A Summer Road Trip With the Intel Core i7 975 EE

We have arrived at our first destination, Air Cooling Canyon. Get out, stretch your legs, and no fighting for the bathroom! Here we will explore the maximum frequencies for our benchmark suite as well as the maximum stable 24/7 overclock we can achieve with our Thermalright Ultra-120 eXtreme CU cooler. We have already seen the hardware lineup so here are a couple quick photos of the setup during this stop of our road trip.

We ran the system with the fans in push/pull with another fan blowing down on the memory and helping move air across the ground level of the motherboard. This is our typical air cooling setup when testing anything. We've said it before and we'll say it again; friends don't let friends run memory without a fan. So when you are out drinking with your friends this summer, enjoying a few too many fine import beers on a patio, don't forget that piece of advice and keep those fans on high. Seriously though, do everyone a favor in that situation and just crawl home. Your knees won't love you the next day but it is the right thing to do.

We can also see that the temperature just above the setup was around 25C +/- a degree or two for irregularities in the digital thermometer. Here is what we managed to squeeze out of this setup for our 24/7 stable overclock:

Overclocking the i7 975 is not unlike any other i7 processor. With an unlocked multiplier you can simply increase the CPU multi until you hit your limit, or you can play with base clock and the CPU multiplier to find that absolute maximum the processor can handle. We managed a lofty 4364MHz overclock using a rather high base clock. There is no performance gain to the high base clock, we just wanted to show that the processor had no problem running up that high. We also found that the 23X CPU multiplier really didn't mind being overclocked at all. A general rule of thumb with i7 processors has been to stick to odd CPU multipliers as they tend to require less vCORE to stabilize than even multipliers do. This continues to hold true with the new D0 stepping processors as we witnessed with this chip, and other retail 920's.

To say all Core i7 975's will reach this kind of a stable overclock is ridiculous, some will, some won't. There have been a number of examples in forums around the internet where users have already clocked retail 920's and 975's alike to these levels and more. We were using our air cooling setup - obviously - and temperatures really were not a big issue despite the high vCORE. We only crossed 90C for a short period of time during our LinX run. During Prime Blend and OCCT, the core temperatures - as read by Realtemp - would never cross north of 85C and often hung out around 80C. In the screen shot above, we can see that the voltage displayed in CPU-Z is 1.458v but this is with vDROOP enabled and under load, we would only see 1.406v for vCORE when measured by a digital multi-meter. Here is a chart of the temperature reading from Realtemp for the individual cores during our LinX run for the stable screen shot above.

You can sort of see the heat cycle of the A/C unit in the testing room. About 10 minutes into the test the A/C cycled on for a few minutes then off, and back on again close to the half way mark where the temperatures leveled right out. This is a great indication of how much ambient temperatures effect the core temperature of your processor when using air cooling.

The ambient temperature reading never dropped below 23C during this testing period and didn't go above 25.3C at any time. Like overclocking results, temperatures should be considered on a case by case basis. Just because this processor runs cool at 4.36GHz and 1.406v under load for vCORE, doesn't mean all i7 975 D0's will. We have an i7 920 D0 sitting here that would melt the TRUE CU at 1.40v...and we are not kidding in any way about that. Temperature characteristics can vary greatly from silicon batch to silicon batch. This is simply a graph of what our particular Core i7 975 did during our testing as far as temperatures are concerned.

One thing that does seem to be consistent about the D0 stepping processors, is the fact that our 4th core reads temperatures noticeably lower than the 1st core. On average, there is a 10C discrepancy across the cores from 1st to 4th. This is consistent with other i7 D0 processors we have tested so far and a lot of reports coming from end users in forums. So if you see this type of temperature difference across your cores, don't be alarmed, it seems to be the norm. Now, let's have a look at the real reason we stopped here, the all out top frequencies we could achieve on air cooling with our suite of benchmarks.

CPU-Z Validation click for full size... CPU-Z Validation	SuperPi 1M click for full size...	SuperPi 32M click for full size...
Hexus PiFast click for full size...	Aquamark 3 click for full size...	Futuremark: 3DMark 01 click for full size...
Futuremark: 3DMark 03 click for full size...	Futuremark: 3DMark 05 click for full size...	Futuremark: 3DMark 06 click for full size...
Futuremark: 3DMark Vantage click for full size...	wPrime 32M click for full size...	wPrime 1024M click for full size...

There is a lot of data up there so let's sum it up in one sentence. The Core i7 975 EE is fast, real fast. Even with just air cooling, this chip excelled, staying cool enough to push over 4.8GHz for SPi 1M and CPU-Z validation. We put a cap on voltage of 1.500v as anything over that would be getting higher than we are even comfortable with. Keep in mind the vCORE droop this board suffers from is still in effect, especially with multi-core benchmarks. The scores put up for a single GTX 295 in Aquamark and 3DMark 05 are especially impressive for having the CPU simply on air cooling. To put things into perspective, our first Core i7 965 we bought here at Hardware Canucks is only capable of 3DMark 06 at 4.8GHz...under LN2 cooling. Running 4.5GHz kind of surprised us at this point in our road trip. We knew we would get there, just not on our first stop. Here is a chart outlining the clocks we reached through our entire suite of benchmarks.

There you have it: one spot on the map checked off and we are already impressed. Unfortunately the old overclocking adage goes: if the chip is good on air, it often struggles under sub-zero conditions. This of course isn't always the case, but is something a lot of elite benchmarkers go by when estimating the abilities of a processor. We are about to find out either way so everyone back in the bus and get your mitts and toques out for the next stop on our journey, Phase Change City.