CPU and RAM speed drop after startup?

[hellob]

I have an Intel E7300 and an nforce 750i motherboard and I am trying some different overclocks. I had it running at 3.66GHz(10x366MHz) and it worked fine but with a high voltage(close to 1.4v) and I had my ram running at something like an 11:12 FSB : DRAM ratio. I decided today to try something different. I dropped the multiplier to 9 and raise the FSB to 400MHz(1600MHz QDR) and get that 1:1 FSB : DRAM ratio. I raised the FSB voltage to 1.44v and the northbridge, although I can't remember what I set it to.

The problem is that It should be running at 3.6GHz whereas when I check CPUz it downclocks the CPU and RAM shortly after startup.(meaning I can see that it does run at 3600MHz) It downclocks it 3300MHz with a 1466FSB QDR and downclocks my ram to 732MHz DDR. I had the same problem getting past the 1333FSB but raising the voltage of the FSB and Northbridge fixed that and now I am stuck at 1466MHz.

Does anyone know why it is doing this?

ALSO: RAM=Corsair XMS2 DDR2 800 5-5-5-15 8GB(4x2GB)

NOTE: The "sweet spot" for this CPU is 3.6+GHz which is around the limit for this CPU.

Why does this board always become inactive after I post in it.(emphasis on the "I") 30 minutes between my post and the next most recent one.

[TheMG]

It's your motherboard decreasing the clock speeds when the CPU isn't doing very much, to conserve power and reduce noise and heat.

You can disable this feature in your motherboard BIOS, but there is no need to, it doesn't affect performance as it goes back up to full clock speed as soon as you do something that requires lots of CPU power.

If you wish to see your overclock status in CPU-Z, just run something like Prime95 in the background, that should cause the CPU to go to full speed.

[hellob]

It's your motherboard decreasing the clock speeds when the CPU isn't doing very much, to conserve power and reduce noise and heat.

You can disable this feature in your motherboard BIOS, but there is no need to, it doesn't affect performance as it goes back up to full clock speed as soon as you do something that requires lots of CPU power.

If you wish to see your overclock status in CPU-Z, just run something like Prime95 in the background, that should cause the CPU to go to full speed.

When I run Prim95 in the background, the speed doesn't go back up to where it should be. It stays low and I have all power saving features disabled in my BIOS.

[renegadeafk]

Are you using vista? Sounds like your power plan could have been put on "power saver" Go into the control panel and go to Power options and choose "high performance"

[hellob]

Are you using vista? Sounds like your power plan could have been put on "power saver" Go into the control panel and go to Power options and choose "high performance"

The power option was set to high performance while overclocking.

[renegadeafk]

What are your temps? Could be downclocking due to heat.

[hellob]

What are your temps? Could be downclocking due to heat.

I know this thread is a day old, but I don't even have a way of measuring the NB/FSB temps on this board. I can only really measure the CPU and GPU and a few other spots and the CPU under load hits around 65C running Prime95. I have contacted the board manufacturer, but I don't expect a response for a while.(more like another couple days)

EDIT: I was able to get a 1600FSB with a multiplier of 10, but unable to boot into windows. I could boot to windows with around a 1560FSB, but that was unstable. now with a multi of 9, I can't get it to stay at 1600.

[renegadeafk]

hmmm at 65 it shouldn't be downclocking either.

[Bad_Motha]

actually anything over 60C is high and the motherboard will trigger the downclocking for this reason. 60C is not high for a video card, which has a much higher threshold, but for a CPU that is high. Most motherboards are setup to turn off once it hits around 70/80C mark to save it from frying. Even if you could run at full speed, if you're hitting 60C+ while gaming or benchmarks, you're most likely gonna get some data corruption, slowing of the system or BSOD.

[hellob]

hmmm at 65 it shouldn't be downclocking either.

I am going to try seeing if I can change it via the nvidia control panel. Maybe that will allow me to force it to this higher clock frequency.(maybe it is set to load this on startup?)

EDIT: I think it was the nvidia control panel running on startup and dropping the FSB to it's saved default.(367MHz) Now I have it running at a pristine 1600MHz with a 9x multi and with my FSB : DRAM ratio at 1:1 :)

Isn't overclocking just fun? /rhetorical Now to try at 9.5x multi for 3800CPU clock :D

[Bad_Motha]

Try these settings...
CPU VCORE @ 1.488v
(stick to the normal rated voltage for your ram and pci-x slot)
multiplier 9.5x400mhz
Set the memory to 1:1 or AUTO

This setup should give you a CPU setting of 3800mhz
Be sure to setup your ram and pci-x voltages manually.

[Bad_Motha]

If you want to use the software, you're going to have to setup all those BIOS settings you changed to defaults or auto-settings so that the software can be able to have full control. If you know about how to tweak the BIOS, don't worry about doing it through software. If it locks up in Windows while doing it, doing this over and over can cause you other issues. It's best to try through the BIOS. If you go too high and the system doesn't boot/post, reset the CMOS to boot back up again.

