Labs

Target Attack 1v1 Overclocking Concept – Test report

It’s been in our minds for more than a year an half : How can we extend in a practical way the capabilities of the current 1v1 live overclocking competition format? We’ve […]

It’s been in our minds for more than a year an half : How can we extend in a practical way the capabilities of the current 1v1 live overclocking competition format? We’ve actually conducted some private tests earlier this year under the internal code name of ‘Versus OC’ but this is the first time that we tested one of the possible formats in live in public. Let us share with you some of the opinion and feedback from this test-session.

The gamification of overclocking

How else could we do live competitive overclocking ?

The possibilities are almost endless. Seriously! All sorts of “OC Games” can be imagined. But the key isn’t to just find the longest list of ways to play, but figuring out the right formats that are both technically challenging for the overclockers competing, comprehensible for the overclocking-beginners as well as entertaining for by-standers to the likes of a live stream audience or show crowd.

Figuring the right format boils down to finding out the right dosage of :

  • Technical challenge and skills needed (is skill and knowledge require to win?)
  • Simplicity of the format (can people understand the rules within one game?)
  • Rules of games (how long, how short, how many steps, what rules govern the game?)
  • Random variable ( some people call it luck ) – in all games, there is some sort of wildcard that affects the game outcome to spice up the outcome – and no, condensation isn’t one of them ^^

So yes, the whole thing boils down to : How can we “gamify overclocking” while keeping intact the roots of this hobby? To this date, the questions remains open – but these first real-life tests are our take on how we see it possible in a practical way.

Target Attack 1v1 Scoreboard

We are not claiming to have invented anything here, these are simply our thoughts that probably others had before. This is the game played according to our rules – anyone can come up with his own variations and join the discussion on our next tests. ( The OCTV Crew is looking for your feedback, let us know :) )

The importance of testing these alternative formats “live”

Conducting a live tests (like here at the Summer 2016 OC Party) was a thrilling experience. It’s a passionate challenge that required two main ingredients to be present for it to work:

  1. Stream setup: OCTV’s side of things, usually under-the-hood but very important for this test. It requires us to be able to capture both screens (we can go up to 8 simultaneous currently, overlay telemetry data, combine visual cues to the game progress (scoreboard / lives info and timers) and blend-in a simplistic commentary (more complex and evolved types of commenting will follow). We will talk about the telemetry overlay used in these test via our new project “overlay.live” later.
  2. Two volunteer players: This is about the overclockers themselves. For these test, marc0053 and steponz both volunteered to be our lab rats (well fed with pizza & beer !). They came prepared with two X99 systems (specs were up to their preferences) and a huge amount of time and patience to go through the testing protocols until 3:30am – big up to them!

With such a team, we had the perfect environment to test the format in a similar fashion as it would be used for example in matches for tournaments like the HWBOT World Tour. 2 overclockers, 1 game, 1 audience, 1 winner.

This way, live, we could quickly test, asses through direct feedback and validate the tested format. It worked like a charm :)

Our take on target competitions

Target competitions refer to competitions type in which you have to get the closest to a certain benchmarks score arbitrary defined.

The current way of doing things

This format have been used over the past years such as in the GIGABYTE Target OC competitions and during live event in the HWBOT World Series 2015 Competition

The current limitations from target overclocking:
Time: The amount of time required to hit a target was either too long or too short (read : you either hit the exact target quickly or spend 3 hours just being close to it)
Audience interest: Simple, it is non existent. Because there is no tracking of score evolution overtime (and no obligation to submit for qualifiers as a score appears) – it is a little bit like a Nascar race without the crashes and loop time tracking. Zzzzzzzz (even other overclockers would fall asleep waiting for the end of the round)

The new way of doing things

From the learning that came out of the HWBOT World Tour events we did in 2015 and 2016, we realized that a lot of overclockers appreciate the target overclocking stages as these usually are the ones where hardware luck does not impacts the outcome (unless target is a Global Top Score or World Record). Previously one of the major drawbacks was the lasting duration of the stage – usually an hour or more. Our take: “live matches can solve this” – let’s see how.

New format proposed: Target Attack 1v1

The idea behind the target attack actually came from a completely unexpected route : not from a “how to improve the existing target score mode” but by brainstorming from a “1v1 time limited format” perspective.

Trouffman: “The target attack was definitely an interesting format and proved overclockers to be challenging and fun”

So this is what we came up with:

  1. Multiple waves of targets : we decided that going for one single target per gamer (or match) would not work in a 1v1 approach, so as we confirmed by our internal testing, we went with the multiple sets per game approach. In our tests, 1 set meant 1 target with a maximal set time limit of 5 minutes. A cool-down period of 1 minutes is enforced between each set. For a score to be valid, the benchmarks must be started after the beginning of the set.
  2. Limited lives : 1v1 overclocking is fun, but repeated failure (BSOD, cold bug, etc) with close to 0 consequences over a certain battle duration does not make sense. We introduce the concept of lives. At the start of each game, both players get two lives. The way you can loose or earn lives is up to the rules of the game. We went with:
    1. Win a life by being the first to hit the exact target (max count set to 3 lives)
    2. Loose a life if both players miss the target and you are the furthest away
    3. Both Loose a life if both players miss the target and are the same distance (relative difference to target score)
    4. Loose a life if no score
  3. End of game : the game ends when one player has no more lives – in that case the player with remaining lives wins. Or the game ends after a set limit of 10 is reached. In such case, the player with the most lives wins. If the count is identical an tie breaker set is initiated.

