据TechPowerup报道，AMD降低了Radeon RX 6000系列显卡的官方建议零售价（MSRP），新的价目表在9月15日敲定，并开始在零售商中生效。最新的官方建议零售价显示，属于高端型号的Radeon RX 6950 XT为949美元（约合人民币6728.6元）、Radeon RX 6900 XT为699美元（约合人民币4956.05元）、Radeon RX 6800 XT为599美元（约合人民币4247.03元）、Radeon RX 6800为549美元（约合人民币3892.52元）。

中端型号中，Radeon RX 6750 XT为419美元（约合人民币2970.79元）、Radeon RX 6700 XT为379美元（约合人民币2687.19元），Radeon RX 6650 XT取代了Radeon RX 6600 XT的位置，为299美元（约合人民币2119.97元），Radeon RX 6600为239美元（约合人民币1694.56元）。Radeon RX 6500 XT和Radeon RX 6400终于回到了属于它们的位置，作为低端型号，官方建议零售价分别为169美元（约合人民币1198.24元）和149美元（约合人民币1056.44元）。

此次调整价格以后，Radeon RX 6000系列显卡的定位分布及价格都显得比较合理。

Ryzen/Athlon 7020系列处理器采用了6nm工艺制造，配备了一个带有四个Zen 2架构内核的CCX，每个内核都有512KB的L2缓存，以及共享4MB的L3缓存，通过Infinity Fabric与核显相连。其核显部分为Radeon 610M，包含了两个RDNA 2架构的CU，也就是128个流处理器。同时核显采用了最新的媒体引擎，支持H.264、H.265（HEVC）和AV1等视频格式，提供硬件解码功能。

此外，Ryzen/Athlon 7020系列处理器可以使用多种Windows 11功能，比如语音唤醒和设备电源管理等，同时硬件安全功能包括Pluton和安全生物通道。AMD宣称，Ryzen/Athlon 7020系列处理器能够提供12个小时的续航时间。

首批Ryzen/Athlon 7020系列处理器有三个型号，分别是Ryzen 5 7520U、Ryzen 3 7320U和Athlon Gold 7220U，均采用相同的核显，TDP范围在8W到15W之间，具有相同的平台和I/O功能。其中Ryzen 5 7520U为4核8线程，基础频率为2.8GHz，加速频率为4.3GHz；Ryzen 3 7320U同样是4核8线程，不过频率低一些，基础频率为2.4GHz，加速频率为4.1GHz；Athlon Gold 7220U则削减到2核4线程，基础频率为2.4GHz，加速频率为3.7GHz。

显然Ryzen/Athlon 7020系列处理器的规格并不高，不过AMD认为这样的配备对于网页浏览和在线视频等日常使用来说足够了。Zen 2+RDNA 2这样的组合也并不是第一次出现，Valve的Steam Deck采用的定制SoC也是同样的组合。

AMD至少会与三个笔记本电脑制造商合作，包括了联想（IdeaPad 1）、惠普（一款未有名字的17英寸产品）和宏碁（Aspire 3），相关产品会在今年第四季度开始出货。

在今年6月，Sam Naffziger曾接受采访，表示下一代GPU的总功耗将增加，效率的提升是为了在此基础上最大限度地提高性能。Sam Naffziger已经在AMD工作了16年，曾负责多个产品领域，专注于推动每瓦性能的提升，以提高AMD在CPU和GPU上的竞争力，同时也是AMD小芯片架构背后的主要推动者之一，类似的思路已经在Ryzen和EPYC系列CPU上取得了成功。

Sam Naffziger称AMD作为当今唯一一家开发高性能CPU和GPU的公司，具有独特的优势，可以利用工程团队的学习成果，并在旗下产品组合中利用最好的IP。目前AMD将CPU微架构的一些设计方法整合到GPU设计中，比如简化芯片的物理构成，使更高的频率成为可能，Infinity Cache就是利用CPU的L3缓存设计思路实现。Infinity Cache是一种高密度、低功耗的缓存，使游戏工作负载中的常用数据更易于访问，从而显着增加带宽，同时降低内存所需的功率和减少延迟。

为了进一步完善和改进RDNA 2架构，提供更高的效率和性能提升，开发团队进行了几项关键的改进，包括：

优化切换 - 改进架构的基本设计，消除任何多余的操作和布线，以优化图形管道。

高频设计 - 针对高频调整设计，使得RDNA 2架构比RDNA架构提高30%以上，同时GPU能够以更低的电压运行，以确保新架构保持相同频率的时候有较低的功耗。

更智能的电源管理 - 在GPU中实施智能电源管理，确定利用更高频率的最佳机会，并且仅在直接提高性能时这样做，减少回退，从而减少功耗。

Sam Naffziger展望了即将到来的RDNA 3架构，将推动更高效的游戏，重申了在5nm工艺和小芯片设计加持下，RDNA 3架构将比RDNA 2架构提高50%以上的每瓦性能，为游戏玩家提供一流的游戏性能，并带来安静和节能的设计。

RDNA 3架构改进了RDNA 2架构的自适应电源管理技术，以设置特定于工作负载的操作点，确保GPU的每个组件仅使用实现最佳性能所需的功率。新架构还引入了新一代Infinity Cache，预计将提供更高密度、更低功耗的缓存，可降低显存的功耗需求，帮助RDNA 3架构实现领先的能耗表现。

不过肯定有很多读者也想知道这些最新一代GPU如果按照光追性能来对比的话，相对来说会是个什么情况？对于现有的光追游戏大概能跑多少帧？而且，从发展的眼光来看，未来的3A大作会越来越多的普及光线追踪，毕竟追求越来越真实的游戏画面是3A大作绝不会错的发展路线之一。所以，这篇文章我们就来看看这些最新一代GPU的光追性能相比较是个什么样的情况。

测试平台与说明我们的测试平台是基于酷睿i9-12900K搭建，尽量减少CPU瓶颈，显卡选择方面尽量包括所有的NVIDIA的Ampere架构桌面显卡以及AMD这边的所有RDNA 2架构桌面显卡，不过与之前的21款最新一代GPU横评相比情况还有一点点变化。

首先是Radeon RX 6400和Radeon RX 6500 XT这两者卡的光追性能实在羸弱，对于游戏测试如果显卡跑不动、帧数太低的话，数据的可读性就很差了，因为误差幅度会显的很大，所以本次光追性能横评把Radeon RX 6400排除了，仅留下相对性能更好一点的Radeon RX 6500 XT参加测试。

