Jump to content
Hero
Sign in to follow this  
Huniken

AMD RX 480 Killer of GTX 1080 @200$

Recommended Posts

I don't think it is going to "kill the 1080". First of all from what I've seen it's the 4GB version that's 200$ and 2nd of all they show a "benchmark" (if you can call it that) from AoS which is the only game/benchmark that favors AMD :P (that and the test was probably done with the 8GB version not the 4GB one).

It might be a great card for crossfire but we all know what a b***h crossfire and SLI are :P

So for me, I'm not that impressed, would love to see more than just the AoS "benchmark".

Will it dethrone the 1080? Probably not, maybe the 1070 though.

Share this post


Link to post
Share on other sites

Wait just a few months for the clouds to clear a little, let the non-reference cards materialise and then see what the real world gains are. 

At this point its all bluster and smoke screens. 

Share this post


Link to post
Share on other sites

So one update this when they 'know'!

 

i need a new GPU to run Star Citizen, anyone know what would be the best? 1070-80 or this new rx480? Or somthing different?

Personally probably gonna end up with a 1080 cause the 3rd party cards come sooner than the other 2.

The 1070 should probably be enough for SC at 1080p or even 1440p (I say probably since it's going to take a while before it gets properly optimized)

As for the RX480... I wouldn't take the chance, since enough games already have problems with Crossfire and SLI and you'd kinda have to run 2 RX480s for it.

Plus it will be a while before SC gets DX12 and Nvidia has proved it destroys AMD on DX11. Looking at the future though when it does have DX12, then AMD might be nice (don't know how much Nvidia will improve their DX12 support)

Having said all that, the RX480 definitely is the cheapest out  of all of them, so if you're short on cash, just go with that one.

Edited by Akanoes

Share this post


Link to post
Share on other sites

The RX480 is being marketed as an affordable step into VR.  In no way will it dethrone the 1080, in SLI is beats out the 1070 sure but as said above, ill wait for official benchmarks.

Share this post


Link to post
Share on other sites

1080 killer no way. MAYBE the 490 will be a 1070 killer. 

Polaris 10 was designed to deliver 970 and 980(ti) preference at half the cost. They are not even aiming for 1080 performance yet.

Honestly Nivida sounds like slime bags with and they gimp AMD cards by shoving gameworks into tons of AAA games. I hope AMD gives them a good run their money this time. Like I said in an earlier post here, what ever card run BF1 the best when it launches is the one I will get. The sad thing BF is an AMD sponsored game and the 980ti still beats out the few by 2 or 3 FPS. 

You take an Nvidia game and 980ti usually smashes fury by 25% because of dirty tricks nvidia puts in the game that is only optimized for Nvidia cards. 

Share this post


Link to post
Share on other sites

So if I had limited funds would 2 rx480s be a workable solution?

 

Also as a 3D designer I run a lot of cpu and GPU rendering, any more thoughts?

Edited by The_Director

Share this post


Link to post
Share on other sites

The rx480 is a great card and for the price it seems to pack a lot of value. Like the others have said this card isnt meant to kill the 1080 (that's Vega's job), its real opponent is the 900 series and any future budget cards nvidia releases. The 480's secondary purpose is to plug up the leak the 1070 would've caused Amd.

For half the price of a 1070 - or still 100-$150 less for 8gb - your getting up to 84% of the performance (5.5 tflops ÷ 6.5 tflops). What this card did was it made the entire maxwell series up to the 980 inferior and gave the 1070 (and the 980ti/Titan for that matter) a run for their money. For a budget card you cant ask for anything more.

To me I find the value persuading, whether or not you should buy one is up to you.

Edited by DeadlySniper007

Share this post


Link to post
Share on other sites

So if I had limited funds would 2 rx480s be a workable solution?

 

Also as a 3D designer I run a lot of cpu and GPU rendering, any more thoughts?

IMO I think SLI and Crossfire are for bench-marking and bragging rights. Too many games have bad support. I have no idea about 3D rendering programs, check out the programs. we still have 480x 490 and 490x and Vega to look forward to. Set a date when you absolutely want/need to upgrade and see whats available then. There are rumors Vega might be out in October. If thats true, that would be amazing and way before we see 1080ti or Titan GP100 

But I agree. AMD is bringing some SERIOUS value to the table. I was all set to blow my wallet on a 1080... but even the 1070 just offers so much better performance for value. Here comes AMD undercutting even that. AMD said from the start they wanted to try and cut into the largest section of the market, the mid-range. I read the 970s sold 70% of all 900 series cards. Its around that area of performance AMD is aiming for... I am going off topic. Bye 

Share this post


Link to post
Share on other sites

i need a new GPU to run Star Citizen, anyone know what would be the best? 1070-80 or this new rx480? Or somthing different?

I think we have pleeeeenty of time and people to figure out what Star Citizen will run the best on, GPU wise i may have an ideer (1080 TI) but for CPU ? CryEngine is a damn b*tch, my 5960X is getting (obviously) better score than my 4970K, but my AMD FX 9590 absolutely rekt the 4790K on Mechwarrior Online, but not on Crysis (?¿) if you add the fact that the Star Citizen Engine is (almost) not a CryEngine anymore and the game won't be optimized until (at least) SQ42 release plus some question like :

- DX12 "native" (full migration) or "mix" from DX11

- OpenGL ? Vulkan ? Direct3D ?

- We already know that the game will be x64 bits native, so Core > Clock speed or Clock Speed > Core ?

- What about Turbo boost 3.0 ?

 

Yep, that's a damn hell to figure out...

 

So if I had limited funds would 2 rx480s be a workable solution?

 

Also as a 3D designer I run a lot of cpu and GPU rendering, any more thoughts?

Like always, depends on budget, i don't think in a righteous mind in 3D Design or GPU rendering a (future) 1080 TI or a true GP104 with HBM2 (in my little book, PU will Hit live late 2017 - early 2018) will be outclassed by a Polaris, but Vega/Navi in the other hand...

 

whatever you do, don't order now, wait at least a good 6 months or even a year to have specific model (now whe have the stelears founders edition and i hope but don't have faith that builder will drop us good GPU at a non-founders price).

Share this post


Link to post
Share on other sites

So if I had limited funds would 2 rx480s be a workable solution?

 

Also as a 3D designer I run a lot of cpu and GPU rendering, any more thoughts?

 I have no idea about 3D rendering programs, check out the programs. we still have 480x 490 and 490x and Vega to look forward to. Set a date when you absolutely want/need to upgrade and see whats available then. There are rumors Vega might be out in October. If thats true, that would be amazing and way before we see 1080ti or Titan GP100 

Basically took the words right out of my mouth. If you can hold off till October or early 2017 at the latest I'd recommend waiting till then when Vega cards featuring HBM2 come out. It's the memory of the future and it would be ideal for 3d rendering. If you need a card right now though the 1080 will do perfectly

Share this post


Link to post
Share on other sites

 For m e great news In my country a 27" Vsync monitor is $350 more than a Freesync one. lets see the 490 and 490x price.

Edited by DiOlll

Share this post


Link to post
Share on other sites

I have a 1080 Founders. If this card even comes close to the performance of the 1080 its worth it. The 1080 is an absolute dream but i would like to point out Nvidia boasted lots of things to before the card came out. Till the card is out and people are running and testing it i dont believe anything the MFR says about a card. Get it in peoples hands first. The 1080 was supposed to be the solution to 4k gaming... It is not. You still need 2 to game at 4k effectively in most games. But, if this card performs close to the 1080 it will be a great value. The 1080 is the best purchase i have made when it comes to computer parts.

Share this post


Link to post
Share on other sites

Uhh.. No. The 480's will not come anywhere close to the 1080. AMD outright lied in their benchmark when they showed the 480CF vs 1080. Skip to 0:50 in this video to see the most obvious example. Left is the 2 480's, and right is the 1 1080:

http://www.pcgameshardware.de/AMD-Polaris-Hardware-261587/Videos/Ashes-of-the-Singularity-1197047/

Don't get your hopes up. 

 

Basically took the words right out of my mouth. If you can hold off till October or early 2017 at the latest I'd recommend waiting till then when Vega cards featuring HBM2 come out. It's the memory of the future and it would be ideal for 3d rendering. If you need a card right now though the 1080 will do perfectly

Just wait for the 1080TI and Pascal Titan. Both of those have HBM2. I wouldn't hold out hope for Vega. It is being rushed by 6 months, which was a decision made the day after the Nvidia showcase. Its going to underperform as well, since AMD is scrambling to not be left behind again like they were with Fury.

