has been my personal experience. You must do what you feel works best for you. ****************************** [B][U]MY INTENT AND WHAT IS RELEVANT[/U][/B][/CENTER] [LIST]I WILL be using [COLOR=Red][B]Prime95[/B][/COLOR] as my gauge for success as I [COLOR=Red][B]Over[/B][/COLOR]Clock. If you don’t like that fact, too bad, this is my thread and that IS what I have chosen to do. If [B]Prime95[/B] torture tests immediately go bust on you all the time, then the settings being used ARE NOT STABLE. I will cover this more later.[/LIST] [LIST]I desire to [COLOR=Red][B]Over[/B][/COLOR]Clock my processor to its fullest extent WITHOUT exceeding [B]1.60 volts[/B] at the absolute most. I would not use [B]1.60 volts[/B] continuously anyway so it would be best to stay below that for me. This is MY decision, My thread, Created by me, for MY purposes. If you agree with me great, you'll be a wonderful asset to my thread and you will benefit from it. If you disagree with me, that is not relevant to MY thread or me. I did not make this thread for people to tell me what maximum volts I should throw at my processor if I have already decided what I want as MY set limit. And NO ONE can tell me I'm wrong with my voltage limit because it is NOT WRONG FOR ME. If you have feelings on things that take you in a different direction that fits your style, then by all means pursue that interest; just do so in a different thread.[/LIST] [LIST][U]This[/U] thread is about staying at or below that [B]1.60 volts[/B] threshold, on [COLOR=Yellow][B]AIR-cooling[/B][/COLOR] [B][U][I]only[/I][/U][/B]. If you have something useful to say in these regards, please feel free to tell me and us your personal experiences. That is what I created this thread for in the first place. If you have some unique approach with timings and or frequency/divider combination that you think might help, please list the data. Listing real data is the MOST helpful tool here.[/LIST] [LIST]My ultimate intent is to get the most for my money out of this [B][COLOR=Green]$6000[/COLOR]+[/B] rig, balancing performance keenly with longevity. I [B]do[/B] want to [COLOR=Red][B]Over[/B][/COLOR]Clock and after all, that is why we are all here. I also want my hardware to last more than just one or two years as well. That is why I will only risk it so far. If anyone here can't understand that or is so wealthy that they cannot relate to what I'm saying, simple answer, do not post in MY thread.[/LIST] [CENTER][B][I]A COURSE IN ETIQUETTE[/I]: [U]WHAT IS NOT RIGHT FOR ME AND WHAT IS INAPPROPRIATE HERE[/U][/B][/CENTER] [LIST]NEVER, NEVER, NEVER bully someone by “[U]commanding[/U]” them to run their rig and set-up the way you want. IT IS NOT YOUR COMPUTER! If they ask, respond with respect. If they tell you to stop, have certain guidelines or tell you that they don't think it is right for [U]THEIR[/U] rig, DO NOT ARGUE WITH THEM! It's THEIR machine to do with as they please, even if they really ARE wrong themselves. Even so, you'll never make someone agree with you by forcing rude opinions down their throat when unsolicited. If you cannot refrain from behaving in this manner, not only do I not want you posting in my thread, but you should log out of [COLOR=Orange][B]DFI-Street[/B][/COLOR] right away! Don't bother logging back in until you get your attitude right.[/LIST] [LIST]If [U]you[/U] want to throw [B]1.7[/B], [B]1.8[/B], [B]1.9[/B], [URL=http://demonical.com/wordstoliveby.htm][COLOR=Red][B]HELL[/B][/COLOR][/URL] even [B]2.x volts[/B] at your San Diego CPU then you go right ahead and do it. If you post that kind of OFF-TOPIC nonsense in MY thread and I express that I think that sort of thing is ridiculous, please do not whine, gripe or complain because I expressed MY "[B]freedom[/B]" to disagree with something that I consider foolish for my rig and my equipment and MY thread. You ask for it by posting here. Do you know what though, it really shouldn't be a big deal to you just because my vision differs from yours. As I said before, if you have the money and time to waste on swapping out processors every 2 to 4 months, that is [U]your[/U] prerogative, [U]NOT[/U] mine. I still say, as would the GREAT VAST MAJORITY OF PEOPLE OUT THERE, that such an investment is very [B]UN[/B]wise, if not straight-up [B]stupid[/B], to say the least. However, you can do that if you want because that's your right, no one ever said that it wasn't. But do not expect me or those vast majority of regular users or even us [COLOR=Red][B]Over[/B][/COLOR]Clockers to listen to what you have to say when what you are saying is not [B]remotely[/B] economically sound or viable.[/LIST] [LIST]Cutting people down, making it personal and attacking someone for disagreeing, calling them childish names like "[COLOR=Sienna][B][I]moron[/I][/B][/COLOR]", listing numerous posts in short spans of time that are all solely engineered to stir a wholly negative reaction by degrading that persons character simply because they have a difference of opinion with you in THEIR OWN THREAD, is in NO WAY RELEVANT, CONSTRUCTIVE or ON-TOPIC by any stretch! That behavior accomplishes nothing but making a menace of yourself, and it doesn't teach people anything either. So not only is it a [COLOR=SeaGreen][B][I]dick[/I][/B][/COLOR] move but it's also expressly forbidden by the FORUM RULES! Besides, how can anyone learn in a grossly hostile and hypocritical environment that drives people away from asking good and relevant questions? The answer is obviously that it doesn't. Moreover, boasting about whom you know and who said this or that, and then saying that the other person who never said those things is "arrogant" is the deepest form of hypocrisy yet. You don't chastise someone you assert as "arrogant" and then turn around and talk about all the forums you belong to, all the “handles” (people) you [I]think[/I] you "[I]know[/I]" at said forums, then talk down to the person you're "speaking-at" and then say "Oh, you're arrogant!" Calling others names and calling them arrogant, then behaving that way yourself makes you doubly arrogant. "Practice what you preach!" Furthermore, what you are doing then is not “relevant” or on-topic or free-speech, it is vindictive, belittling vicious HARASSMENT that is OFFENSIVE in the very purest form.[/LIST] [LIST]NEVER, EVER tell someone that their UNIQUE experiences are worthless! Talking down to others like they’re stupid, especially when you really don’t know them at all or anything about them besides one of their threads is a really stupid thing to do! As with the whole program thing, you don’t know what programs I have or know, so it shows NO patience or intelligence on any person’s part to jump down another’s throat before getting even any facts at all. I know all about [B][email protected][/B] and [B]Prime95[/B] and [B]CPU Burn-In[/B] and [B]SuperPI[/B] and [B]OCCT[/B] and [B]Stress Prime 2004[/B] and how [B]MemTest86+[/B] can be rigged to do [B]RAM burn-in[/B]. DO NOT TALK TO ME ABOUT THIS JUNK because I know a LOT about it, as I have all of them and am using or have used them all. For example, [B][email protected][/B] is very similar in its function to [B]Prime95[/B]’s “[B]PrimeNet[/B]”, of which I am an active member. It’s all about an organization giving your CPU on-line work in the form of data crunching. Should the web connection fail some of those programs can continue crunching the data for the project they were on, for a time. In MY experience, [B]PrimeNet[/B] pushes my computer overall harder than [B][email protected][/B] buy utilizing more physical-memory. That has been my own experience. Then there are all the Torture Tests. [B]OCCT[/B] sucks-up the most resources when it’s at the peak of its operation but both [B]Prime95[/B] and [B]Stress Prime 2004[/B] ([I]which is basically the[/I] [B]Prime95[/B] [I]software engine with an enhanced[/I] GUI) is consistently more “peaked-out”. Now if you want to ‘nit-pick’ over every little bit of minutia that’s involved with all these programs just to make me or anyone else look bad then you are a jerk and are wasting everyone’s time. We really don’t care about that unimportant junk and it really is nothing more than a waste of my time and yours. So please, don’t come around acting like you know what you [I]think[/I] you know, acting like you know what I know, because you don’t have a clue. If you want to know something, ask politely and I’ll most likely divulge the information you seek, should I have it.