AMD Catalyst 14.1 delayed due to a last minute bug

[SeekaX]

There "may" still be a chance of the driver coming this weekend.

Just seen an article on Guru3D and AMD have released a not quite finished version of the driver to them.

Here are some early results that they got

AMD a10 7850k (r9-290x) mantle same scores in 720p and 1080p and both lower than dx11 on medium settings?
typo?

 

[SieB]

AMD a10 7850k (r9-290x) mantle same scores in 720p and 1080p and both lower than dx11 on medium settings?
typo?

Not sure, there could be some explanation to it though.

 

[RadeonHDx]

But if you also look at the 2 DX11 scores for 720p and 1080p, there is only 2FPS drop between the 2?

 

[SieB]

I don't know how credible Golem.de is, maybe some Germans on here may know?
Anyway, they have done some test on the MP on 64 man Siege of Shanghai using a 3770k and a 290x and these are the results.

Thing to note this is during CPU intensive situations where Mantle does it's thing and improves performance. Take it with a pinch of salt though because I don't know if this site is credible, but seen as AMD gave them the drivers to play with then they must at least be well known.

 

[SeekaX]

But if you also look at the 2 DX11 scores for 720p and 1080p, there is only 2FPS drop between the 2?

that isn't that weird. the 290x on medium settings probably gets bottlenecked by the CPU in both scenarios, hence the difference isn't that big.

 

[WYP]

Hmm, if data is being release by the press then the previously stated 24 hour press exclusive period must have started .

We should have it in about a day lads!!!

 

[dladdin]

hurry i need you mantle !!!!!

 

[WYP]

There "may" still be a chance of the driver coming this weekend.

Just seen an article on Guru3D and AMD have released a not quite finished version of the driver to them.

Here are some early results that they got

The data for the 7850K and a 290x results are switched around in your pic, here is the results here.

 

[Zoot]

As usual 'techpowerup' are a bunch of dumbasses.

Indeed. There's a whole lot of misinformation going around.

AMD just can't seem to catch a break.

 

[barnsley]

Indeed. There's a whole lot of misinformation going around.

AMD just can't seem to catch a break.

It's always the case, not surprised really. Pretty sure some sites out there carry a bias.

 

[NeverBackDown]

So in the case of BF4. The only real benefits so far are for CPUs. Hopefully they can get it to work for the GPUs soon as well. Lower end cards would be able to perform better and lower end PCs would be needed which could help the PC market gain some more traction it needs.

 

[Zoot]

It's always the case, not surprised really. Pretty sure some sites out there carry a bias.

Yup, that's a rant for another day though.

To be fair though, I quite like TechPowerUp, for the reason that their reviews are very heavy on plots/results while keeping opinions to a minimum. If only more sites took that approach.

 

[barnsley]

Yup, that's a rant for another day though.

To be fair though, I quite like TechPowerUp, for the reason that their reviews are very heavy on plots/results while keeping opinions to a minimum. If only more sites took that approach.

Yeah, they're quite good although they do occasionally end up saying some mad things that have no basis.

 

[Jinxxter]

Ladies and gentlemen the driver has been released...
And I'm stuck at work until 5:30pm...

 

[Excalabur50]

Had to go back to 13.30 drivers due to BF4 crashing with 14.1 so guess I have to wait for 14.2 or something

 

[SieB]

Had to go back to 13.30 drivers due to BF4 crashing with 14.1 so guess I have to wait for 14.2 or something

Use this to completely remove any traces of the old driver, then install the 14.1 driver. I was having the same problem till I did this.

http://www.guru3d.com/files_details/display_driver_uninstaller_download.html

 

[WYP]

I will be back in Belfast around 5, I'm looking forward to see how mantle works

 

[SieB]

Here's what I got so far.

I used the same sequence for the SP as I did in my 290 FPS bench test sequence.

​

For the MP I played two rounds of 64 man Siege of Shanghai, one round running on DX11 and one round on running on Mantle.
No matter what you do in the MP though you will never be able to get an exact reading because there are to many variables and situations which effect performance.

During the two rounds I played I stuck to the same areas, taking the same routes and tried to keep engagements around the same each time. Luckily, people like to camp around the same areas on Siege of Shanghai so keeping engagements around the same wasn't to hard to do. Also, during both rounds the skyscraper was taken down at the beginning of the round.

One difference between DX11 and Mantle that I noticed straight off the bat is that Mantle is extremely smoother than DX11, both in the MP and SP. This isn't because of any FPS gains either, because the FPS gains aren't big enough where one would be smooth and the other wouldn't.

Considering this is the first iteration of Mantle, and given the fact that it is still in the Beta stages, then it can only improve. When drivers mature, developers find better ways of using it and games are built with it from the ground up not patched in like with BF4, who knows what performance gains we will be seeing. But so far things are looking good.

 

[cooperman]

^^^ i used to get huge frame drops on siege of shanghai (MP) when i start running through the skyscrapers at the spawn point (down to 35-40fps at times) now i dont have any drops in that area running the mantle update now its a map i can play and not avoid.

 

[SPS]

For the MP I played two rounds of 64 man Siege of Shanghai, one round running on DX11 and one round on running on Mantle.
No matter what you do in the MP though you will never be able to get an exact reading because there are to many variables and situations which effect performance.

I did however read somewhere that due to MP being more CPU intensive that it would show a higher increase in performance, like you said we can't use this as solid results because the test isn't deterministic but it would be interesting to throw a few numbers out there for a comparison.