Share this post


Link to post
Share on other sites

Uhh.. No. The 480's will not come anywhere close to the 1080. AMD outright lied in their benchmark when they showed the 480CF vs 1080. Skip to 0:50 in this video to see the most obvious example. Left is the 2 480's, and right is the 1 1080:

http://www.pcgameshardware.de/AMD-Polaris-Hardware-261587/Videos/Ashes-of-the-Singularity-1197047/

Don't get your hopes up. 

Just wait for the 1080TI and Pascal Titan. Both of those have HBM2. I wouldn't hold out hope for Vega. It is being rushed by 6 months, which was a decision made the day after the Nvidia showcase. Its going to underperform as well, since AMD is scrambling to not be left behind again like they were with Fury.

How are they lying? Ashes of Singularity is the only game AMD has been using (exclusively) to boast that they are better optimized for DX12. You'll see countless benchmarks supporting that conclusion. When the game pool extends to say Rise of the Tomb Raider the disparity becomes much less noticeable.

Sure you could wait a year or more for Nvidia to release their flagship cards but not everyone can wait that long. Will they be better or worse who knows, but having extra time to work on a product is never a bad thing. The rest of what you said is all speculation and bias. Until the official release date is announced and people can actually compare the products for themselves, all of that conjecture is useless.

Share this post


Link to post
Share on other sites

How are they lying? Ashes of Singularity is the only game AMD has been using (exclusively) to boast that they are better optimized for DX12. You'll see countless benchmarks supporting that conclusion. When the game pool extends to say Rise of the Tomb Raider the disparity becomes much less noticeable.

Sure you could wait a year or more for Nvidia to release their flagship cards but not everyone can wait that long. Will they be better or worse who knows, but having extra time to work on a product is never a bad thing. The rest of what you said is all speculation and bias. Until the official release date is announced and people can actually compare the products for themselves, all of that conjecture is useless.

Sorry about the late reply, it didn't go to my email like they usually do.

They are lying about the 480 performance specifically.Watch the benchmark video I posted, which is directly from AMD. If you don't want to watch the whole thing to see all the differences, just skip to 50 seconds into the video. The left screen is 2 480's in crossfire, and the right is a single 1080. You can see clearly that the 480's have lower graphic settings, less units on-screen, and less effects to render. That is a lie right from AMD's presentation. They were probably hoping nobody would spot it, which is ironic since he was talking about how people can see the difference between 30 and 60 fps in games. 

Nvidia flagship cards have never come out a year after the other releases. At max, its 6 months. They generally wait until Christmas time and release them, since kids will start asking for them. You can guarantee they will be better than what is currently offered. They always are. 

The release date for Vega was confirmed in Q1 2017. Suddenly Nvidia has a presentation showing this cool tech, and they move it to October 2016 the day after? Lets be real here. Fury missed the boat on the last cycle, then under-delivered massively even after drivers were worked out. AMD doesn't want that to happen again. (2017 release confirmation: http://wccftech.com/amd-unveils-polaris-vega-navi-graphics-architectures-hbm2/) (2016 6-month push: http://www.game-debate.com/news/20146/amd-has-reportedly-pushed-vega-gpu-launch-forward-to-october-2016) 

Nothing in the Polaris line will be able to directly compete with the 1080 1-to-1, which is why the only comparison was two of the top polaris cards vs the 1080, and they still had to fudge the benches to get the performance they wanted.

The only actual conjecture I made was on Vega's performance. Nobody can know that yet. However you can use some deductive reasoning to come to my conclusion. Do you know what AMD did again this year? They cut their R&D budget further, and gave more raises to their execs. They also let their exclusivity with Hynix run out with nothing special to show for it. No new tech, and their new cards don't have the new memory. Meanwhile, Nvidia has already released the P100 with HBM2 for datacenters and supercomputers, and they sold out immediately over AMD's "equivalent", the Pro Duo, which only has HBM1. The P100, a single chip GPU, utterly destroys the Pro Duo, a dual chip GPU, in their respective applications. Now considering the flagship gaming cards from both companies are built off of the architecture from these professional cards, what exactly do you think the outcome will be? Keep in mind that Vega's chip (Greenland) is identical to the Fiji chipset, but in the 14nm size. Pascal is 16nm, and we've already proven that die shrinks that close are not enough to push performance over the edge without other major modifications, none of which AMD has talked about. AMD is always very, very forward with whatever new stuff they have, since their entire Desktop GPU profit is driven by hype instead of performance, which is why they haven't made a profit on Desktop GPU's for over 5 years. 

