Reminder about RX 480 benchmarks

[neokhoth]

Read this post here. Drivers are not finalized. Even the benchmarks that are very possibly real are inaccurate until we get our AMD drivers. 

 

We could see performance improvements of 10% or more (not to mention stability like was NOT showcased in the GTA V benchmark). Don't let yourself down looking at some benchmarks that seem to be "disappointments."

 

 

[JamGorby]

Did this really need to be a new topic? 

[xAcid9]

50 minutes ago, neokhoth said:

Read this post here. Drivers are not finalized. Even the benchmarks that are very possibly real are inaccurate until we get our AMD drivers. 

 

We could see performance improvements of 10% or more (not to mention stability like was NOT showcased in the GTA V benchmark). Don't let yourself down looking at some benchmarks that seem to be "disappointments."

I take that post from a random guy on the internet with a truckload of salt. Even AMD them self hinting at 390s level of performance which most recent leaked somehow proved that.

[neokhoth]

44 minutes ago, JamGorby said:

Did this really need to be a new topic? 

Apparently so. 

[ivan134]

13 minutes ago, xAcid9 said:

I take that post from a random guy on the internet with a truckload of salt. Even AMD them self hinting at 390s level of performance which most recent leaked somehow proved that.

What hint? Please tell me you're not taking about Steam VR. 

[xAcid9]

1 hour ago, ivan134 said:

What hint? Please tell me you're not taking about Steam VR. 

Exactly that and AOTS performance.

[ivan134]

33 minutes ago, xAcid9 said:

Exactly that and AOTS performance.

You mean the test that is so reliable that an 860K + gtx 580 is considered VR ready?

 

[xAcid9]

5 minutes ago, ivan134 said:

You mean the test that is so reliable that an 860K + gtx 580 is considered VR ready?

Care to find any result on AMD card that is lower than 380 but show VR Ready?

What about AMD's AOTS number?

[ivan134]

3 minutes ago, xAcid9 said:

Care to find any result on AMD card that is lower than 380 but show VR Ready?

What about AMD's AOTS number?

I don't know what AotS number you're talking about except the xfire 480s they had at the reveal. I pointed out to you how inconsistent the VR test is and you're asking for other numbers of an unreliable test. An 860k + gtx 580 is apparently faster than RX 480 + whatever CPU they used and you're asking me for numbers on a test that makes no sense. Okay buddy.

[ivan134]

7 minutes ago, xAcid9 said:

Care to find any result on AMD card that is lower than 380 but show VR Ready?

What about AMD's AOTS number?

In case you missed it this morning.

Let me know if you can get a 390 to put out these numbers.

[xAcid9]

1 hour ago, ivan134 said:

I don't know what AotS number you're talking about except the xfire 480s they had at the reveal. I pointed out to you how inconsistent the VR test is and you're asking for other numbers of an unreliable test. An 860k + gtx 580 is apparently faster than RX 480 + whatever CPU they used and you're asking me for numbers on a test that makes no sense. Okay buddy.

I believed they ran it many times before they posted the result on an event. These inconsistent SteamVR result mostly happen on Nvidia card and mostly on GTX580.

I'm talking about AOTS benchmarks that AMD ran before Macau event. Which is slower than what i get with my 290x.

 

 

 

http://www.ashesofthesingularity.com/metaverse#/personas/b0db0294-8cab-4399-8815-f956a670b68f/match-details/824ce3e8-67a6-4b68-889f-21c426cdf031

1 hour ago, ivan134 said:

In case you missed it this morning.

Let me know if you can get a 390 to put out these numbers.

I only have GTA5 and Witcher 3 to compare. Can't find anything on the web that use that fucked up setting in GTA 5 and i can't compare my benchmark to his because my benchmark sequence is different from him. I believe he'll get better average and minimum than me because of that DDR4 2133mhz RAM and Skylake IPC.

 

For Witcher 3 i chose this screenshot because it's the only scene i can similarly replicate

 

At stock speed which just slightly slower than a 390 i get 54.9 which only 2+ fps lower with the same graphical settings. 

