Wow, that is some good advice dividebyzero. Thanks! I never would have thought of half of that (C1E, Memtest x86, performance enhance... I have all this stuff but I wouldn't have thought to use it).

Now there's memory profiles in the BIOS for my ram (Profile1) do I disable that? And as you said I'll change it from Turbo to standard.

For interests sake about the PCIe frequency-- would setting it to auto run it over the 100mhz spec with the increased FSB, or is it something else?

And 1.1ghz clocks run just fine on my GSKILL DDR2 1066 (Aqua blue heat-sinks; They're 2 1gb modules, so their timings can be better as well, right?).


EDIT: the Sunbeam CR-CCTF wouldn't fit my case lol.

Auto should be 100, and you probably don't want to change it.

Yeah, I've heard that changing it cause some serious stability problems. I'm just wondering if Auto would ever cause it to operate above 100mhz.

well mine runs @ 101mhz and that's the Auto on this board... nothing has happened so far and i run two video cards. my board is a little nuts though so i would still recommend leaving it where it is; there's really NO need to change it.

Absolutely agree.

Auto for PCIe frequency SHOULD be 100...But thats not necessarily the case with GIGABYTE boards based on the P45 chipset. If it's manually set to 100 then the doubt about it's frequency is moot. Gigabyte EP45-DS and -UD boards can have OC stability hassles with the auto setting- quite the opposite of what you've heard
The RAM profile you already have will be either the standard JEDEC profile (usually with loose timings) or an XMP profile if they're running at 1066 automatically. I assume you have the F2-8500CL5D-2GBPK kit- if so they should run ok at 1100 - you may need to push through a big voltage for it to be stable tho 2.1-2.2v. -don't know if you can keep them at 5-5-5-15 timings tho. What is your Channel A/B Static Read Value (under the advanced timing control tab- or use Memset which calls it "Performance level") ? if you need to have it at 7 or higher then you're wasting your time having the RAM clocked high and you will be better off running your RAM at 1:1 with your FSB -this should allow you to run around 850-900MHz at 4-4-4-12 and a perf. level of 6 or 7 -also you wouldn't need to overvolt.
If you're in any doubt or want further verification I'd check out the main/official Gigabyte forum over at Tweaktown. With a bit of patience you should be able see some nice timings with 2 x 1Gb sticks. I just pulled up old screenshot from my backup computer (see attachment) that runs Q9400/EP45-DS4P/Crucial 2x2Gb Tracer DDR2-800 and running at 1000 to give you an idea of the timings you should achieve (or better)- the perf level is set at 8 as I hadn't yet finished with the OC'ing- I reset the memory latch to 2.00D once I reached 425FSB.
Also note that tWTR and tWR in Memset is your BIOS value PLUS 8 if using the 2.66B/D memory latch, or PLUS 7 if using the 2.00B/D latch, i.e. if your BIOS says tWTR = 3 then Memset will list it as either 10 or 11.
Keep your current BIOS profile and save a few of the stable profiles that you create - for instance 400FSB/400 mem, 400FSB/533 mem at the best timings you can achieve. During oc'ing DISABLE "XMP" (under the advanced clock contol tab) ,leave (G)MCH latch at "Auto" and if your DRAM Voltage (under the MCH/ICH tab) isn't at 2.0 - 2.05volts (i.e.set at JEDEC's 1.8v) then you will most likely need it if your running the RAM I quoted earlier.

Attached Images

3.33GHz.JPG (176.1 KB, 7 views)

I just made a supposition about the PCIe frequency while in auto->auto-adjusting itself (>100mhz in respect to fsb >333mhz). It was a hypothesis worth answering anyways. I have no sources for this information except myself. Thanks for your clarification though.

You are correct with the RAM model. Despite their recommended voltage of 2.1 at 5-5-5-15 1066, I am running it at a lower 2.0v with no problems. So if I have to put it up to the reference 2.1v to attain stability, no problem!

According to Memset4.1: Performance Level: 6
I took a screen shot of it, and attached it. *I don't know why its called DDR3

When it comes to the timing of my RAM, would the effort put in to figuring out the BEST possible timings on the Tweaktown Gigabyte forum be worth it? As in, would the performance increase be that much greater? This is not to say that I won't do it, rather, how much time I should set aside.

I have to admit though, I know little to nothing about the advanced timings. So your tips are invaluable!

Final note: I have, in the past, run my RAM at 1092mhz at 2.080V with no problems (Crysis, and stress testing; Never performed a Memtest x86 though). Also, I now have access to Windows 7 64-bit, would this aid my overclocking persuit? (I currently have installed Windows 7 32-bit).

Attached Images

Untitled.jpg (121.9 KB, 4 views)

Last things first-I don't think 64 bit is going to help you- More for utilising 4+Gb RAM. If you do decide later on 4Gb (or 8) you'd be better advised to go to 2 x 2Gb rather than 4 x 1Gb- a lot of Gigabyte P45 mobo's get touchy with all four DIMM's populated.
Regarding your memory timings-it depends on how much time you want to put into it. You could probably get away with leaving your timings stock as you raise the cpu's FSB- just run Memtest86+ for a few passes (5-10 -each pass takes about 30 min with 4Gb, probably a fair bit less with 2Gb) before you reboot into the OS and stress test the cpu.
Is it worth it ? For comparison I ran SuperPi (1M) after each memory timing change (back when I had too much time on my hands) Memory testing took about a week. You decide if it's worth it....
(EPP profile is similar to the XMP/JEDEC profiles stored in the RAM)

Mem speed...........800 (EPP1).....900.....1000(EPP2)....1000.......1100
Mem latch.............2.00...............2.40........2.40.............2.40..... .....2.66
FSB......................400.................375.........417..............4 17..........414

CAS.......................4...................5.............5.............. ...5..............5
tRCD.....................4...................4..............5.............. ...5..............5
tRP........................4...................4..............5............ .....5.............5
tRAS.....................12................12............15................ 15..........15
All precharge.........5...................5.............6...................6.. ...........6
tRFC.....................52.................52...........64................ 52...........54
Perf. level..............6...................9.............9...................8. ............8
tWR......................13.................14...........14................ 14...........14 .....(Note 1)
tWTR....................10.................12...........13................1 2...........11.......(Note 2)
tRRD.....................3...................3.............3............... ...3...............3
tRD_WR................8...................8.............8.................. 9..............9
tRTP......................5...................5.............5.............. ....5..............5........(Note 3)
all precharge to
refresh delay.........5.....................5............6.................6....... ........6
tREF..................3117..............3510.......3900..........3903...... ...4452
CR........................2T.................2T...........2T.............2T ..............2T

SuperPi (in sec) 13.23..........12.02.......13.19........11.77............12.11

Note 1: This appears as "Write to precharge delayed" in Memset and the number is 8 or 7 higher than it appears in the BIOS depending on memory latch used.
Note 2: This is also 8 or 7 higher than in your BIOS ( 8 for 2.40/2.66 latch, 7 for 2.00 latch)
Note 3: This is 2 higher than it appears in your BIOS

Not huge differences but not too insignificant either-just remember that I use Crucial RAM (with the good D9 chips) but they are 2gb sticks so as the saying goes-your mileage may vary.

Hello, I'm back and this time with Coolermaster CCTF and a Raidmax Smilodon case in my possession.

Results-CPU Q8200 at 2.8Ghz peak temperature is 62. Completely stable.