Nvidia has quietly eliminated among the concurrent video encoding limitations from its shopper graphics processing models, to allow them to now encode as much as 5 streams. The transfer will simplify the lifetime of video fans, however Nvidia’s information heart grade {and professional} GPUs will proceed to have an enormous edge over shopper merchandise as now Nvidia doesn’t limit the variety of concurrent periods on them.
Nvidia has elevated the variety of concurrent NVENC encodes on shopper GPUs from three to 5, in response to the corporate’s personal Video Encode and Decode GPU Help Matrix (opens in new tab). That is efficient for dozens of merchandise based mostly on the Maxwell 2nd Gen, Pascal, Turing, Ampere, and Ada Lovelace microarchitectures (besides some MX-series merchandise) and launched within the final eight years or so. Nonetheless, the variety of concurrent NVDEC decodes in these GPUs remains to be restricted to at least one stream.
The change doesn’t have an effect on the variety of NVENC and NVDEC {hardware} models activated in Nvidia’s shopper GPUs. For instance, Nvidia’s newest AD102 graphics processor, based mostly on the Ada Lovelace structure, options three {hardware} NVENC encoders (opens in new tab) and three NVDEC decoders. All three are enabled on Nvidia’s RTX 6000 Ada and L40 boards for workstations and information facilities, however solely two are lively on consumer-grade GeForce RTX 4090.
Traditionally, Nvidia has restricted the variety of concurrent decoding and encoding periods on all of its graphics playing cards. Client-oriented GeForce boards supported as much as three simultaneous NVENC video encoding periods and as much as one NVDEC video decoding stream. In contrast, Nvidia’s workstation and information heart options operating the identical silicon and aimed toward ProViz, video streaming providers, video games streaming providers, and digital desktop infrastructure (VDI) may assist 11 – 17 concurrent NVENC periods relying on high quality and {hardware}. It turned out a few years in the past that Nvidia’s restrictions may very well be eliminated by a comparatively easy hack.
However Nvidia’s stance on NVENC and NVDEC limitations has evidently modified. Now, shopper GPU assist as much as 5 concurrent NVENC encode periods and as much as 1 NVDEC decode session without delay. In contrast, workstation-grade and information center-grade boards should not have any restrictions and the precise variety of concurrent periods depends upon precise {hardware} capabilities and video high quality.