AMD Polaris GCN 4.0 Macau China Event

For DX12 it is very relevant.

For async compute, sure, but isn't DX12 more than just async compute? Developers don't seem to be in any hurry to implement other DX12 aspects, such as MGPU support, are we sure async compute will be widely implemented? Are any AAA game titles (I don't consider AOTS to be in this category) currently implementing async compute?
 
For async compute, sure, but isn't DX12 more than just async compute? Developers don't seem to be in any hurry to implement other DX12 aspects, such as MGPU support, are we sure async compute will be widely implemented? Are any AAA game titles (I don't consider AOTS to be in this category) currently implementing async compute?
aots does mgpu already, see below. and people have to remember how long it took to implement EVERY past DX update. shit, dx11 how been out for years and there are still dx9 games out. dx12 adoption will take time and in a few years people will be saying the same things about dx13...

Ashes of the Singularity DirectX 12 Mixed GPU Performance
http://www.anandtech.com/show/9740/directx-12-geforce-plus-radeon-mgpu-preview/4
 
Last edited:
aots does mgpu already, see below. and people have to remember how long it took to implement EVERY past DX update. shit, dx11 how been out for years and there are still dx9 games out. dx12 adoption will take time and in a few years people will be saying the same things about dx13...

Ashes of the Singularity DirectX 12 Mixed GPU Performance
GeForce + Radeon: Previewing DirectX 12 Multi-Adapter with Ashes of the Singularity

How many other AAA DX12 games have MGPU implementation? Let's forget AotS for a moment (the gist of my original post), we already know it's a fine showcase benchmark for AMD, what other titles?
 
Looks like some of you clearly missed my point with DX12 Warhammer.

It's not to say Fury X = 1080.

It's to make the point that this site tests very few games. Each title makes a big difference in the outcome and if they are all skewed to one side, NV or AMD, it leads to an obvious conclusion. Currently there's 5 out of 6 games that are sponsored by NV. If they used 5 out of 6 games sponsored by AMD, I would bring up the exact same point.

Example, on TPU, with more games, they find Fury X now = 980Ti at 1440p, and beat it a little at 4K. Yet here on this site, it's the 980Ti in front by a fair margin, why? Cos again, most of the titles skew to NV.

perfrel_2560_1440.png


perfrel_3840_2160.png


As it currently stands, there's zero benefit for AMD to send their GPUs to [H] for reviewing. I hope that point is understandable.
I guess then PCGameshardware (who do some of the best analysis out there) should not get AMD cards then....

Using the summary figures unfortunately hides any trend-behaviours and why individual games should be used rather than that chart, just emphasise any caveats.

Try PCGameshardware where they use custom AIB vs custom AIB if you want to compare between AMD and Nvidia.
Will need to use a translator such as the one in Google Chrome, also benchmark charts defaults to 1080p:
Geforce GTX 1080 im Test: Der erste 16-nm-König mit 2 GHz im OC-Betrieb [Test der Woche]
Hitman (2016) Episode 2: DX12 mit bis zu 60 Prozent Leistungsplus - und Problemen
Total War: Warhammer is optimised with AMD and yet custom 980ti still competes with Fury X: Total War: Warhammer - Vorab-Benchmarks mit DirectX 12
Homefront: The Revolution im Test mit Benchmarks der Cryengine
etc.
They use a standard FuryX as custom models are same clocks.
Here the trend (meaning majority but not all games) is that 980ti is still outperforming the Fury X at 1440p, at 4k it gets much closer and can swap positions for some games but they are not realistic and practical so 1440p makes more sense unless specifically going SLI/xfire 4k (which adds other technical scope-focus headaches).
Cheers
 
Last edited:
It's not just a showcase for AMD it shows what can be done with dx12. My points were that aots has other dx12 features and that we will see more dx12 games as adoption picks up. Ignoring aots, no other game has proper dx12 yet. when a new dx version is released it takes years for it to become the go to API. Just like ever release in the past. Your argument is the same as when it moved from 9 to 11. Adoption takes time. I remember when they're was no such thing as DirectX...

Edit
I guess war hammer has dx12.
 
It's not just a showcase for AMD it shows what can be done with dx12. My points were that aots has other dx12 features and that we will see more dx12 games as adoption picks up. Ignoring aots, no other game has proper dx12 yet. when a new dx version is released it takes years for it to become the go to API. Just like ever release in the past. Your argument is the same as when it moved from 9 to 11. Adoption takes time. I remember when they're was no such thing as DirectX...

Edit
I guess war hammer has dx12.