I could keep ranting, but I think I've gone far enough. Vega will probably fall flat on its face. Fiji->Greenland is just an increment, and not a leap like Maxwell->Pascal. AMD is very bad at leaps.

Share this post


Link to post
Share on other sites

Sorry about the late reply, it didn't go to my email like they usually do.

They are lying about the 480 performance specifically.Watch the benchmark video I posted, which is directly from AMD. If you don't want to watch the whole thing to see all the differences, just skip to 50 seconds into the video. The left screen is 2 480's in crossfire, and the right is a single 1080. You can see clearly that the 480's have lower graphic settings, less units on-screen, and less effects to render. That is a lie right from AMD's presentation. They were probably hoping nobody would spot it, which is ironic since he was talking about how people can see the difference between 30 and 60 fps in games. 

Nvidia flagship cards have never come out a year after the other releases. At max, its 6 months. They generally wait until Christmas time and release them, since kids will start asking for them. You can guarantee they will be better than what is currently offered. They always are. 

The release date for Vega was confirmed in Q1 2017. Suddenly Nvidia has a presentation showing this cool tech, and they move it to October 2016 the day after? Lets be real here. Fury missed the boat on the last cycle, then under-delivered massively even after drivers were worked out. AMD doesn't want that to happen again. (2017 release confirmation: http://wccftech.com/amd-unveils-polaris-vega-navi-graphics-architectures-hbm2/) (2016 6-month push: http://www.game-debate.com/news/20146/amd-has-reportedly-pushed-vega-gpu-launch-forward-to-october-2016) 

Nothing in the Polaris line will be able to directly compete with the 1080 1-to-1, which is why the only comparison was two of the top polaris cards vs the 1080, and they still had to fudge the benches to get the performance they wanted.

The only actual conjecture I made was on Vega's performance. Nobody can know that yet. However you can use some deductive reasoning to come to my conclusion. Do you know what AMD did again this year? They cut their R&D budget further, and gave more raises to their execs. They also let their exclusivity with Hynix run out with nothing special to show for it. No new tech, and their new cards don't have the new memory. Meanwhile, Nvidia has already released the P100 with HBM2 for datacenters and supercomputers, and they sold out immediately over AMD's "equivalent", the Pro Duo, which only has HBM1. The P100, a single chip GPU, utterly destroys the Pro Duo, a dual chip GPU, in their respective applications. Now considering the flagship gaming cards from both companies are built off of the architecture from these professional cards, what exactly do you think the outcome will be? Keep in mind that Vega's chip (Greenland) is identical to the Fiji chipset, but in the 14nm size. Pascal is 16nm, and we've already proven that die shrinks that close are not enough to push performance over the edge without other major modifications, none of which AMD has talked about. AMD is always very, very forward with whatever new stuff they have, since their entire Desktop GPU profit is driven by hype instead of performance, which is why they haven't made a profit on Desktop GPU's for over 5 years. 

I could keep ranting, but I think I've gone far enough. Vega will probably fall flat on its face. Fiji->Greenland is just an increment, and not a leap like Maxwell->Pascal. AMD is very bad at leaps.

Little info here.

AMD Responds to Recent Ashes of the Singularity Benchmark Controversy

AMD responds to the controversy surrounding its Ashes of the Singularity benchmarks between 2XRX480 & GTX1080

 

I’m sure that most of you have already seen AMD’s Computex presentation by now, especially the part where two RX480 graphics cards were compared to one NVIDIA GTX 1080 using two Ashes of the Singularity benchmarks. The benchmarks saw AMD’s Crossfire solution coming ahead of Nvidia’s GTX 1080, which is definitely interesting since two of these AMD cards will reportedly cost less than one GTX 1080. However, the community noticed that the two side-by-side benchmarks were not exactly identical when it came to graphics quality, which is why many have speculated that the one for the RX480s was actually running at lower settings when compared to the other.

Not too long ago, AMD’s Technical Marketing Lead, Robert Hallock has decided to address this controversy by claiming that NVIDIA’s GPUs do not render Ashes of the Singularity properly, which might explain the visual differences spotted during the event.

