I didn't write a VEGA white paper that claims that new features will improvise graphics everywhere.
AMD did.
Did the white paper claim that? Really? Lets see....
The introduction to the White Paper states the following:
"The “Vega” architecture is intended to meet today’s needs by embracing several principles:
flexible operation, support for
large data sets,
improved power eficiency, and
extremely scalable performance. “Vega” introduces a host of innovative features in pursuit of this vision, which we’ll describe in the following pages. This new architecture promises to revolutionize the way GPUs are used in both established and emerging markets by
offering developers new levels of control, flexibility, and scalability." Nowhere is mentioned in the introduction that Vega features are targeted at "brute force" performance.
Regarding HBCC it states that "The availability of a higher-capacity pool of hardware-managed storage can help enable game developers to create virtual worlds with higher detail, more realistic animations, and more complex lighting effects without having to worry about exceeding traditional GPU memory capacity limitations." This is not exactly about performance but raising capacity limitations. While it has necessarily an effect in performance, that ia not something that could be immediately seen since games available today are designed around current limitations.
Concerning the "Next Generation Geometry Engine", the White Paper does say "To meet the needs of both professional graphics and gaming applications, the geometry engines in “Vega” have been tuned for higher polygon throughput by adding new fast paths through the hardware and by avoiding unnecessary processing." Now, it seems we can see the impacts of these changes on CAD and other 3D workstation workloads (where it is competitive with GP102), not just in games. This is a big red flag for me towards the stance that something in Vega that is good for professional graphics/compute is not for gaming. Otherwise why would the gains on geomtery be seen in one case and not the other?
Then we have Rapid Packed Math, which as we all know can only be relevant if games start making extensive use of FP16. Again this is a feature whose impact on performance that cannot be seen on today's games. AMD recognises this on the paper when it says a)"The programmable compute units at the heart of “Vega” GPUs have been designed to address this
changing landscape with the addition of a feature called Rapid Packed Math." and b) "
For applications that can leverage this capability, Rapid Packed Math can provide a substantial improvement in compute throughput and energy efficiency."
There is also this: "In addition to Rapid Packed Math, the NCU introduces a variety of new 32-bit integer operations that can improve performance and efficiency in
specific scenarios. These include a set of eight instructions to accelerate memory address generation and hashing functions
(commonly used in cryptographic processing and cryptocurrency mining), as well as new ADD/SUB instructions designed to minimize register usage. This explains why Vega is so strong at cryptography. There are also additonal tidbits about improvements that are focused at video and image processing algorithms for AI.
After it goes into the DSBR technology which is arguably one of the features that could have a bigger impact on graphics after clock speed and for all we know its not working in current drivers. Still, the impact as stated by AMD themselfs is not huge: "In the case of “Vega” 10, we observed up to 10% higher frame rates". AMD also includes caveats like "Even larger performance improvements are possible
when developers submit geometry in a way that maintains screen space locality or in cases where many large overlapping polygons need to be rendered."
Clock Speed is next and we all know the huge impact it obviously has. Still its part of the architecture and not just a "natural metric" as some make it seem: "One of the key goals for the “Vega” architecture was achieving higher operating clock speeds than any prior Radeon™ GPU. Put simply, this effort required the design teams to close on higher frequency targets.
The simplicity of that statement belies the scope of the task, though. Meeting “Vega’s” substantially tighter timing targets required some level of design effort for virtually every portion of the chip.". The white paper goes into a lot of detail on this.
Next is power efficiency, specifically about reducing IDLE power consumption as well as in video transcoding. It barely mentiones power efficiency while performing graphics tasks.
- "Meanwhile, an improved “deep sleep” state allows “Vega” 10 to scale down its clock speeds dramatically at idle in order to achieve substantially lower power consumption."
- "In “Vega” 10, this fabric is clocked separately from the graphics core. As a result, the GPU can maintain high clock speeds in the Infinity Fabric domain in order to facilitate fast DMA data transfers in workloads that feature little to no graphics activity, such as video transcoding."
The white paper concludes with information about "Display and Multimedia Improvements" which is all about Free Sync, HDR, etc, with zero regard to graphics improvements.
So I ask you now, where did you see the White Paper claiming "that new features will improvise graphics everywhere"? Because AGAIN, I cannot see any of it, after not seeing your claims that VEGA has worse perf/Flops than AMD previous chips. The White Paper is actually quite cautious in what it states with plenty of context and caveats given. Like I said a few posts ago, it seems that you (and not only you) made up your mind about VEGA even before looking at the data available.
You make good points. In the white paper, AMD significantly backpedals on all their pre-release marketing hype.
