安装 Steam
登录
|
语言
繁體中文（繁体中文）
日本語（日语）
한국어（韩语）
ไทย（泰语）
български（保加利亚语）
Čeština（捷克语）
Dansk（丹麦语）
Deutsch（德语）
English（英语）
Español-España（西班牙语 - 西班牙）
Español - Latinoamérica（西班牙语 - 拉丁美洲）
Ελληνικά（希腊语）
Français（法语）
Italiano（意大利语）
Bahasa Indonesia（印度尼西亚语）
Magyar（匈牙利语）
Nederlands（荷兰语）
Norsk（挪威语）
Polski（波兰语）
Português（葡萄牙语 - 葡萄牙）
Português-Brasil（葡萄牙语 - 巴西）
Română（罗马尼亚语）
Русский（俄语）
Suomi（芬兰语）
Svenska（瑞典语）
Türkçe（土耳其语）
Tiếng Việt（越南语）
Українська（乌克兰语）
报告翻译问题
If you have some time please rate my game as well. Thank you very much!
http://psteamcommunity.yuanyoumao.com/sharedfiles/filedetails/?id=674394887
I'm going to refrain from correcting grammatical typos in my previous comments, as far as I know all the information below this is factually accurate (you could argue that most games are CPU bound but in my experience that is not the case (I'm not a game developer (for now))).
I find it annoying that I cannot edit previous comments (or if it is possible, the UI makes discovering that capability a rather frustrating exercise).
Keep in mind that Piledriver has 1 FPU per integer core so effectively you have 4 FPUs compared with the 5820K's 6, note that AMD was sued for false advertising in november because of this. Also Bulldozer and later architectures have a higher clock rate than Intel processors, but this is deceptive as Intel processors perform more instructions per clock as well as having capabilities like instruction pipelining that enable more efficient branch prediction.
This isn't to say that AMD architectures are strictly worse than Intel's. AMD processors are generally cheaper than equivalent Intel processors, and for workloads that are _not_ CPU bound (eg most games or server applications) they are probably the better choice, this map is one of a few edge cases where this becomes obvious to a consumer.
"[lun3dm5 is] an interesting test case in that it winds up stressing the CPU far harder than the GPU. Modern GPUs handle lots of triangles sharing one texture (the very image of this level) very well, but Quake3's need to touch every vert on the CPU every frame makes for a big bottleneck in roughly all the same cases and thus it simply doesn't scale well to crazy projects like this"
For comparison, my 5820K handles this like a champ and I get 250 fps throughout the entire map
I'd love to port this to reflex which should handle this a lot better, but the sheer number of vertices makes recreating this at least a months work of man hours. The alternatives would be: 1) using the q3toreflex conversion tool that the reflex developers provide, but it is useless as my character won't move in the converted map
2)generating the vertices programmatically but the map editor doesn't allow scripting :(