NVIDIA RTX30 New Product Launch Conference

 Presumably everyone is still in the horrible performance improvement of NVIDIA's latest RTX30 series graphics cards and the price of making friends (the 3070 with performance comparable to 2080Ti only starts at 3899 yuan); of course, the card is very strong and the price is very affordable. Hi, but there are actually more details hidden in this press conference. It’s just that everyone’s eyes are mostly attracted by the new card, so they haven’t paid attention to it; let’s give you a list below. 1. NVIDIA RTX IO bypassing the CPU Here we make a bold conclusion: In fact, this generation of RTX30 graphics cards supports PCIe 4.0 by default, and can benefit a lot from it! Why do you say that? I don’t know if you have noticed the RTX IO introduced by Lao Huang before showing the real card. He first explained the workflow of a traditional PC, that is, the GPU needs to process image data through the CPU to copy the data to the memory and decompress the image to the video memory. This process, and this processing method currently has caused a bottleneck to the GPU. The new RTX IO API can quickly load data directly from the SSD and stream it to the GPU memory. The GPU is decompressed without loss, which greatly simplifies The process of transferring data from storage devices to GPU memory not only greatly improves data throughput, but also saves CPU resources and reduces latency. The reason why RTX30 graphics cards can benefit a lot from PCIe 4.0 is because PCIe 4.0 SSD can meet the massive data transmission needs of RTX30 series GPUs to a great extent. It may not be obvious in low-resolution scenes such as 1080P and 2K, but when it comes to 4K high frame rate or even 8K environment, the gap in experience is only Will be revealed! And I think this is also the key to this year's RTX3090 can bring users an 8K experience (although I don't know how many users can use 8K, XD), but as the number one graphics chip manufacturer, it is indeed right to go ahead. But what makes me more curious is that now that we all know that the new RTX30 series cards support PCIe4.0 is a certainty, but why doesn't NVIDIA use this as a core selling point to promote it? (Just like when AMD released the RX5700 series of new cards back then) and Lao Huang almost didn't mention a word at the press conference. I personally believe that the possible reasons are as follows: the Intel platform has not yet supported PCIe 4.0. If the high-end NV graphics cards push PCIe 4.0 with great fanfare at this time, it is tantamount to directly denying the status of Intel's high-end platform, and all high-end players will immediately turn to AMD ( Because PCIe4.0 SSD can be used on AMD's platform at present). Of course, NVIDIA has to take into account Intel's face (or Intel has negotiated with NV). After all, it is also a partner. When Intel also supports PCIe 4.0, this selling point will definitely be released by NV. (Of course, this is just my guess. Just listen to it. XD) 2. Another major breakthrough in the world's fastest GDDR6X memory Ampere card jointly developed with Micron is the use of PAM4 jointly developed by NVIDIA and Micron. Signaling GDDR6X video memory, its transmission data is twice that of GDDR6, can provide graphics cards with ultra-high bandwidth greater than 1TB/s, paving the way for future 4K high frame rate and 8K application rendering scenes. 3. NVIDIA Reflex: The e-sports weapon that counts every second! This is a response optimization technology specially launched by NVIDIA for e-sports gamers to optimize and measure system latency. Among them, NVIDIA Reflex low-latency mode is being used in popular e-sports games such as "Apex Legends", "Call of Duty: Warzone", "Fortnite" and VALORANT. Reduce delay by 50%;