[WCCFtech] AMD and NVIDIA DX12 big picture mode

[lilltesaito]

Old and new data was included in the graphs. But thanks.

I can't seem to find the 1440 and 4k new data, mind telling me where I can find that please.

 

[piesquared]

How is Venus these days?

Sounds like a personal attack. I suggest you stop or get out of this thread NOW.

 

[Silverforce11]

Old and new data was included in the graphs. But thanks.

Old and new *drivers* were included in the graphs.

The *benchmark* (Ashes) has been updated (by Oxide) since then to give all GPUs an uplift in performance.

An accurate comparison is made when we compare the old data and new one. The result?

Same order at 1080p. Missing is the 1440 and 4K results which shows AMD leading and NV GPUs tanking in DX12 vs DX11.

 

[littleg]

Impressive increases from both companies to be fair. 55 to 72 fps is quite an achievement in such a short space of time.

 

[Silverforce11]

Impressive increases from both companies to be fair. 55 to 72 fps is quite an achievement in such a short space of time.

Oxide updated the benchmark, all GPUs performed better.

Compare the results for the *older* drivers:

980Ti: 65.3 fps
Fury X:67.8 fps

To the original benchmark (first release of Ashes bench):

980Ti: 55.4 fps
Fury X:55.3 fps

The recent drivers add 6-11% on top of what Oxide has optimized.

980Ti: 72.5 fps
Fury X:72.1 fps

AMD improved performance in Ashes with 15.8 and 15.10, it was a two step process.

Ultimately now it still matches the original result. It's ~tie at 1080p.

It's a shame Computerbase.de didn't test 1440 and 4K.

It seems some people prematurely got over-excited about NV's 1080p result, growing the lead from 0.1 FPS to 0.4 FPS... they forgot about the performance at higher resolutions.

 

[LTC8K6]

Unplayable is unplayable. Does it really matter who is the fastest at unplayable in DX12?

When we get to playable, it looks like a tie?

 

[Silverforce11]

Note the 1440/4K data were from the old version of Ashes bench, the newer version improved performance by around 15-20% across the board.

I wouldn't call 50 fps unplayable for an RTS. If the 15-20% gains are seen at 1440p, it would be ~60fps and at 4K ~50fps.

 

[Azix]

Computerbase.de was lame for doing the article like that. Showing the only resolution that favors nvidia made people think things changed. They also went ahead and compared gains to make it seem nvidia could overcome the problems with drivers. Tin foil

BTW for those saying its not playable, this is an actual game people are playing (backers). Hits early access steam soon too. Still beta, but playable.

http://www.pcgamer.com/ashes-of-the-singularity-hitting-early-access-next-week/

if there is a benchmark more people should be showing results.

 

[Keysplayr]

Sounds like a personal attack. I suggest you stop or get out of this thread NOW.

It was meant to show that what is "clear as day" to one person, may not be so clear and a bit muddy to others.

Not a personal attack. Did not think I had to explain this, but whatever.

 

[Keysplayr]

Old and new *drivers* were included in the graphs.

The *benchmark* (Ashes) has been updated (by Oxide) since then to give all GPUs an uplift in performance.

An accurate comparison is made when we compare the old data and new one. The result?

Same order at 1080p. Missing is the 1440 and 4K results which shows AMD leading and NV GPUs tanking in DX12 vs DX11.

Old and new drivers is considered old and new data. Why wouldn't it be?
The data you get from old drivers is performance you'd get in the "past". Older. What's the deal?

 

[Keysplayr]

I can't seem to find the 1440 and 4k new data, mind telling me where I can find that please.

I guess search?

 

[Silverforce11]

Old and new drivers is considered old and new data. Why wouldn't it be?
Please stop the symantecs man, it is sooo counterproductive in a discussion.

Because the benchmark is no longer the same. It's been updated since.

Compare to the old data, Fury X still matches 980Ti at 1080p. It doesn't fit your narrative.

 

[Keysplayr]

Because the benchmark is no longer the same. It's been updated since.

Compare to the old data, Fury X still matches 980Ti at 1080p. It doesn't fit your narrative.

Ohhhh.. I get it. Your way shows AMD better. No prob.

 

[ShintaiDK]

At least DX12 offers multithreaded drivers for AMD. Then a top end CPU in terms of ST performance isn't as needed anymore.

