P45 Refresher


The GA-EP45-UD3P is based on the Intel P45/ICH10R chipset combination. The P45 is manufactured on a 65nm process, which makes it not only smaller but more energy efficient than 90nm chipsets like the X48. It typically runs much cooler than the X48 and P35. Officially, the P45 does not support the 1600Mhz front-side bus used on the Core 2 Extreme QX9770 quad-core processor. However, just about all motherboard manufacturers are unofficially supporting 1600FSB.

The P45 supports either DDR2 or DDR3 memory. Once again, official support is limited to JEDEC approved memory speeds up to DDR2-800 or DDR3-1066. Unofficial memory support by the manufactures is available for speeds up to DDR2-1366 and DDR3-2000 depending on the supplier. Gigabyte claims support for DDR2-1366+ on this board and we were able to hit DDR2-1300 with a less than impressive memory kit.

The P45 MCH is limited to 16 PCI-E lanes; unlike the P35, these lanes support the latest PCI Express 2.0 specification. PCI Express 2.0 doubles the standard bus bandwidth from 2.5 Gbit/s to 5 Gbit/s. We have not noticed any real performance differences between PCIe 1.0 and PCIe 2.0 in most cases, but as graphics technology improves, we expect to cross over that line shortly. Also keep in mind that two x8 PCI-E 2.0 slots are equivalent to two x16 PCI-E 1.x slots, provided you have PCI-E 2.0 cards installed.

The big news is that the P45’s PCI-E 2.0 slots can work in a dual x8 configuration for CrossFireX. The P35 had a slow x16/x4 CrossFire setup with the x4 slot running off the Southbridge. Compared to the dual x16 setup on the X38/X48 boards, we have not found any appreciable performance differences in GPU testing with AMD’s latest single GPU video cards.

With the P45 chipset comes the new ICH10R Southbridge. Except for official Turbo Memory support, consider this Southbridge to be a slightly revised ICH9R with support for six SATA II ports, AHCI, Matrix RAID, and twelve USB 2.0 ports.

Index Features Features Features
POST A COMMENT

73 Comments

View All Comments

  • Norm VH - Monday, August 17, 2009 - link

    I am rebuilding in an old computer case using the Gigabyte EP45-Ud3P motherboard. Have new e7500 cpu. New evga 9500 Gt video card. new Corsair ddr2 ram and a new antec earthwatts psu. Before I add all the rest of the items I am checking fan features. Board lights come on. Case fans work. video card fan works. But the Zalman 9700 fan just jumps a little and stops. The Intel fan that came with the CPU does the same thing. Tried cpu fans in both cpu_fan 4 pin and the sys_fan2. Is this normal? Is the fan waiting for the CPU to heat up? I returned a motherboard to Fry's for the same behavior. They said fans should start. I tested the fans on an old asus motherboard and they are working. Reply
  • Isme - Thursday, December 03, 2009 - link

    Yes On my EP45-UD3L there is a warm up time before the CPU fan needs to turn on. Its not long for most heat sinks. Maybe 5-10 seconds. You see the CPU fan twitch then sit idle for a time related to how big the heat sink itself is.

    For modern CPUs you should not need to worry about CPU overheat while messing around with the CPU fan unless you already have cranked voltages for extreme overclocking. The CPU should shut itself off within 10C and 30 seconds of reaching maximum normal operating temperature. I would not do any overclocking until you get fans and other basics working.
    Reply
  • lancerocke - Friday, July 24, 2009 - link

    so there is no fix for this crackling audio?
    i kinda want to return this board now.
    Reply
  • Voldenuit - Tuesday, August 18, 2009 - link

    Sadly, I just got an EP45-UD3P Rev1.1, and the onboard audio still crackles. :(

    Running Win7 Ultimate RC.
    Reply
  • syseng - Friday, June 12, 2009 - link

    My second Gagabyte board of this model has died. The first one was DOA and this second one died more gradually. The problem is that it woudl sometimes refuse to turn on or off with the power button. Pressing the button mutliple times or holding it in sometimes worked. Finally, it would not power up at all. I have grounded the pin 16 of the system power connector to start and run the system until I can get a replacement. I have read many comments from varous forums of people having power control problems with this board. Mine was never over-clocked and no power management control was on.

    Gigabyte will not ship a replacement until they receive the dead board and test it. Estimated round trip time to get the board back is about 3 weeks. Probably the last Gigabyte for me. When I built this system in March, I ordered a Gigabyte 4850 graphics card (all from Newegg) and it was DOA too.

    I am just wondering if I am incredibilty unlucky with Gigabyte or if they really have a quality problem. They certainly have a customer service problem if they think it is acceptable to have a system down for 3 weeks.
    Reply
  • keithh - Wednesday, May 13, 2009 - link

    I decided to give Gigabyte a shot for my latest system build and purchased the EP45-UD3L for the kid's linux desktop. I was using a PATA DVD drive to install to an older IDE HD. After the installation started, the system abrupty died. After resetting, the bios was unable to see the drives. Cutting a long story short, I concluded that the motherboard was bad and exchanged it.

    The second board had exactly the same problem.

    After switching to an ASUS motherboard, I had no problems.

    I eliminated the following possibilities during my investigation:
    - master/slave settings
    - IDE cable
    - DVD drive
    - IDE drive (tried a SATA drive from another machine)
    - RAM test
    - BIOS revision (F2 -> F4 didn't make a difference)
    - installing Windows XP from a disk that is known to be good
    - installing two different Linux distributions including Fedora 10.

    It uses the JMicron 368 PATA chipset which has a dodgy reputation so conceivably using a SATA DVD drive and a SATA HD would have worked.