With slight overclock i could easily get 60fps.

[ivan134]

26 minutes ago, xAcid9 said:

 

You are using a 4 core 8 thread CPU while he's using a shitty 6400 and it's still faster than your 290x. A 390 can't do that.

[xAcid9]

3 hours ago, ivan134 said:

You are using a 4 core 8 thread CPU while he's using a shitty 6400 and it's still faster than your 290x. A 390 can't do that.

So you not going to comment about AOTS result i posted?  

 

His shitty i5-6400 perform similar to mine when gaming especially with AMD driver which can't utilize 8 threads.

[App4that]

Wait, are we supposed to ignore the leaks and wait for the real benchmarks to hit after the NDA is lifted? Or lose our shit?

 

Seems like the lost and found is full of shit in this topic. And salt. 

[Briggsy]

There are now three 1080p crazy scores for the 480. They are all fairly consistent at the 36 FPS mark, while the lowest 390 score (out of dozens) is 39FPS at 1080p crazy. For games that aren't built around AMD hardware, I would imagine driver improvements are incoming, but for Ashes... I think its probably the best case scenario at this point for performance speculation.

 

Result 1

Result 2

Result 3

[Big-G]

16 minutes ago, App4that said:

Wait, are we supposed to ignore the leaks and wait for the real benchmarks to hit after the NDA is lifted? Or lose our shit?

 

Seems like the lost and found is full of shit in this topic. And salt. 

Need's more tequila and lime!!!!

[ivan134]

1 hour ago, xAcid9 said:

So you not going to comment about AOTS result i posted?  

 

His shitty i5-6400 perform similar to mine when gaming especially with AMD driver which can't utilize 8 threads.

The 6400 is only on par with a 4460 so no, it's not the same as yours and since when is this a thing?

[xAcid9]

1 hour ago, ivan134 said:

The 6400 is only on par with a 4460 so no, it's not the same as yours and since when is this a thing?

and 4460 perform similar with my proc when gaming. Show me how many games will give me much higher FPS compare to 4460, you can use i7-3770 or even i7-4770 to prove your point.

 

It's a thing when AMD poor DX11 overhead exposed to the world. My CPU usage peaked just over 50% in some CPU intensive games(NFS 2016, AC:U, BF4, Dying Light, GTA5) with some GPU usage drop when my friends with 970 and 4770k able to utilized 60%-70% on that very same game.

 

 

And to your claim about my CPU give me extra FPS in Witcher 3

Spoiler

The game is GPU bound especially after recent patch 1.22 where CDProjekt optimized drawcall bundling.

 

2 hours ago, Briggsy said:

There are now three 1080p crazy scores for the 480. They are all fairly consistent at the 36 FPS mark, while the lowest 390 score (out of dozens) is 39FPS at 1080p crazy. For games that aren't built around AMD hardware, I would imagine driver improvements are incoming, but for Ashes... I think its probably the best case scenario at this point for performance speculation.

 

Result 1

Result 2

Result 3

Funny that i also get 36fps but in Crazy 1440p preset.

 

http://www.ashesofthesingularity.com/metaverse#/personas/02274b44-2b0e-4909-b186-b3657b420e5a/match-details/db1dbb15-a4ff-4af4-88e1-a5113dba7562

[Tech Inquisition]

11 hours ago, neokhoth said:

Read this post here. Drivers are not finalized. Even the benchmarks that are very possibly real are inaccurate until we get our AMD drivers. 

 

We could see performance improvements of 10% or more (not to mention stability like was NOT showcased in the GTA V benchmark). Don't let yourself down looking at some benchmarks that seem to be "disappointments."

 

 

Agreed in principle... anything that's not commercially available yet will be running on very new and possibly buggy drivers/software. That's why it's a good idea to most likely do a "follow-up" performance review on new tech say after 5 driver versions post-release. Unless the specific tech item becomes part of the day to day bench-marking regime...

[killcomic]

Now, don't forget that it could also drop in performance. It could be possible that because the drivers are not finalised, it's not rendering everything correctly and it's running artificially faster because some things are missing in the rendering process.

