Help out a Epox 4g4a newbie

[Xtasy]

I just got my refurb epox 4g4a+ from newegg and luckily it worked, even tho it came with arctic silver smeared all over near the socket. After cleaning it up, i got it up and running. I heard that this mobo is a topnotch overclocker, and it seems to be so. I heard that people unlocked amazing overclocking potential on this board by moving the ram to dimm 2. Is that true for my situation, i'm using samsung single-sided 256mb ddr333. Also, is there anything else i need to be aware of? Thanks.

 

[Ketchup]

Slot 2 is the best place for memory when overclocking.

Everyone swears the 2424 bios overclocks the best, but I get better performance (3dMark, etc) from the latest bios and the overclock seems to be about the same. The 2424 bios is the only one that doesn't undervolt a little, so that may be why so many people swear by it. I make sure the bios doesn't report over 1.75 volts in the bios, so I set mine to 1.775.

If you want a ton of info on this board, go see the "The 4G4A+ seems very promising" thread.

If you don't wanna read all that, just remember that you need to set the FSB jumper on the board to 100 to get the 3:4 Memory ratio, and to 133 to get the 4:5 ratio. There is an issue with running the 3:4 ratio at high FSB on the 845G chipset. I am one of the few people that have gotten it to work. I am hitting a wall right now with 157 FSB and DDR 418. It freezes in games when I go up to 158, but I am not giving up yet. It runs faster at 157 with the 3:4 than at 161 with the 4:5 (where my CPU tops out at 1.75 volts.)

This is a great board and I still think it's one of the best out there. Have fun with it and let us know how it goes!

 

[Xtasy]

Do you know where i can get the 2424 bios? Thanks. So far i seem to be stable at 155mhz fsb. 160 is a bit iffy, i will need to try to up the vcore some more.

 

[Ketchup]

http://www.aoaforums.com/files/index.php?dir=189

 

[Johnnychangs]

The new 4G4A2B04.BIN Bios file does allow me to use the 3:4 ratio over 140FSB (older Bios revisions didn't). I set 150FSB (400DDR). However, it only worked on my WinXP Pro-Lite setup.

I have one drive with WinXP Pro with all drivers (Intel, Nvidia, XP SP1, etc). The other "Lite" setup is a clean install with no driver or OS updates. I could boot into the Lite XP fine and ran Prime95 other tests, etc. However, the same FSB and voltages on the normal XP Pro locks up while loading the desktop.

I guess not installing the Intel and Nvidia drivers do let you overclock more. However, I prefer having the latest drivers and running stable, so I' back to 145FSB and the 4:5 ratio.

 

[Xtasy]

I just tried the 2424 bios and it is more unstable than the latest ones at the same overclocked speed. It takes more vcore to get stability on the 2424 (actual vcore, not the setting) than the latest one. Could it be that the 2424 is also reporting the wrong vcore? It took me 1.750 volt setting in the latest bios to get stability at 155mhz fsb, which is about 1.69 volts undervolted. I need 1.700 volts to get stability on the 2424 bios, which is 1.71 volts overvolted. Doesn't make sense.

 

[Ketchup]

You've hit the nail on the head Xtasy. All bioses save the 2424 report less voltage than what you set it at. I think a lot of people are going by what they set the voltage at, and not what the setting actually is. I am glad you and Jonnychangs agree with me that the 2B04 bios is the best.

I think the 2424 bios reports all the voltage wrong, not the vcore. Did you happen to notice that the 12-volt and 5-volt lines seemed a lot higher when using the 2424 bios over the others? Just one of my observations. Also if you run 3dMark 2001 with the 2424 bios and then run it again with the 2B04, I can almost guarantee you will see an improvement.

Happy tweaking!

 

[Xtasy]

Wish i could get 160 mhz fsb stable. I set it to 1.8 volts (1.74 volt undervolted) and it was still unstable. What vcore would u say is the reasonable highest for a northwood with a good copper cooler?