    The power supply was a brand new Corsair.

    That doesn't leave a lot of room for interpretation - I was 0/2 on the Gigabyte MB. I wasn't impressed.
    Reply
  • syskin - Saturday, February 21, 2009 - link

    Hi, I have one question which is very important to me.
    If I:
    - overclock FSB and
    - set a higher Vcore voltage and
    - activate EIST/speedstep,

    ...will this motherboard still lower Vcore at idle (as per speedstep) or will it keep it at bios setting all the time?

    Thank you in advance.
    Reply
  • foofoo - Tuesday, February 17, 2009 - link

    I have 2 x 2GB RAM installed and winXP pro 32 bit only sees 2.5GB. The BIOS (f7) sees all 4GB. Changed RAM and slots with the same results. I put in a trouble ticket to Gigabyte support and got the following response (or non-response)

    "You will need to use 64 bit OS in order to see the full 4gb capacity or higher
    Windows will automatically reserve an certain amount of memory being used and there are no way to adjust it"

    All other gigabyte boards that I have used (965 and P35 chipsets) show 3.5GB with 4 installed under XPpro. This is actually a game stopper for me since losing that much RAM affects the usefulness of my PC.

    Fair warning for all wanting to use a 32 bit OS with this board.

    Reply
  • Isme - Thursday, December 03, 2009 - link

    Make sure you are looking at BIOS RAM reports especially the POST. Or I'd use the Task Manager Physical Memory report or the built-in Windows Accessories-System Tool-System Information-System Summary tab-Total Physical Memory line before I panic about hardware issues. Note that on one of my computers I have only 1.19GB available out of 2029MB. Application reports of RAM are pretty much useless except for that application at that point in time with whatever else happens to be running. Know how 3rd party utility tools define available RAM.

    Well the amount of RAM visible varies. Depending on OS and reporting tool 1GB may be reserved for the OS. That is the standard 32-bit Windows OS only has 3GB RAM for user purposes. With older very basic peripheral devices and XP that often included memory mapped IO space. But with newer larger OSes and huge numbers of motherboard ports and video cards sporting 1GB+, dedicated SATA port buffer frames, dedicated buffers for fancy network ports...

    Remaining RAM space, especially user space, may be more dependent on how much of the 4GB space is mapped for peripherals in your particular setup. I would not be surprised if some video cards are mapping 512MB or more to main memory as a window now that 1GB+ video cards are common. I suppose a less considerate video BIOS might even automatically reserve space for an X-fire setup even if you are not using one.

    Reply
  • pudlglum - Sunday, February 15, 2009 - link

    Thanks for all the excellent knowledge on the AnandTech Forum. I believe this is my first post on any forum, and I've been into PCs since purchasing my brand new Compaq Portable 286, with 256K ram and a single 1.2 Mb floppy.

    I've had a GA-EP45-UD3R running for about a week. All components are conventional, all good quality, nothing is "bleeding edge", and I'm not overclocking. I have an ATI Radeon HD 4350 PCI-e graphics card, connected to a Hitachi CM751 CRT monitor. Two WD SATA hard drives are connected to the Intel (yellow) connections, configured in the BIOS to IDE mode. My reliable Plextor PX-760A CD/DVD burner is connected as "master" to the single IDE connection.

    Everything was smooth except for two things: occasionally the screen resolution was corrupted when resuming from standby, and I was getting disk write failures from both an old (4X) CD burner and my recent Plextor PX-760A CD/DVD burner.

    I downloaded whatever recent drivers I could find; the video problem hasn't recurred, but ImgBurn or CDBurnerXP both froze while attempting to write an iso image (Windows 7 Beta) to a DVD. ImgBurn locked both itself and Explorer, and required a forced reboot. CDBurnerXP Pro provided the following error message:

    Writing Error: (3) Error occurred writing data to disc.
    The SCSI/ATAPI bus was reset and caused a write failure (1026).
    Error Sense Data: SENSE KEY: 6 ASC: 29 ASCQ: 0

    Apparently both IDE and the purple SATA2 connectors are controlled by the "Gigabyte SATA2" chip, enabled in my BIOS in IDE mode. It is listed in the Device Manager "SCSI and RAID controllers" section as "GIGABYTE GBB36X Controller", with driver provider "JMicron Technology Corp". I was distracted by several suggestions that the JMicron driver was buggy, but that was just a wild-goose-chase.

    I then pursued some marginal comments about DPC latency. The comments were typically in the context of audio crackling issues, which I had not (yet) noticed. However, several posts also referred to the Gigabyte Energy Saver service. I had not knowingly installed that package from the Gigabyte motherboard CD, but I found the GSvr.exe service running, with description "GEST service for power management". It also appeared in my Start menu, under GIGABYTE > EnergySaver.

    I downloaded DPC Latency Checker V1.1.0 and discovered that my latency was averaging around 500µs, with all "foreground" applications closed, and with Windows Desktop Search and Avast Antivirus disabled. Running ImgBurn in this condition resulted in a another DVD coaster.

    After hard-rebooting, I disabled the GSvr.exe service, and confirmed that the DPC latency dropped to around 5µs. This time, ImgBurn successfully wrote the iso file to the DVD (yay!).

    I have now uninstalled GSvr.exe (GIGABYTE > EnergySaver > Uninstall), and re-enabled Windows Desktop Search and Avast. In the last half hour or so, my DPC Latency has continued to average about 5µs, with an absolute maximum of 177µs. I suspect removing the Gigabyte EnergySaver service will solve both my disk burning and standby/wakeup problems.
    Reply

Log in

Don't have an account? Sign up now