Please report all spam threads, posts and suspicious members. We receive spam notifications and will take immediate action!
Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 23

Thread: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)




  1. #11
    Join Date
    Aug 2008
    Location
    La Jolla, CA
    Posts
    175

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    As I said (if you could find it in the middle of my rant...sorry), I only got stable once I bumped the FSB to 400 and went 1:1. That made it easy to clock up the multiplier to get as high as 3.4G on the CPU, which is fine by me. I still think it shouldn't be that difficult to get stable in dual channel.
    Gigabyte GA-EP45-DS3L (BIOS F9); E7200 Core2Duo 2.53Ghz 1066Mhz FSB 3MB L2 (OC to 3.4Ghz); SuperTalent 2x2G DDR2-800 PC6400 CL 5-5-5-12 (T800UX4GC5); Seagate 500GB SATA2 7200 32mb cache HD; EVGA 8800GT 512MB Superclocked Edition
    Ultra X-connect X2 550-watt PSU; Vista x64 SP1 and OpenSUSE 11.0 (KDE 4.1)

  2. #12
    Join Date
    Sep 2008
    Posts
    10

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    Quote Originally Posted by mxcrowe View Post
    As I said (if you could find it in the middle of my rant...sorry), I only got stable once I bumped the FSB to 400 and went 1:1. That made it easy to clock up the multiplier to get as high as 3.4G on the CPU, which is fine by me. I still think it shouldn't be that difficult to get stable in dual channel.

    haha sorry I read your post but missed the obvious, I was more focused on your agreeing with me

  3. #13
    Lsdmeasap's Avatar
    Lsdmeasap is offline GIGABYTE Guru
    Join Date
    Jan 2008
    Location
    PCB Island
    Posts
    25,940

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    So how are you coming long with this now?

  4. #14
    Join Date
    Sep 2008
    Posts
    10

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    Quote Originally Posted by Lsdmeasap View Post
    So how are you coming long with this now?
    I've got it stable at a 2.4B multiplier, (with the last timings you suggested. I haven't tried setting the advanced timing control settings back down yet) with the FSB at 400 so the mem freq. is at 960Mhz.

    Not too bad, I am sure I could go much higher, but I didn't aim for an OC'ing rig as it is for a friend and is being shipped a long distance, and the heatsink is stock.

    if I bump it just one notch above 2.4B (To 2.5) even at 333 FSB it errors or freezes in memtest86. Other wise seems good to go now. I will not RMA this over 100 Mhz lol.

  5. #15
    Lsdmeasap's Avatar
    Lsdmeasap is offline GIGABYTE Guru
    Join Date
    Jan 2008
    Location
    PCB Island
    Posts
    25,940

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    Well sounds good then if you did not plan to overclock it anyway. If you do decide you want help with getting it further before you ship out feel free to post back

    Hey, When you ship it, if you think the person can handle it, it would be best to not ship with the heatsink on the CPU because if it comes loose it can break all kinds of things as they toss boxes around very hard in shipping warehouses!

  6. #16
    Join Date
    Sep 2008
    Posts
    10

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    Quote Originally Posted by Lsdmeasap View Post
    Well sounds good then if you did not plan to overclock it anyway. If you do decide you want help with getting it further before you ship out feel free to post back

    Hey, When you ship it, if you think the person can handle it, it would be best to not ship with the heatsink on the CPU because if it comes loose it can break all kinds of things as they toss boxes around very hard in shipping warehouses!

    Yea, that's why its just the tiny stock HSF which I'll be zip-tying in place. I'd like to get this ram to run at its rated speed, but I've tried a lot of settings and can't get it to go. Any other suggestions?

  7. #17
    Lsdmeasap's Avatar
    Lsdmeasap is offline GIGABYTE Guru
    Join Date
    Jan 2008
    Location
    PCB Island
    Posts
    25,940

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    Yes, I can suggest some more settings. Please post me again what you have set in there now and I will post back some suggested.

    Did you ever use the above 400FSB with the 2.66 Mutli with any luck? That is 1065Mhz ram. If you do not see 2.66, set MCH latch to 333 or 400. And I would advise you use FSB 402 or so as 400 is right on the strap line and can be unstable. Always best to use right above or below if you want to stay in that general range
    Last edited by Lsdmeasap; 09-12-2008 at 05:39 PM.

  8. #18
    Join Date
    Sep 2008
    Posts
    990

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    You can also just wait for a newer BIOS, the board ist still relatively new and BIOS work ist mostly being done on the more expensive boards, the cheaper boards obviously aren't the main priority.

