My 3.0 GHz haswell never breaks 60 on [I]stock[/I]. Granted, I use Arctic Silver 5.

3.6 GHz at 1.18V hitting 100C under stress sounds a bit wrong. Reapply the paste would be my first thing to try. I was thinking the installation of the heatsink might not be completely perfect either but by reapplying the paste you'll get a chance.

In the mean time, you could try to twist the heatsink around just a few degrees or as much as you can without breaking anything and something might slip into place. I did that once with an older computer and when I looked at a log of CPU temperature there was a sudden 10C drop during the time I was fiddling with the cooler.
[LIST][*]You could always upgrade to the 120mm version of your heatsink (a Hyper 212) for more cooling power.[*]You could use two fans on that heatsink, I am quite sure. You could buy one (or upgrade and buy two). A local shop would be best if possible since shipping time and cost is a lot for two small fans. Something like [URL="http://www.newegg.ca/Product/Product.aspx?Item=N82E16811999617"]this[/URL] would work well. [URL="http://www.newegg.ca/Product/Product.aspx?Item=N82E16811999613"]Something crazy like this[/URL] could work as well if you're not needing this thing to be super quiet.[*]It sounds like the voltage controller is directly on the CPU as well. What were your first gen i7 temperatures like? Voltage? Frequency?[/LIST]

[QUOTE=Prime95;352961]I'm upgrading my i7-860 computer by replacing the motherboard, CPU, and memory. The CPU is the Haswell 4670K, memory is DDR3-2400.

I recycled the i7-860 CPU cooler: [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16835103064[/url]

Right now, with CPU at 3.6GHz, voltage at 1.08 (I can't turn off adaptive voltage, so this should mean 1.18V running AVX instructions), and memory at DDR3-2400, I'm getting temps of 90C running prime95 (up to 100C running the small torture test).

Does this sound right?

Tomorrow, I'll try either 1) reapplying the thermal compound, or 2) using the stock CPU cooler, or both to see if temperatures get any better. I had hoped to get this running at least somewhat close to my first Haswell box running at 4.1GHz, 1.2V (adaptive voltage off), DDR3-2400 and temps of around 80C.[/QUOTE]

That's the kind of temperatures I saw with the stock Haswell cooler at 3.4 GHz. It couldn't keep up. I have a Hyper 212 Evo and it keeps it at 80, give or take a few degrees. Well worth the small amount of money.

Anyone tried de-lidding their Haswells and running prime95?

Given intel used different mechanism to attach the lid to the chip for ivy bridge/haswells, and shown not to be as efficient.

I'm keen to try this myself.

-- Craig

[QUOTE=Mark Rose;352965]That's the kind of temperatures I saw with the stock Haswell cooler at 3.4 GHz. It couldn't keep up. I have a Hyper 212 Evo and it keeps it at 80, give or take a few degrees. Well worth the small amount of money.[/QUOTE]

Was that running v28 or v27? What T-diff do you see between the 2, running same #threads/cores/FFT-lengths?

[QUOTE=ewmayer;353007]Was that running v28 or v27? What T-diff do you see between the 2, running same #threads/cores/FFT-lengths?[/QUOTE]

I didn't see any difference between v27 and v28. I assume the fan spins faster.

RMA time?
 

[QUOTE=Prime95;352961]Tomorrow, I'll try either 1) reapplying the thermal compound, or 2) using the stock CPU cooler, or both to see if temperatures get any better.[/QUOTE]

I reapplied the thermal paste, no better. I tried the stock cooler and it still runs way too hot.

Finally, I reset the motherboard to its defaults. With the stock cooler, torture test temps hit 100C. At this point, I think RMA is my only option.

For now, I changed the VCore to 1.0V, underclocked to 3.2 GHz with no turbo, but memory at DDR3-2400. Temps are at 85C, we'll see if it is stable overnight.

On Haswell, if you're going to hold 100% load for any more than a few seconds, the stock heatsink is hopeless. The Hyper TX3 is a step up, but check the fan speed is ramping up to 2,800rpm during the toture test to be sure it's working correctly.

Rather than looking at temperature, get a copy of HWMonitor (it's a free download) and look at the power drawn by the CPU. You can compare it to your other Haswell, at the same settings, to determine if the power drawn is similar.

If the power drawn is a few watts within each other, then you'll need to beef up the heatsink. Given your aim of 100% load endurance runs, I'd recommend something watercooled.

Oh well I could have told you the stock cooler on the 4670K was gonna be BAAAAAAAAAAD back when Sandy Bridge was hoity toity.

To be frank I don't even know what you are RMA'ing. If it's the chip, I would hold onto it until you get some slightly better cooling. It would be really nice if a person could slip a tiny wire for a thermocouple between the CPU and the heatsink just to see how much cooler the lid is compared to the die. By the sounds of things, Intel has used the same crappy internal heat spreader system as they did in Ivy Bridge, AND they've added another heat source.

I'm looking for AMD to come racing back into first place in the enthusiast market really, really soon, with the way Intel has been designing its stuff lately.

Stock cooler, stock settings, can't hold 100% load? Honestly, WTF? How does that make any bleeding sense?

Actually, I have noticed that the Intel stock coolers are getting smaller and smaller. Back in the Pentium D times, I remember getting a pretty heavy cooler with a large copper core. The copper core got smaller with each generation, and the aluminium funs getting shorter too. Perhaps for K series processors they should just stop bundling them together, it doesn't make sense for this target segment.

[QUOTE=db597;353106]Rather than looking at temperature, get a copy of HWMonitor (it's a free download) and look at the power drawn by the CPU. You can compare it to your other Haswell, at the same settings, to determine if the power drawn is similar.[/QUOTE]
Seeing my 4770K without OC reaching more than 140W for a few seconds reminded me how little TDP means.