You can find the rules of the Target Attack 1v1 over here: http://www.overclocking-tv.com/games/target-attack-1v1-rev-1-0/

Watch the replay of the match and give us your feedback

Target attack 1v1 format evaluation

These were our three main angles of evaluation : fun, entertainment & technically possible.

  • Fun : Our two volunteers (Steponz & Marc0053) enjoyed the format and to a great extend this triggered another discussion on the benchmarks designs and choice (more on that in another piece)
  • Entertaining : As the two contestant were battling with a big smile even tho the lack of sleep was definitely a challenge, the viewers on the “TEST STREAM” had a lot of fun. As for the commentary, we kept it simple on purpose. Of course there are a lot of possibilities here and improvements to be made with more active / technical commentary and post-game analysis set by set.
  • Technically possible : the purpose of testing this live at an event was to validate the broadcasting setup, technical solutions and viability overall. All of these have been confirmed and from this point of view, the format could be replicated to a World Tour style tournament without apparent challenges.
Target Attack 1v1 testing with Steponz and marc0053

Target Attack 1v1 testing with Steponz and marc0053

We went to ask to the participating overclockers what they thought about the format, here is what they said:

Marc0053’s feedback:

Last Saturday at the 2016 PA, USA event myself and Joe Stepongzi participated in a beta competition called target attack. The goal was to be the first person to achieve a random Cinebench R15 score chosen by Xyala and Trouffman within a set time of 5 minutes. The set would be over as soon as the target was hit or the closest score to the target after 5 minute would win. We were given 1 minute to prepare between matches. The idea of target attack was quite unique and exciting as 5 minutes goes really quick and you can’t waste too much time tweaking memory, uncore and core clocks in bios, in fact I lost many rounds by doing this. The time it takes simply to boot up your pc can easily lose you 1 minute of the round. After a few trials I quickly learned that booting up into the OS and tweaking simple parameters inside the OS was key. You need to react quick to the bench score and change parameters on the fly to get closer to the target. The fact that each round was random this meant that sometime you need to go full out and reach max frequencies and sometimes near stock.

We started this beta competition near 2 am and were tired but once we started both me and Joe had a sudden adrenaline boost to be the first to get the target scores! We tried both Cinebench R15 and Geekbench. Since this competition style only allows 5 minute per round, quick benches are required such as R15. Geekbench was simply too slow to run and you could get maybe 3 or 4 runs. I think most overclockers who participated in the 30 minute showdowns in this year’s world tour will love this style of benching as it requires quick thinking from beginning to the end and simple errors can cost you the round.

I have to say the amount of effort being put by the HWBot staff and volunteers is simply amazing to make overclocking competitions better and better. Keep up the great work guys!

Steponz’ feedback:

I thought the concept was interesting and had possibilities, but when we were actually doing it.. I really enjoyed it and kept wanting to play… r15 worked very well but geekbench not so much. I think 3d11 physics (Ed. 3DMark11 Physics sub-test) would work very well in that concept or FS  (Ed. 3DMark FireStrike).

Format challenges

With such formats, we noted certain challenges that will have to be addressed on the next occasion:

  1. Benchmark choice – As the second run using GeekBench proved it, ( we would need more testing ) the duration of a given benchmark has to be on par with the actual set duration time. Cinebench R15 is in the 15-20sec average duration but Geekbench is around 65 to 75 seconds. This might not be much difference but on a 5 min (300 sec) set time it changes a lot the dynamic. A proposed solution could be to adjust the set time to something like : set time = 5x benchmark run duration + 3min (for tweaking)
  2. Score update & follow up : The score update as it’s done today for the broadcasting backend is a manual process. This locks in an important amount of manpower for such a low-value task. This challenge is also faced on the classic 30min 1v1 matches as we’ve seen it on the world tour. An ideal solution would be to have a communication done by the benchmark itself to a “game server backend” for instantaneous score collection and scoreboard updating. Having this would permit more complex formats with results base on calculations etc.
  3. Rules adjustment & future mods : This format appear to be a winner as a “Base format”. We can use this format to build up game variations and moving around pieces to satisfy a different game need.

Target Attack 1v1 Test Conclusion

Overall the testing of the Target Attack 1v1 concept format went really well. No major technical issues were face on the broadcasting side of things. And no issues were face on the player side of things. The rules as they’ve been engineered at this point allows both players and the audience to have fun during the game and even ask for more ones it finished (at the exception of the geekbench episode ^^)

As introduced, these were test runs and does not reflect actual competition format as they are currently used in official competitions. Our hope nevertheless is is lay the foundation of new live formats for competitive overclocking that could apply not only to offline tournaments, but also online as overclocking streamers will grow and benchmarks will add support (we hope) to enable such formats in a more practical way.

We will be looking forward to our next testing opportunity. We are also considering to host and online live championship later in 2017 with select overclockers – let us know if you are interested about it!

Keep pushing it!

About Timothée Pineau

Vice-President, Producer and Editor for OverClocking-TV since 2008, Xyala is mainly in charge of developing OCTV's video activities by supervising the video productions and organizing overclocking live streaming shows on the most prestigious competitions out there as the GOOC or the MOA. He reported for OCTV from all IT corners of the world on fairs as the CeBIT, Computex and also CES. For any questions and inquiries about live streaming possibilities and published videos: he's the guy!