Intel Challenges Nvidia, AMD With Trio Of Workstation GPUs

    3
    Intel Challenges Nvidia, AMD With Trio Of Workstation GPUs

    This week, Intel revealed a trio of new graphics cards targeted at business applications including architectural design, engineering, and content production. These new graphics cards are Intel’s response to AMD’s FirePro and what was once Nvidia’s Quadro workstation GPUs.

    The single slot A40 and the dual slot A50 are among the mobile workstation GPUs in Intel’s professional portfolio, which is referred to as the Arc Pro series. The only distinction between the two desktop cards, which both have 6GB of GDDR6 memory, is that the A50’s dual-slot cooler supports a higher 75W power envelope. The A30M, however, has 4GB of GDDR6 and a TDP of 35 to 50 W.

    Curiously, Nvidia also offers an A40 GPU, but that model comes with 48GB of GDDR6 and advertises FP32 performance of between 37.4 and 74.8 teraflops while using 300W.

    Despite having similar names, Intel’s Arc Pro cards primarily target low-end workstations and applications that can benefit from the platform’s advantages.

    Have you heard AV1?

    According to Moor Insights and Strategy analyst Anshel Sag, Intel’s top-tier Arc Pro A50 boasts peak performance of 4.8 teraflops of FP32 performance, placing the card between Nvidia’s T1000 (2.5 teraflops) and its recently introduced A2000 (8 teraflops).

    However, a portion of that performance difference could be explained by a restriction necessary to stay inside the PCIe slot’s 75W power allowance. For instance, the Nvidia GA106 utilised in the A2000 is also used in the RTX 3060, which offers the performance of more than 12 teraflops for FP32 but uses around twice as much power.

    Sag points out that the performance of Intel’s first-generation device isn’t particularly shocking or unexpected.

    In general, he claimed, “Arc was never designed to compete with the best performance graphics cards from AMD or Nvidia, at least at launch. As far as consumers are concerned, “they were always going to start entering the market at a more mainstream target profile” since “creating a flagship GPU is really tough, highly costly, and actually a pretty low-volume business.”

    While items like Nvidia’s 3090 TI could get people’s attention, he added, they are relatively low volume, have a good margin, and ultimately account for a very little portion of the company’s revenues. The majority of the money and profit, in actuality, are found in the centre, according to Sag.

    Sag was taken aback by Intel’s choice to introduce a workstation-class chip at this time, even if he doesn’t think it strange that the firm targeted the low-to-mid-tier graphics market for its initial foray into Arc.

    But he adds that features like AV1 encoding, which Intel claims to be the first to offer, may be useful to app developers and content producers who want to get a head start with the codec. As long as you don’t sacrifice quality, Sag said, “faster is always better when it comes to rendering of any type.”

    Additionally, Arc Pro enables Intel to approach OEMs with a workstation component that is reasonably priced and suitable for a range of workloads. However, Intel makes a point of highlighting the cards’ PCIe 4.0 8x interfaces and PCIe 3.0 backward compatibility, indicating that it obviously anticipates these cards won’t be restricted to OEM designs.

    As AMD has done with several of its recent entry-level gaming and workstation cards, Intel might have gotten away with a PCIe 4.0 4x interface given the relatively modest performance objective for these cards. Performance on previous computers would have most certainly been affected by doing so.

    Can AMD be replaced by Intel in workstation graphics?

    Sag points out that while Intel faces tough competition from AMD and Nvidia in the gaming graphics industry, the company has a far greater chance of succeeding there.

    According to Sag, AMD’s efforts in this area pale in comparison to those of Nvidia. Because it has “previously invested a lot of time and effort certifying its CPUs for these precise workloads and has established ties with all the ISVs,” Intel has an edge over its rivals.

    Intel emphasised compatibility for a number of well-known software systems and frameworks in their marketing materials, including Handbrake, Premier Pro, Davinci Resolve, and Gigapixel AI.

    This is especially true with regards to Intel’s QuickSync media-encoding features for its onboard graphics. Additionally, Arc Pro, when used in conjunction with a suitable GPU, allows Intel to distribute encoding workloads across the onboard and discrete graphics.

    Don’t write Intel Arc off just yet.

    The launch also dispels reports that Intel’s graphics branch was going to be eliminated in the wake of dismal second-quarter financial results. The company shut down its Optane persistent memory and SSD businesses and posted a $454 million net loss before the end of last month.

    The reports of average performance and weak launch drivers did not assist these concerns.

    They are obviously having some difficulties with the drivers, Sag said, adding that it is typical for drivers to significantly advance throughout the length of a GPU’s life.

    He went on to say that “this workstation launch kind of negates a lot of the rumours that they’re going to shutdown their discrete business that they haven’t even really spun up yet,” contending that many of these rumours are probably the result of concern that Intel might threaten Nvidia and AMD’s duopoly.

    According to Sag’s calculations, Intel’s graphics business is on a distinct trajectory than its other product divisions and was created as a separate product division for this specific purpose.