[Zaryabzww]

Basically ive lost all hope in the 480, i saw all the tests they were 100% legit, whether the performance will increase with driver updates is the only thing to see and wait for otherwise those benchmarks with high/ultra were VERY DISSAPOINTING. Since my gtx 970 gets better fps on all ultra no hairworks... pathetic why would i downgrade when upgrading 

[DXMember]

Reviewers got the drivers just yesterday, the NDA is over on the launch day,

I doubt that anyone that hasn't signed an NDA will get the drivers sooner than June 29th, so by then even if they have the card, they probably can't do much more than drive a screen or two off it with Windows generic VGA driver

so I'd say all the 'leaked' benchmarks are probably fabricated

[Briggsy]

1 hour ago, DXMember said:

Reviewers got the drivers just yesterday, the NDA is over on the launch day,

I doubt that anyone that hasn't signed an NDA will get the drivers sooner than June 29th, so by then even if they have the card, they probably can't do much more than drive a screen or two off it with Windows generic VGA driver

so I'd say all the 'leaked' benchmarks are probably fabricated

When the 29th arrives and the benchmarks start coming in with similar performance to the leaks, what then? Launch drivers were buggy? reviewers used the wrong driver? give it a few months for drivers to mature? AIB cards will be better? 

 

Polaris has been supported on official AMD drivers for some time, there's a strong chance that all the big game titles that people would normally benchmark are driver optimized already. We have the luxury right now of believing that current AMD drivers could be buggy with polaris, but so far we've seen fairly solid gpu utilization in firestrike results and from various youtubers playing games with their 480. If the gpu is getting 99-100% utilization, then the drivers are not buggy. There could be further optimization work, but they are certainly not buggy.

[DXMember]

36 minutes ago, Briggsy said:

When the 29th arrives and the benchmarks start coming in with similar performance to the leaks, what then? Launch drivers were buggy? reviewers used the wrong driver? give it a few months for drivers to mature? AIB cards will be better? 

 

Polaris has been supported on official AMD drivers for some time, there's a strong chance that all the big game titles that people would normally benchmark are driver optimized already. We have the luxury right now of believing that current AMD drivers could be buggy with polaris, but so far we've seen fairly solid gpu utilization in firestrike results and from various youtubers playing games with their 480. If the gpu is getting 99-100% utilization, then the drivers are not buggy. There could be further optimization work, but they are certainly not buggy.

I mean it's not hard to fabricate the results that are pretty accurate - we know the core count which is basically set in stone now and is +/- around Hawaii Pro core count and we have pretty accurate core clock estimation by now, so considering the architectural improvements we could guess that it'll perform like a Hawaii Pro card at around 1200~1300MHz and that probably wouldn't be too far from truth

 

what I'm saying is that I doubt that those people "leaking" actually are running the hardware they say they're running

[xAcid9]

2 hours ago, Zaryabzww said:

Basically ive lost all hope in the 480, i saw all the tests they were 100% legit, whether the performance will increase with driver updates is the only thing to see and wait for otherwise those benchmarks with high/ultra were VERY DISSAPOINTING. Since my gtx 970 gets better fps on all ultra no hairworks... pathetic why would i downgrade when upgrading 

The card perform like 390 series on release, question now is how well will it scale with overclock and how far can you push it.

My 290x get from below 970 at stock to over 980 with just 120mhz extra on the core so i hope Polaris 10 will scale just as well.

2 hours ago, DXMember said:

Reviewers got the drivers just yesterday, the NDA is over on the launch day,

I doubt that anyone that hasn't signed an NDA will get the drivers sooner than June 29th, so by then even if they have the card, they probably can't do much more than drive a screen or two off it with Windows generic VGA driver

so I'd say all the 'leaked' benchmarks are probably fabricated

They will get their press release driver so they can publish their reviews on 29th. This press release driver probably still buggy but i doubt it will increase the card performance significantly with official driver. Expect 12 months time frame for the driver to improve significantly just like 7900 series and 290 series.