 

[Johnnychangs]

I've never been able to get over 155 stable on my 4G4A+. I realize there could be a number of factors limiting the overclock, but I've tried Mushkin PC3200, Corsair XMS PC3200 (two different sticks), a P4 1.6A, a P4 2.66, blah blah. I just think that some boards do have a limit. Our boards must be from the same batch.

I ran my 1.6A at 1.725v for almost 6 months. Just replaced it with a P4 2.66. The 1.6A is still fine, I'll probably install it into another system to keep or sell. Ive got an Alpha 8942 heatsink with a Mechatronics 53CFM 80mm fan. In the summertime (no Air conditioning) Max temp under load was about 53C. Technically the P4s will start throttling down speed at 65C+, but I never let mine get that hot. I would say as long as you're not going over 55C you should be OK. Also figure in how long you plan on owning your current system. If you plan on buying a new processor within the next year, then overclock it to the highest stable speed/voltage. In my case, I usually replace components at least once or twice a year, so pushing them to their potential is my goal.

I have an Asus P4PE on the way, which should be nice since I hear its overclockable with two DDR modules. I've never been able to run my two Corsair sticks on the 4G4A over 133FSB.

 

[Ketchup]

Intel says you can take a Northwood up to 1.75 volts without killing it. I agree with Jonnychangs that you could go higher if you upgrade your CPU often.

Mine takes around 1.72 volts to hit 2.88 GHz, 160 FSB. Even if I raise the voltage to max, it won't be stable much above that. I notice that above 2.8 GHz, the CPU voltage starts to get pretty unstable (with 1.775 it will sometime go down to 1.69 while I am watching it in the bios) so I am thinking the voltage to the CPU is limiting overlcocking when the CPU starts drawing lots of power. I got a Vantec Tornado today, and even that wouldn't help me overclock. I tried running 3 GHz but it crashed in 3DMark.

Right now I am running 2.82 GHz and 157 FSB with the 3:4 ratio for DDR 418. If I go up to 158, it ALWAYS crashes in 3DMark. If I use the 4:5 divider, I get get to 2.88 GHZ and DDR 400, but my 3DMarks are a bit lower, so I figure I might as well not push my CPU any harder than I have to.

I really like the memory I am using. I got it up to DDR 446 at 2.7 volts and got a Sandra score of 3339/3376.

This board may be holding me back a little, but I think it's the CPU more than anything. I still really like this board, and with the 3:4 divider, it is just about as fast as PC1066 RDRAM and Granite Bay according to some reviews I have read.

 

[Xtasy]

Would 1.74 ~ 1.76 volts be too much at 2.56Ghz? My Coolermaster fujiyama copper cooler keeps it at around 41C idle. After running 3dmark2k1, the cpu is at 48C.

 

[Xtasy]

Ketchup, which bios are u using to keep a stable high fsb with the 3:4 ratio. I have samsung pc2700 original and it can take a beating, all the way up to 420 or higher! Using the latest bios, I am able to boot into windows and run benchmarks with 155mhz fsb and 3:4, but when i restart the computer, computer can't boot and the port led says that it is sending the bios code to dram. ?HUH? How could it be that the system could boot fine from a cold boot, but can't boot from restarting.

 

[Ketchup]

That's funny, mine is the opposite. When running high FSB, the computer will not cold boot. When I press the start button, it hands on code 26. I just push the reset button, and everything boots fine. When I do a restart, it also boots fine. It is kind of a pain, but I don't mind, as the performance with the 3:4 ratio is excellent.

I have been using the 2822, 2A31 and 2B04 to hit higher speed. I couldn't get the 2424 stable above 152. I have not gotten above 157 with the 3:4 so it isn't a big difference for me. If you are using the 4:5 ratio, all official bioses work fine.

The problem seems to stem around the 3:4 ratio and the AGP controller. The system is perfectly stable at 160 with the 3:4 until I play a game and the computer freezes. The new Asus P4PE with the 845PE also has a 3:4 ratio, and I am curious as to whether it has the same problems.