AMD Announces The Radeon Pro VII

[agd5f]

Being a good architecture for compute means nothing if AMD says the opposite and doesn't provide the drivers. Has everyone forgotten Terascale already? Even after a decade, OGL4 support is still blocked because AMD says so. Being open source changes nothing, AMD controls that part of Mesa. GCN has been out for more than what Terascale ever got. It can be shut down tomorrow. Even today. Just to boost RDNA sales. They don't even need a valid reason.

Huh? Mesa is an open source project. We are major contributors, but we don't stop others from contributing. OpenGL 4.x is more or less complete barring the 64 bit extensions that very few, if any, games use. They have to be emulated on certain chips because those chips only support 32 bit operations. To support it, mesa needs to emulate them using 32 bit operations. That support didn't exist until more recently. Anyone could wire it up in r600. I'd be great if someone would. Even if AMD or any other vendor suddenly disappeared tomorrow, that doesn't take away the existing code or your ability to contribute to it.

[bridgman]

Wait, none of them? So AMD hasn't released a card with SR_IOV for 4 years?

I think it's actually a bit under 3 years with MI-25:

https://www.amd.com/en/products/prof.../instinct-mi25

[coder]

Can someone give me a recommendation (for my sister) as to how much VRAM is recommended for professional video editing with Davinci Resolve Pro on Windows 10?

Try asking on a forum specific to that product, or at least focused on professional video editing.

[coder]

A few differences I noticed:
1. TDP: 295W->250W, likely due to the improvements from TSMC

No, it's probably due to a market difference between gaming and workstations.

Maybe also a warranty thing, since lower clocks translate to better longevity, and I'm pretty sure it has a longer warranty than Vega VII. Also, it's more likely to be running 24/7 workloads that Radeon VII, which was marketed as a gaming card.

I'm not sure FP64 is still a thing for GPU computing...

Of course it is! GPUs don't implement denormals, so you really want fp64 for any serious floating point number crunching.

[moriel5]

Can someone give me a recommendation (for my sister) as to how much VRAM is recommended for professional video editing with Davinci Resolve Pro on Windows 10?

My country does not import any AMD Pro or Vega-based cards at all, with the cheapest 16GB NVidia option being the equivalent of ~2200$ (with a Radeon VII, or a used Vega Frontier on Amazon, including shipping from the US, being the equivalent of ~900$). Also, given the issues outlined here with GCN/CRDN, would they still be a recommendation for video editing, or does that suffer the same way as gaming (my sister does not game)?

If so, the next best option would be a 11GB GTX 1080 Ti for the equivalent of ~750$).

[Geopirate]

No, PRO cards don't support SR-IOV.

Wait, none of them? So AMD hasn't released a card with SR_IOV for 4 years?

[coder]

Speaking of which, is that the only difference between this and the Radeon VII?

Radeon VII:

• is 300 W (this is 250)
• runs at higher clocks (boosts to 1750 MHz instead of 1700 MHz)
• has only PCIe 3.0 (instead of 4.0)
• has half of the fp64 performance
• has no over-the-top Infinity Link connector for direct card-to-card communication.
• costs $700 (this lists for $1900)

Also... wasn't the Radeon VII basically just a binned workstation GPU? If so, does that mean this is a bin of a bin?

Before now, you could only get Vega 20 in:

• Mac-specific Radeon Pro II & Pro II Duo
• MI50 & MI60 (which lack any display outputs)

So, this is the first workstation card to feature it.

That's not to say it wasn't binned at all, since it has only 60 CUs, yet the MI60 and Radeon Pro II have 64.

[bridgman]

How will it allow for multi-GPU (using passthrough) -> single display?

Do you mean Crossfire or something different ?

The links are general purpose - they just give each GPU the ability to access HBM on a linked GPU with very high performance - although I expect they will be used for compute at least as much as for graphics.

[deusexmachina]

How will it allow for multi-GPU (using passthrough) -> single display?

[bridgman]

Making a pro version of an already failed product which is also on the verge of being artificially moved to artificial legacy state due to being a too old architecture.

With respect, what the heck does this even mean ?

Being a good architecture for compute means nothing if AMD says the opposite and doesn't provide the drivers. Has everyone forgotten Terascale already? Even after a decade, OGL4 support is still blocked because AMD says so. Being open source changes nothing, AMD controls that part of Mesa. GCN has been out for more than what Terascale ever got. It can be shut down tomorrow. Even today. Just to boost RDNA sales. They don't even need a valid reason.

Maybe I'm missing something here, but Terascale parts have been supporting GL 4.5 for a while (see MesaMatrix). Can you be more specific ?

What do you think we have been blocking via controlling "that part of Mesa" ?

How can we "shut down tomorrow" an upstream open source driver ? I didn't think any company was that powerful.