其次就是GeForce RTX 3080 12GB这一款GPU在现在有了新的消息——业界现在已经基本放弃了GeForce RTX 3080 12GB型号的生产，可以说现在的GeForce RTX 3080 12GB型号都是存货，卖完了就不会有了，所以这次横评没有GeForce RTX 3080 12GB这一款。（其实更主要的原因是这次没有收集到GeForce RTX 3080 12GB型号的卡......）

那么本次横评一共是19个GPU型号，其中N卡方面GeForce RTX 3060 Ti到GeForce RTX 3090之间都是公版显卡，GeForce RTX 3050和GeForce RTX 3060本来就没有公版卡，这两个型号我们都收集到了影驰的金属大师Mini显卡，最后对于GeForce RTX 3090 Ti，用的是一张影驰GeForce RTX 3090 Ti名人堂超频实验室限量版。

AMD显卡方面，除了Radeon RX 6950 XT是一张技嘉Radeon  RX 6950 XT GAMING OC 16G显卡之外，其他都是XFX讯景的显卡，大多也是OC版显卡，所以频率相对更高，代表的性能也更好，总得来说AMD这边的代表显卡可能平均有优势一点。

预览var tableid = settableid(); document.write(''); Table().init({ align: 'left', width: '82%', padding: '8px 15px', header: ["测 试 平 台", ""], columnWidth: 'AUTO', //表格单元宽度，"MAX"最大内容显示，""AVG"平均，"AUTO"自动，指定[15,16,20] id: tableid, data: [  ["硬件平台"],                     ["CPU","Intel Core i9-12900K"], ["主板","华硕ROG Strix Z690-A GAMING WiFi D4"], ["显卡","Radeon RX 6400Radeon RX 6500 XTRadeon RX 6600Radeon RX 6600 XTRadeon RX 6650 XTGeForce RTX 3050GeForce RTX 3060Radeon RX 6700 XTRadeon RX 6750 XTGeForce RTX 3060 TiGeForce RTX 3070GeForce RTX 3070 TiRadeon RX 6800Radeon RX 6800 XTRadeon RX 6900 XTRadeon RX 6950 XTGeForce RTX 3080 10GBGeForce RTX 3080 TiGeForce RTX 3090GeForce RTX 3090 Ti"], ["内存","芝奇 皇家戟 DDR4-3600 16GB*2"], ["硬盘","影驰 HOF Pro 20 2TB SSD"], ["散热器","ROG STRIX LC 360 RGB一体式水冷"], ["电源","ROG 雷神电源1200W"],                     ["软件配置"],                    ["操作系统","Microsoft Windows 11 64-bit"], ["驱动","Geforce Driver 516.59Adrenaline 22.5.1"] ] });

测试方面笔者将通过3DMark理论性能测试、实际游戏测试两个主要方面进行测试。其中实际游戏测试的情况有必要说明一下，实际游戏测试部分会将显卡分为三组，分别是面向1080p游戏环境的、面向2K分辨率游戏环境的以及面向4K分辨率游戏环境的。

面向1080P分辨率游戏环境的显卡中A卡主要是Radeon RX 6650 XT及以下的显卡，N卡主要是GeForce RTX 3050和GeForce RTX 3060。

面向1440P也就是2K分辨率游戏环境的显卡中A卡主要是Radeon RX 6700 XT及Radeon RX 6750 XT这两张，N卡主要是GeForce RTX 3060 Ti到GeForce RTX 3070 Ti这三张。

面向4K分辨率游戏环境的显卡中A卡主要是Radeon RX 6800及以上的显卡，N卡主要是GeForce RTX 3080及以上的显卡。

然后在各种定位的卡只跑其目标分辨率的基础上，将RX 6800和RTX 3060 Ti也分别参加1440P和1080P分辨率的游戏测试，这样就可以以其为锚最后将所有19张显卡的性能准确的统一反映出来。

3DMark理论光追性能测试3DMark中的Port Royal测试的就是混合了光线追踪渲染和传统光栅化渲染情景下的性能，因为游戏画面也不可能完全是光线追踪渲染，实际也是混合的，所以3DMark中的Port Royal测试已经能较好的反映实际光追游戏性能的相对表现。

这个测试之前的21款最新一代GPU横评中笔者已经跑了，这里还是放到这里便于大家跟后面的游戏实测的情况想对比。其具体成绩见下表，表中所列成绩均为3DMark显卡单项的得分。

预览var title_text = "3DMark Port Royal理论性能测试"; // 标题文字热管 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "Score "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF']; var data_arr =[ {"type":"Port Royal","RX 6400":"335","RX 6500 XT":"485","RX 6600":"3882","RX 6600 XT":"4607","RX 6650 XT":"4772","RTX 3050":"3430","RTX 3060":"5009","RX 6700 XT":"6055","RX 6750 XT":"6273","RTX 3060 Ti":"6885","RTX 3070":"8213","RTX 3070 Ti":"8654","RX 6800":"8022","RX 6800 XT":"9676","RX 6900 XT":"10340","RX 6950 XT":"10759","RTX 3080 10GB":"11235","RTX 3080 Ti":"12508","RTX 3090":"13180","RTX 3090 Ti":"14732"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);就光追性能来说，N卡这边完全是大大的领先，如果看重光追性能的话同预算无脑选N卡就是了，另外光追性能方面由于显存太小以及本身芯片性能羸弱，Radeon RX 6400和Radeon RX 6500 XT这两者卡的光追性能显的非常差劲，与其他型号有非常大的差距。

实际光追游戏性能测试游戏画质均设置为预设里面的最高画质（《地铁：离去》选择Ultra画质），然后开启光线追踪至“高”（《赛博朋克2077》预设里面的最高画质就打开了光线追踪到“超级”，于是就不改变这个设置），为保证画面一样所以不开FSR和DLSS，此外默认不是全屏的手动改为全屏，默认开启了垂直同步的手动关闭垂直同步，除此之外其他选项均为默认设置，均采用游戏自带的Benchmark输出结果。