Also, even if the system does post, if at anytime you get a BSOD while Windows is trying to load, reboot and go to BIOS and turn the clocks down, as they're too high for the hardware to handle.

[hellob]

If you want to use the software, you're going to have to setup all those BIOS settings you changed to defaults or auto-settings so that the software can be able to have full control. If you know about how to tweak the BIOS, don't worry about doing it through software. If it locks up in Windows while doing it, doing this over and over can cause you other issues. It's best to try through the BIOS. If you go too high and the system doesn't boot/post, reset the CMOS to boot back up again.

I am trying 400x9.5 right now for stability.(instant BSOD @ 1.384v) It was the nvidia control panel interfering with it and setting it to it's stored values. Now testing it at 1.4v.

ALSO: This mobo has a clrCMOS button on the rear IO panel so it makes this fairly painfree.

[Bad_Motha]

You may need to up the clock on the north-bridge but be careful, it's a much more fragile chip, it can burn out @ 90C

[hellob]

You may need to up the clock on the north-bridge but be careful, it's a much more fragile chip, it can burn out @ 90C

I am just going to drop back to 3.6. Prime95 crashes almost immediately with a "prime95 application has stopped working" message. CPU temps are almost exactly the same though.

[Pickanewnameplzkthx]

It's not stable. Turn it down...way way down. Also, for that level of a FSB you'd need some faily high voltages. I wouldn;t recommend that at this point. 60C really isn't that high. It is hot, but not hot enough to trigger the CPU Thermal Throttling. I'm pretty sure the TJMax on that CPU is 90C, but whatever you do don;t let that thing get near 90C.

I can tell you this. Nvidia chipsets can be trouble. It's quite possible you could have a bum NB or a Bum mem controller. Those Chipsets get damn hot. So hot that people who watercool automatically go with a NB waterblock when they have a Nvidia chipset. They can be buggy also, but they are notrious for getting hot...very hot. Hot enough to burn up if you push one without proper cooling and few boards have good enough cooling stock for them.

I'm a beliver that Intel chipset belong with Intel CPU's, but you already have this board now. I'd definitely downclock it. Then ssssslllllllloooooowwwwllllyyyy work your way back up and find the piont where ther system become unstable using only the FSB. That will be your initial FSB wall. Keep the ram out of the equation by lowering it multi. After you do this, test for atability. If it doesn;t pass there, it will not pass if you go any futher.

I cannopt tell hardly anything about this system by the descriptions here. You need to work up in steps and isolate the issue down to one setting only. Isolate the initial FSB wall first. After that we can go from there. If it still fails at stock voltage, and with the mem out of the equation, then I;d suggest testing with MemTest with those same settings and see if you get errors. if you do that would point to the NB being bad or overheating. 750i's are notorious for running very very hot. They also are FSB limited compared to the Intel chipsets.

[Bad_Motha]

The temp triggers aren't based on the CPU installed, it's the motherboard and it's bios that trigger it. For example I have mine setup to shut off @ 80C, this way when it's in use as a server and I'm away, if the cpu fan should fail, soon as it hits this temp the machine shuts off to prevent burning the CPU.

Every motherboard is different though but I'm sure there is some sort of setting in your bios that is causing the CPU/FSB to turn itself down. Either that or it's due to the NVIDIA windows software.

[Pickanewnameplzkthx]

That's technically correct, but most motherboards default to the thermal threshold that Intel recommends. The TJMax on the CPU would be 90C, and the motherboard should be defauted to that unless it's been changed. You can change this, but I doubt he did. I don't mess with them, becasue i control the heat myself through cooling. If I loose a waterpump, the system shuts down anyway, which would be the same as a fan fail detection on a air cooled system.

I highly suspect the chipset though, but he needs to isolate things first. Isolate the FSB, and test RAM, then isolate the RAM, and then we can really isolate that thing down to the chipset. It'll be a dead giveaway when you do that. This is why I always suggest that people OC slowly in steps and test at each step so you can geta feel for the system, and make sure nothing is going wrong along the way. Alot of people will take numbers they find on the Internet and just plug them in, and 9 out of 10 times it spells trouble.

Something tywells me this thing has overheated the NB at some point and may have damaged it, but it's back to square one since he has no progressive info on how it was OC'd, and what it acted like along the way.

[Bad_Motha]

Some boards are just easier to work with then others. I had to RMA one of my MSI motherboards a while back. After 3 good years from it. They replaced it and when I got it I made sure to update it to the latest BIOS version, but for whatever reason the newer replacement board just wouldn't hold up to the same overclocking settings I had been using with my old board. Yet it's the same exact board.