The First Radeon Vega Frontier Linux Benchmark Doesn't Tell Much

[phoronix]

Phoronix: The First Radeon Vega Frontier Linux Benchmark Doesn't Tell Much

We have some OpenGL numbers for Radeon Vega Frontier Edition on AMDGPU-PRO under Linux...

http://www.phoronix.com/scan.php?page=news_item&px=Vega-Frontier-Linux-Quickie

[chuckula]

At this point even the Windows drivers are in such bad shape that there are rumors that major features (like tiled rendering) aren't even turned on.
Given that level of heartburn, don't expect miracles on Linux anytime soon.

[microcode]

At this point even the Windows drivers are in such bad shape that there are rumors that major features (like tiled rendering) aren't even turned on.
Given that level of heartburn, don't expect miracles on Linux anytime soon.

Why would there be tiled rendering on desktop?

[juno]

Why would there be tiled rendering on desktop?

To reduce bandwidth requirements and boost efficiency. Basically what Nvidia does to be that competitive since Maxwell

[Shevchen]

Why would there be tiled rendering on desktop?

In games... its Vega. AMD said, tiled-rendering is a feature so you would expect to be active when benching games. It isn't, cause the driver does not support it (for whatever reasons)

The net right now is full of rumors and honestly - all those speculations slowly give me a headache. What we can say for now:

The per-shader performance is about Polaris level (meaning Vega arch-improvements are not active). These are expected to kick off with the RX Vega and its "real" gaming drivers. 3dcenter said, no binnig on Vega cards, tiled rasterization is off, HBCC is off, the new geometry pipeline is off/nerfed/whatever, gaming-part of the driver is from January and so on...

In general, its possible that the card runs with 50% handbrake nerf due to the drivers and we have no way to guess if it stays on 1080 performance or will beat the Titan Xp.

[juno]

The per-shader performance is about Polaris level (meaning Vega arch-improvements are not active). These are expected to kick off with the RX Vega and its "real" gaming drivers. 3dcenter said, no binnig on Vega cards, tiled rasterization is off, HBCC is off, the new geometry pipeline is off/nerfed/whatever, gaming-part of the driver is from January and so on...

In general, its possible that the card runs with 50% handbrake nerf due to the drivers and we have no way to guess if it stays on 1080 performance or will beat the Titan Xp.

But why would AMD release a card that underperforms by 50%?

IMHO tiled caching is either turned off, not working or working in another way what Nvidia does so that Nicolas Guillemot's tool doesn't show it. HBCC shouldn't improve performance when everything fits inside the VRAM anyway and I see the "new geometry pipeline" more like internal rearrangements that simplify the pipeline and doesn't result in a huge performance boost.

[Shevchen]

But why would AMD release a card that underperforms by 50%?

IMHO tiled caching is either turned off, not working or working in another way what Nvidia does so thatNicolas Guillemot's tool doesn't show it. HBCC shouldn't improve performance when everything fits inside the VRAM anyway and I see the "new geometry pipeline" more like internal rearrangements that simplify the pipeline and doesn't result in a huge performance boost.

The most agreed answer to the first part is:
They promised their share-holders to release Vega in H1, so they delivered Vega. It didn't matter how well it ran, how mature the drivers were or if even all features were active - they did not want to break their promise, so they delivered.

For the geometry pipeline I have to disagree - this thing could give a performance boost if it solves the "wide arch" bottleneck, which AMD has been plagued with. But lets wait and see. All in all, I find the 50% handbrake picture not so far-fetched and wouldn't be surprised, if we end up with 2x the performance we see right now. But this is also me hoping for it, so yes - I'm a little biased.

[vortex]

But why would AMD release a card that underperforms by 50%?

IMHO tiled caching is either turned off, not working or working in another way what Nvidia does so that Nicolas Guillemot's tool doesn't show it. HBCC shouldn't improve performance when everything fits inside the VRAM anyway and I see the "new geometry pipeline" more like internal rearrangements that simplify the pipeline and doesn't result in a huge performance boost.

There is simply no way that the card is nerfed that much. Maybe, 5-10%, but, that is about it.
AMD even said that the drivers are not gimped for the FE, they are just "older".

As for tiled rendering, it IS doing that, it was already proven it is doing that on the pcper 'live' benchmark.

It sure seems like they just hit a brick wall with the process tech from GloFlo. Once they push up the voltages to get their desired speed, TDP skyrockets. This is why pretty much all the cards are only rarely hitting 1600MHz, it just is guzzling too much power to stay under the 300w envelope they wanted.

[Shevchen]

As for tiled rendering, it IS doing that, it was already proven it is doing that on the pcper 'live' benchmark.

Nope, its not:





Source (german)
https://www.3dcenter.org/news/amds-r...und-damit-im-p

Rough translation:
Main features are off and performance drop could bring Vega FE even below Polaris shader-to-shader performance, due to the fact that Polaris driver are made for Polaris and not for GCN 5 (which could invoke further performance losses)

[Stoatally]

As for tiled rendering, it IS doing that, it was already proven it is doing that on the pcper 'live' benchmark.

This is the exact opposite of what happened, he used the tool and it exhibited the same rasterisation as all other GCN cards, but was closest to how a Fury tests.