1080P分辨率预览var title_text = "1080P光追游戏测试"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "FPS "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF']; var data_arr = [ {"type":"孤岛惊魂6","RX 6500 XT":"34","RX 6600":"74","RX 6600 XT":"82","RX 6650 XT":"82","RTX 3050":"60","RTX 3060":"78","RTX 3060 Ti":"97"}, {"type":"看门狗：军团","RX 6500 XT":"5","RX 6600":"30","RX 6600 XT":"33","RX 6650 XT":"34","RTX 3050":"30","RTX 3060":"40","RTX 3060 Ti":"52"}, {"type":"赛博朋克2077","RX 6500 XT":"4.67","RX 6600":"16.05","RX 6600 XT":"17.7","RX 6650 XT":"17.92","RTX 3050":"21.58","RTX 3060":"29.76","RTX 3060 Ti":"39.62"}, {"type":"尘埃5","RX 6500 XT":"20.1","RX 6600":"76","RX 6600 XT":"86.9","RX 6650 XT":"88.1","RTX 3050":"53.6","RTX 3060":"71","RTX 3060 Ti":"97.2"}, {"type":"光明记忆benchmark","RX 6500 XT":"7","RX 6600":"23","RX 6600 XT":"27","RX 6650 XT":"28","RTX 3050":"24","RTX 3060":"34","RTX 3060 Ti":"47"}, {"type":"边境benchmark","RX 6500 XT":"8.8","RX 6600":"18.8","RX 6600 XT":"22.2","RX 6650 XT":"23","RTX 3050":"22","RTX 3060":"30.7","RTX 3060 Ti":"41.5"}, {"type":"地铁：离去（增强版）","RX 6500 XT":"12.6","RX 6600":"46.48","RX 6600 XT":"54.8","RX 6650 XT":"56.73","RTX 3050":"40.5","RTX 3060":"55.64","RTX 3060 Ti":"76.35"}, {"type":"古墓丽影：暗影","RX 6500 XT":"22","RX 6600":"64","RX 6600 XT":"72","RX 6650 XT":"73","RTX 3050":"57","RTX 3060":"77","RTX 3060 Ti":"103"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);传统光栅化游戏场景的时候，在面对不同款的游戏时有可能差距不太大的显卡会出现帧数排名不一样的顺序，但是相对来说还是较为稳定的，但是对于光线追踪游戏场景的时候，可以看到面对不同游戏的时候，显卡的帧数排名不一样的情况更为浮动剧烈。

这其中的主要原因还是现在的不同光线追踪游戏其中的光线追踪渲染部分的压力不一样，像是《尘埃5》、《孤岛惊魂6》这样的就只是加入了相对较少的光线追踪效果，所以光追部分的渲染占比相较于一般的光线追踪游戏较低。而《赛博朋克2077》和《看门狗：军团》则是相对来说引入了较多的光线追踪效果，其光线追踪渲染的部分要比《尘埃5》、《孤岛惊魂6》这样的游戏要多。

可是即便如此，在《尘埃5》与《孤岛惊魂6》中，各个显卡的帧数的相对排序也是有较大不同，而《赛博朋克2077》和《看门狗：军团》中也是这样，比如在《赛博朋克2077》中RTX 3050要帧数明显好于RX 6650 XT，而在《看门狗：军团》中则相反。

另外，通过测试可以看到，如果不借助FSR和DLSS之类的技术，像《赛博朋克2077》和《看门狗：军团》这样压力比较高的游戏即使RTX 3060 Ti也难以达到60帧。

另外，RX 6500 XT这个型号在光线追踪游戏环境下确实几乎没有游戏体验，也就光线追踪效果不那么强的《孤岛惊魂6》中达到了刚能玩的水平，其他像是《尘埃5》也都只有20.1帧，而像是《赛博朋克2077》和《看门狗：军团》这样引入了较多的光线追踪效果的游戏就已经是卡成PPT了，毫无游戏体验。

2K分辨率预览var title_text = "2K光追游戏测试"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "FPS "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF'];  var data_arr = [ {"type":"赛博朋克2077","RX 6700 XT":"14.83","RX 6750 XT":"15.97","RTX 3060 Ti":"24.67","RTX 3070":"28.79","RTX 3070 Ti":"31.8","RX 6800":"19.78","RTX 3080 10GB":"40.17"}, {"type":"孤岛惊魂6","RX 6700 XT":"78","RX 6750 XT":"83","RTX 3060 Ti":"77","RTX 3070":"85","RTX 3070 Ti":"89","RX 6800":"93","RTX 3080 10GB":"101"}, {"type":"看门狗：军团","RX 6700 XT":"30","RX 6750 XT":"32","RTX 3060 Ti":"38","RTX 3070":"44","RTX 3070 Ti":"48","RX 6800":"38","RTX 3080 10GB":"58"}, {"type":"尘埃5","RX 6700 XT":"84.4","RX 6750 XT":"90.9","RTX 3060 Ti":"77.8","RTX 3070":"86","RTX 3070 Ti":"96.2","RX 6800":"103.7","RTX 3080 10GB":"115.5"}, {"type":"光明记忆benchmark","RX 6700 XT":"21","RX 6750 XT":"22","RTX 3060 Ti":"28","RTX 3070":"33","RTX 3070 Ti":"36","RX 6800":"27","RTX 3080 10GB":"46"}, {"type":"边境benchmark","RX 6700 XT":"16.8","RX 6750 XT":"18.1","RTX 3060 Ti":"26.5","RTX 3070":"30.8","RTX 3070 Ti":"33.3","RX 6800":"22.4","RTX 3080 10GB":"42.5"}, {"type":"地铁：离去（增强版）","RX 6700 XT":"49.62","RX 6750 XT":"52.39","RTX 3060 Ti":"55.78","RTX 3070":"64.57","RTX 3070 Ti":"70.68","RX 6800":"63.17","RTX 3080 10GB":"86.74"}, {"type":"古墓丽影：暗影","RX 6700 XT":"66","RX 6750 XT":"70","RTX 3060 Ti":"68","RTX 3070":"78","RTX 3070 Ti":"84","RX 6800":"81","RTX 3080 10GB":"104"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);1440P也就是2K分辨率下，即便是设计面向4K游戏环境的RTX 3080 10GB在面对开启高级设置的光线追踪而且不借助FSR和DLSS，也只能在《看门狗：军团》中刚刚达到58帧，而在《赛博朋克2077》中则只有40.17帧，对于RTX 3070 Ti，则在《赛博朋克2077》中只有31.8帧，只能到刚刚能玩的地步。