“Ashes uses procedural generation based on a randomized seed at launch. The benchmark does look slightly different every time it is run. But that, many have noted, does not fully explain the quality difference people noticed.

At present the GTX 1080 is incorrectly executing the terrain shaders responsible for populating the environment with the appropriate amount of snow. The GTX 1080 is doing less work to render AOTS than it otherwise would if the shader were being run properly. Snow is somewhat flat and boring in color compared to shiny rocks, which gives the illusion that less is being rendered, but this is an incorrect interpretation of how the terrain shaders are functioning in this title.

The content being rendered by the RX 480–the one with greater snow coverage in the side-by-side (the left in these images)–is the correct execution of the terrain shaders.

So, even with fudgy image quality on the GTX 1080 that could improve their performance a few percent, dual RX 480 still came out ahead.”

Apparently, AMD ran the test ten times before presenting it to the audience in order to make sure that the figures would check out. As for the system that was used for the benchmarks, Robert also went ahead and revealed its exact specifications, as well as the precise results of the testing sessions.

System specs:

CPU: i7 5930K.
RAM: 32GB DDR4-2400Mhz.
Motherboard: Asrock X99M Killer.
GPU config 1: 2x Radeon RX 480 @ PCIE 3.0 x16 for each GPU.
GPU config 2: Founders Edition GTX 1080.
OS: Win 10 64bit.
AMD Driver: 16.30-160525n-230356E.
NV Driver: 368.19.

In Game Settings for both configs: Crazy Settings | 1080P | 8x MSAA | VSYNC OFF

Ashes Game Version: v1.12.19928

Benchmark results:

2x Radeon RX 480 – 62.5 fps | Single Batch GPU Util: 51% | Med Batch GPU Util: 71.9 | Heavy Batch GPU Util: 92.3%
GTX 1080 – 58.7 fps | Single Batch GPU Util: 98.7%| Med Batch GPU Util: 97.9% | Heavy Batch GPU Util: 98.7%.

Share this post


Link to post
Share on other sites

Little info here.

AMD Responds to Recent Ashes of the Singularity Benchmark Controversy

AMD responds to the controversy surrounding its Ashes of the Singularity benchmarks between 2XRX480 & GTX1080

 

I’m sure that most of you have already seen AMD’s Computex presentation by now, especially the part where two RX480 graphics cards were compared to one NVIDIA GTX 1080 using two Ashes of the Singularity benchmarks. The benchmarks saw AMD’s Crossfire solution coming ahead of Nvidia’s GTX 1080, which is definitely interesting since two of these AMD cards will reportedly cost less than one GTX 1080. However, the community noticed that the two side-by-side benchmarks were not exactly identical when it came to graphics quality, which is why many have speculated that the one for the RX480s was actually running at lower settings when compared to the other.

Not too long ago, AMD’s Technical Marketing Lead, Robert Hallock has decided to address this controversy by claiming that NVIDIA’s GPUs do not render Ashes of the Singularity properly, which might explain the visual differences spotted during the event.

“Ashes uses procedural generation based on a randomized seed at launch. The benchmark does look slightly different every time it is run. But that, many have noted, does not fully explain the quality difference people noticed.

At present the GTX 1080 is incorrectly executing the terrain shaders responsible for populating the environment with the appropriate amount of snow. The GTX 1080 is doing less work to render AOTS than it otherwise would if the shader were being run properly. Snow is somewhat flat and boring in color compared to shiny rocks, which gives the illusion that less is being rendered, but this is an incorrect interpretation of how the terrain shaders are functioning in this title.

The content being rendered by the RX 480–the one with greater snow coverage in the side-by-side (the left in these images)–is the correct execution of the terrain shaders.

So, even with fudgy image quality on the GTX 1080 that could improve their performance a few percent, dual RX 480 still came out ahead.”

Apparently, AMD ran the test ten times before presenting it to the audience in order to make sure that the figures would check out. As for the system that was used for the benchmarks, Robert also went ahead and revealed its exact specifications, as well as the precise results of the testing sessions.

System specs:

CPU: i7 5930K.
RAM: 32GB DDR4-2400Mhz.
Motherboard: Asrock X99M Killer.
GPU config 1: 2x Radeon RX 480 @ PCIE 3.0 x16 for each GPU.
GPU config 2: Founders Edition GTX 1080.
OS: Win 10 64bit.
AMD Driver: 16.30-160525n-230356E.
NV Driver: 368.19.

