Please report all spam threads, posts and suspicious members. We receive spam notifications and will take immediate action!
Page 1 of 5 123 ... LastLast
Results 1 to 10 of 43

Thread: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles




  1. #1
    Join Date
    Feb 2009
    Posts
    42

    Default Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Hi all.

    I just recently purchased an EP45-UD3R along with an E8400 and 2x 2GB Corsiar Dominator 1066. Initially I was planning just to run everything at stock (it's pretty fast as it is!) However, having used AMD systems consistently since around 2002 I'd forgotten about the whole mismatched FSB vs. memory bandwidth issue.

    So now that I have this zippy dual-channel RAM setup capable of pushing 128 bits at 1066MT/s, it just seems plain silly to leave the FSB at stock where it has nearly 40% less bandwidth. I'm not interested in overclocking my CPU or memory, I just want to get the most out of the bandwidth between the two.

    My ultimate goal:

    * 533MHz FSB with a 6x multi (3.2GHz), keeping my CPU as near stock as possible
    * RAM running 1:1 at 533MHz, exactly at stock

    Is this reasonable? I purchased the EP45-UD3R on Anandtech's budget system builder guide recommendation, and they had mentioned being able to "easily" reach 540MHz FSB<http: www.anandtech.com="" guides="" showdoc.aspx?i="3486&p=5">, so I'm hoping 533 will be quite possible. I'd like to keep voltages for everything as low as possible, and insist on a 24-hr prime and 24-hr memtest stable system.

    I made an attempt to just set 533 for the FSB and a 6x CPU multi while everything else at "auto". I was having some success, but then noticed to my horror that the BIOS had goosed my CPU to 1.44V vcore and it was idling at 50C! That's one astoundingly stupid (and dangerous!) auto setting if you ask me, since the CPU was only running 6% faster than stock. Luckily I was only doing some Memtest86+ runs, so the system was only running at that voltage for half an hour or so. I also have a stock cooler on the chip from an old Q6600 system I built for someone (one of the larger, copper-core stock coolers), so I don't think it hurt anything.

    Anyway, after that experience I'm reluctant to trust any of the BIOS "auto" settings if I'm going to change anything. I tried setting all main voltages to "normal" with 533MHz FSB, which promptly failed requiring a CMOS reset. So, that leaves me with a whole slew of settings that I'm really not sure about how to set reasonably, primarily various voltages and the MCH strap.

    The MCH strap was one of the main things I was hoping to get help with. Here is how I currently understand it, please correct me if I'm wrong.

    The MHC bases it's internal operating clock on the FSB, similar to the CPU. The MHC straps are sets of timing configurations that are intended to correspond with the FSB running at a given speed. So, the 266 strap is what would be used for a stock 266MHz FSB setup, 333 strap for 333MHz FSB, etc. I would assume that, since higher straps are intended to be used with higher bus speeds, it stands to reason that the higher the strap the looser the timing profile associated with it.

    Is that correct?

    If so, I would guess that for any FSB of 400MHz or higher you'd always want to go with the 400 strap, since that's what's intended for use with 400MHz FSB configurations, and will provide the loosest and most forgiving (hence stable) setup at 400+ speeds. Is this also right?

    Lastly, if anyone has an idea of what voltages that I'd need to change beyond stock to get 533MHz FSB (I'm guessing just MCH-related things, yes?) and any suggested values to start off with, I'd greatly appreciate it!

    Full system outline
    CPU: Core2 Duo E8400 (using a Q6600 heatsink)
    Motherboard: Gigabyte GA-EP45-UD3R PCB v1.1, BIOS F6
    RAM: Corsair Dominator TWIN2X4096-8500C5D (2x2GB modules)
    Video: HIS Radeon 4870 1GB
    Audio: SB Audigy2 ZS PCI (in PCI3)
    HD: 2x Seagate Barracuda 7200.9 ST3808110AS 80GB 7200 RPM SATA-II (RAID 0)
    Optical: LiteOn DH-20A3H-08 20X IDE DVD+/-RW
    PSU: Enermax EG565AX-VEFMA2.0-SLI 535W
    +3.3v: 32A
    +5V: 32A
    +12V1: 18A
    +12V2: 18A
    -12V: 0.8A
    +5VSb: 2.5A
    +12V Total: 34A
    Case: LianLi PC-61 USB (2x 80mm lower front intake, 80mm blowhole, 80mm rear exhaust, and 80mm exhaust in PSU, lower rear of left side panel is also perforated, so air is drawn in over expansion cards)
    </http:>