4K分辨率预览var title_text = "4K游戏测试"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "FPS "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF']; var data_arr = [ {"type":"赛博朋克2077","RX 6800":"9.07","RX 6800 XT":"11.29","RX 6900 XT":"12.11","RX 6950 XT":"12.59","RTX 3080 10GB":"18.42","RTX 3080 Ti":"21.27","RTX 3090":"22.24","RTX 3090 Ti":"24.87"}, {"type":"孤岛惊魂6","RX 6800":"57","RX 6800 XT":"67","RX 6900 XT":"71","RX 6950 XT":"76","RTX 3080 10GB":"65","RTX 3080 Ti":"70","RTX 3090":"73","RTX 3090 Ti":"79"}, {"type":"看门狗：军团","RX 6800":"21","RX 6800 XT":"25","RX 6900 XT":"27","RX 6950 XT":"28","RTX 3080 10GB":"33","RTX 3080 Ti":"36","RTX 3090":"38","RTX 3090 Ti":"41"}, {"type":"尘埃5","RX 6800":"66.6","RX 6800 XT":"80.6","RX 6900 XT":"84.3","RX 6950 XT":"90.9","RTX 3080 10GB":"75","RTX 3080 Ti":"82.1","RTX 3090":"85.6","RTX 3090 Ti":"99"}, {"type":"光明记忆benchmark","RX 6800":"13","RX 6800 XT":"15","RX 6900 XT":"17","RX 6950 XT":"17","RTX 3080 10GB":"22","RTX 3080 Ti":"25","RTX 3090":"25","RTX 3090 Ti":"28"}, {"type":"边境benchmark","RX 6800":"10.7","RX 6800 XT":"13.2","RX 6900 XT":"14.1","RX 6950 XT":"14.1","RTX 3080 10GB":"20.4","RTX 3080 Ti":"22.9","RTX 3090":"24","RTX 3090 Ti":"26.6"}, {"type":"地铁：离去（增强版）","RX 6800":"35.2","RX 6800 XT":"42.22","RX 6900 XT":"44.38","RX 6950 XT":"46.09","RTX 3080 10GB":"51.13","RTX 3080 Ti":"57.33","RTX 3090":"59.58","RTX 3090 Ti":"66.73"}, {"type":"古墓丽影：暗影","RX 6800":"43","RX 6800 XT":"52","RX 6900 XT":"55","RX 6950 XT":"59","RTX 3080 10GB":"57","RTX 3080 Ti":"64","RTX 3090":"66","RTX 3090 Ti":"72"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);4K分辨率下有个相对前面两个分辨率呈现较大不同的是显卡的帧数排名相对固定了。在《尘埃5》与《孤岛惊魂6》这种光追渲染低的场景中，这8张卡的排序是一样的，而《赛博朋克2077》和《看门狗：军团》这种光追渲染相对高的场景中，这8张卡的排序也是一模一样的。

另外，对于4K分辨率的光线追踪游戏体验，即使是目前地表最强的RTX 3090 Ti在《看门狗：军团》中不借助FSR或DLSS这样的帧率提升技术也只有41帧，而在《赛博朋克2077》中甚至不到30帧，可以看到目前的顶级游戏体验对显卡性能的要求仍然有很大的缺口需要填补。

实际光追游戏性能指数预览var title_text = "实际光追游戏性能指数"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "Score"; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF']; var data_arr =[ {"type":"超能指数","RX 6500 XT":"0.188866908","RX 6600":"0.587413683","RX 6600 XT":"0.668367173","RX 6650 XT":"0.682450919","RTX 3050":"0.552025402","RTX 3060":"0.749303529","RX 6700 XT":"0.83926632","RX 6750 XT":"0.894283297","RTX 3060 Ti":"1.014370467","RTX 3070":"1.165641862","RTX 3070 Ti":"1.268705679","RX 6800":"1.059463913","RX 6800 XT":"1.27362314","RX 6900 XT":"1.363739348","RX 6950 XT":"1.420548068","RTX 3080 10GB":"1.621743335","RTX 3080 Ti":"1.81525876","RTX 3090":"1.883344986","RTX 3090 Ti":"2.089372089"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);将以上的游戏实测结果汇总之后，就可以得到这19张显卡之间相对的光追游戏性能幅度了，拿这个结果与3DMark中的Port Royal测试结果对比的话会发现其实是相当符合的，顺序完全没有变化，仅仅是各自相对的幅度略有出入，而这个是与游戏选择相关的，总得来说，Port Royal的测试结果还是能反映尽量多的游戏的综合情况的。

DLSS游戏状况测试从上面的结果还可以看到，这一代的显卡面对高质量的光线追踪游戏的话，如果不借助FSR或DLSS这样的帧率提升技术还是很难获得较好的游戏体验的，而这也正是AMD推出FSR或英伟达推出DLSS技术的原因，可以说这一代的光线追踪游戏体验在实际应用中还是与FSR或DLSS技术分不开的。

而由于FSR技术A卡和N卡都支持，这里就不测试了，仅仅测试一下N卡额外具有的DLSS技术加持下的游戏帧数表现，对于一些仅有DLSS技术支持的游戏来说，N卡的游戏体验将会明显具有优势。

另外由于这类技术有个特点是帧数提升幅度并非线性（原始帧数越高提升幅度越小），所以并没有用来横向对比性能的意义，所以这部分也就没有像上面"实际光追游戏性能指数"这样的19张卡统一对比了，只是给大家看看对于仅有DLSS技术支持的光线追踪游戏在不同分辨率下N卡可以获得的游戏帧数体验大概是个什么情况。

这里的游戏画质均设置为预设里面的最高画质（《地铁：离去》选择Ultra画质），然后开启光线追踪至“高”（《赛博朋克2077》预设里面的最高画质就打开了光线追踪到“超级”，于是就不改变这个设置）。

其中1080p分辨率下的测试将DLSS开启至质量模式，2K分辨率下的测试将DLSS开启至平衡模式，4K分辨率下的测试将DLSS开启至性能模式（这三种对应关系也是官方推荐的组合搭配），此外默认不是全屏的手动改为全屏，默认开启了垂直同步的手动关闭垂直同步，除此之外其他选项均为默认设置，均采用游戏自带的Benchmark输出结果。《孤岛惊魂6》和《尘埃5》没有DLSS，这里也放进来了，做个参考。

