数天前，英伟达推出了基于Ada Lovelace架构的新一代主流GPU，包括GeForce RTX 4060、RTX 4060 Ti 8GB和RTX 4060 Ti 16GB三款。其中RTX 4060搭载的是AD107-400，有别于其他两款的AD106 GPU。
近日MEGAsizeGPU就曝光了AD107-400 GPU的照片，这是目前英伟达在桌面平台上尺寸最小的Ada Lovelace架构GPU。据了解，AD107芯片的尺寸大概为146mm2，比起上一代Ampere架构的GA107（200mm2）缩小了约27%。
AD107-400采用了PCIe 4.0 x8接口，具有24MB的L2缓存，拥有24组SM，即3072个CUDA核心，相比AD106的32MB的L2缓存、36组SM和4608个CUDA核心少了三分之一。其显存位宽与AD106一样都是128位，搭配的GDDR6容量为8GB，显存速率为17Gbps，对应的显存带宽为272 GB/s，显存配置上似乎比AD106更为合理。
搭载AD107-400的GeForce RTX 4060显卡功耗并不高，一般在115W至120W之间，对于小型主机会比较友好。
document.write(‘
‘);
Table().init({
title: “NVIDIA GeForce RTX 40系列规格”,
header: [“型号”, “TITAN”, “RTX 4090 Ti”, “RTX 4090”, “RTX 4080”, “RTX 4070 Ti”, “RTX 4070”, “RTX 4060 Ti”, “RTX 4060”, “RTX 4050”],
width: ‘100%’,
columnWidth: ‘AVG’, //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: [
[“架构”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”, “Ada Lovelace”],
[“工艺”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”, “TSMC 4N”],
[“GPU”, “AD102-450?”, “AD102-350?”, “AD102-300”, “AD103-300”, “AD104-400”, “AD104-250/251”, “AD106-350/351”, “AD107-400”, “AD107?”],
[“PCB”, “PG137-SKU0”, “PG139/136-SKU310”, “PG139/136-SKU330”, “PG139-SKU360”, “PG141-SKU331”, “PG141-SKU336/337”, “PG190-SKU361/363”, “PG173-SKU371”, “TBC”],
[“SM数量”, “142”, “142”, “128”, “76”, “60”, “46”, “34”, “24”, “20”],
[“CUDA数量”, “18176”, “18176”, “16384”, “9728”, “7680”, “5888”, “4352”, “3072”, “2560”],
[“显存容量”, “48GB G6X”, “24GB G6X”, “24GB G6X”, “16GB G6X”, “12GB G6X”, “12GB G6X”, “8/16GB G6”, “8GB G6”, “6GB G6”],
[“显存位宽”, “384bit”, “384bit”, “384bit”, “256bit”, “192bit”, “192bit”, “128bit”,”128bit”,”96bit”],
[“显存速率”, “24Gbps”, “24Gbps”, “21Gbps”, “23Gbps”, “21Gbps”, “21Gbps”, “18Gbps”, “18Gbps”, “TBC”],
[“TGP”, “800W”, “600W”, “450W”, “320W”, “285W”, “200W”, “160W/165W”, “115W”, “TBC”],
[“发售时间”, “TBC”, “TBC”, “2022年10月”, “2022年11月”, “2023年1月”, “2023年4月?”, “2023年5/7月”, “2023年7月”, “TBC”]
]
});
微信扫一扫 
支付宝扫一扫 