I think Quantum Break has it too, but that game is riddled with way too many bugs and is too unoptimized to warrant testing.

Edit: I honestly don't care about any dx12 benchmarks until Deus Ex comes out and, when hell freezes over, Star Citizen.
 
Warhammer is full dx12 including async and battlezone will be too. like I said adoption is slow but it is starting to pick up. in a year we may have dozens of games and devs may be saying fuck dx11.
 
As it currently stands, there's zero benefit for AMD to send their GPUs to [H] for reviewing. I hope that point is understandable.

Sorry to post on this again but I thought it is pertinent as your coming to a strong conclusion using potentially wrong data.
I say wrong data because you used the 1080FE as a way to compare the Fury X to the 980ti, but that is not a reliable way to do this as all % are relative to 1080 and its scale.

I have just looked at every game they used when reviewing the Asus 980ti back in Feb 2016 and in majority of games the reference 980ti comes up comfortably in front, 2 games FuryX is faster, and 2 games pretty close to each other.
The custom 980ti is well in front in all games, which is a better comparison tbh for reasons I mentioned earlier and that the FuryX has a designed performance envelope closer to that of a custom AIB rather than reference; one reason the gap grows so much from a reference 980ti to an ok custom AIB 980ti.
ASUS GeForce GTX 980 Ti Matrix 6 GB Review
However again the % gap between FuryX and reference 980 is skewed with them relative to the custom review where it has over 20% gain compare to reference 980ti; their gap going through the games is averaged around 6-7% when calculating just reference 980ti and FuryX.

Also I think you are being overly critical to the reviewers here at HardOCP and the games they use.
Check out the few games they used for Nvidia 1080FE and it is pretty balanced inlcuding ones really not designed for Nvidia:
They used Doom (OpenGL game so worth having), AoTS, The Division, RoTR (has technology from both AMD-Nvidia), Fallout 4, Witcher 3 (rendering engine was optimised towards AMD consoles but also includes Nvidia tech), and Battlefield 4.

One could say maybe Fallout 4 could be replaced with Chapter 2 of Hitman (Chapter 2 is also designed to work with Nvidia where Chapter 1 has issues).
But it is impossible to keep everyone happy on what games should be used.

Cheers
 
Last edited:
Sorry to post on this again but I thought it is pertinent as your coming to a strong conclusion using potentially wrong data.
I say wrong data because you used the 1080FE as a way to compare the Fury X to the 980ti, but that is not a reliable way to do this as all % are relative to 1080 and its scale.

I have just looked at every game they used when reviewing the Asus 980ti back in Feb 2016 and in majority of games the reference 980ti comes up comfortably in front, 2 games FuryX is faster, and 2 games pretty close to each other.
The custom 980ti is well in front in all games, which is a better comparison tbh for reasons I mentioned earlier and that the FuryX has a designed performance envelope closer to that of a custom AIB rather than reference; one reason the gap grows so much from a reference 980ti to an ok custom AIB 980ti.
ASUS GeForce GTX 980 Ti Matrix 6 GB Review
However again the % gap between FuryX and reference 980 is skewed with them relative to the custom review where it has over 20% gain compare to reference 980ti; their gap going through the games is averaged around 6-7% when calculating just reference 980ti and FuryX.

Also I think you are being overly critical to the reviewers here at HardOCP and the games they use.
Check out the few games they used for Nvidia 1080FE and it is pretty balanced inlcuding ones really not designed for Nvidia:
They used Doom (OpenGL game so worth having), AoTS, The Division, RoTR (has technology from both AMD-Nvidia), Fallout 4, Witcher 3 (rendering engine was optimised towards AMD consoles but also includes Nvidia tech), and Battlefield 4.

One could say maybe Fallout 4 could be replaced with Chapter 2 of Hitman (Chapter 2 is also designed to work with Nvidia where Chapter 1 has issues).
But it is impossible to keep everyone happy on what games should be used.

Cheers
screenshot-www pcgameshardware de 2016-05-30 21-41-42.png
PNaaoZc.jpg
 
Sorry to post on this again but I thought it is pertinent as your coming to a strong conclusion using potentially wrong data.
I say wrong data because you used the 1080FE as a way to compare the Fury X to the 980ti, but that is not a reliable way to do this as all % are relative to 1080 and its scale.