In Game Settings for both configs: Crazy Settings | 1080P | 8x MSAA | VSYNC OFF

Ashes Game Version: v1.12.19928

Benchmark results:

2x Radeon RX 480 – 62.5 fps | Single Batch GPU Util: 51% | Med Batch GPU Util: 71.9 | Heavy Batch GPU Util: 92.3%
GTX 1080 – 58.7 fps | Single Batch GPU Util: 98.7%| Med Batch GPU Util: 97.9% | Heavy Batch GPU Util: 98.7%.

Ok, assuming some of this is true, and we know some of it is not, then it still doesn't get better for AMD. These reasons they give still show they are being shady, and trying to exploit certain circumstances to make themselves look good. They did this with the 900 series, where they got caught using very, very old drivers to make the 980 have terrible FPS. I'm not surprised they are doing it again. 

1: I own Ashes, and I can tell you that the terrain is not procedurally generated. I've played the snow maps many times, and they look the same every time. Graphics are turned up all the way for me. Google the features of the game. Absolutely nothing about procedural generation comes up until AMD messed up their benchmark. Its complete BS. Now, if it was actually Proc-Gen, then this game is unviable as a benchmark anyway. Benchmark programs are supposed to be exactly the same every time they are run, to establish a base. So its either AMD lying, or AMD being bad at their job. Both are possible.

2: They say the 1080 is incorrectly rendering shaders. That's funny, because 1) the 1080 side looked better. Probably because it had a higher quality. 2) Again, the snow isn't dynamic in the game. It doesn't change. 3) Are they really saying that the 1080 didn't render the shader properly, so it got a higher quality visual from it? Are they stupid? What AMD is implying here, is that they went into the game code and disabled some of the shaders to get this result. That is BS. Especially since their FPS result is below what every other 1080 AOTS benchmark did, which put the 1080 at 70+ in 1080p using the correct drivers.

3: Guess what? Nvidia driver 368.19 has no support for the 1080, so the entire bench is invalidated anyway. The driver that supports the 1080 is 368.39. So again, AMD is caught using older drivers so they get the result they wanted. 368.19 has no support for Nvidia's Asynch compute, or any of the new tech that helps the 1080 work.

4: If they ran a faulty test 10 times, then its still a faulty test. I noticed they said nothing about the reduced unit counts, or the reduced amount of effects, or the lower quality of the graphics overall for the 480's. Let me get my surprised face out. 

So even with this response, AMD is still bullshitting very, very hard. They do it every year. They try to get away with something, and get caught for it. Last year it was using drivers that didn't support the card or the game they were benching, this years its using an AMD optimized game with out of date drivers and different graphics settings.

Share this post


Link to post
Share on other sites

Ok, assuming some of this is true, and we know some of it is not, then it still doesn't get better for AMD. These reasons they give still show they are being shady, and trying to exploit certain circumstances to make themselves look good. They did this with the 900 series, where they got caught using very, very old drivers to make the 980 have terrible FPS. I'm not surprised they are doing it again. 

1: I own Ashes, and I can tell you that the terrain is not procedurally generated. I've played the snow maps many times, and they look the same every time. Graphics are turned up all the way for me. Google the features of the game. Absolutely nothing about procedural generation comes up until AMD messed up their benchmark. Its complete BS. Now, if it was actually Proc-Gen, then this game is unviable as a benchmark anyway. Benchmark programs are supposed to be exactly the same every time they are run, to establish a base. So its either AMD lying, or AMD being bad at their job. Both are possible.

2: They say the 1080 is incorrectly rendering shaders. That's funny, because 1) the 1080 side looked better. Probably because it had a higher quality. 2) Again, the snow isn't dynamic in the game. It doesn't change. 3) Are they really saying that the 1080 didn't render the shader properly, so it got a higher quality visual from it? Are they stupid? What AMD is implying here, is that they went into the game code and disabled some of the shaders to get this result. That is BS. Especially since their FPS result is below what every other 1080 AOTS benchmark did, which put the 1080 at 70+ in 1080p using the correct drivers.

3: Guess what? Nvidia driver 368.19 has no support for the 1080, so the entire bench is invalidated anyway. The driver that supports the 1080 is 368.39. So again, AMD is caught using older drivers so they get the result they wanted. 368.19 has no support for Nvidia's Asynch compute, or any of the new tech that helps the 1080 work.