  2. #2
    Join Date
    Sep 2008
    Posts
    990

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    I don't have time to reply to the whole post right now, but I can give you a few quick answers: At 1:1 FSB and RAM bandwidth is exactly the same in dual channel mode. Increasing the RAM frequency basically improves latency a little bit, but that's not really a big deal, neither is fine-tuning the timings. You'll see improvements in things like SuperPi or memory bandwidth benchmarks, but start up a game benchmark and you'll hardly see a difference. At least in current ones that are somewhat GPU limited, in old CPU limited ones you may see a bigger difference, but it doesn't really matter whether you have 200 or 220 FPS there.

    If you're considering spending time tuning your timings you should instead consider overclocking. Even a small overclock will bring a much bigger improvement than tuning timings ever will.


  3. #3
    Join Date
    Feb 2009
    Posts
    42

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Quote Originally Posted by Nickel020 View Post
    I don't have time to reply to the whole post right now, but I can give you a few quick answers: At 1:1 FSB and RAM bandwidth is exactly the same in dual channel mode.
    Right, this is exactly what I'm going for.

    Currently my available memory bandwidth is:
    533MHz x 2 transfers/clock (DDR) x 16 bytes (2x 64-bit, dual channel) = 17,056 MB/s

    At the moment my stock FSB bandwidth is only:
    333MHz x 4 transfers/clock (quad pumped) x 8 bytes (64-bit bus) = 10,656 MB/s

    For FSB that's a deficit of 6,400 MB/s, or about 37% less than the memory is capable of delivering, which is what I want to correct.

    Quote Originally Posted by Nickel020 View Post
    Increasing the RAM frequency basically improves latency a little bit, but that's not really a big deal, neither is fine-tuning the timings. You'll see improvements in things like SuperPi or memory bandwidth benchmarks, but start up a game benchmark and you'll hardly see a difference.
    I'm not looking to increase RAM frequency. I want it to stay at stock (533MHz, or 1066MT/s). I just want to boost FSB to match it so that I can use all the available memory bandwidth, while holding everything else as close to stock as possible. I'm interested in learning what BIOS controls and voltages to adjust to achieve that.

    Are you saying that bringing the FSB up from 333MHz to 533MHz while holding my CPU and memory at/near stock won't make that much real-world difference? I'd think an extra 37% in usable bandwidth between CPU and RAM would make a significant impact.
    Last edited by Dalamar; 02-21-2009 at 02:57 AM. Reason: Clarification

  4. #4
    Join Date
    Sep 2008
    Posts
    990

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    If you increase the FSB as well then you might as well overclock, as technically you're not running your CPU within its specificatiosn anymore and you've voided your warranty (Intel can't tell though :))
    You will see see mroe of an icnrease in performance than by pure performance, but you will probably have to increase the CPU voltage if you're aiming for 533 FSB. All in all, a 100 to 200 MHz will bring you more performance though I think.
    The Core 2 CPUs have so much cache that they don't really benefit from extra RAM bandwidth all that much afaik.

    But if you want to try it go ahead, you can't really break anything. Would be nice if you were to post some benchmarks!


  5. #5
    Join Date
    Feb 2009
    Posts
    42

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Quote Originally Posted by Nickel020 View Post
    If you increase the FSB as well then you might as well overclock, as technically you're not running your CPU within its specificatiosn anymore and you've voided your warranty (Intel can't tell though :))
    You will see see mroe of an icnrease in performance than by pure performance, but you will probably have to increase the CPU voltage if you're aiming for 533 FSB. All in all, a 100 to 200 MHz will bring you more performance though I think.
    The Core 2 CPUs have so much cache that they don't really benefit from extra RAM bandwidth all that much afaik.

    But if you want to try it go ahead, you can't really break anything. Would be nice if you were to post some benchmarks!
    Well, I'll be slightly overclocking (from 3.0GHz to 3.2GHz) even in this setup, since I can't go lower than a 6.0 multiplier.