But all the DX12 nonsense claims are dead now. Not only did Fable show it, now AOTS does it as well.

And I can see its a bitter pill for some that already claimed victory on the base of an alpha benchmark.

 

[Silverforce11]

Ohhhh.. I get it. Your way shows AMD better. No prob.

No mate, my way shows the truth, backed by more data.

If that doesn't fit your narrative, or the ones who were so quick to jump on board thinking "lololol NV driver boost performance! so much for no Async Compute!"... because they failed to analyze what they were looking at properly, then so be it. -_-

Before or after, Fury X still ~= 980Ti at 1080p.
390 still faster than 970.

No 1440/4K data to suggest anything is different either.

 

[Silverforce11]

But all the DX12 nonsense claims are dead now. Not only did Fable show it, now AOTS does it as well.

You clearly haven't paid attention to UE4 powered games where AMD normally is 25-50% slower at each competitive SKU. Or maybe you have but choose to ignore it, as if that kind of performance in a UE4 game from AMD GPU is considered normal.

For AMD to come out on top in Fable on their entire stack (even R290X > 980!) up to the 390X and Fury X matching the 980Ti, is a huge showing for DX12's potential on GCN.

It's quite sad that a bunch of you here got so excited over a delta of 0.4 fps with this new driver versus the status quo and use it as a trumpet on NV's DX12 performance beating AMD.. do a quick google search on Ashes in the past few days, so many tech sites re-post the Computerbase 1080p data claiming NV wins!! These fools don't even realize the performance went from ~55fps to ~72fps for BOTH the Fury X and 980Ti, with a few decimal difference that is within the margin of error.

Yet the biggest issue with DX12 for the past months has been the poor performance NV has at 1440/4K, especially the worse DX12 performance at those resolutions than DX11. Why didn't Computerbase publish those data with the latest drivers? Hmm. Doesn't fit their attention grabbing headline I bet.

 

[Keysplayr]

No mate, my way shows the truth, backed by more data.

If that doesn't fit your narrative, or the ones who were so quick to jump on board thinking "lololol NV driver boost performance! so much for no Async Compute!"... because they failed to analyze what they were looking at properly, then so be it. -_-

Before or after, Fury X still ~= 980Ti at 1080p.
390 still faster than 970.

No 1440/4K data to suggest anything is different either.

We can't have that now. Thinking that NV submitted a driver that boosted performance is taboo and should not be talked about. Or, it could be talked about, but only in the way YOU see it.
What is your deal anyway?

 

[stuff_me_good]

You clearly haven't paid attention to UE4 powered games where AMD normally is 25-50% slower at each competitive SKU. Or maybe you have but choose to ignore it, as if that kind of performance in a UE4 game from AMD GPU is considered normal.

This. :thumbsup:

 

[Silverforce11]

We can't have that now. Thinking that NV submitted a driver that boosted performance is taboo and should not be talked about. Or, it could be talked about, but only in the way YOU see it.
What is your deal anyway?

If it helps your narrative, feel free to add the 0.4 FPS delta to the 980Ti results for 1440 and 4K:

 

[Mondozei]

If you look at this chart, it's quite amazing just how much Fury X improved from DX11:

So if this remains a pattern for DX12 games, AMD's relatively understaffed and underfunded driver team will no longer be at such a disadvantage compared to NV. Which is what I hoped for from the beginning for DX12.

Very impressive, but also very telling just how much DX11 drivers held AMD back. The hardware was never the issue.

 

[ShintaiDK]

If you look at this chart, it's quite amazing just how much Fury X improved from DX11:

You assume AMD did DX11 optimizations.

 

[Techhog]

You assume AMD did DX11 optimizations.

And you assume that they didn't. You're just as guilty of forcing things into your personal narrative as the people you accuse of doing so.

 

[Bobisuruncle54]

You assume AMD did DX11 optimizations.

You're full of crap.

We can't have that now. Thinking that NV submitted a driver that boosted performance is taboo and should not be talked about. Or, it could be talked about, but only in the way YOU see it.
What is your deal anyway?

And so are you.

Infraction issued for personal attack.
-- stahlhart

 

[stahlhart]

And we're done here.
-- stahlhart