4: If they ran a faulty test 10 times, then its still a faulty test. I noticed they said nothing about the reduced unit counts, or the reduced amount of effects, or the lower quality of the graphics overall for the 480's. Let me get my surprised face out. 

So even with this response, AMD is still bullshitting very, very hard. They do it every year. They try to get away with something, and get caught for it. Last year it was using drivers that didn't support the card or the game they were benching, this years its using an AMD optimized game with out of date drivers and different graphics settings.

Sorry for the late reply as well, have been away from the forums for a few days

Look I'm not arguing for either side, just trying to bring the facts to the table. 

This is just a link that goes even deeper than Yodel's information, but generally the same thing. http://wccftech.com/article/rx-480-gtx-1080-ashes-singularity-details/

1. Doing a quick google search proves that statement false. As you can tell both are coming straight from the developer and both are written within a year of AMD's presentation (June 1st). From what I gather the maps they make now are procedurally generated at the beginning and then that seed is used as the map. The technology they're hoping for post-release is what you are talking about, which is that the map changes everytime you play it. At that point I'd have to agree with you, benchmarks need to analyze the same thing and if the graphics cards aren't measuring the same thing than it is worthless. Now in this demo not everything was equal; the missing snow shader, more units on one side or in different places, etc. So in that regard the benchmark is too randomized for my liking. 

First link: http://forums.ashesofthesingularity.com/469449

Early on, we’re just manually making these maps (they’re procedurally generated via our in-game editor).  This way, we can nail down balance and find out what kinds of maps people like.  Once we have a better idea of what is “fun” with these maps we then move into having more maps be created via algorithm so that you can have potentially random maps."

2nd link: https://steamcommunity.com/app/228880/discussions/1/490123938428206916/

Draginol  [developer] Oct 23, 2015 @ 8:59am 
You ready for something really crazy?

Ashes has the holy grail for RTS map makers: Those maps your'e playing? They're procedurally generated. No artist ever worked on those maps. We use a very simple map editor that then generates a fairly realistic looking map (including simulating erosion).

This means we one day hope to provide random map generation. The only limitation now is that generating a map requires 64GB of memory. We'll have to do a lot of work on the editor before it can be made available to the public but we plan to release it post-release.

2. While I'd have to agree that the missing shader does make the game contrast better, the fact remains that the snow was actually meant to look like that.

3. They used the press driver version 368.19 instead of the public driver available for the 1080 at the time 368.25 which would've fixed the issue. 368.39 is actually the 1070 driver. So again poor decision by amd by ignoring the latest drivers and using only one benchmark to prove they can beat/keep up with a 1080.

Conclusion: In the end AMD ran a somewhat faulty benchmark in which the 1080 wasn't given its latest driver, hence what caused the shader glitch. Given that whoever had to prepare this presentation valued their career over procrastination, it would explain why the results would be a week or 2 old. So while they performed the test 10x at the event it was most likely that they stuck with the press driver or that they saw little to no performance gain from the latest driver enough to change their presentation last minute. 

Below is a tweet from the founder of the game, which if you click on it and actually go look at the conversation you'll see that the Senior Product Manager from Nvidia actually got involved in the conversation and only questioned why they used the press driver and how they got it. Then later went on to confirm a .6% performance difference between the press version and 368.39 ran on a 1070.

Edited by DeadlySniper007

Share this post


Link to post
Share on other sites

Sorry about the late reply, it didn't go to my email like they usually do.

They are lying about the 480 performance specifically.Watch the benchmark video I posted, which is directly from AMD. If you don't want to watch the whole thing to see all the differences, just skip to 50 seconds into the video. The left screen is 2 480's in crossfire, and the right is a single 1080. You can see clearly that the 480's have lower graphic settings, less units on-screen, and less effects to render. That is a lie right from AMD's presentation. They were probably hoping nobody would spot it, which is ironic since he was talking about how people can see the difference between 30 and 60 fps in games. 

Nvidia flagship cards have never come out a year after the other releases. At max, its 6 months. They generally wait until Christmas time and release them, since kids will start asking for them. You can guarantee they will be better than what is currently offered. They always are. 

