If you are so smart, tell everybody what affects MAXIMUM performance, in the pipeline? Which part of the pipeline is important for maintaining highest framerate, and therefore, achieving higher average framerates?This here is the embodiment of what I am talking about, this here is the definition of straw grasping, overhyping, wishful thinking that plagues this whole launch.
I will start by admitting I didn't read the whitepaper, I did now and indeed the white paper clearly says: DSRB can do up to 10% higher frame rates, finally we have a winner from AMD, UP TO 10%, meaning max percentage, meaning 0% as well, meaning typical 1~2%. meaning it's not really a game changer for Vega. Meaning BS marketing crap, anybody can come up with a new tech and claim it improves fps by up to 10%! and no one can challenge that claim! If it truly improved fps by any measurable amount they would have stated it quite clearly, eg, 5% to 8%, not this lame up to 10%. Their lame 10% maybe just 1fps in a theoretical test that runs at 10fps before DSBR!
Proof? Right under that very paragraph is a graph for bandwidth saving in games due to DSRB, but no graph for fps increases in those games, or any game whatsoever. Meaning they really did not record or achieve any fps improvements. They were keen enough and accurate enough to measure BW savings, but no fps? Yep no fps savings alright!
When you put this lame statement next to the statements they gave to Anandtech, it all fits together. They really don't expect anything to change from these technologies.
Man, your post is so epic it needs to be put in a picture frame!
Ignoring the continuing attempts to derail the discussion, would anyone happen to know if there is a way to tell which games that are commonly used for benchmarking are the most and least demanding of geometry performance (tessellation or otherwise)? In looking at this chart:
How many of the games near the bottom are known to be demanding of geometry performance and how many near the top are known to be undemanding of geometry performance? From what I've already been able to gather, GTA V and Watch Dogs 2 are both tessellation heavy, but I'm not too sure about the rest.
Previously we found Vega 56 to be 2% faster than the MSI GTX 1070 Gaming X on average across 25 games. We've since downgraded slightly to the GTX 1070 Founder Edition which is a few percent slower and we've also removed Crysis 3 and replaced it with eight other games. So this has changed the margin slightly and now Vega 56 is 5% faster, though only about 3% faster than the MSI model.
First, there is that suspicious win in Dirt 4 but even if you remove that result from the equation, Vega is still 1% faster. Not exactly a pantsing but at least it’s not slower.
Someone is back home...and filling in some gaps...
They made the 1070 look worse since their original 56 review:
Ignoring the continuing attempts to derail the discussion, would anyone happen to know if there is a way to tell which games that are commonly used for benchmarking are the most and least demanding of geometry performance (tessellation or otherwise)? In looking at this chart:
How many of the games near the bottom are known to be demanding of geometry performance and how many near the top are known to be undemanding of geometry performance? From what I've already been able to gather, GTA V and Watch Dogs 2 are both tessellation heavy, but I'm not too sure about the rest.
Unfortunately I don't think you can just look at the geometry complexity of a game and determine how it affects performance.
Edit: posted in the builders thread.
How do you know it actually reaches those voltages?
Try selecting 900mv what happens?
Unfortunately I don't think you can just look at the geometry complexity of a game and determine how it affects performance. Pretty sure overwatch isn't heavy on geometry, and neither is Hellblade probably (it's by a small time studio iirc). There are other things to consider like if the game already does geometry calling effectively or how well the draw calls can fill the wavefronts.
I think Dirt 4 is just somehow an excellent case for Vega where everything is perfect and Vega's compute performance is fully exposed.
Ignoring the continuing attempts to derail the discussion, would anyone happen to know if there is a way to tell which games that are commonly used for benchmarking are the most and least demanding of geometry performance (tessellation or otherwise)? In looking at this chart:
How many of the games near the bottom are known to be demanding of geometry performance and how many near the top are known to be undemanding of geometry performance? From what I've already been able to gather, GTA V and Watch Dogs 2 are both tessellation heavy, but I'm not too sure about the rest.
These tweets convinced me. I'm sold...
On firing Raja.
This kind of marketing BS that he speaks is a big part of the problem with Vega/RTG.
The reviews didn't highlight the perf/Watt dynamic well?
Ugh. When your product utterly fails to deliver in some category, stop trying to doublespeak your way out of it. Just saying nothing would be better, or admitting Perf/Watt is not good and moving on.
Firing may be a bit much, but otherwise bang on.
Vega is a huge failure in perf/watt to anyone with a set of eyes able to read benchmarks, if he cannot see this and thinks it actually competes with Nvidia in this metric just reviews are not highlighting it then he is not the man for this job. Get someone in there who knows what its performance is, and knows what Nvidias performance is, and go from there. Focus on what you are actually competitive in, being ignorant of your own products performance metrics is just sad from PR standpoint.
Sure, but that isn't related to the thing I'm trying to get an insight into. I'm interested in knowing the extent to which games near the bottom of the chart are very demanding of geometry performance and those near the top are not because I'm trying to investigate the extent to which Vega is presently front end bottlenecked by its polygon throughput being limited to 4 triangles per clock as primitive shaders (which are supposed to significantly boost polygon throughput) have not been enabled yet in the public drivers. If For example, Dirt 4 seems to perform spectacularly well on all AMD cards, so it would be relevant to know if that engine is more demanding of shader performance relative to geometry performance (for example), or if's outlier status is unrelated to GCN's front end bottleneck issues.
Still winning on compute when it comes to perf/watt, I believe, which is what he was hired to do, and what RTG's mission is. So, I don't see how he has failed at his job.
Raja's comments do sound like nonsense when you are looking at gaming performance, but that isn't Vega's goal. [...]
It's weird that the Vega 56 costs less than the GTX1070 in the US ($399 vs $499) but costs more than the GTX1070 in Canada ($699 vs $629).
Raja's comments do sound like nonsense when you are looking at gaming performance, but that isn't Vega's goal.
When I shop around, Vega 56 is way more than a 1070, if you can find a Vega 56.It's weird that the Vega 56 costs less than the GTX1070 in the US ($399 vs $499) but costs more than the GTX1070 in Canada ($699 vs $579).
Wrong, RX is the gaming line of cards, RX vega is 100% a gaming card, or at least AMD is marketing it as such.
Wrong, RX is the gaming line of cards, RX vega is 100% a gaming card, or at least AMD is marketing it as such.
You must have missed the part where they used a single die to go after GP104, GP102 and GP100 in different markets.
Thats all fine and dandy, they still made the decision to both add it to the RX line, and market it as a gaming card.
It's weird that the Vega 56 costs less than the GTX1070 in the US ($399 vs $499) but costs more than the GTX1070 in Canada ($699 vs $579).