i5 750 Results, Post your comparisons.

Ok, since I was needlessly hijacking a thread, I am starting this one. 87dtna and I were making comparisons on our systems.

They are not identical, but we were trying to get as close as possible.

SameSaveMax pointed out something I thought was interesting, he noticed that even at a lower Bclk overclock, he was getting similar bench results. It turns out his board was pushing a 24 Multiplier. Wickedout also pointed out that 8 GB of Memory would make it harder to reach top clock speeds, which turns out to be correct. I will explain later.

So I started messing with different settings on my Gigabyte P55-UD4 and came across a setting called C4/C6/C7 State Support, which when enabled along with C1E and EIST causes the Motherboard to scale to a 24 Multiplier for short bursts during CPU usage. It is assumed at this point that it only occurs on one or two cores and not all 4 because when Prime is enabled, it stays stable at a 21 Multiplier.

So with that little tidbit found out, I started making comparisons from Raw Overclock of a 21 Multi and 160 Bclk. 160 X 21 is the standard Overclock using Gigabyte's Easy Tune 6. So with that in mind I started to up the Bclk with all of the aboved named settings disabled, except for turbo (for the 21 Multi) and was able to get to about 170 before I needed to raise Vcore. So pretty much 170 is a gimme. After that I started raising my Vcore, I stopped initially at 1.5 because I read somewhere that Intel spec was 1.55V.



According to this chart, I did not want to exceed 1.55V especially only on air, which turns out to be a good thing. At 1.5V I am able to achieve an overclock of as much as 185 X 20 (Turbo Boost) which gives me 3885 MHz with a boost of 4440 MHz. One problem, this is boarderline stable. Yes... I can run it at this speed, but after about 20 min. of Prime or running Intel Burn tests the Temps will exceed 90 degrees. I tried 190, but with all those settings turned out I can barely make it into windows without locking up.

So Next I tried w/o all of the C1E/C4/C6/C7 and EIST enabled.While I had 8 GB of ram in, I was unable to go past 200 Bclk and a 21 Multi for a 4200 MHz Overclock. With only 4GB of ram (I did forget to put Ram in Interleaved Mode so I don't know if this would have helped at all) I was able to achieve an Overclock of 4410 MHz. At this speed temperatures were obviously an issue, so I won't even discuss that. Running even prime would have been futile. I did however run Super Pi and scored 9.550s. Here is my CPUz Validation:

CPUz Validation

So back to my theory. While running at 185 Bclk X 21 with C1E/C4/C6/C7 and EIST Turbo Boost 3885MHz (4440) enabled, I was able to achieve a 9.698s on 1M Super Pi. When compared to the 4410 Overclock without those settings I got a 1M 9.550s Overclock. Both of them are pretty high temperature wise, with one exception. With all of the power saving functions, lets call this Soft-overclock, the processor throttles back so that idle temps are at about 27-30 degrees Celsius. On the other hand the Overclock with the power saving functions turned off, lets call them a Hard-overclock, the temperatures idle at about 45-61 degrees Celsius.

So it would seem that whether you choose the Hard-overclock or the Soft-Overclock, you end up with almost the exact same results speed wise. Obviously more testing needs to be done using a handful of popular bench-tests. Given the "Turbo Boost" Mode hangs more hand in hand with lower core usage it will probably be very popular with the average Overclocker. Because the speed of the Turbo Boost switching is so quick, it is hard to say what is going on inside the processor, it is hard to map because of how quickly it switches. I would say the differences will become apparent whether your trying to do a Gaming type Bench (Speed) or a Encoding type Bench (Power). If your trying to do both and/or you are a Hardcore enthusiast, then you will probably end up using the Hard-Overclock, however a good cooling set-up will be mandatory. If However you are using it for both and Heat is an issue for you that you do not want to spend another 400 dollars on an WC setup, then the Soft-Overclock might be a better route.

I know I skipped a bunch of small details like what happened from increments between 180 and 190 for example, but I didn't want to bore anyone. If anyone has any thoughts feel free to exchange them.

Well, thanks for memtioning of what I have started. Though I'm not a professional bencher nor gamer so I only tested what is "real world" test to me. I would thought of it as a foundadtion for anyone whom would takean interest about Intel's Turbo Boost feature.
Throughout the process overclocking of both Turbo OFF/ON, I do believe with Turbo ON has advantages for everyday usage. Energy, Cooling and Speed. Those are the three advantages that I found.
Well, as far as for Brutal-Force's testing. It seems that he has heat issue and possibily auto voltage setting. Eventhough his and mine are both from Gigabyte, but it seems the P55-UD3R that I use must be a solid overclocker. Which I've used AUTOVCORE throughtout the entire testing from 165~185bclk with Turbo Boost enabled. It proved lower heat, voltage and maintained stable through the test.

Rightnow, I'm finalizing of another alternative which pushed a litte further with Turbo which I will update in my post very soon. Well, not much of a testing but it is a little faster. Cheers.

Yeah, Auto Vcore hasn't worked out too well for me , would have been nice though. It may be due largely in part to my 8 GB of Ram. If It was a huge difference, I would consider dropping to 4 GB, but realistically 10 Bclk isn't going to do it for me.