TRESSFX: A NEW FRONTIER OF REALISM IN PC GAMING

flopper · 6 Mar 2013 at 13:48

Those numbers dont look for nvidia.
and hardly able to go past a 7870...

LtMatt · 6 Mar 2013 at 13:56

Rusty0611 said:
lol stop going on about this, Matt. They didn't.

The graphs were quite clear what they were saying with regards to max playable settings. I admit I didn't read the comments in depth to see if there was implicit bias but the graphs showed (from memory) marginally higher FPS on nVidia cards at LOWER settings.

The Apples to Apples comparison showed the AMD card being faster.

I dislike HardOCP but I think although I don't like that particular way of displaying information it isn't an automatic reason to suspect bias.

I compared their performance and results to what i got on my system. It was bull.

Rusty0611 · 6 Mar 2013 at 13:58

They looked about right to me for Apples to Apples. Unless you use Extreme AA the 680/7970 were even.

LtMatt · 6 Mar 2013 at 14:04

Rusty0611 said:
They looked about right to me for Apples to Apples. Unless you use Extreme AA the 680/7970 were even.

I know that my performance numbers were far in excess of what they came up with regardless of which settings they used. High SSAA also uses compute so for a 680 to be any where near a 7970 is bull.

Rusty0611 · 6 Mar 2013 at 14:05

LtMatt said:
I know that my performance numbers were far in excess of what they came up with regardless of which settings they used. High SSAA also uses compute so for a 680 to be any where near a 7970 is bull.

Where's that Sleeping Dogs thread gone?

Tone and I ran High AA and Extreme AA both while overclocked and he removed my underwear at Extreme but I was marginally quicker at High.

LtMatt · 6 Mar 2013 at 14:13

Rusty0611 said:
Where's that Sleeping Dogs thread gone?

Tone and I ran High AA and Extreme AA both while overclocked and he removed my underwear at Extreme but I was marginally quicker at High.

Go find it then.

I'm talking about H's results. I know their 7970 ghz performance was not correct compared to what i was seeing on a slower system.

I'd be happy to bench up against a 680 as well.

Rusty0611 · 6 Mar 2013 at 14:45

LtMatt said:
Go find it then.

I'm talking about H's results. I know their 7970 ghz performance was not correct compared to what i was seeing on a slower system.

I'd be happy to bench up against a 680 as well.

We all did it ages ago when it was flavour of the month mate. You've fiddled with your 7970 so much since then you've gone blind and forgot

.

LtMatt · 6 Mar 2013 at 14:52

Rusty0611 said:
We all did it ages ago when it was flavour of the month mate. You've fiddled with your 7970 so much since then you've gone blind and forgot .

I've been looking through the sleeping dogs benchmark thread and the ocuk benchmark thread. 680 results are a fair way behind my results at extreme + high.

Rusty0611 · 6 Mar 2013 at 14:57

Tonester0011 said:
Because I'm bored - Click for all scores so far.

1920x1080 - Extreme AA, High presets. (Single GPU)

1. LtMatt - HD7970 - 54.2FPS
2. Nickolp1974 - HD7970 - 52.1FPS
3. Tonester0011 - HD7970 - 49.5FPS
4. Rusty0611 - GTX680 - 46.1FPS

1920x1080 - High AA, High presets. (Single GPU)

1. Rusty0611 - GTX680 - 89.7FPS
2. Tonester0011 - HD7970 - 89.5FPS
3. LtMatt - HD7970 - 85.3

Edit: things might have changed slightly since then but not enough to suddenly give AMD a stonking lead at High AA. Extreme AA as you say is always going to be better on 7900.

LtMatt · 6 Mar 2013 at 15:06

Something wrong with those HIGH results Rusty.

I don't think i need to spell it out when you compare them to the extreme results.

However i will reinstall sleeping dogs to run the bench again at some point as it makes no sense that my score is a lot lower than tones.

No mentions of clock speeds either.

Rusty0611 · 6 Mar 2013 at 15:08

LtMatt said:
Something wrong with those HIGH results Rusty. I don't think i need to spell it out when you compare them to the extreme results.

No mentions of clock speeds either.

I think you're confused. They're grouped by Extreme and then High.

