No announcement yet.

Gigabyte G41M-Combo Vcore BIOS setting does not work

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

    Originally posted by profJim View Post
    Nice!!
    Higher end Gigabyte LGA775 motherboards often overvolt the Cpu Termination (aka VTT) and Cpu PLL bios settings with high overclocks when these settings are set to AUTO.
    I can confirm that this board behaves like this as well. I was about to post about my Vcore being set higher than my pin modded VID when Vcore is set to "Auto". I just did not understand why. I wish that they 'd at least document such features (or do they?). Only after reading your reply was the mystery solved but hey, I guess that's also a way to learn.

    I figured the voltage increase depended on the base clock increase so to check and verify the Vcore "Auto" setting, I tried 266 which yielded a voltage of 1.400V and 333 which yielded a voltage of 1.450V. Obviously these values are based on my pin modded VID of 1.325V.

    I honestly would like to see what Vcore values would be produced by "Auto" with my E5200 not pin modded at a stock VID of 1.150V. I think I did not try that configuration when I was doing my initial tests, that is overclock the CPU with Vcore set to "Auto". I was focused on setting a specific value which the board didn't do. I am indeed curious to find out what values "Auto" would yield, but I am not willing to reverse my pin mod for that, also removing and reinstalling the cooler is a hassle. I believe "Auto" would provide increased values but they would be much more coarse grained, in bigger steps. Also uncontrolled, which could probably mean higher than necessary (or lower than necessary, but with much less probability).

    Obviously, taking the board's Vcore "Auto" behavior into account, there is a risk involved when pin modding a CPU for VID, especially if the mod is for a much higher value than stock. One could argue that using "Normal" is predictable and safe, but a CMOS clear or loading defaults will set Vcore to "Auto". If the mod is too high your CPU may get fried. In my case, voltage maxed out at 1.450V with a high overclock, as mentioned earlier.

    So after all, concerning Vcore, the board does not correctly set specific Vcore settings, but does have a nice auto overvoltage feature when base frequency is increased from stock.

    Originally posted by profJim View Post
    I didn't see a Cpu PLL setting in your motherboard user manual. Higher Cpu Termination settings often help with system stability, but it's recommended that that VTT should be at least 0.05 volts lower than the Cpu Vcore setting. I don't know if ET6 displays the VTT voltage settings with your motherboard.
    I believe that upon overclocking VTT is also increased accordingly when set to "Auto", but there is no corresponding sensor available to verify its actual value.

    Another weird occurrence concerning "Auto": DIMM voltage setting. BIOS detects a 1.500V value, and this is the setting that is used when "Normal" is selected. Upon selecting "Auto" however, the voltage used is 1.600V regardless of overclocking. I believe this is within specs, it is just very unnecessary with 1.35V memory.
    Last edited by blacktruth; 09-08-2016, 12:19 PM.

    Comment


    • #17
      Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

      With all AUTO bios settings, this is what ET6 --> Tuner --> Advance --> Voltage looks like with my Q9550 @ 3.91GHz + EP45-UD3P motherboard:



      I don't recall if all of the voltage settings were set to AUTO, but Vcore, Cpu PLL and MCH Core were purposely set to AUTO.
      CPU PLL is badly overvolted at 1.7500 volts, the highest recommend setting is less than 1.6 volts.
      VTT (CPU Termination) is set higher than the CPU Core voltage.
      CPU Vcore is close to optimal.
      Dram Voltage is too high, but my DDR2 memory is rated for up to 2.20 - 2.30 volts if needed.
      MCH Core = 1.30 volts is too high, with my 3.825GHz setup, AUTO sets MCH Core = 1.20 volts.

      AUTO settings are often somewhat (or significantly) higher than necessary to help ensure that your system will boot.
      As you can see from my ET6 screenshot, in some cases the AUTO setting is dangerous.

      With my setup, all AUTO settings were stable from the 2.83GHz stock settings up to 3.20GHz, higher overclocks required additional bios tweaks.

      I've included my stable 3.825GHz, 8GB of memory at 1080MHz full screenshot at the end of this post.

      Manually set VTT to a higher value than the default 1.20 volt setting and retest your system for stability, but don't run IBT.
      You want the lowest VTT setting that is stable with your 3.60GHz overclock.