1080P分辨率预览var title_text = "1080P光追游戏测试"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "FPS "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF']; var data_arr = [ {"type":"孤岛惊魂6","RTX 3050":"60","RTX 3060":"78","RTX 3060 Ti":"97"}, {"type":"看门狗：军团","RTX 3050":"40","RTX 3060":"55","RTX 3060 Ti":"70"}, {"type":"赛博朋克2077","RTX 3050":"38.33","RTX 3060":"51.59","RTX 3060 Ti":"65.25"}, {"type":"尘埃5","RTX 3050":"53.6","RTX 3060":"71","RTX 3060 Ti":"97.2"}, {"type":"光明记忆benchmark","RTX 3050":"42","RTX 3060":"59","RTX 3060 Ti":"80"}, {"type":"边境benchmark","RTX 3050":"39.3","RTX 3060":"50.5","RTX 3060 Ti":"72.9"}, {"type":"地铁：离去（增强版）","RTX 3050":"67.99","RTX 3060":"71.26","RTX 3060 Ti":"96.71"}, {"type":"古墓丽影：暗影","RTX 3050":"77","RTX 3060":"102","RTX 3060 Ti":"130"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);可以看到在1080p分辨率下，借助DLSS质量模式，确实达成了“RTX 3060可以流畅体验光线追踪游戏”的定位，其帧数最低的游戏是《赛博朋克2077》，但是也有51.59帧了，说是流畅游玩是差不多的，如果必须超过60帧，那么选择RTX 3060 Ti就可以。

2K分辨率预览var title_text = "2K光追游戏测试"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "FPS "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF'];  var data_arr = [ {"type":"赛博朋克2077","RTX 3070":"60.35","RTX 3070 Ti":"65.21","RTX 3080 10GB":"79.49"}, {"type":"孤岛惊魂6","RTX 3070":"85","RTX 3070 Ti":"89","RTX 3080 10GB":"101"}, {"type":"看门狗：军团","RTX 3070":"71","RTX 3070 Ti":"74","RTX 3080 10GB":"87"}, {"type":"尘埃5","RTX 3070":"86","RTX 3070 Ti":"96.2","RTX 3080 10GB":"115.5"}, {"type":"光明记忆benchmark","RTX 3070":"70","RTX 3070 Ti":"75","RTX 3080 10GB":"95"}, {"type":"边境benchmark","RTX 3070":"64.9","RTX 3070 Ti":"68.9","RTX 3080 10GB":"86.5"}, {"type":"地铁：离去（增强版）","RTX 3070":"97.65","RTX 3070 Ti":"104.01","RTX 3080 10GB":"121.36"}, {"type":"古墓丽影：暗影","RTX 3070":"120","RTX 3070 Ti":"125","RTX 3080 10GB":"151"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);在1440P也就是2K分辨率下，借助以DLSS平衡模式，RTX 3070在压力最大的《赛博朋克2077》中也能达到流畅的60帧，其他游戏或者说更强的RTX 3070 Ti要超过60帧流畅运行更是不在话下。

4K分辨率预览var title_text = "4K游戏测试"; // 标题文字 var subtitle_text = "www.expreview.com | 数值越大越好 "; // 副标题文字 var unit_text = "FPS "; //单位 var leftside = "8%"; //左边空白 var topside = 50; //每组间距 var mblur = 0.05; //模糊效果 var bgopacity = 0; //背景透明度，0为全透明 var mysort = 1; //是否排序 var myper = 1; //是否显示百分比 var mylengend = true; //是否显示图例 var toplegend = 50; //图例上端空白 var labelcolor = '#fff'; // 标签文字颜色 var legendwidth = '90%'; //图例宽度 var titlecolor = '#222'; //标题文字颜色 var barheight = 35; //柱高度,当为7倍数时，不显示刻度线 var bargap = 3; //柱间距,相当于百分值，即2% var xaxismax = null; //指定X轴最大值，null时为自动 var autocolor = 1; //是否自动判断颜色，0为否，1根据品牌生成颜色 ，如果 >1 则为AMD NV Intel 标准色, var mycolor = []; //自动生成随机颜色 也可ANIColor("NVIDIA",3)可以生成三种NV绿，可替换成AMD Intel //var mycolor = ANIColor('AMD',4); //这样自动生成4种AMD红色，其它如NVIDIA、Intel同理 //var mycolor=['#953f61', '#72355f', '#d77169']; //自定义颜色 AMD红 '#9F1417', NV绿 '#82B41B', Intel蓝 '#0075C5', //var mycolor = ['#0dcc47', '#969696', '#6f7b80','5A94DF']; var data_arr = [ {"type":"赛博朋克2077","RTX 3080 10GB":"52.8","RTX 3080 Ti":"58.53","RTX 3090":"60.34","RTX 3090 Ti":"67.99"}, {"type":"孤岛惊魂6","RTX 3080 10GB":"65","RTX 3080 Ti":"70","RTX 3090":"73","RTX 3090 Ti":"79"}, {"type":"看门狗：军团","RTX 3080 10GB":"68","RTX 3080 12GB":"72","RTX 3080 Ti":"74","RTX 3090":"77","RTX 3090 Ti":"81"}, {"type":"尘埃5","RTX 3080 10GB":"75","RTX 3080 12GB":"79.3","RTX 3080 Ti":"82.1","RTX 3090":"85.6","RTX 3090 Ti":"99"}, {"type":"光明记忆benchmark","RTX 3080 10GB":"60","RTX 3080 12GB":"65","RTX 3080 Ti":"68","RTX 3090":"71","RTX 3090 Ti":"79"}, {"type":"边境benchmark","RTX 3080 10GB":"51.2","RTX 3080 12GB":"54.1","RTX 3080 Ti":"58.5","RTX 3090":"60.3","RTX 3090 Ti":"67.3"}, {"type":"地铁：离去（增强版）","RTX 3080 10GB":"97.05","RTX 3080 12GB":"102.68","RTX 3080 Ti":"107.68","RTX 3090":"110.79","RTX 3090 Ti":"120.57"}, {"type":"古墓丽影：暗影","RTX 3080 10GB":"109","RTX 3080 12GB":"116","RTX 3080 Ti":"119","RTX 3090":"124","RTX 3090 Ti":"136"} ]; TDInit(title_text, subtitle_text, unit_text, leftside, topside, mblur, bgopacity, mysort, myper, mylengend, toplegend, labelcolor, legendwidth, titlecolor, barheight, bargap, mycolor, data_arr, autocolor, xaxismax);而在在4K分辨率下，借助以DLSS性能模式，RTX 3080 10GB在压力最大的《赛博朋克2077》可以达到52.8帧，在《边境》中平均有51.2帧，其他游戏则是都至少达到了60帧，其他更高性能的卡相应的表现要更好一点，总得来说，定位4K游戏解决方案的这些型号在DLSS性能模式下面对光线追踪游戏效果也是有不错的游戏体验的。