    Are you saying that the CPU will need a vCore increase even if I'm lowering the multiplier to keep the core speed within 200MHz of stock?

    Thanks for your replies, by the way. Sorry for all my questions, but I just want to make sure I'm understanding things correctly.

  6. #6
    Join Date
    Sep 2008
    Posts
    990

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Are you saying that the CPU will need a vCore increase even if I'm lowering the multiplier to keep the core speed within 200MHz of stock?
    Will probably need it.


  7. #7
    Join Date
    Feb 2009
    Posts
    42

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Quote Originally Posted by Nickel020 View Post
    Will probably need it.
    Got it.

    Ok, so I'll probably want to bump MCH core voltage and perhaps CPU vCore.

    Do you think I'd need to worry about termination or reference voltages for either CPU or MCH to get to 533, or is it mainly the MCH core voltage that will determine stability at high FSB speeds?

    Also, if you can confirm my understanding about the MCH strap functions that I mentioned in the first post, that'd be greatly helpful as well. I'm assuming I should stick with the 400 strap, since I'll definitely be above 400MHz on the FSB and would want the loosest timings possible to keep the MCH stable. Is that right?

    Thanks!

    Oh, and once I've actually got something working I will definitely post before vs. after benchmarks. I'm curious now too to see what difference it'll make.
    Last edited by Dalamar; 02-21-2009 at 05:25 AM. Reason: Typos

  8. #8
    Join Date
    Sep 2008
    Posts
    990

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    I hiope you have an E0 stepping, if you have a C0 you might be unlucky and not even hit it at all.
    You will probably need some Termiantion, MCH maybe +0.1V. But there's not much point speculating, just try it.


  9. #9
    Join Date
    Feb 2009
    Posts
    42

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Quote Originally Posted by Nickel020 View Post
    I hiope you have an E0 stepping, if you have a C0 you might be unlucky and not even hit it at all.
    You will probably need some Termiantion, MCH maybe +0.1V. But there's not much point speculating, just try it.
    Yup, I have an E0 (SLB9J). Going to try some settings now. Hopefully I'll be back later tonight with successful results.

  10. #10
    Join Date
    Feb 2009
    Posts
    42

    Default Re: Help understanding GA-EP45-UD3R BIOS settings & Core2 FSB OC principles

    Well, I've found a few things that sort of work.

    I can achieve 533MHz FSB (x6 for a 3.2GHz CPU) that will boot into Windows, but isn't Prime stable, by doing the following:

    Set Host Freq manually to 533
    Set multiplier manually to 6
    Leave (G)MCH Latch at auto
    Set Memory Multiplier to 2.00B or 2.00D (both result in 1066 Mem)
    Set (G)MCH core voltage to 1.20v
    Manually specify my memory timings and voltage (5-5-5-15, 2.1V)
    Leave all other voltages at "normal" and all other options at default or auto settings.

    It'll only prime for about 10 min though, so there's still something that needs tweaking. I tried bumping MCH core to 1.30, and even CPU vcore to 1.30 as well, but it still wouldn't prime more than 10-15 min, so I suspect one of the other timing or voltage controls needs manipulating.

    Oh, and I did see a hefty bump in Sandra (v2009.1.15.72) memory bandwidth. Here are the averages of three runs for before and after.

    Stock
    (333MHz FSB x 9.0 for 3.0GHz CPU, 533MHz RAM at 5:8 FSB:RAM ratio)
    Int: 6773 MB/s
    Float: 6767 MB/s

    FSB Overclock
    (533MHz FSB x 6.0 for 3.2GHz CPU, 533MHz RAM at 1:1)
    Int: 8831 MB/s (a 30.3% increase)
    Float: 8807 MB/s (a 30.1% increase)

    Considering the CPU speed only increased 6% it looks like I'm successfully making use of more of the available bandwidth of the DDR2-1066 memory.

    Now I just need to figure out how to get the configuration 24-hr Prime stable.

    Any suggestions on what settings to turn to next?
    Last edited by Dalamar; 02-21-2009 at 06:48 PM. Reason: Clarification

Page 1 of 5 123 ... LastLast

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •