throughput
US /ˈθruˌpʊt/
・UK /ˈθru:pʊt/
C2 高級
n. (u.)不可數名詞(某一時期內的)吞吐量
The computer's hard drive crashed due to the overloaded throughput
影片字幕
最討厭塞車！到底塞車是從哪台車開始的呢？ (The Simple Solution to Traffic)
- A solid lane of self-driving cars vastly increases throughput.
就很容易長出一隻可以持需幾個小時的車陣貪食蛇
超微半導體 AMD 2024 年電腦展演講，12 分鐘濃縮重點！ (AMD's 2024 Computex Keynote: Everything Revealed in 12 Minutes)
- It also enables us to deliver much more.We also designed Zen5 with a wider CPU engine and instruction window to run more instructions in parallel for leadership compute throughput and efficiency.
在設計 Zen5 時，我們還採用了更寬的 CPU 引擎和指令窗口，以並行方式運行更多指令，從而提高計算吞吐量和效率。
- As a result, compared to Zen4, we get double the instruction bandwidth, double the data bandwidth between the cache and floating point unit, and double the AI performance with full AVX-512 throughput.
是以，與 Zen4 相比，我們獲得了雙倍的指令帶寬、雙倍的高速緩存和浮點運算單元之間的數據帶寬，以及雙倍的人工智能性能和完整的 AVX-512 吞吐量。
Nvidia 2024 年電腦展主題演講，15 分鐘濃縮精華！(Nvidia's 2024 Computex Keynote: Everything Revealed in 15 Minutes)
- It has to distribute the workload across multiple GPUs, tensor parallelism, pipeline parallelism, data parallelism, all kinds of parallelism, expert parallelism, all kinds of parallelism, distributing the workload across multiple GPUs, processing it as fast as possible because if you are in a factory, if you run a factory, your throughput directly correlates to your revenues.
但是，運行人工智能的計算堆棧複雜得令人難以置信。
- Your throughput directly correlates to the quality of service,
當你使用哈拉 GPT 時，他們的堆棧下是一大堆軟件。
Intel 英特爾 AI 晶片發佈會，10 分鐘濃縮精華帶你看！(Intel's Lunar Lake AI Chip Event: Everything Revealed in 10 Minutes)
- We're projected to deliver 40% faster time to train than H100s, and 1.5x versus H200s, and faster inferencing than H100s, and delivering that 2.3x performance per dollar in throughput versus H100s.
預計我們的訓練時間將比 H100s 快 40%，是 H200s 的 1.5 倍，推理速度比 H100s 快，每美元吞吐量性能是 H100s 的 2.3 倍。
- We're projected to deliver 40% faster time to train than H100s and 1.5x versus H200s and faster inferencing than H100s and delivering that 2.3x performance per dollar in throughput versus H100s.
我們還支持所有預期的開源框架，如 PyTorch 和 VLLM。
Accton 交換器伺服器 | CSP-9550 (Accton Switch Server | CSP-9550)
- Designed with two sockets, two Intel Xeon Scalable processors, an Intel Tofino Switch chip, CSP 9550 significantly expands throughput performance and capacity between CPUs.
CSP 9550 採用雙插槽設計，搭載兩顆 Intel Xeon 可擴充處理器和 Intel Tofino Switch 晶片，大幅提升 CPU 間的傳輸效能與容量。
- Intel Tofino Switch chip, CSP-9550 significantly expands throughput performance and capacity between CPUs.
具備高達 28 個核心、六個 Intel UPI 通道，以及
乙太網路規模擴展網路的互連選項！ (Interconnect Options for Ethernet Scale Up Networks)
- Um, you know, I think that Meta said that they, they could see up to a 90% increase in the throughput or in the performance of their training jobs based on that reduction of link flaps.
所以這真的證明瞭它的可靠性。
- Um, you know, I think that Meta said that they, they could see up to a 90% increase in the throughput or in the performance of their training jobs based on that reduction of link flaps.
嗯，你知道，我認為 Meta 表示，由於連結中斷的減少，他們可以看到訓練工作的輸送量或效能最高可提升 90%。
現在我只推薦 MacBook 了… (I Can Only Recommend Macbooks Now…)
- The Pro and the Max configurations both come with a new neural processor and a nice little upgrade to peak unified memory, throughput and GPU performance, and
Pro 和 Max 配置都配備了新的神經處理器，以及統一記憶體、輸送量和 GPU 效能的小幅升級，
- The Pro and the Max configurations both come with a new neural processor and a nice little upgrade to peak unified memory throughput and GPU performance,
說真的，Apple 甚至默認了有多少 Apple Silicon 使用者會升級，方法是強調與 Intel CPU MacBook 相比，電池續航力會有多大的提升。
AMD Instinct MI455X 準備大展身手，效能與價值全面輾壓 NVIDIA！🔥 (AMD Instinct MI455X Is Turning Up the Heat on NVIDIA — Big Performance & Value!)
- Now MI455X aims right at the sweet spot: more throughput, smarter memory, aggressive price-performance.
H100、H200 還有 Blackwell 雖然在望，但 AMD 的 Mi300 已經讓人驚豔。
- More throughput, smarter memory, aggressive price performance so what is the Mi 455X?
更多的吞吐量、更聰明的記憶體、更具競爭力的價格效能，所以 Mi 455X 到底是什麼？