总结游戏开发者一直在追求带给玩家更加逼真的游戏画面，毕竟更加逼真的游戏画面与更加沉浸的游戏体验是正相关的，而与很多事物的发展过程一样——电子游戏的画面经过这么多年的发展，越来越成熟，也意味着提升的难度是越来越大的，就跟百米运动员从11秒到10秒再到9.9秒一样。

在已经很逼真的游戏画面基础上继续提高逼真度，哪怕一点点都需要很大的算力来支撑，这样的情况也就是现在光线追踪技术的情况。于是不论N卡还是A卡，在光线追踪这里都靠的是通过专用的计算单元来提高效率，但即便如此我们也可以看到，如果不借助FSR或DLSS这样的帧率提升技术，目前的显卡性能还是不能在合适的价格和功耗下具有良好的光线追踪游戏体验，而FSR或DLSS尽管是对于画质影响没那么大，但是终究是不如原始渲染的，原始渲染的光线追踪才是目标。

消费级显卡的实时光线追踪始于英伟达2018年的图灵架构显卡，距今已经4个年头，即使经历了安培架构的性能大增，但是在实现实时光线追踪游戏体验方面仍然摆脱不了FSR或DLSS，而在光线追踪方面起步更晚的AMD更不用说。

光线追踪确实是一项革命性的3D图形技术，但是目前的情况借用孙先生的一句话来说就是——“革命尚未成功，同志仍须努力”。而这，或许这就是下一代显卡让我们翘首以盼的原因吧。

Ryzen 7000系列的核显规模并不大，仅有2个CU，总计128个流处理器，频率最高可达2200MHz，可提供0.563 TFLOPs的浮点运算性能，略高于Nintendo Switch。近日，有网友就测试了Ryzen 7000系列核显的性能，使用的是Ryzen 9 7950X，搭配了64GB的DDR5内存，由于插满了四条内存槽，使得速率降低到3600Mbps，对核显的性能有一定的影响。

Ryzen 7000系列核显的设备ID为“GFX1036”，在Geekbench 5的OpenCL基准测试中成绩为7808分。相比之下，采用7nm工艺的Vega 6（6个CU）在同一基准测试中的成绩为7003分，Vega 8（8个CU）则是8000分到8500分之间。如果测试的系统在内存速率上能高一些，预计Ryzen 7000系列核显的成绩与Vega 8应该处于同一水平。考虑到两者在规模上存在一定差距，这样的结果不错了。

虽然在很多人看来，Ryzen 7000系列的核显仅能作为亮机卡使用，相比英特尔Alder Lake和Raptor Lake配备32个EU的核显在基准测试中的表现也逊色一些，不过性能并非那么弱不禁风。

AMD目前已在官网上公布了Radeon RX 6700的规格，确认了该款显卡的存在。其搭载了Navi 22核心，配备了36个CU，即2304个流处理器，Infinity Cache为80MB，采用了10GB的GDDR6显存，位宽为160位，速率为16Gbps，基本配置与移动平台的Radeon RX 6700M相同。

蓝宝石的PULSE系列与官方规格相比，频率有所提高，比如加速频率由2450MHz提高到2495MHz，TDP也由175W提高到220W。同时蓝宝石也更改了官网的信息，重现添加了“RX”的字样。不知道是蓝宝石的新显卡计划打乱了AMD的部署，从而“先斩后奏”，还是其他原因所致了。

预览var tableid = settableid(); document.write(''); Table().init({ title: "AMD Radeon RX 6000系列 GPU 参数表", header: ["型号","Radeon RX 6950 XT","Radeon RX 6900 XT","Radeon RX 6800 XT","Radeon RX 6800","Radeon RX 6750 XT","Radeon RX 6700 XT","Radeon RX 6700","Radeon RX 6650 XT","Radeon RX 6600 XT","Radeon RX 6600","Radeon RX 6500 XT","Radeon RX 6400","Radeon RX 6300"], width: '92%', columnWidth: 'MAX', //表格单元宽度，"MAX"最大内容显示，""AVG"平均，"AUTO"自动，指定[15,16,20] id: tableid, data: [ ["GPU","Navi 21 XTXH","Navi 21 XTX","Navi 21 XT","Navi 21 XL","Navi 22 XT","Navi 22 XT","Navi 22 XL","Navi 23 XT","Navi 23 XT","Navi 23 XL","Navi 24 XT","Navi 24 XL","Navi 24"], ["CU数量","80","80","72","60","40","40","36","32","32","28","16","12","TBC"], ["流处理器数量","5120","5120","4608","3840","2560","2560","2304","2048","2048","1792","1024","768","TBC"], ["Infinity Cache","128MB","128MB","128MB","128MB","96MB","96MB","80MB","32MB","32MB","32MB","16MB","16MB","16MB"], ["游戏频率","2100MHz","2015MHz","2015MHz","1815MHz","2495MHz","2424MHz","2174MHz","2410MHz","2359MHz","2044MHz","2610MHz","2039MHz","TBC"], ["Boost频率","2310MHz","2250MHz","2250MHz","2105MHz","2600MHz","2581MHz","2450MHz","2635MHz","2589MHz","2491MHz","2815MHz","2321MHz","TBC"], ["显存容量","16GB G6","16GB G6","16GB G6","16GB G6","12GB G6","12GB G6","10GB G6","8GB G6","8GB G6","8GB G6","4GB G6","4GB G6","2GB G6"], ["显存位宽","256-bit","256-bit","256-bit","256-bit","192-bit","192-bit","160-bit","128-bit","128-bit","128-bit","64-bit","64-bit","64-bit"], ["显存速度","18 Gbps","16 Gbps","16 Gbps","16 Gbps","18 Gbps","16 Gbps","16 Gbps","17.5 Gbps","16 Gbps","14 Gbps","18 Gbps","14 Gbps","TBC"], ["TGP","335W","300W","300W","250W","250W","230W","175W","180W","160W","132W","107W","53W","TBC"], ] });

