（原文） This is by design. When a workunit is aborted (either because the task crashed, the process was terminated, or a time-limit expired), the Agent intentionally transmits the amount of time spent to the server so that you will receive credit for the time spent. You will not receive credit for a workunit/result having been fully completed. Additionally, there is no scientific benefit gained, since partial result files are not usable by us and are discarded by the Agent. As has been noted, this activity also wastes bandwidth, which is one of the primary costs involved in UD's sponsorship of the project.
>>58 私の、公式掲示板のアカウント、Aim@NextCentury のシグネチャの、 Sorry,My English ability is the same level as Woodstock of PEANUTS. :- ( で、最後の一文の " Woodstock of PEANUTS " は、 スヌーピーに出てくる「ウッドストック」というキャラクターを指しています。 （新聞の４コマ漫画のタイトルが " PEANUTS " なのです。） こんなのもあります。 http://www.snoopy.co.jp/
「途中で終了したリザルトは、まれにリザルトに計上される可能性がある （サーバ側で）が、その結果は無意味であり、また、ネットワークに影響を与える」 「同様、CPU Time と point も、計上されてしまう」 「しかしプロジェクトにおいて重要なのは、ポイントやリザルト数の正確さではなくて、 正確に最後まで処理されたＷＵがUDに返信されることだ。」
こんな感じですかね。 UD Agent のバグではなくて、サーバ側の処理によるものという事の様です……。
So, in summary (in my humble opinion)... The statistic that is the most important when comparing one device to another, or one member's statistics to another, is UD's Points Generated calculation. Again, it is just a simple formula that takes your overall CPU Time contributed and gives weight to the device resources you choose to allocate. On the other hand, Results Returned, while being a interesting statistic to look at, really is nothing more than a listing of the number of "breaks" your device(s) took during the overall "stream" of work for a given project or combination of projects.