Enermax: When did you begin overclocking? Why did you get interested in overclocking?
Massman: I began overclocking about 8 years ago when I bought my first desktop computer. At the time, I couldn’t afford to buy the latest high-end hardware, so I had to settle with an Athlon XP 2600+ and Geforce3 Ti200. Unlike many other overclockers I did not get involved in overclocking because I wanted better performance in games, but actually just to get more FPS* in the 3Dmark01 benchmark, which was at that moment the most recent 3D benchmark available. Very soon, I joined the local overclocking forum and was able to compete against people with pretty much the same hardware. The competition itself was (and still is) the biggest incentive for me to keep pushing the different platforms I test.
*FPS = Frames per second: frames frequency rate
E: You demonstrated in European GOOC final that you currently belong to the best overclockers around. What are your basic principles to be successful in overclocking?
M: To start with: there’s no such thing as the best overclocking method. As I meet more and more international top overclockers and I see them do their thing in real-life and it became apparent that each and everyone seems to have his own method of getting the most of out their system. Some spend hours and hours trying out each setting on air cooling, while others go extreme from the minute they receive the hardware. Even the order in which parts of the configuration are tested varies from person to person, not mentioning the way voltages are set and tweaked.
Personally, I’m a fan of the so-called zone-method. Basically, when I start overclocking I try to find a few combinations of frequencies and voltages that are stable. Each of those combinations is a zone from which you can start tweaking again. Preferably, the number of changed settings should be as low as possible in each zone, so that you have a good overview on how the system is configured. Too much changed lead too many difficulties trying to tweak more. This is the same method Thomas and I applied in the GOOC contest: in a few minutes time we find our safe settings and from that point we see where extra cooling and voltage led us.
E: Which components are the most important for overclocking?
M: Having luck! No, I’m joking, although luck is a significant part of the overclocking process as it’s pretty obvious that certain samples just overclock better than others. However, lately it seems that more and more people are stressing this luck-factor, whereas I’m still convinced that no hardware overclocks itself. In other words, even if you have the best sample in the world, if you don’t know how to overclock it, it’s worthless.
Apart from that, it seems that having a good CPU becomes more and more important these days, especially if you want to score in the really popular benchmarks. Note that your configuration is as powerful as the bottleneck of it. Is your GPU weak, then that will be the most important component to overclock.
E: Is there a difference between Intel and AMD CPU? Which CPU is the right one for overclocking? Which graphics card would you recommend for overclocking?
M: For the extreme overclockers, there’s a significant difference between AMD and Intel CPU’s: the latest AMD’s do not have this thing called coldbug (CPU fails to work below a given temperature), which means that you can cool it down even below minus 200°C (with liquid helium, for instance), whereas the latest Intel CPU’s often have this bug around minus 100°C, which means that the overclocker has to actively control the temperature to keep the system stable. Although the latest Phenom-II is not that bad in terms of performance, I’d choose a good Core i7 over the Phenom-II, simply because a good sample will lead you to 5GHz stable on single stage cooling (~ -35°C) which is good enough to run a few decent scores. However, for different benchmarks, there are different platforms you can use to reach the highest scores. A good overclocking E8600 is still priceless in benchmarks such as SuperPi and Aquamark.
For people who want to try LN2 for the first time, I’d recommend the Phenom-II, for those who want good scores very fast, I’d say i7 and for those not shy of a bit of hard work I’d say go for the E8600.
For high-end scores, only the fastest GPU’s will provide you the power to play along the other top-overclockers. However, middle-end cards are just as fun to play with thanks to HWBot.org. Basically, you can compete with any type of graphics card in its own hardware category, which allows people on a tight budget to compete as well. It’s actually very much fun to overclock the non-high-end cards: recently, I pulled out a 9500GT and modified the voltage controllers to compete in the online overclocking competition of HWBot.org!
E: Before the start of an overclocking contest, is it possible to make a difference in overclocking potential of graphics cards/CPU?
M: Ah, this is a very interesting question because participants have been complaining about bad samples messing up the competition and in particular the ranking of the competition. I know that one of the Gigabyte employees tested each 965 processor for its overclockability and matched them all in terms of raw cpu frequency, which probably was a very stressful task knowing he had to test 20 samples or more. However, it’s incredibly difficult to find 15 samples that overclock exactly the same, because all samples react differently to temperature and voltages. Furthermore, there are certain samples that have a better overclocking Uncore frequency, which allows high memory frequency. For contests, we also have the graphics cards and memory to be tested if we want a really fair competition. That’s just not doable.
For me, these live competitions bring a different aspect into the overclocking scene: not only we have the overclocking, modding and tweaking skills of an overclocker, there’s now also the time restriction which plays a big role. It’s very different from overclocking at home, which makes it more fun. At least, for me it is.
But, when overclocking at home, it’s sometimes interesting to check the stepping/batch* of a CPU. Certain combinations will give you a higher chance to have a good overclock.
* Due to the stepping (version number) of a CPU you can identify its “overclockability”.
E: Choosing the right power supply for overclocking, which technical features and performances data are the most important?
M: The most important aspect of a PSU is its stability: if the PSU is not stable, it’s very difficult to reach high frequencies and high scores. So, the amount of amps available per voltage rail and number of different rails are pretty important. Also, to have a clean working desk, it’s easier to have a modular PSU, so you can only connect the wires you need.