不过据Cowcotland报道，蓝宝石即将推出Radeon RX 6700显卡，包括了PULSE及一个未确定的系列。其搭载了Navi 22核心，配备了36个CU，即2304个流处理器，Infinity Cache为80MB，采用了10GB的GDDR6显存，位宽为160位，速率为16Gbps，基本配置与移动平台的Radeon RX 6700M相同。

据悉，Radeon RX 6700的游戏频率为2230MHz，加速频率为2495MHz。此外，蓝宝石为Radeon RX 6700显卡配备了一个8Pin外接供电接口，据推测，功耗在220W左右。据了解，Radeon RX 6700可能会在6月9日发布。

预览var tableid = settableid(); document.write(''); Table().init({ title: "AMD Radeon RX 6000系列 GPU 参数表", header: ["型号","Radeon RX 6950 XT","Radeon RX 6900 XT","Radeon RX 6800 XT","Radeon RX 6800","Radeon RX 6750 XT","Radeon RX 6700 XT","Radeon RX 6700","Radeon RX 6650 XT","Radeon RX 6600 XT","Radeon RX 6600","Radeon RX 6500 XT","Radeon RX 6400","Radeon RX 6300"], width: '92%', columnWidth: 'MAX', //表格单元宽度，"MAX"最大内容显示，""AVG"平均，"AUTO"自动，指定[15,16,20] id: tableid, data: [ ["GPU","Navi 21 XTXH","Navi 21 XTX","Navi 21 XT","Navi 21 XL","Navi 22 XT","Navi 22 XT","Navi 22 XL","Navi 23 XT","Navi 23 XT","Navi 23 XL","Navi 24 XT","Navi 24 XL","Navi 24"], ["CU数量","80","80","72","60","40","40","36","32","32","28","16","12","TBC"], ["流处理器数量","5120","5120","4608","3840","2560","2560","2308","2048","2048","1792","1024","768","TBC"], ["Infinity Cache","128MB","128MB","128MB","128MB","96MB","96MB","80MB","32MB","32MB","32MB","16MB","16MB","16MB"], ["游戏频率","2100MHz","2015MHz","2015MHz","1815MHz","2495MHz","2424MHz","2230MHz","2410MHz","2359MHz","2044MHz","2610MHz","2039MHz","TBC"], ["Boost频率","2310MHz","2250MHz","2250MHz","2105MHz","2600MHz","2581MHz","2495MHz","2635MHz","2589MHz","2491MHz","2815MHz","2321MHz","TBC"], ["显存容量","16GB G6","16GB G6","16GB G6","16GB G6","12GB G6","12GB G6","10GB G6","8GB G6","8GB G6","8GB G6","4GB G6","4GB G6","2GB G6"], ["显存位宽","256-bit","256-bit","256-bit","256-bit","192-bit","192-bit","160-bit","128-bit","128-bit","128-bit","64-bit","64-bit","64-bit"], ["显存速度","18 Gbps","16 Gbps","16 Gbps","16 Gbps","18 Gbps","16 Gbps","16 Gbps","17.5 Gbps","16 Gbps","14 Gbps","18 Gbps","14 Gbps","TBC"], ["TGP","335W","300W","300W","250W","250W","230W","220W","180W","160W","132W","107W","53W","TBC"], ] });

据了解，代号Mendocino的APU最多为四核心八线程配置，显然这并不适合高性能系统。该款SoC主要针对二合一平板电脑和低功耗的笔记本电脑而设计，电池使用方面会有相关优化，续航时间在10个小时以上。据称，搭载Mendocino APU的笔记本电脑售价在399美元到699美元之间，可适配Windows以及ChromeOS操作系统。

AMD也并不是第一次采用Zen 2+RDNA 2这样的计算/图形组合，Valve的Steam Deck掌机上搭载的APU也有着同样的架构。Steam Deck采用了AMD代号为Van Gogh的APU，定制的SoC也称为“Aerith”，基于Zen 2架构的CPU拥有4核8线程，频率在2.4 GHz到3.5 GHz之间，集成的RDNA 2架构GPU具有8个计算单元，频率在1.0 GHz到1.6 GHz之间，整个SoC的功耗在4W到15W之间。以目前Mendocino的情况来看，或许与Van Gogh有着不少相似之处。

图表上每一款Radeon RX 6000系列GPU都有对应的一款英伟达GPU（缺了RTX 3090 Ti），上面列明了每款GPU的售价、功耗和性能，同时引入了两个指标：一个是性价比，即perf/$，另外一个则是能耗比，即fps/watt。此外，AMD将游戏目标设置分为四个等级，分别是1080P中等画质、1080P最高画质、1440P最高画质及4K最高画质。

从结果来看，最高端Radeon RX 6950XT与GeForce RTX 3090对比中，每美元的性能高出了80%，功耗低了15W，每帧所需功耗领先22%。由于目前市场上显卡的价格具有比较大的波动性，所以这种对比数据的结果可能每天都不同，该图表是基于5月10日某电商平台的报价来制作的。

类似的营销手段过往并不少见，寻找一个合理的切入角度，以此显示自己的产品更具有竞争力，且更值得玩家选购。AMD和英伟达在几个月后，将发布新一代显卡，届时肯定会有新一轮的营销大战，或者又会有新的营销方式。

虽然目前Radeon RX 6x50 XT系列的规格大致已经清楚，但一直没有定价方面的消息，以其规格来说，只能猜测大概与原版接近。据VideoCardz报道，Radeon RX 6x50 XT系列的官方建议零售价会比之前对应的产品略有提高，其中Radeon RX 6950 XT的MSRP为1099美元，相比Radeon RX 6900 XT提高了100美元；Radeon RX 6750 XT的MSRP为549美元，相比Radeon RX 6700 XT提高了70美元；Radeon RX 6650 XT的MSRP为399美元，相比Radeon RX 6600 XT提高了20美元。

据了解，AMD会在其官方商店直接销售Radeon RX 6950 XT和Radeon RX 6750 XT公版，均采用公版的散热设计，不过并没有Radeon RX 6650 XT，这意味着该款显卡很可能没有公版散热设计的零售款式。