I have just looked at every game they used when reviewing the Asus 980ti back in Feb 2016 and in majority of games the reference 980ti comes up comfortably in front, 2 games FuryX is faster, and 2 games pretty close to each other.
The custom 980ti is well in front in all games, which is a better comparison tbh for reasons I mentioned earlier and that the FuryX has a designed performance envelope closer to that of a custom AIB rather than reference; one reason the gap grows so much from a reference 980ti to an ok custom AIB 980ti.
ASUS GeForce GTX 980 Ti Matrix 6 GB Review
However again the % gap between FuryX and reference 980 is skewed with them relative to the custom review where it has over 20% gain compare to reference 980ti; their gap going through the games is averaged around 6-7% when calculating just reference 980ti and FuryX.

Also I think you are being overly critical to the reviewers here at HardOCP and the games they use.
Check out the few games they used for Nvidia 1080FE and it is pretty balanced inlcuding ones really not designed for Nvidia:
They used Doom (OpenGL game so worth having), AoTS, The Division, RoTR (has technology from both AMD-Nvidia), Fallout 4, Witcher 3 (rendering engine was optimised towards AMD consoles but also includes Nvidia tech), and Battlefield 4.

One could say maybe Fallout 4 could be replaced with Chapter 2 of Hitman (Chapter 2 is also designed to work with Nvidia where Chapter 1 has issues).
But it is impossible to keep everyone happy on what games should be used.

Cheers


I think that [H]ardocp picks games that push video cards. I also believe that they would love to add in a few more into the testing suite. The problem is this:

AMD and Nvidia's PR teams are jerks. It's the truth. Here's why.

First they send cards out with a week or less to test them. It's like they want to limit the testing time with a NDA for some special day to launch them. All the consumer wants is a card in the store and a test to know if it is worthwhile. How they suckered the powers that be at these companies into making a special job title for them amazes me sometimes. [H]ardocp tests are extremely time consuming; you can't rush a play through. A week is not enough!

Then in the middle of the review AMD and Nvidia like to send over updated drivers. Strictly to f*ck with the reviewers I believe. You would think that they had a team of testers playing the latest games to notice that the game crashes when you hit jump or whatever. Nope! Just update the drivers and make the reviewers run every test again. So now instead of having 7 days to run tests you have 3. I wonder how much hair does Brent_Justice lose when they do this? I'm malicious when stuff like that happens to me. I would ding them a point if I were testing. :sneaky:

Then when the review comes out they want to argue the results because they are a PR team designed to cover up mistakes. Fury X doesn't overclock? Make up a reason why or not help Wizzard to update MSI Afterburrner to be able to OC the thing. If you can't use the program to OC it then it might still OC well! GTX 1080 can't hit 2100 consistently as shown on stage? Don't tell people that the consumer cards are voltage locked and hopefully everyone will not realize it until they have sold a boatload. Wood screws on stage. Dual Fury with an Intel CPU. The Cat and Mouse game goes on and on.

Look the truth is that people play a wide assortment of games. After I have the [H]ardocp opinion, I know how to read other opinions also as they give me a sliver of information on games that I play. I was taught to seek knowledge. It's not that hard.
 
I think that [H]ardocp picks games that push video cards. I also believe that they would love to add in a few more into the testing suite. The problem is this:

AMD and Nvidia's PR teams are jerks. It's the truth. Here's why.

First they send cards out with a week or less to test them. It's like they want to limit the testing time with a NDA for some special day to launch them. All the consumer wants is a card in the store and a test to know if it is worthwhile. How they suckered the powers that be at these companies into making a special job title for them amazes me sometimes. [H]ardocp tests are extremely time consuming; you can't rush a play through. A week is not enough!

Then in the middle of the review AMD and Nvidia like to send over updated drivers. Strictly to f*ck with the reviewers I believe. You would think that they had a team of testers playing the latest games to notice that the game crashes when you hit jump or whatever. Nope! Just update the drivers and make the reviewers run every test again. So now instead of having 7 days to run tests you have 3. I wonder how much hair does Brent_Justice lose when they do this? I'm malicious when stuff like that happens to me. I would ding them a point if I were testing. :sneaky:

Then when the review comes out they want to argue the results because they are a PR team designed to cover up mistakes. Fury X doesn't overclock? Make up a reason why or not help Wizzard to update MSI Afterburrner to be able to OC the thing. If you can't use the program to OC it then it might still OC well! GTX 1080 can't hit 2100 consistently as shown on stage? Don't tell people that the consumer cards are voltage locked and hopefully everyone will not realize it until they have sold a boatload. Wood screws on stage. Dual Fury with an Intel CPU. The Cat and Mouse game goes on and on.

Look the truth is that people play a wide assortment of games. After I have the [H]ardocp opinion, I know how to read other opinions also as they give me a sliver of information on games that I play. I was taught to seek knowledge. It's not that hard.