    Also, you will not notice any real world performance difference bettween DDR2 800 and 1066 anyway, same as DDR3 gets you almost no noticeable performance increase. This is because FSB bandwidth is smaller than RAM bandwidth. Data form RAM gets to the northbridge at RAM speed and is passed on the CPU at FSB speed. Now if you increase only the RAM speed, the bandwidth from the NB to the CPU will remain unchanged. If you use the 1:1 memory multiplier (called 2.0 on GB boards) FSB to NB and NB to RAM bandwidth will be equal. If you use a higher multiplier RAM bandwidth will just be bigger than FSB bandwidth.

    Here's a basic example how the bandwidth is calculated:

    FSB at 400MHz quad pumped -> 400 x4 = 1600 MHz effective rate

    RAM at 400 MHz (DDR2 800) double data rate & dual channel -> 400 x2 x2 = 1600 MHz effective rate

    (This actually only gets you the frequency, bandwidth would be calcualted by multiplying with MB/Hz (amount of data per effective clockcycle).

    So faster RAM doesn't improve the abdwidth to the CPU, but it does slightly improve latency (same as better timings improve latency). As latency isn't all that important in most real-world applications you will only see slight performance improvements, of 0-5% when running faster RAM.

  9. #19
    Join Date
    Sep 2008
    Posts
    10

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    Quote Originally Posted by Nickel020 View Post
    You can also just wait for a newer BIOS, the board ist still relatively new and BIOS work ist mostly being done on the more expensive boards, the cheaper boards obviously aren't the main priority.

    Also, you will not notice any real world performance difference bettween DDR2 800 and 1066 anyway, same as DDR3 gets you almost no noticeable performance increase. This is because FSB bandwidth is smaller than RAM bandwidth. Data form RAM gets to the northbridge at RAM speed and is passed on the CPU at FSB speed. Now if you increase only the RAM speed, the bandwidth from the NB to the CPU will remain unchanged. If you use the 1:1 memory multiplier (called 2.0 on GB boards) FSB to NB and NB to RAM bandwidth will be equal. If you use a higher multiplier RAM bandwidth will just be bigger than FSB bandwidth.

    Here's a basic example how the bandwidth is calculated:

    FSB at 400MHz quad pumped -> 400 x4 = 1600 MHz effective rate

    RAM at 400 MHz (DDR2 800) double data rate & dual channel -> 400 x2 x2 = 1600 MHz effective rate

    (This actually only gets you the frequency, bandwidth would be calcualted by multiplying with MB/Hz (amount of data per effective clockcycle).

    So faster RAM doesn't improve the abdwidth to the CPU, but it does slightly improve latency (same as better timings improve latency). As latency isn't all that important in most real-world applications you will only see slight performance improvements, of 0-5% when running faster RAM.


    I understood the math just not the "function" of what it does, thanks for the information. I think I'll just leave it at stock FSB or slightly above, if the owner decides he likes to OC down the road (which I doubt) Ill help him buy an aftermarket heatsink for it and perhaps by then a future BIOS will make things easier. At least he will have some heardoom with the ram

    I am just thrilled its stable now That was a bit of a pain in the butt, especially since I RMA'd the first RAM sticks without really thinking, lol.

    Thanks, I didn't know that about setting to an exact number could make it unstable. I am just leaving it at 375 FSB and 2.4B (Nothing else will run stable,at least without mroe tweaking :p) even at 333 as soon as I boot into Windows at the 2.66 multiplier and try running Prime95 it errors immediatly.


    Thanks for all the help.

  10. #20
    Join Date
    Sep 2008
    Posts
    990

    Default Re: Gigabyte EP45-DS3L ... the board from Hell? (Unstable, need some tweak advice)

    On current Core 2 systems, the memory controller sits on the northbridge. The RAM is linked to the memory controller and the memory controller forwards data to the CPU (via the FSB). But the memory controller can only forward as much data to the CPU as will "fit through" the FSB (higher FSB means more data can fit through = higher bandwidth).
    Even at the smallest memory multiplier the badwidth between RAM and the memory controller is already the same as the bandwidth of the FSB (i.e. RAM bandwidth is always equal or greater than FSB bandwidth). Because the FSB always has the smallest bandwidth here there's not much point in fast RAM that could deliver data very fast to the memory controller, as the data would have to "wait" there (the RAM can get data faster to the memory controller than the memory controller can pass on the data to the CPU).

    The FSB limiting memory bandwidth was also why Athlon 64 was so good at it's time and why people are excited about Nehalem/Core i7: They both have the memory controller on the CPU meaning data will actually get to the CPU at the speed of the RAM, and not at the speed of the FSB.

Page 2 of 3 FirstFirst 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
  •