      Last edited by profJim; 09-08-2016, 03:10 PM. Reason: typo
      Q9650 @ 4.10GHz [9x456MHz]
      P35-DS4 [rev: 2.0] ~ Bios: F14
      4x2GB OCZ Reaper PC2-8500 1094MHz @5-5-5-15
      MSI N460GTX Hawk Talon Attack (1GB) video card <---- SLI ---->
      Seasonic SS-660XP2 80 Plus Platinum psu (660w)
      WD Caviar Black WD6401AALS 640GB (data)
      Samsung 840 Pro 256GB SSD (boot)
      SLI @ 16/4 works when running HyperSLI
      Cooler Master 120XL Seidon push/pull AIO cpu water cooling
      Cooler Master HAF XB computer case (RC-902XB-KKN1)
      Asus VH242H 24" monitor [1920x1080]
      MSI N460GTX Hawk (1GB) video card
      Logitech Z-5500 Digital 5.1 Speakers
      win7 x64 sp1 Home Premium
      HT|Omega Claro plus+ sound card
      CyberPower CP1500PFCLCD UPS
      E6300 (R0) @ 3.504GHz [8x438MHz] ~~ P35-DS3L [rev: 1.0] ~ Bios: F9 ~~ 4x2GB Kingston HyperX T1 PC2-8500, 876MHz @4-4-4-10
      Seasonic X650 80+ gold psu (650w) ~~ Xigmatek Balder HDT 1283 cpu cooler ~~ Cooler Master CM 690 case (RC-690-KKN1-GP)
      Samsung 830 128GB SSD MZ-7PC128B/WW (boot) ~~ WD Caviar Black WD6401AALS 640GB (data) ~~ ZM-MFC2 fan controller
      HT|Omega Striker 7.1 sound card ~~ Asus VH242H monitor [1920x1080] ~~ Logitech Z-5500 Digital 5.1 Speakers
      win7 x64 sp1 Home Premium ~~ CyberPower CP1500PFCLCD U.P.S
      .

      Comment


      • #18
        Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

        Wow awesome stuff there.

        Honestly, it would drive me nuts having all those parameters available for tweaking.

        I just had a OCCT/FurMark failure at 3.665GHz (349MHz) with my 1.325V pin mod. I thought that frequency was stable. At the moment I am retrying the test with a VTT=1.300, the next higher setting. This breaks your 0.050V difference rule, but I am trying it anyway. I am considering testing a Vcore of 1.350V with a VTT of 1.300V to see what it would give performance wise and temperature wise.

        About ET6: I am probably asking the obvious, but what do the two columns "Power on" and "Target" indicate? I mean where do the values come from? I know that I can change a parameter on the "Target" column, press Set and change a setting at OS run time.

        Another question: When you find your overclock's limits, the point that it starts to fail, do you back up frequency wise? If yes, what is your rationale/strategy for ensuring absolute stability?
        Last edited by blacktruth; 09-08-2016, 01:47 PM.

        Comment


        • #19
          Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

          Originally posted by blacktruth View Post
          Wow awesome stuff there.
          Yep, I'm just full of it, that is, awesome stuff. At least that's what I like to think when people say that I'm full of it.

          From what I've seen, ET6's Power On column displays the default bios settings for each setting that allows adjustments and the Target column shows the applied voltage settings for each displayed setting.

          When I reach an overclocking limit, I always "retreat" to a stable setting that is several hertz lower. For example, my memory is stable up to 1092MHz {455 x 2.4) with 5-5-5-15 settings and with Performance Enhance = Turbo. I don't like to run at the ragged edge of stability, especially when my hardware is almost 10 years old. With my setup, the difference in performance would be minimal with:
          Cpu = 3.825GHz (8.5 x 450) + memory = 1080MHz (2.4 x 450) versus
          Cpu = 3.868GHz (8.5 x 455) + memory = 1092MHz (2.4 x 455) which is 1% faster

          Didn't you say that your system was stable with a 3.602GHz overclock?

          My stress test procedure is:
          1. run LinX using 4096MB of memory for about 15 minutes as a preliminary stress test. I've seen many cases where LinX or IBT run with no errors, sometimes for a lot longer than my test, only to have the system get Prime95 errors even after Prime95 had no errors for several hours.
          2. I run each of the three Prime95 tests (small FFT, large FFT and blend) for 8 - 12 hours and each test is run individually. In the good old days, each test had to run for at least 24 hours before we considered our systems stable. I'm not getting better, I'm getting older and I have lowered my test standards to an 8 hour or somewhat longer test for each of the Prime95 tests.
          3. I run my overclocked graphics tests where I might run Furmark for about 30 minutes, but the majority of my gpu testing is with intensive GPU tests such as Unigine Heaven or Valley.


          I haven't used OCCT in ages, but it does have some demanding stress test options.

          I always have HWiNFO running at all times and if necessary, I will enable HWiNFO's sensor logging option during a stress test where the log file might help diagnose a temperature or voltage or other problem.

          edit: Our former Gigabyte guru found that the Intel specs specified that VTT should be at least 0.05 volts lower than cpu vcore. While I respect his extensive knowledge and experience, I doubt that setting VTT somewhat closer to cpu vcore would be fatal. Keep in mind that we are often exceeding some of Intel's specs for our systems when we are applying very high overclocks. In the past I had several bios profiles where VTT was only 0.04 volts lower than cpu vcore for an extended period of time. I've seen several overclocks where VTT was set higher than cpu vcore but doing this might have shortened the life of their systems.
          Last edited by profJim; 09-08-2016, 03:30 PM. Reason: Additional VTT Voltage Info
          Q9650 @ 4.10GHz [9x456MHz]
          P35-DS4 [rev: 2.0] ~ Bios: F14
          4x2GB OCZ Reaper PC2-8500 1094MHz @5-5-5-15
          MSI N460GTX Hawk Talon Attack (1GB) video card <---- SLI ---->
          Seasonic SS-660XP2 80 Plus Platinum psu (660w)
          WD Caviar Black WD6401AALS 640GB (data)
          Samsung 840 Pro 256GB SSD (boot)
          SLI @ 16/4 works when running HyperSLI
          Cooler Master 120XL Seidon push/pull AIO cpu water cooling
          Cooler Master HAF XB computer case (RC-902XB-KKN1)
          Asus VH242H 24" monitor [1920x1080]
          MSI N460GTX Hawk (1GB) video card
          Logitech Z-5500 Digital 5.1 Speakers
          win7 x64 sp1 Home Premium
          HT|Omega Claro plus+ sound card
          CyberPower CP1500PFCLCD UPS
          E6300 (R0) @ 3.504GHz [8x438MHz] ~~ P35-DS3L [rev: 1.0] ~ Bios: F9 ~~ 4x2GB Kingston HyperX T1 PC2-8500, 876MHz @4-4-4-10
          Seasonic X650 80+ gold psu (650w) ~~ Xigmatek Balder HDT 1283 cpu cooler ~~ Cooler Master CM 690 case (RC-690-KKN1-GP)
          Samsung 830 128GB SSD MZ-7PC128B/WW (boot) ~~ WD Caviar Black WD6401AALS 640GB (data) ~~ ZM-MFC2 fan controller
          HT|Omega Striker 7.1 sound card ~~ Asus VH242H monitor [1920x1080] ~~ Logitech Z-5500 Digital 5.1 Speakers
          win7 x64 sp1 Home Premium ~~ CyberPower CP1500PFCLCD U.P.S
          .

          Comment


          • #20
            Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

            Originally posted by profJim View Post

            When I reach an overclocking limit, I always "retreat" to a stable setting that is several hertz lower. For example, my memory is stable up to 1092MHz {455 x 2.4) with 5-5-5-15 settings and with Performance Enhance = Turbo. I don't like to run at the ragged edge of stability, especially when my hardware is almost 10 years old. With my setup, the difference in performance would be minimal with:
            Cpu = 3.825GHz (8.5 x 450) + memory = 1080MHz (2.4 x 450) versus
            Cpu = 3.868GHz (8.5 x 455) + memory = 1092MHz (2.4 x 455) which is 1% faster

            Didn't you say that your system was stable with a 3.602GHz overclock?
            Indeed, meaning it has not failed a test yet. 3.602GHz (343MHz) is the frequency I "retreated" to after considering 3.655GHz (349MHz) stable. It seems our retreat tactic is similar as I retreated 6 vs 5 MHz, heh.

            It seems however that 3.655GHz is not stable after all so I may be closer to the edge than I thought. Further testing is needed. At the moment running OCCT/Furmark at 3.655GHz but with a VTT of 1.300V. I am always talking about 1.325V of Vcore. It has clocked 2 hours and counting but we 'll see.

            For sure I will be integrating your testing strategy as well. More testing never hurts. Well, perhaps IBT does, hehe.

            You may find the following interesting. It explains my obsession with testing CPU/GPU at the same time:
            The right way to stress-test an overclocked PC « Trying To Be Helpful

            About performance, I have not done a lot of rigorous testing, but it seems when the frequency is close to maxed out, that the gains are less than expected (normally linearly proportional gains depending on frequency). My 3.700-ish benchmark tests (with the system barely stable) were only slightly higher than 3.600-ish. This did not happen at lower frequencies, 3.300 was clearly better than 3.200. Perhaps some law of diminishing returns here, or some kind of internal errors that delay the system but it can recover from them. Perhaps it was just a bad performance test run, who knows.

            I will most probably go into 1.350V territory and see what happens there, but most likely will not stay there long term. Will report any nice results.

            Thanks again, I am immensely enjoying the exchange.
            Last edited by blacktruth; 09-08-2016, 03:45 PM.

            Comment


            • #21
              Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

              You're welcome, it's so much easier working with you with your well above average experience and knowledge.

              re: The right way to stress-test an overclocked PC « Trying To Be Helpful

              The author's approach is similar to mine, but he didn't include the Prime95 large FFT test.

              Why don't I follow his approach? Because I don't play games on my computer.

              I see his approach as not being terribly realistic, unless I was abusing my cpu by running IBT or LinX for hours on end while gaming. If I were a gamer and if I sometimes ran other demanding/intensive programs while gaming, I would then run my most demanding games while encoding videos or run other demanding applications at the same time.

              Here's another excellent stress test; run your favorite zip archiving program where you archive a large folder (e.g. 10GB - 25GB) such as your My Documents folder. After you have created the archive, use the archiving program to test the integrity of the newly created archive file. If the integrity check has no errors, run the archiving program on the newly created archive file and unzip the archive file to a temporary folder to ensure that there aren't any errors.

              BTW, how about buying an ssd and using it as a boot disk!!
              I would go with a 250GB or 500GB ssd unless you are also running Photoshop or other demanding programs where you might need to buy a larger drive..

              Buyer Beware: Some of the least expensive ssd's have performance issues when under heavy load.
              Do your homework and read ssd reviews from the most reputable review sites.

              Based on many reviews, the Crucial BX100 ssd got rave reviews as an inexpensive, budget ssd with performance that was almost on par with more expensive ssds. BUT, and this is a big one, Crucial's replacement for the BX100 was the BX200 and all of the review sites showed how the BX200 had significantly worse performance than the BX100 models.

              Check out the final thoughts in Crucial BX200 480GB & 960GB SATA III SSD Review

              You will probably find that spending $10 or $20 more will overcome the deficiencies in the lowest price models.
              Last edited by profJim; 09-08-2016, 05:03 PM. Reason: tiepoh
              Q9650 @ 4.10GHz [9x456MHz]
              P35-DS4 [rev: 2.0] ~ Bios: F14
              4x2GB OCZ Reaper PC2-8500 1094MHz @5-5-5-15
              MSI N460GTX Hawk Talon Attack (1GB) video card <---- SLI ---->
              Seasonic SS-660XP2 80 Plus Platinum psu (660w)
              WD Caviar Black WD6401AALS 640GB (data)
              Samsung 840 Pro 256GB SSD (boot)
              SLI @ 16/4 works when running HyperSLI
              Cooler Master 120XL Seidon push/pull AIO cpu water cooling
              Cooler Master HAF XB computer case (RC-902XB-KKN1)
              Asus VH242H 24" monitor [1920x1080]
              MSI N460GTX Hawk (1GB) video card
              Logitech Z-5500 Digital 5.1 Speakers
              win7 x64 sp1 Home Premium
              HT|Omega Claro plus+ sound card
              CyberPower CP1500PFCLCD UPS
              E6300 (R0) @ 3.504GHz [8x438MHz] ~~ P35-DS3L [rev: 1.0] ~ Bios: F9 ~~ 4x2GB Kingston HyperX T1 PC2-8500, 876MHz @4-4-4-10
              Seasonic X650 80+ gold psu (650w) ~~ Xigmatek Balder HDT 1283 cpu cooler ~~ Cooler Master CM 690 case (RC-690-KKN1-GP)
              Samsung 830 128GB SSD MZ-7PC128B/WW (boot) ~~ WD Caviar Black WD6401AALS 640GB (data) ~~ ZM-MFC2 fan controller
              HT|Omega Striker 7.1 sound card ~~ Asus VH242H monitor [1920x1080] ~~ Logitech Z-5500 Digital 5.1 Speakers
              win7 x64 sp1 Home Premium ~~ CyberPower CP1500PFCLCD U.P.S
              .

