Radical_53

Well, normally the CoreTemp reading is much more realistic. And yes, the difference between the cores is there, you can get it away with a different heatsink or so (depends on the heatsink how big the difference is, at least in water cooling it does). If you're running without IHS and with water cooling, then the reading of the board is pretty much the same than on CoreTemp. But like right now, you should rather believe the core 1/2 reading than the normal sensor...

Thanks for the input. Honestly, I haven't yet understood this RAM. I've read reports that sometimes it runs better with really low voltage (somewhere in the modest speed range) and is more stable then (without heat being an issue), then the errors it gives when clocked too high are really weird to me, like the screen flickering and so on. I'll definitely try to start all over with your settings, some things are totally different from what I've started off so far, maybe (and hopefully) it can do the trick. About the timings, can you explain that a little further, or do you have a link to something you might have said earlier, or someone else? It makes sense to me that if it's stable, more relaxed timings won't make it "more stable". But if it's not totally stable with relaxed timings, are you saying that somewhat tighter timings might work then?

Thanks The CPU is stable at 3GHz, runs Prime 24.14 (2 instances) for hours. I used it on a normale Expert board before and it worked perfectly fine there too. The problem just occurs if the machine was off for a while and then is turned on. After 5-10 mins "heat up phase", it works again. Really weird. Yet I think you're right on the memory as I only tested those maximums on Memtest. I first test it there, than go back 10-15MHz and test it on Windows. Or the other way around, if I need a specific speed for Windows (like for a divider), I test this speed + 10-15MHz on Memtest, play around with the settings until it works and then go back to Windows with the lower speed, but same other settings (as Memtest was never to accurate for me when I tried something on the edge). I'll give your settings a try for sure. I'm just a little confused as you use way less voltage on the RAM, but still have way tighter RAM timings. Let's see what that does

I'm running my Venus board with the most actual 406 bios right now, still it's giving me some problems. First, it has some kind of cold boot issue. It keeps restarting when I boot up at 300x10 directly. 295x10 works no matter what. Second, the chipset won't work above 305MHz. Even there it's not 100% stable. Chipset voltage is at 1.59V, more didn't help eiter (310 was not possible to boot). I tried some different settings on the RAM, tighter/looser timings, 166/140 divider, nothing really helped me so far. Now, is there anything *special* that might help? I've seen too many boards working at well beyond 300MHz, so normally it shouldn't be the board itself holding me back. Of course, I also tried to run it with a lower CPU multiplier, didn't help (CPU is at only 1.39V, tried 1.46 but that definitely wasn't the problem either). PSU is not the problem, 12V rail drops from 12.06V idle to 12.04V under full load and stay their (measured, not with the sensors). Is there any special option concerning the CPU that might do this? Or maybe any special RAM option? I fuddled around with *all* the settings to get the RAM to run at really high clocks 1:1 (~280MHz), not enough but normally it should keep them stable when I'm using a divider right? Has anyone seen improvements from a better cooling on the chipset? It doesn't get hot right now, even though the stock cooler is still on it (with AS5 as TIM). Any help would be greatly appreciated. I'm trying a burn in right now, default clocks and raised voltages, but after that I really don't have a clue anymore...

Well, I had the same problem (one core got much warmer than the other, always failed Prime earlier), and the IHS removal did fix the temperature problem. But: Didn't help the Prime thing much, gain was ~50MHz from removing the IHS. So not really recommended if you plan on selling it some day.

Sounds good Let us know if it does the trick and if it's stable for a long run

Well, you might want to try if the chipset temps are ok so far. 1.8V is max as far as I remember, so maybe you could try 1.65-1.7V.

Yes, you've hit max if it still isn't stable with more voltage. If you have some good cooling, then this is the max your chip might want to do.

Did you raise the chipset voltage? My board won't even go to 300 without raising the voltage. If your mem runs fine with the divider and the CPU isn't the problem, well... did you check the temps? Maybe something is getting really hot without the sensors reporting it correctly. *maybe*... PS: I'm stuck at 305MHz myself.. so far no clue.

I pulled myself out of this by setting a temperature offset in Speedfan 4.28. I set mine to 7°C, as then idle temps are just a bit over water temp, which seems so far "ok". Something I can live with at least

... but the never versions mix it up again, like build 550. Yesterday I took some time to measure CPU temps with different water temps. Ambient was 25°C, later 26°C. With fans @12V, my water temp stayed 5° above ambient. Idle, my readings for CPU (C&Q) are 5° below water temperature. Load, they go up to water temperature, ususally staying 1-2° below water. With fans @6V, my water stayes 7° above ambient. CPU reads 4 to 0° below water, usually staying at 1-2° below water. These temps are off at least 5°, more realistic would be somewhere between 5-10°. Also I'm not sure why my temps are jumping this much... it doesn't really make sense.

I think the main thing about dividers is not which one is best, but which dividers run and which don't. Also you should take into consideration that if you're using C&Q, your memory might be clocked differently depending on the p-state of the CPU. Example: If I run my CPU with a 166 devider at 300MHz HT, RAM always stays at 250MHz through the full range of multipliers from 5 to 10. But if I use the 180 divider, RAM will run at 250MHz at multiplier 5, going up to 272MHz at multiplier 10 under full load. Also, there's more than one stick of ram which will run better 1:1 compared to dividers. Max. clock without a divider is usually higher.

Well, you "age" your components with this. Some manufacturers do even burn-in their components before delivery to have lower failure rates or better performance. Like for me, a fresh board or CPU normally doesn't work too well. But after some time they behave much better or work at higher clocks. If you don't want to wait, you do a burn-in to speed-up the aging and to have the components work well right away.

Did you raise voltage on chipset and LDT? Might help you out on that one. I can't get my board to work "correctly" above 310MHz, which normally shouldn't be too much for these kinds of boards (I've seen scores of 350MHz+ more than once).

So far, I was only lucky with 166 and 133 dividers. All others didn't work properly... But: This also might be a matter of your CPUs IMC... you might want to try a burn-in on those components, like giving high CPU and DRAM voltage to them at standard clocks for a couple of hours. Might help.