Competition Benchmarks
First a little bit of an explaination as to what I mean by “Competition Benchmarks”.
Benchmarking has changed over the years, it used to be simply a way to see if your system was performing as it should. Benchmarks are still used that way now, but now there are also competitions based on benchmarks.
The largest is hosted by a site known as HWBot.org, they run an ongoing online competition in a wide variety of benchmarks, compile the results and let you see how you stack up against the rest of the world. This is the world of Competitive benchmarking, it even goes beyond HWBot to live competitions with fairly substantial prizes and money for the winners.
I haven’t made it that far yet, but I’m currently 134th in the world (out of 2363 other “extreme” benchmarkers. Extreme means using dry ice or liquid nitrogen to cool the CPU/GPUs). This is by far my favorite part of reviewing benchmarkable parts, so I am going to run rather more of this sort of benchmark than strictly necesary for an ITX motherboard likely destined for a HTPC.
The benchmarks tabulated by HWBot are the Competition Benchmarks.
For this review I’ll be using the following: SuperPi 32m (CPU speed and Memory speed/stability), WPrime 32m (raw CPU power), Cinebench 11.5 (CPU raytracing, not a HWBot benchmark), 3DMarkVantage GPU score (Mostly GPU). All benchmarks will be run both at stock speeds and at the maximum I can overclock to for each benchmark. This will give an excellent idea of how much more performance this Zotac Z68-ITX WiFi can drag out of a 2600k.
Moving right along, we arrive at the…
Stock Scores
All OC related settings at defaults, XMP enabled for ram. It is important to remember that Turbo Boost will bump the CPU speed up to 3700MHz for single thread loads (like SuperPi32m), and to 3500MHz for highly threaded loads (Cinebence, WPrime).
- WPrime32m: 7.32s
- SuperPi 32m: 8m 48.498s
- Cinebench 11.5: 6.87
- 3dMarkVantage GPU score: 1753
All of which are perfectly reasonable stock scores, they match up well with the stock scores of pretty much every other 2600k motherboard out there.
Overclocked Scores
With those out of the way we can move on to overclocking! The first thing to do before doing any serious overclocking is to ditch the stock Intel cooler and get something with some real cooling power. In this case I’ll be using my trusty ThermalRight Ultra-120 Extreme cooler, with a pair of not-at-all-subtle 5300rpm 120x38mm Nidec server fans. It’s nothing you’d want to watch a movie around, but it is enough cooling power that I don’t have to worry about the CPU overheating even when OC’d to the moon.
I managed to get this CPU up to 4831MHz, and the onboard GPU up to 1563MHz, with the ram running at 972MHz with 7-9-7 timings. This is a CPU OC of 38%, and a GPU OC of 15%. Not at all shabby for a motherboard smaller than a salad plate!
The overclocked scores (and percentage increase over stock) are as follows:
- WPrime32m: 5.164s, 38% faster than stock.
- SuperPi 32m: 7m 8.002s, 23% faster than stock.
- Cinebench 11.5: 9.57, 39% faster than stock
- 3dMarkVantage GPU score: 2116, 22% faster than stock.
All in all, overclocking yieded some very good results! In my opinion it’s definitely worth it to do some overclocking, though aiming for 4.8ghz may be a bit much.
Video Transcoding
The last benchmarking section is a HTPC specific look at video transcoding, I’ll be using Handbrake to convert a ten minute 1080p HD video from h.264 to mp4, and noting the average number of frames transcoded per second.
Initially I was going to test playback as well, but given this board and a 2600k (or any other CPU that fits this board, really) the performance will be through the roof, you would have to be watching a video at something like 6000×4000 and 120hz to stress this setup. As such it isn’t really worth testing.
At totally stock settings, this Zotac Z68-ITX WiFi + 2600k combo manages an impressive 41.4FPS.
Overclocked however it is a staggering 50.9FPS an increase of 23%! Not too many HTPCs around are going to beat that! One thing to note is that I backed the overclock down to a rather more reasonable 4.4ghz(25% OC) for this test, as that is a clock speed that can be run on a 24/7 basis. You can run 4.8 24/7, but it isn’t an especially good idea.