              Comment


              • #22
                Re: Gigabyte G41M-Combo Vcore BIOS setting does not work

                Oops, I overlooked the last part of your post dealing with cpu speed and performance.

                At 2500MHz (2.50GHz) a 100MHz increase in cpu speed is a 4.0% increase.
                At 3000MHz (3.00GHz) a 100MHz increase in cpu speed is a 3.3% increase and the increase from 2500MHz is 20%.
                At 3600MHz (3.60GHz) a 100MHz increase in cpu speed is a 2.8% increase and the increase from 2500MHz is 44%.
                At 3700MHz (3.70GHz) a 100MHz increase in cpu speed is a 2.7% increase and the increase from 2500MHz is 48%.
                At 4000MHz (4.00GHz) a 100MHz increase in cpu speed is a 2.5% increase and the increase from 2500MHz is 60%.

                An interesting side note, IBT and LinX GFlops scores will closely increase the same amount as the percentage increase in cpu speed.
                Using more memory in an IBT or LinX test will almost always give higher GFlops scores.
                I usually use 4096MB of memory in these tests with my 8GB memory setup, but I only run with3, 4, or 5 test loops.
                With your current temperatures I would not run either test.

                Even though there is a 4.8% difference in cpu speed with both of the following setups
                • Q9550 @ 4.012GHz, ram = 944MHz
                • Q9550 @ 3.825GHz, ram =1080MHz

                there is very little difference in performance.
                There might be a more noticeable performance difference if I was running any of the following applications:
                • encoding videos or music
                • performing major image editing with very large RAW photo files in Photoshop
                • image rendering or a Cinebench benchmark


                BTW, I forgot that I had installed PassMark's Performance Test evaluation software a while ago but never used it.
                I ran the full benchmark with my 3.825GHz setup but did not overclock my video card.



                My uploaded Performance Test summary results: PassMark Software - Display Baseline ID# 677577


                These are the detailed results that compare my almost 9 year old hardware with some newer hardware:

                Q9650 @ 4.10GHz [9x456MHz]
                P35-DS4 [rev: 2.0] ~ Bios: F14
                4x2GB OCZ Reaper PC2-8500 1094MHz @5-5-5-15
                MSI N460GTX Hawk Talon Attack (1GB) video card <---- SLI ---->
                Seasonic SS-660XP2 80 Plus Platinum psu (660w)
                WD Caviar Black WD6401AALS 640GB (data)
                Samsung 840 Pro 256GB SSD (boot)
                SLI @ 16/4 works when running HyperSLI
                Cooler Master 120XL Seidon push/pull AIO cpu water cooling
                Cooler Master HAF XB computer case (RC-902XB-KKN1)
                Asus VH242H 24" monitor [1920x1080]
                MSI N460GTX Hawk (1GB) video card
                Logitech Z-5500 Digital 5.1 Speakers
                win7 x64 sp1 Home Premium
                HT|Omega Claro plus+ sound card
                CyberPower CP1500PFCLCD UPS
                E6300 (R0) @ 3.504GHz [8x438MHz] ~~ P35-DS3L [rev: 1.0] ~ Bios: F9 ~~ 4x2GB Kingston HyperX T1 PC2-8500, 876MHz @4-4-4-10
                Seasonic X650 80+ gold psu (650w) ~~ Xigmatek Balder HDT 1283 cpu cooler ~~ Cooler Master CM 690 case (RC-690-KKN1-GP)
                Samsung 830 128GB SSD MZ-7PC128B/WW (boot) ~~ WD Caviar Black WD6401AALS 640GB (data) ~~ ZM-MFC2 fan controller
                HT|Omega Striker 7.1 sound card ~~ Asus VH242H monitor [1920x1080] ~~ Logitech Z-5500 Digital 5.1 Speakers
                win7 x64 sp1 Home Premium ~~ CyberPower CP1500PFCLCD U.P.S
                .

                Comment

                Working...
                X