Well mine was max OC on my reference card. Tones was whatever his 7970s go to so around 1200-1250 from memory. Some might go higher, some might not. What the point is that they are round about even at the High AA.

LtMatt · 6 Mar 2013 at 15:12

Rusty0611 said:
I think you're confused. They're grouped by Extreme and then High.

Well mine was max OC on my reference card. Tones was whatever his 7970s go to so around 1200-1250 from memory. Some might go higher, some might not. What the point is that they are round about even at the High AA.

I can see that but i know my score is wrong, I'm reinstalling now so should have some accurate results up in comparison to see if they really are as close at HIGH as you say.

Rusty0611 · 6 Mar 2013 at 15:15

LtMatt said:
I can see that but i know my score is wrong, I'm reinstalling now so should have some accurate results up in comparison to see if they really are as close at HIGH as you say.

Well it's just what you posted at the time and it is roughly in line with Tonester's.

Not saying things ain't changed at all - indeed performance has improved for nVidia a bit reading the driver notes (which aren't always true

). This was all discussed on the thread at the time if you care to read it. That is, nVidia even with AMD at High AA lagging at extreme AA.

LtMatt · 6 Mar 2013 at 15:24

Rusty0611 said:
Well it's just what you posted at the time and it is roughly in line with Tonester's.

Not saying things ain't changed at all - indeed performance has improved for nVidia a bit reading the driver notes (which aren't always true ). This was all discussed on the thread at the time if you care to read it. That is, nVidia even with AMD at High AA lagging at extreme AA.

Pretty sure no other 680 got any where near your score either in that thread. So i wouldn't say that would represent the general performance.

If both cards were at stock id expect a victory for the 7970 ghz without doubt at High, even more at Extreme.

Thinking back my card would have only been clocked at 1200 when i ran that HIGH result.

Rusty0611 · 6 Mar 2013 at 15:29

LtMatt said:
Pretty sure no other 680 got any where near your score either in that thread. So i wouldn't say that would represent the general performance.

All I did was overclock it. Not like I could add extra volts to it lol.

Image is still up:

http://img833.imageshack.us/img833/1382/680aahigh.png

LtMatt said:
If both cards were at stock id expect a victory for the 7970 ghz without doubt at High, even more at Extreme.

Interesting theory, but that would implicitly suggest that the 680 gains(ed) more from overclocking to allow it to pull ahead in the benchmark above when clocked up.

LtMatt said:
Thinking back my card would have only been clocked at 1200 when i ran that HIGH result.

Probably, yeah. Mine would have been 1280 ish more like likely, so it isn't a true like for like max OC comparison as 1280 was more towards the top end for 680s whereas 1200-1225 is more towards the low end for 7970s. We're talking small margins though which isn't going to dramatically alter the results.

Edit: just realised we've gone wildly off the topic of Lara Croft's new hair do. Shall we finish over email when you've re-ran your benchie?

LtMatt · 6 Mar 2013 at 15:38

Rusty0611 said:
All I did was overclock it. Not like I could add extra volts to it lol.

Interesting theory, but that would implicitly suggest that the 680 gains(ed) more from overclocking to allow it to pull ahead in the benchmark above when clocked up.

Probably, yeah. Mine would have been 1280 ish more like likely, so it isn't a true like for like max OC comparison as 1280 was more towards the top end for 680s whereas 1200-1225 is more towards the low end for 7970s. We're talking small margins though which isn't going to dramatically alter the results.

We'll see. Your results were far above any others ive seen. I'm looking at other 680 single card results on High and yours are 10-15 fps faster than any others i can find. However the ones i have found were not clocked as high as yours.

EDIT

Anand did a review recently. This puts an end to the discussion as its using latest drivers from both sides.

Stock 7970 almost 10fps faster at 1080p with HIGH SSAA. Ghz edition much much faster.

EDIT

Before you come back with they were closer originally. AMD had their drivers optimised for this game months before it was released. Gaming evolved title and they worked so closely with the devs square enix. So the AMD gains would be no more than a few fps at most if that. I know AMD made large gains in this at 12.11 and beyond but their performance was already good. After 12.11 the gap, which might have been closer than it is now widened.

