AMD Arcturus Is Probably a Vega-Based Professional GPU (Updated)

Linux publication Phoronix spotted a few patches to the AMDGPU Linux graphics driver that are related to AMD's next-generation Arcturus graphics cards.

(Image credit: AMD)

When AMD announced its latest Radeon DNA (RDNA) graphics card architecture at Computex 2019, the chipmaker made it very clear that it would co-exist along with the existing Graphics Core Next (GCN) architecture. The general consensus is that AMD would probably base its future gaming products around the RDNA architecture while relegating the GCN architecture exclusively to its workstation products. Therefore, Arcturus, not Navi, was Vega's successor all along.

Current AMD professional-grade offerings, such as the Radeon Pro Vega IIRadeon Instinct MI60 and Radeon Instinct MI50 employ the Vega 20 silicon, which TSMC produces for AMD on the 7nm manufacturing process. There's a high possibility that Arcturus will probably use a variation of the Vega silicon, and there is some evidence to support the rumor.

(Image credit: Freedesktop.org)

Phoronix previously discovered mentions of an unreleased AMD graphics card going by the "GFX908" ID. GFX9 is Vega and GFX10 is Navi. The Vega 20 die, in particular, has two IDs: GFX906 and GFX907. Assuming that GFX908 is Arcturus, it's definitely based on Vega, but with a small twist. We know from AMD's 2017 Financial Analyst Day slide presented by Mark Papermaster that the chipmaker's post-Navi graphics products will be fabricated under the 7nm+ process node. Up to this point, Arcturus seems to be a rewarmed Vega chip coming out of the 7nm+ microwave.

The Linux code shows up to three different device IDs so we could see at least three Arcturus models at launch. Twitter user 0x22h dug deeper into the Linux code and found out that Arcturus lacks display IP blocks, which means it's a pure compute graphics card making it akin to the Radeon Instinct models. We can expect the graphics card's feature set to include a benevolent amount of HBM2 memory, ECC (Error-Correcting Code) support, among other goodies. AMD currently utilizes Fiji, Polaris and Vega silicons for its Radeon Instinct graphics cards. Why the chipmaker is still selling Fiji-based products in 2019 is beyond us. Our guess is that Arcturus will most likely knock of the lower bound.

AMD could announce Arcturus at SIGGRAPH 2019 in July or Hot Chips in August. Either convention would be a great place to reveal an enterprise graphics card. As hinted in the AMD PowerPoint slide, Arcturus could launch next year.

Zhiye Liu
RAM Reviewer and News Editor

Zhiye Liu is a Freelance News Writer at Tom’s Hardware US. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.

  • bit_user
    We can expect the graphics card's feature set to include a benevolent amount of HBM2 memory
    Had to LOL at that phrase - "a benevolent amount of HBM2"!

    An amount can't be benevolent. AMD can be benevolent, if they provide a generous, copious, abundant, or sizable amount. Perhaps a plentitude, even.
    Reply
  • alextheblue
    This should come as no surprise given that Vega was great at compute, but only passable at gaming. RDNA is definitely more optimized for gaming workloads.
    Reply
  • bit_user
    alextheblue said:
    RDNA is definitely more optimized for gaming workloads.
    Presumably, most of the additional transistor budget went towards improving gaming efficiency. However, I've not actually seen any GPU compute benchmarks for Navi - have you?

    I think the compute driver stack is still catching up. Navi isn't yet supported on the latest Linux release.

    Not that I disagree about GFX908 being Vega-based nor about RDNA possibly even representing the bifurcation of their consumer & professional lines, as Lisa Su recently alluded. I just don't want to prejudge Navi's compute capabilities before we have some good data on it. I'm also not sure we won't eventually see RDNA in server products.

    Consider that Nvidia's Volta core is faster at both gaming and compute workloads. AMD certainly could've achieved a similar accomplishment.
    Reply
  • bit_user
    Zhiye,

    Though I doubt you'll read this...

    bit_user said:
    Had to LOL at that phrase - "a benevolent amount of HBM2"!
    If a single word choice is the biggest complaint I have about your articles, consider it a compliment!
    : )
    Thanks for your articles & keep up the good reporting.
    Reply
  • alextheblue
    bit_user said:
    Presumably, most of the additional transistor budget went towards improving gaming efficiency. However, I've not actually seen any GPU compute benchmarks for Navi - have you?

    I think the compute driver stack is still catching up. Navi isn't yet supported on the latest Linux release.

    Not that I disagree about GFX908 being Vega-based nor about RDNA possibly even representing the bifurcation of their consumer & professional lines, as Lisa Su recently alluded. I just don't want to prejudge Navi's compute capabilities before we have some good data on it. I'm also not sure we won't eventually see RDNA in server products.

    Consider that Nvidia's Volta core is faster at both gaming and compute workloads. AMD certainly could've achieved a similar accomplishment.
    Regarding what they can accomplish, they have to make the most of their smaller R&D budget.

    Frankly AMD's actions with their Instinct line (and their focus on gaming with Navi) had me doubting Navi's compute chops more than anything else. That aside, perhaps you're right. Their drivers are early, and the only compute I'd really seen was LuxMark. That has Vega beating the snot out of Navi at Luxmark but there's probably some serious software stack problems there. When I crunch the numbers, Vega and Navi have the same FLOPs per MHz per shader. At least on paper.

    So perhaps they just don't want to spend the time and effort (and precious R&D budget) releasing a Vega 20 sized Navi, only to end up with merely the SAME compute performance. If that's the case they may be waiting until their future architecture is baked.
    Reply