[/LIST] [LIST]I have unequivocally chosen [COLOR=Red][B]Prime95[/B][/COLOR] as my main testing bed for my [COLOR=Red][B]Over[/B][/COLOR]Clocking. Yelling at me for that, calling me names for that and harassing me because of that, is not only childish and idiotic but also shows the rest of us exactly how small minded you are! My personal experience with [B]Prime95[/B] is very extensive and I have found it to be DAMN accurate at uncovering system instabilities. I’ve been harassed and personally attacked just for saying that I like to use this program as a tool that works for me. I’ve been [B][I]ORDERED[/I][/B] to use [B][U]SUPER[/U][/B] [COLOR=Red][B]Over[/B][/COLOR]Clocked and [B][I]non[/I][/B]-[COLOR=Red][B]Prime[/B][/COLOR] stable settings for “REAL-WORLD” apps. No joke? You don’t say... [U]Of course[/U] I [B]do[/B] use my rig in “REAL-WORLD” applications and I have personally found that when my [COLOR=Red][B]Over[/B][/COLOR]Clocked rig fails [B]Prime95[/B], in under 5 hours, my settings “REALLY” [U]are[/U] [B]UN[/B]stable. I can’t even begin to tell you how many crashes have occurred and/or huge amounts of data have been lost because of unexpected crashes due to [B]UN[/B]stable [COLOR=Red][B]Over[/B][/COLOR]Clocked settings. I use Word a lot, Archiving applications, video encoding and raw data storage and some gaming. DON’T TELL ME WHAT I SHOULD USE MY COMPUTER TO DO BECAUSE IT’S NOT YOURS! When I play around with the BIOS settings, to try something new and more “[B][I]risky[/I][/B]” and then I start getting random crashes here and there, I run [B]Prime95[/B]. Sure enough, if I’ve been getting even one crash every week, Prime usually catches that within an hour or two. This is why I'm so "[B][I]slavish-to-the-program[/I][/B]" and trust it and rely upon it so heavily. It is not the only test I use per the big list I delineated for you earlier. Additionally, [B]Prime95[/B] is NOT all encompassing in that it does not tell you the exact nature and/or location of the problem that caused the test to fail. I really wish it could do that, as do many other [B]Prime95[/B] advocates. So don’t tell me not to use a program that has proven itself to me personally many, many times over.[/LIST] The bottom line is that [B]1.60 volts[/B] on [COLOR=Yellow][B]AIR-ONLY-COOLING[/B][/COLOR] is [U]MY[/U] limit in this thread. I want the temps kept at or below [COLOR=Red][B]50C[/B][/COLOR]. It needs to be [B]Prime95[/B] stable for at least 5 to 8 hours. I want my processor that cost [COLOR=Green][B]$1000.00[/B][/COLOR] to last [U]at least[/U] two years. All in all, I am looking for a balance of these worlds. If you can help me fantastic! If your intent is to make personal attacks and argue with what I have laid-out as the [COLOR=Red][B]Over[/B][/COLOR]Clocking foundation for MY personal “[I]safe-zone[/I]”, WHO ARE YOU to tell me I can’t run [B][I][U]LESS[/U][/I][/B] voltage??? If I ramp my processor up to [B]1.8[/B] or [B]1.9 volts[/B] on [COLOR=Yellow][B]AIR-cooling[/B][/COLOR] and leave it like that for an extended period, then it fries in a month or two, are you going to give me [COLOR=Green][B]$1000.00[/B][/COLOR] to buy a new one since you said it was “[I]safe[/I]”? If you aren’t willing to pony-up [B][I]your[/I][/B] cash because you gave me bad advice, then DO NOT ARGUE WITH THE PERSONAL DECISIONS I HAVE CHOSEN AS MY SOMEWHAT-SAFE LIMIT! [B][U]READ[/U] [U]THE[/U] [U]FORUM[/U] [U]RULES[/U][/B] FOR [COLOR=Red][B]Over[/B][/COLOR]Clocking and [COLOR=DarkOrange][B]Over[/B][/COLOR]Volting: "[COLOR=RoyalBlue][B]There IS NO 'right' setting for everyone; there is only a right setting for you![/B][/COLOR]"[/INDENT][/FONT][/SIZE] ~ ~ ~ ~ ~ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ~ ~ ~ ~ ~ P.S. - Don't be so gullible as to believe every screen shot you see some braggart boasting about either. Case in point: ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ WE’VE GOT UPDATES, AND THEY’RE WAY TOO HOT TO HANDLE!!! BEST OverClocked Genie BIOS Settings (65 hours, Prime95 stable) FSB Bus Frequency...................................... - 250 LDT/FSB Frequency Ratio............................. - 4 CPU/FSB Frequency Ratio............................. - 12 PCI eXpress Frequency................................ - 116Mhz CPU VID StartUp Value................................ - 1.425v CPU VID Control......................................... - 1.400v CPU VID Special Control.............................. - 110% LDT Voltage Control.................................... - 1.20v Chip Set Voltage Control............................. - 1.50v DRAM Voltage Control................................. - 2.80v DRAM+ .03V If it’s not 3.2V......................... - Disable DRAM Configuration Settings: DRAM Frequency Set................................... - 200=RAM/FSB:01/01 Command Per Clock (CPC)........................... - Enable CAS Latency Control (Tcl)............................ - 2.5 RAS# to CAS# delay (Trcd).......................... - 03 Bus Clocks Min RAS# active time (Tras)........................ - 06 Bus Clocks Row precharge time (Trp)............................ - 03 Bus Clocks Row Cycle time (Trc)................................... - 7 Bus Clocks Row refresh cyc time (Trfc).......................... - 14 Bus Clocks Row to Row delay (Trrd).............................. - 02 Bus Clocks Write recovery time (Twr)............................ - 02 Bus Clocks Write to Read delay (Twtr).......................... - 02 Bus Clocks Read to Write delay (Trwt).......................... - 03 Bus Clocks Refresh Period (Tref).................................. - 1560 Cycles DRAM Bank Interleave................................ - Enabled DQS Skew Control...................................... - Decrease Skew DQS Skew Value........................................ - 0 DRAM Drive Strength.................................. - normal 4 DRAM Data Drive Strength........................... - Level 3 Max Async Latency..................................... - 07.0 Nanoseconds DRAM Response Time................................. - Fast Read Preamble Time................................... - 05.0 Nanoseconds IdleCycle Limit........................................... - 64 Cycles Dynamic Counter........................................ - Disable R/W Queue Bypass..................................... - 04 x Bypass Max............................................... - 02 x 32 Byte Granularity..................................... - Enable (2 Bursts) ++++++++++++++++++++++++++++++++++++++++++++++++++ HIGHEST OverClocked Genie BIOS Settings (15 hours, Prime95 stable with very cold ambient room temperatures) FSB Bus Frequency...................................... - 218 LDT/FSB Frequency Ratio............................. - x 5 CPU/FSB Frequency Ratio............................. - x 14 PCI eXpress Frequency................................ - 116Mhz CPU VID StartUp Value................................ - 1.500v CPU VID Control......................................... - 1.500v CPU VID Special Control.............................. - 104% LDT Voltage Control.................................... - 1.50v Chip Set Voltage Control............................. - 1.50v DRAM Voltage Control................................. - 2.70v DRAM+ .03V If it’s not 3.2V......................... - Enable DRAM Configuration Settings: DRAM Frequency Set................................... - 200=RAM/FSB:01/01 Command Per Clock (CPC)........................... - Enable CAS Latency Control (Tcl)............................ - 2.0 RAS# to CAS# delay (Trcd).......................... - 03 Bus Clocks Min RAS# active time (Tras)........................ - 05 Bus Clocks Row precharge time (Trp)............................ - 02 Bus Clocks Row Cycle time (Trc)................................... - 07 Bus Clocks Row refresh cyc time (Trfc).......................... - 14 Bus Clocks Row to Row delay (Trrd).............................. - 02 Bus Clocks Write recovery time (Twr)............................ - 02 Bus Clocks Write to Read delay (Twtr).......................... - 02 Bus Clocks Read to Write delay (Trwt).......................... - 03 Bus Clocks Refresh Period (Tref).................................. - 1560 Cycles DRAM Bank Interleave................................ - Enabled DQS Skew Control...................................... - Decrease Skew DQS Skew Value........................................ - 0 DRAM Drive Strength.................................. - Normal 4 DRAM Data Drive Strength........................... - Level 3 Max Async Latency..................................... - 07.0 Nanoseconds DRAM Response Time................................. - Fast Read Preamble Time................................... - 06.0 Nanoseconds IdleCycle Limit........................................... - 64 Cycles Dynamic Counter........................................ - Disable R/W Queue Bypass..................................... - 04 x Bypass Max............................................... - 02 x 32 Byte Granularity..................................... - Enable (2 Bursts) The only reason I was able to "Prime" the 3051MHz setting as long as I did was because I cracked open the window in the room where I keep my rig. Since it is now winter, the ambient room temperature dropped down to about 29F degrees. The CPU, under these "ideal" conditions, never got warmer than 38C degrees at load. Once I closed the window, it took less than an hour for the ambient temperatures to rise significantly enough to cause the Prime95 Test to fail. I have always and consistently noticed that when this CPU's temperature reaches anywhere between 52C to 55C degrees, Prime95, OCCT and Stress Prime 2004 all crap-out, as is to be expected. It is EXACTLY for this reason that I CANNOT run more than 1.6 volts through my CPU at NORMAL ambient room temperatures! If I meet or exceed the 1.6 volts on my processor, it gets way too HOT, not only causing my various Torture Tests to fail, but my rig starts crashing frequently and getting many BSoD's! Though the last test was obviously skewed, it is still a fairly stable setting and would be rock-solid if I had liquid or phase-cooling (which I can't afford right now) at my disposal. Here are a few of the various programs I used to test and validate the latter settings titled "HIGHEST OverClocked Genie BIOS Settings": To go directly to the CPU-Z Validator Database for these settings, click HERE. ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------ Again, thanks for the advice! Since this post I have tried the new FX60TEST Beta BIOS, for the DFI LP nF4 SLI-DR motherboard, in the hope that I could get better OverClocks. Simple answer; no dice! In fact, this is one of the crappiest Beta BIOS I have tried to date! Even the fairly modest and extremely stable settings I used for my 3000MHz OverClock would not properly post with this new Beta BIOS piece of junk! I also tried many other more modest settings and various Torture Tests and all I got from this FX60TEST BIOS were multiple, random crashes and freeze-ups, BSoD's and failures to post. If you'd like to give this Beta BIOS a go for yourself, simply click on any of the crap-brown colored FX60TEST links to download the ".exe" version. ++++++++++ [COLOR=SandyBrown][SIZE=3][B]Like I said at the top of this post, "THIS NEEDS TO BE SAID..." The [I]FINAL WORD[/I] is that this is a [U]SUPPORT[/U] [U]FORUM[/U]! If you are not here for support or to lend it [I][U]graciously[/U][/I], you should [U]not[/U] be posting here at [COLOR=Orange][COLOR=Red]D[/COLOR]iamond [COLOR=Red]F[/COLOR]lower [COLOR=Red]I[/COLOR]ncorporated[/COLOR][COLOR=Red]-Street[/COLOR] at all! And no matter [I]who[/I] you are, [I]who[/I] you think you "know" or [I]what[/I] you think you "know", you [I]really[/I] don't! [U]Every[/U] person's rig, components and setup IS [U]unique[/U] to [U]that[/U] [U]individual[/U], so you can't even possible begin to pretend that you "[I]know[/I]" more about someone else's rig than that rig's owner does, [U]PERIOD[/U]! The best you or anyone else can do is to offer up "[I]suggestions[/I]" or "[I]recommendations[/I]" based on [U]your[/U] [I]personal[/I] and [I]unique[/I] experiences. Do so [I]POLITELY[/I] and with [I]RESPECT[/I]. Thank You [COLOR=Yellow][B]:^)[/B][/COLOR] ![/B][/SIZE][/FONT][/COLOR][/INDENT][/SIZE][/SIZE]
  12. Just so everyone knows, I DID call AMD Tech-Support and asked them if there was an alternate way to find out the "week" or "manufacturing date" of my processor. As expected and as usual, their response was "Duh, I don't know... :tooth: ". I will have a major update coming to this thread this week, including a HUGE post with tons of data. Stay tuned :nod: !
×