E: We saw during setup period overclockers doing a lot of BIOS configurations. Which configurations do I have to check or to change?
M: For the Core i7, there are a few main factors that have to be tuned to get a good result: temperature, vcore, QPI/Dram (Cpu VTT), QPI PLL and vdimm*. The vcore is maybe the most important one, because this one will help you predict how far you will go. Basicly, it’s important to know at what voltage the cpu is running most stable and if more voltage gives you higher frequencies. For the Core i7, it seems that each cpu has a different sweet spot for the vcore. For instance, the one I have here at home only performs decent at either 1.488v or 1.52v, whereas the really good cpu’s scale up to 1.6v or even 1.65v. The QPI/Dram voltage is important when you want to set the vdimm over 1.65v: Vdimm has to be equal to or less than QPI/Dram + 0.5v. In addition, the QPI/Dram often helps for higher frequency
*vcore: Core voltage; QPI/DRAM (CPU VTT): voltage of CPU busses; QPI PLL: processor clock rate; vdimm: supply voltage of the motherboard’s memory
E: How do you get your system components prepared for extreme overclocking conditions? Why do you for example have to isolate the CPU with sealing putty?
M: The biggest problem with extreme overclocking, and extreme cooling, is that the temperature of the components is lower than the ambient temperature, which causes condensation, which will lead to short-circuiting and most likely hardware death. So, to prevent the condensation to make contact between two electrical points, we insulate the hardware. This can be done with plastic spray, armaflex, vaseline, eraser and some even tried baking flour. Personally, I prefer a coat of plastic spray and paper towel. This has given me the best results in terms of preventing condensation and easy to clean up.
E: There are quiet a lot overclocking software programs. Which are the most important and how do they work?
M: There are only a few programs I really use for overclocking of which Setfsb and Rivatuner are my favourite ones. Setfsb is a very easy-to-use software program that allows you to change the frequency of the FSB* (or BCLK** for i7) on the fly and due to its simple layout it hardly ever makes the system crash. Rivatuner I use for overclocking the videocard, again, because of its simplicity.
New motherboards, like the Asus Rampage (2) Extreme and MSI 790FX-GD70 offer a hardware feature on the motherboard to tune the frequency. If it works well enough, I even prefer that over a software utility.
*FSB = Front side bus: interface between CPU and northbridge, which connects CPU with other components
**BCLK = Baseclock: determines clock rate of i7 CPU
E: Cooling plays a major role in overclocking. Do you have a strategy and some special tricks to reach best possible cooling performance during overclocking contest?
M: The main issue with live overclocking sessions is the time restrictions, so there’s very little room for mistakes. Basically, we start running the cpu on LN2, but with the temperature kept around +20°C, which is pretty much like running on air cooling. In this stage, we test all components, except the CPU, for overclockability and tweak the operating system for optimal performance. Only when we know that everything is ready for subzero temperatures, we decrease the temperature.
With Intel processors, it’s important to find the coldbug and coldboot limitation*. I don’t spend too much time figuring out how low I can go, because you will always find out during the session. Mostly, I start at around -100°C and whilst overclocking I try lower temperatures. You don’t always find the coldbug, though: at the GOOC competition in Prague, we know the cpu was able to run at -125°C, but didn’t go below that value.
*Coldbug and coldboot features of CPU cause a boot fail at low temperatures (usually beneath 0°C).
E: Is it possible to reach outstanding benchmark results without using liquid nitrogen? What is your personal recommendation to an overclocking beginner for best possible cooling performance?
M: It’s definitely possible, but not easy. A lot depends from the platform you’re overclocking with: using an i7, it’s not impossible to reach 4.5GHz on air cooling and that will give you a very decent platform to get high scores with. The Phenom-II only hits the high frequencies on LN2, so without it you won’t be able to reach 6GHz or close to the top scores.
For beginning overclockers, I would suggest Dry Ice. That gives you around -78°C, which is definitely better than plain air or water cooling and lets you get acquainted with the combination of overclocking on the one hand and making sure the temperature is okay on the other hand. However, it requires less attention than LN2 cooling, so there’s much room for beginner mistakes. For the first LN2 session, I’d recommend a joined benchmark session with someone who already has some experience. Just to point out the difficulties and to show the basic principles of LN2 overclocking.
E: Enermax is sponsor of GOOC 2009 and powers participants with high-end power supply Revolution85+. How do you like it? From your point of view, what are the main advantages of the Enermax PSU?
M: When overclocking, I want to spend as much time as possible on the system itself and as less time as possible on anything besides CPU/GPU, so it’s for me very important to have a PSU of which I’m absolutely sure it’s stable and will provide the power I need. How the PSU is designed and how efficient it is, are a less important aspect of the PSU when overclocking. Maybe interesting to note is that the PSU used in the GOOC European finals was a modular one, which is very handy to work with as you only need to hook up the cables you really need. Therefore, the setup is much cleaner. Since the PSU gave us 0 issues at GOOC, I’m very positive about the Enermax Revolution85+!
E: Thank you very much that you took the time to explain us the art of overclocking. Good luck in the GOOC final in Taiwan!
