I've got what appears to be an unusual problem. I've only had this CPU a few days and have been gradually working up my overclocks. At the moment I'm testing 180 x 25 (4.5GHz) at a vcore of 1.3875v and I've noticed that when I run P95 small FFT the multiplier drops from 25 to 24 and stays there dropping the CPU speed from 4.5GHz to 4.32GHz. However if I run the P95 Blend test, LinX or a rendering test the multiplier stays at 25. The temperatures haven't exceeded 60ºC so I don't think it's thermal throttling and the vcore isn't that high. The multiplier still drops whether C1E and/or TM Support is enabled or disabled. After fitting the 970 a new BIOS option appeared for 'Performance/Watt select' but it doesn't matter which option from 'Traditional' or 'Power Optimized' is selected the multiplier still drops under load to 24 only when running P95 Small FFTs. Is this a bug in the BIOS or is there something I've missed? As far as I can tell the RAM stays at the same speed of 1800MHz when the multiplier drops.
Unfortunately I never used Prime, only LinX! This is a weird issue and could have something to do with Prime/Windows. I assume you've got the most recent BIOS for your P6T? I dont think you've missed anything; before I got the 970 I remember reading that there are a number of people who are "against" CPU turbo because it isn't a genuine CPU multiplier, which is why some people stick to the non-turbo maximum. I wouldn't worry about it though - I used the 25x mutli and never once saw it drop in anything I did, and I really put the CPU through its paces (eg booting at 5GHz)!!
Thanks for that. Yes, it's the latest BIOS. I think it's some peculiarity of CPU-Z that reports the multiplier has dropped but only with small FFTs in Prime. Benchmarking shows a clear difference between the 2 settings so performance isn't being affected. The unusual thing is that as LinX stresses it more I would have expected the multiplier drop with that if anything. Asus does use the Turbo in a non-standard way allowing you to use the extra multiplier permanently. I've raised a technical support request with Asus to find out whether it's a bug and on the off chance that it is will release a new BIOS. However I'm not expecting an intelligent, intelligible or quick answer but you never know. I'm aiming to find the speed I'll use 24/7 and it's looking like it'll be 4.5GHz (HT On) as I'm getting a ambient to chip Delta of 41ºC which is what I feel comfortable with. I reckon it'd go to 4.8GHz with HT On but I'm not aiming for the highest result just the most usable one on a day to day basis.
I suppose you can conclude that the multi isn't dropping as a result of heavy stress; it is more likely just a bug with P95. As you said, Asus boards let you set the Turbo multiplier permanently in BIOS (as did my Foxconn Bloodrage) so there shouldn't be anything in Windows that is able to override it. 4.5GHz with HT is a great frequency for that monster of a chip!
I had a few exchanges with an Asus Support Technician and they were a lot better than previous ones in that they were in english. However they said that the problem is caused by Prime95. I then did a bit more research and it turns out that the Asus P6T motherboard series has known throttling issues when pushed hard to the extent that Asus created a special BIOS 0006 which has a power limit disable function added which has then had further microcode updates for it to work with the Gulftowns. It appears to be the way Asus implement the turbo boost function that can cause the throttling. I'm trying to decide now whether to risk flashing this BIOS (and work out how to use AUDOS again) or just stick with a 4.42GHz overclock.
I beg to differ on that one. The CPU itself is designed to throttle back if it starts getting a bit to hot. It's basically self preservation to stop it from frying. It will get to the point where it will shut itself down if you push to hard no matter what bios you're running. My P6TD used to start thottling back once it hit mid 70's, which to be honest is about 15 degrees off what an i7 can actually take, but I'd be happier with it doing that, than getting hot enough to cook breakfast on. I could push my chip to 4.4 so getting 4.5 is pretty much as good as you're going to get with out some serious water cooling.
I did some more testing. If I drop the multiplier by 1 and increase the BCLK then it runs at 4.5GHz and above without throttling. The P6T series of motherboards apparently all throttle at a certain power load, not temperature and drop the turbo boost multiplier http://www.xtremesystems.org/forums/showthread.php?210480-Asus-P6T-Deluxe-Discussion-Thread/page72 I've now hit 4.7GHz without throttling so far. Either dropping the multiplier and increasing the BCLK or flashing the 0006 BIOS gives you the option to bypass it.
Known issues. When using the turbo multi, stress the cpu for too long and it will throttle. Dont use turbo multi
This is very strange - not least that Prime95 small FFTs stress the CPU more than LinX (whisky tango foxtrot). It seems to me that this mysterious "stress threshold" on the P6T is unrealistically high since LinX doesn't tip the scales. Yes I know what throttling is and why it's there, but if you read what Big Elf is saying this has nothing to do with thermal throttling - no i7 CPU will throttle at 60C load (if it does, it's a dud). The behaviour of Big Elf's 970 is something of a conundrum because it throttles immediately with Prime95 but not at all with linpack. Not buying this - I've never seen it happen before, and I have run hour after hour of Prime95 with the turbo multi into the 80s and there has been no throttling, thermal or otherwise; it looks as though it's peculiar to the P6T rather than Turbo Boost itself.
According to the power meter my 920 drew around 190W plus a bit for the wattage it requires at idle at full load. At 4.5GHz the 970 is drawing 35W more than that indicating that the P6TD throttles by dropping the turbo boost multiplier at around 220W. What is surprising (or maybe not) is that the Asus Support Technician didn't know this. Reading the link to the thread I posted above it looks like it's the Asus mobo's that do this and specifically the P6T series. It's probably documented in the manual but I can't find it unless it's the 'Asus CPU Parameter Recall' (which it doesn't describe) function. It may be to stop the Mosfets overheating but as they're watercooled now I don't anticipate a problem with higher loads.
I said it was a known issue, as in its a known issue to that motherboard. Happened to mine. http://www.google.co.uk/search?sour...gc.r_pw.&fp=9fe7b03c7fd138e9&biw=1071&bih=611