Rusty0611 · 6 Mar 2013 at 15:45

LtMatt said:
We'll see. Your results were far above any others ive seen. I'm looking at other 680 single card results on High and yours are 10-15 fps faster than any others i can find. However the ones i have found were not clocked as high as yours.

Are they overclocked? It's not like you can fiddle the benchmark anyway lol. It's a fixed benchmark which shows you the settings/results in one screenshot.

Here's my result:

LtMatt said:
EDIT

http://i.imgur.com/7qOCOhe.png[IMG]

Stock 7970 almost 10fps faster at 1080p with HIGH SSAA. Ghz edition much much faster.[/QUOTE]

Looks suspect as SLI 680 is suddenly faster than CF 7970 GE. :confused:
Suspect AMD drivers? :p

And anyway, we were clearly discussing overclocked comparisons. Stock comparisons are boring as there's already a plethora of them around.

TheRealDeal · 6 Mar 2013 at 15:46

LtMatt said:
We'll see. Your results were far above any others ive seen. I'm looking at other 680 single card results on High and yours are 10-15 fps faster than any others i can find. However the ones i have found were not clocked as high as yours.

EDIT

Anand did a review recently. This puts an end to the discussion as its using latest drivers from both sides.

Stock 7970 almost 10fps faster at 1080p with HIGH SSAA. Ghz edition much much faster.

EDIT

Before you come back with they were closer originally. AMD had their drivers optimised for this games month before it was released. Gaming evolved title and they worked so closely with the devs square enix. So the AMD gains would be no more than a few fps at most if that.

I am pretty sure most off the benches ran in the benchmark thread were run before 12.11 drivers which saw big boosts in most games. Rusty's oc results make sense when looking at anand's results. It does seem amd has given more performance through drivers as the anand ghz edition is giving results way ahead of oc benches from september last year. Shows how wrong H were though.

Could be cpu limited in crossfire/sli as they are pretty close and scaling ain't great on any of them. When you move up in resolution the 7970cf and 7990 pull ahead where cpu speed is less limiting. Note that crossfire and sli scaling is much better which backs up a cpu bottleneck at 1080p.

Rusty0611 · 6 Mar 2013 at 15:52

Yeah to be fair, I haven't factored in whether AMD have provided more performance since we all ran the Sleeping Dogs comparisons. I haven't really followed it closely enough to tell but it would appear so as it was definitely round about even when we ran those benches.

Now it looks less so. I looked at a 12.11 benchmark on Tom's and the 12.11 GHz edition has a large enough lead. Higher than your original FPS results when overclocked in fact so this would point to a bump in performance sometime between when they were ran and now.

LtMatt · 6 Mar 2013 at 15:53

Rusty0611 said:
Are they overclocked? It's not like you can fiddle the benchmark anyway lol. It's a fixed benchmark which shows you the settings/results in one screenshot.

Here's my result:

Looks suspect as SLI 680 is suddenly faster than CF 7970 GE.
Suspect AMD drivers?

And anyway, we were clearly discussing overclocked comparisons. Stock comparisons are boring as there's already a plethora of them around.

Lol didn't notice the SLI/CFX results.

TheRealDeal said:
I am pretty sure most off the benches ran in the benchmark thread were run before 12.11 drivers which saw big boosts in most games. Rusty's oc results make sense when looking at anand's results. It does seem amd has given more performance through drivers as the anand ghz edition is giving results way ahead of oc benches from september last year. Shows how wrong H were though.

Could be cpu limited in crossfire sli as they are pretty close and scaling ain't great on any of them.

Good point TheRealDeal. I have not run the benchmark in ages so will check soon. My runs were certainly a long time before 12.11 drivers.

@Rusty my original point was and still is 7970 is faster than a 680 at HIGH or Extreme. Clearly. That point still stands. However your super clocking 680 result was impressive, but that does not represent general performance. Infact i imagine that if a 7970 was clocked as much as your 680 was the gap would widen, not get closer. I will be testing this theory shortly and will come back with some results. I also believe that at no point was the 680 even. It might have been close but it was never even or ahead. Your super clocking 680 might have made it even but that was with 7970's clocked lower.