Nvidia has quietly removed some of the concurrent video encoding limits from their consumer graphics processing units, allowing them to encode up to 5 streams. The move will simplify life for video enthusiasts, but Nvidia’s data center-grade and professional GPUs will continue to have a significant edge over consumer products as Nvidia no longer limits the number of concurrent sessions. .
Nvidia has increased the number of concurrent NVENC encodes on consumer GPUs from 3 to 5, according to the company itself. GPU support matrix for video encoding and decoding (opens in new tab)It is based on the Maxwell 2nd generation, Pascal, Turing, Ampere, and Ada Lovelace microarchitectures (excluding some MX series products) and has been used in dozens of products released in the last eight years or so. valid. However, the number of concurrent NVDEC decodes on these GPUs is still limited to one stream.
This change does not affect the number of NVENC and NVDEC hardware units activated on Nvidia’s consumer GPUs. For example, Nvidia’s latest AD102 graphics processor, based on the Ada Lovelace architecture, features three pieces of hardware. NVENC Encoder (opens in new tab) and 3 NVDEC decoders. All three are enabled on his Nvidia’s RTX 6000 Ada and L40 boards for workstations and data centers, while only two are active on the consumer-grade GeForce RTX 4090.
Historically, Nvidia has limited the number of concurrent decode and encode sessions on all graphics cards. His consumer GeForce board supported up to 3 simultaneous NVENC video encoding sessions and up to 1 NVDEC video decoding stream. In contrast, Nvidia’s workstation and data center solutions running the same silicon and targeted at ProViz, video streaming services, game streaming services, and virtual desktop infrastructure (VDI) can, depending on quality and hardware, It can support 11-17 concurrent NVENC sessions. A few years ago, it turned out that Nvidia’s restrictions could be removed with a relatively simple hack.
However, Nvidia’s stance on NVENC and NVDEC limits has clearly changed. Consumer GPUs today support up to 5 concurrent NVENC encoding sessions and up to 1 NVDEC decoding session at a time. In contrast, workstation grade and data center grade boards have no limit and the actual number of concurrent sessions depends on the actual hardware capabilities and video quality.