No reason this time that cards are not issued out for testing since NDA is 29 June, so AMD maybe prudent to give HardOCP some (if they will sign NDA) if not Kyle can leak all month what others are finding out about Polaris :)
 
AMD's Roy Taylor Speaks - AMD, Roy Taylor, the Nano, and the Press

thats where it went downhill. I love this site and love what kyle and brent do and the effort they put in to reviews. The nano article was very unprofessional. I know H didn't get a card but to go out and put every email out there and to read too much in to a tweet and call out Roy Taylor on his email. The article wasn't begging for a nano, but it was too personal from a journalism stand point. I mean I understand why AMD doesn't want to deal with Hardocp, after that article just too much bad blood and they probably want to stay away.

Now you can call me out, I am no fanboy I am just being a critic here of an article. As I said I love this site and the in depth reviews but there were other sites that didn't get the nano but I don't remember such harsh article from anyone else. I mean nano was a good product just priced bad, I just really read this article for the second time today and boy it was pretty harsh. It was too personal and too many attacks in the article on AMD, they were responding via email and I think that was also unprofessional to just paste them in an article.

Just making sense of where the trust bridge was burned between the two and that was the article.

Kyle and Brent, I still have great respect for your work and the time you invest in us doing in depth reviews. But I must disagree on that article and say thats why relationship with AMD has gone south. May be its personal to you guys idk, but AMD or not I am still [H] at heart.
 
Last edited:
Why do we literally have 3 separate threads all posting essentially the exact same information?
 
AMD's Roy Taylor Speaks - AMD, Roy Taylor, the Nano, and the Press

thats where it went downhill. I love this site and love what kyle and brent do and the effort they put in to reviews. The nano article was very unprofessional. I know H didn't get a card but to go out and put every email out there and to read too much in to a tweet and call out Roy Taylor on his email. The article wasn't begging for a nano, but it was too personal from a journalism stand point. I mean I understand why AMD doesn't want to deal with Hardocp, after that article just too much bad blood and they probably want to stay away.

No it went down hill when AMD got pissed off about the Fury X and Fury reviews. Other websites like AT and THG posted more favorable reviews. But their results were not all that different from Kyles. But [H] called it the way they saw it when they questioned the value. [H] questioned the 4K claims. And they were right to do so. THG and AT made it look playable at 4K but they kind of downplayed the fact you really had to crank details down. And AMD didn't like that [H] pointed that out and called them on it. [H] also called them out on the lack of future proofing (4 gigs memory, and HDMI 1.3)

Fury X and Fury weren't bad cards. They were just too little, too late, for too much money for what they offered. End of story.

Do I consider the whole Roy Taylor fiasco news worthy? No. It really doesn't affect the end product. That's just bad PR, not engineering. But I consider internal instability news worthy because it shows where the company directs and treats it's R&D.
 
Last edited by a moderator:
No it went down hill when AMD got pissed off about the Fury X and Fury reviews. Other websites like AT and THG posted more favorable reviews. But their results were not all that different from Kyles. But [H] called it the way they saw it when they questioned the value. [H] questioned the 4K claims. And they were right to do so. THG and AT made it look playable at 4K but they kind of downplayed the fact you really had to crank details down. And AMD didn't like that [H] pointed that out and called them on it. [H] also called them out on the lack of future proofing (4 gigs memory, and HDMI 1.3)

Fury X and Fury weren't bad cards. They were just too little, too late, for too much money for what they offered. End of story.

Do I consider the whole Roy Taylor fiasco news worthy? No. It really doesn't affect the end product. That's just bad PR, not engineering. But I consider internal instability news worthy because it shows where the company directs and treats it's R&D.
internal stability is news worthy but I haven't heard it from any place else. I would have to see it through multiple sites to actually start believing such instability is actually there.
 
well tell ya this when a company is going down hill, there is always instability. Everyone will be looking for themselves, when I was consulting for Lionsgate, even though they were producing some good movies, people were like rats.....
 
  • Like
Reactions: NKD
like this
well tell ya this when a company is going down hill, there is always instability. Everyone will be looking for themselves, when I was consulting for Lionsgate, even though they were producing some good movies, people were like rats.....
Exactly my point. I haven't seen anyone or anyone else report or any of these rats speaking out lol. Shit would be all over the place.

But it should be more clear tonight. I think AMD will give out the specs tonight but reviews look like won't be published until end of month.
 
well nah they don't usually talk about it outside the company only after they leave they do. Internally though, lots of back stabbing, looking over ones shoulder and what not.
 
Back
Top