The release date for Vega was confirmed in Q1 2017. Suddenly Nvidia has a presentation showing this cool tech, and they move it to October 2016 the day after? Lets be real here. Fury missed the boat on the last cycle, then under-delivered massively even after drivers were worked out. AMD doesn't want that to happen again. (2017 release confirmation: http://wccftech.com/amd-unveils-polaris-vega-navi-graphics-architectures-hbm2/) (2016 6-month push: http://www.game-debate.com/news/20146/amd-has-reportedly-pushed-vega-gpu-launch-forward-to-october-2016) 

Nothing in the Polaris line will be able to directly compete with the 1080 1-to-1, which is why the only comparison was two of the top polaris cards vs the 1080, and they still had to fudge the benches to get the performance they wanted.

The only actual conjecture I made was on Vega's performance. Nobody can know that yet. However you can use some deductive reasoning to come to my conclusion. Do you know what AMD did again this year? They cut their R&D budget further, and gave more raises to their execs. They also let their exclusivity with Hynix run out with nothing special to show for it. No new tech, and their new cards don't have the new memory. Meanwhile, Nvidia has already released the P100 with HBM2 for datacenters and supercomputers, and they sold out immediately over AMD's "equivalent", the Pro Duo, which only has HBM1. The P100, a single chip GPU, utterly destroys the Pro Duo, a dual chip GPU, in their respective applications. Now considering the flagship gaming cards from both companies are built off of the architecture from these professional cards, what exactly do you think the outcome will be? Keep in mind that Vega's chip (Greenland) is identical to the Fiji chipset, but in the 14nm size. Pascal is 16nm, and we've already proven that die shrinks that close are not enough to push performance over the edge without other major modifications, none of which AMD has talked about. AMD is always very, very forward with whatever new stuff they have, since their entire Desktop GPU profit is driven by hype instead of performance, which is why they haven't made a profit on Desktop GPU's for over 5 years. 

I could keep ranting, but I think I've gone far enough. Vega will probably fall flat on its face. Fiji->Greenland is just an increment, and not a leap like Maxwell->Pascal. AMD is very bad at leaps.

I watched the entire presentation live and immediately noticed the difference between the two. My problem was that it looked like it was running 2 different matches pretending to be the same match. Units weren't missing they were just in different places, effects timed differently, Color/Sharpness difference. 

Nvidia's releases haven't always been cut and dry 1 generation they may release the flagship first then the budget/mainstream cards next (700 series) and the next generation it will be mainstream - budget - flagship - budget (950). I can give you one example where there was a one year difference between cards: Going from Titan to Titan Black. Now of course the Titan was already released first at launch but still. Generally you are correct but a safer bet would be 8 months max.

Emphasis on "Reportedly" - information that is not at all official. Which makes that release information conjecture as well. 

True their bonuses to their execs were more generous than they can afford,  but if you've been paying attention to AMD you'd know that they have fairly high turnover when it comes to their execs/leads due to Nvidia among others snatching them with what I presume more attractive offers than what AMD could provide them. So it could easily be a retention plan, but selfish business practices arent out of the question. Their R&D budget was already at risk of being cut. They're losing money they had to cut expenses. And the numbers aren't all that different especially when one controls 80% of the market and the other controls 20%, when one makes hundreds of millions in profit and the other is in the red.

As for their exclusive deal with Hynix it was for HBM1 only and not HBM2, secondly they had supply problems with Hynix and later switched to Samsung, third comparing a 4gb card to a 12gb or even a 6gb card and calling it professional is not ideal, even though it kept up fairly well with the 980ti. Comparing the performance of a 16gb HBM2 Professional Pascal card that hasn't even been released yet to a 4gb HBM1 card and then calling it "their equivalent" is absolutely hysterical. If your going to compare the Tesla P100 at least compare it to its actual equivalent the Greenland GPU which could have up to 18 billion transistors (as opposed to the 15.3 billion of the P100) and will supposedly release before the Tesla. As for the GPU Architecture Fiji is closer to Polaris not Vega.

http://wccftech.com/amd-vega-10-gpu-4096-stream-processors/

Edit: Seen this pop up a couple weeks ago (June 20th) but I've been a little lazy posting it. Nvidia confirms a Q4 '16 release for the P100 PCIe version. 

http://www.anandtech.com/show/10433/nvidia-announces-pci-express-tesla-p100

Edited by DeadlySniper007
Updating information/moving a quoted conversation to the correct post

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

×
×
  • Create New...