预览var tableid = settableid(); document.write(''); Table().init({ title: "AMD Radeon RX 6000系列 GPU 参数表", header: ["型号","Radeon RX 6950 XT","Radeon RX 6900 XT","Radeon RX 6800 XT","Radeon RX 6800","Radeon RX 6750 XT","Radeon RX 6700 XT","Radeon RX 6700","Radeon RX 6650 XT","Radeon RX 6600 XT","Radeon RX 6600","Radeon RX 6500 XT","Radeon RX 6400","Radeon RX 6300"], width: '92%', columnWidth: 'MAX', //表格单元宽度，"MAX"最大内容显示，""AVG"平均，"AUTO"自动，指定[15,16,20] id: tableid, data: [ ["GPU","Navi 21 XTXH","Navi 21 XTX","Navi 21 XT","Navi 21 XL","Navi 22 XT","Navi 22 XT","Navi 22 XL","Navi 23 XT","Navi 23 XT","Navi 23 XL","Navi 24 XT","Navi 24 XL","Navi 24"], ["CU数量","80","80","72","60","40","40","TBC","32","32","28","16","12","TBC"], ["流处理器数量","5120","5120","4608","3840","2560","2560","TBC","2048","2048","1792","1024","768","TBC"], ["Infinity Cache","128MB","128MB","128MB","128MB","96MB","96MB","96MB","32MB","32MB","32MB","16MB","16MB","16MB"], ["游戏频率","2100MHz?","2015MHz","2015MHz","1815MHz","2495MHz?","2424MHz","TBC","2410MHz?","2359MHz","2044MHz","2610MHz","2039MHz","TBC"], ["Boost频率","2310MHz?","2250MHz","2250MHz","2105MHz","2600MHz?","2581MHz","TBC","2635MHz?","2589MHz","2491MHz","2815MHz","2321MHz","TBC"], ["显存容量","16GB G6","16GB G6","16GB G6","16GB G6","12GB G6","12GB G6","12GB/6GB G6?","8GB G6","8GB G6","8GB G6","4GB G6","4GB G6","2GB G6"], ["显存位宽","256-bit","256-bit","256-bit","256-bit","192-bit","192-bit","192-bit","128-bit","128-bit","128-bit","64-bit","64-bit","64-bit"], ["显存速度","18 Gbps","16 Gbps","16 Gbps","16 Gbps","18 Gbps","16 Gbps","14 Gbps?","17.5 Gbps?","16 Gbps","14 Gbps","18 Gbps","14 Gbps","TBC"], ["TGP","335W?","300W","300W","250W","TBC","230W","TBC","TBC","160W","132W","107W","53W","TBC"], ] });

据VideoCardz报道，Canada Computers & Electronics似乎成为了第一家确认推出Radeon RX 6x50 XT系列公版显卡的零售商，外观设计上与此前泄露的新显卡渲染图一致。

Wccftech则表示，Radeon RX 6950 XT的游戏频率和Boost频率分别达到了2100 MHz和2310 MHz，相比Radeon RX 6900 XT提高了4%，可提供23.65 TFLOPs的FP32峰值性能，功耗将提高到335W；Radeon RX 6750 XT的游戏频率和Boost频率分别达到了2495 MHz和2600 MHz，相比Radeon RX 6700 XT提高了3%，可提供13.31 TFLOPs的FP32峰值性能，功耗将提高到250W；Radeon RX 6650 XT的游戏频率和Boost频率分别达到了2410 MHz和2635 MHz，相比Radeon RX 6600 XT提高了2%，可提供10.79 TFLOPs的FP32峰值性能，功耗将提高到180W。

预览var tableid = settableid(); document.write(''); Table().init({ title: "AMD Radeon RX 6000系列 GPU 参数表", header: ["型号","Radeon RX 6950 XT","Radeon RX 6900 XT","Radeon RX 6800 XT","Radeon RX 6800","Radeon RX 6750 XT","Radeon RX 6700 XT","Radeon RX 6700","Radeon RX 6650 XT","Radeon RX 6600 XT","Radeon RX 6600","Radeon RX 6500 XT","Radeon RX 6400","Radeon RX 6300"], width: '92%', columnWidth: 'MAX', //表格单元宽度，"MAX"最大内容显示，""AVG"平均，"AUTO"自动，指定[15,16,20] id: tableid, data: [ ["GPU","Navi 21 XTXH","Navi 21 XTX","Navi 21 XT","Navi 21 XL","Navi 22 XT","Navi 22 XT","Navi 22 XL","Navi 23 XT","Navi 23 XT","Navi 23 XL","Navi 24 XT","Navi 24 XL","Navi 24"], ["CU数量","80","80","72","60","40","40","TBC","32","32","28","16","12","TBC"], ["流处理器数量","5120","5120","4608","3840","2560","2560","TBC","2048","2048","1792","1024","768","TBC"], ["Infinity Cache","128MB","128MB","128MB","128MB","96MB","96MB","96MB","32MB","32MB","32MB","16MB","16MB","16MB"], ["游戏频率","2100MHz?","2015MHz","2015MHz","1815MHz","2495MHz?","2424MHz","TBC","2410MHz?","2359MHz","2044MHz","2610MHz","2039MHz","TBC"], ["Boost频率","2310MHz?","2250MHz","2250MHz","2105MHz","2600MHz?","2581MHz","TBC","2635MHz?","2589MHz","2491MHz","2815MHz","2321MHz","TBC"], ["显存容量","16GB G6","16GB G6","16GB G6","16GB G6","12GB G6","12GB G6","12GB/6GB G6?","8GB G6","8GB G6","8GB G6","4GB G6","4GB G6","2GB G6"], ["显存位宽","256-bit","256-bit","256-bit","256-bit","192-bit","192-bit","192-bit","128-bit","128-bit","128-bit","64-bit","64-bit","64-bit"], ["显存速度","18 Gbps","16 Gbps","16 Gbps","16 Gbps","18 Gbps","16 Gbps","14 Gbps?","17.5 Gbps?","16 Gbps","14 Gbps","18 Gbps","14 Gbps","TBC"], ["TGP","335W?","300W","300W","250W","TBC","230W","TBC","TBC","160W","132W","107W","53W","TBC"], ] });