• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

**The New Official THIEF-Incorporating Bench mark Thread**

Soldato
Joined
29 Aug 2010
Posts
7,850
Location
Cornwall
It's good that they're continuing to make improvements to the game (rather than dropping support once it's a few months old). Bit annoying for benchmark threads though as it renders a lot of results obsolete.
In this case the scores should go up, but if they 'fixed' something that actually led to slightly worse performance it could mean that people may never beat a result that now couldn't be repeated on the hardware that first got it.

I imagine proper benchmarking apps like Heaven and 3DMark don't update for this very reason.
 
Caporegime
Joined
12 Jul 2007
Posts
40,576
Location
United Kingdom
It's good that they're continuing to make improvements to the game (rather than dropping support once it's a few months old). Bit annoying for benchmark threads though as it renders a lot of results obsolete.
In this case the scores should go up, but if they 'fixed' something that actually led to slightly worse performance it could mean that people may never beat a result that now couldn't be repeated on the hardware that first got it.

I imagine proper benchmarking apps like Heaven and 3DMark don't update for this very reason.

Here is the fix list.

- Performance improvements: Removed unnecessary rendertarget copies (for Mantle and in some cases D3D as well).
- Fixed a rare case where Mantle would be enabled when there are no Mantle capable GPUs, crashing the game.
- Fixed stencil buffer state being set in depthbind while no stencil buffer exists (validation errors but not noticeable in game).
- Fixed transitions to extension image states (wsipresent) being ignored because they were treated as a datatransfer state (validation errors but not noticeable in game).

The last bugfix was to fix Valve’s problems with integrating the steam overlay.
 
Man of Honour
Joined
20 Sep 2006
Posts
34,050
3440x1440

DX11:
K9kpfZL.jpg

Mantle:
kyzRZTE.jpg


Although Mantle got the better score, DX looked smoother and Mantle had some horrific jerk moments. I ran it twice, same both times.
 
Soldato
Joined
29 Aug 2010
Posts
7,850
Location
Cornwall
Here is the fix list.

- Performance improvements: Removed unnecessary rendertarget copies (for Mantle and in some cases D3D as well).
- Fixed a rare case where Mantle would be enabled when there are no Mantle capable GPUs, crashing the game.
- Fixed stencil buffer state being set in depthbind while no stencil buffer exists (validation errors but not noticeable in game).
- Fixed transitions to extension image states (wsipresent) being ignored because they were treated as a datatransfer state (validation errors but not noticeable in game).

The last bugfix was to fix Valve’s problems with integrating the steam overlay.

Like I said, good they're fixing it, but spoils these threads as 75% of the results or more will be the old version. So it's like comparing apples and oranges.

That's because it using more vram than you have S2KIP.

Just beat Alataars score at 1080P. Even 5+fps ahead of the one he posted prior to that using cheats.

You mean the score he got before the fixes that you're now running?
Shocker!

What cheats?
 
Caporegime
Joined
12 Jul 2007
Posts
40,576
Location
United Kingdom
His first score (see the quoted post right after his entry from scotty) was 99 fps, then edited and suddenly it was 95fps.

99fps was a result of me not realizing that I had my benching driver tweaks and other shady business going on. So I went back to normal settings with everything and re-benched.

So basically I had stuff on that was only supposed to be on if running stuff for hwbot on LN2 or DICE. Went and fixed it once I realized what happened.

From the horses mouth.
 
Soldato
Joined
25 Nov 2011
Posts
20,639
Location
The KOP
It's good that they're continuing to make improvements to the game (rather than dropping support once it's a few months old). Bit annoying for benchmark threads though as it renders a lot of results obsolete.
In this case the scores should go up, but if they 'fixed' something that actually led to slightly worse performance it could mean that people may never beat a result that now couldn't be repeated on the hardware that first got it.

I imagine proper benchmarking apps like Heaven and 3DMark don't update for this very reason.

Same could be said for GPU driver support though.. I bet a game that was released close to the 290 would now perform better on latest driver vs what driver was out at that time.

like you said also its great to see they still patching this game though.
 
Soldato
Joined
29 Aug 2010
Posts
7,850
Location
Cornwall
His first score (see the quoted post right after his entry from scotty) was 99 fps, then edited and suddenly it was 95fps.



From the horses mouth.

He didn't say it was cheating, you branded it as that. He said tweaks and 'shady stuff' which could mean anything but is apparently acceptable for HWBot submissions.


Same could be said for GPU driver support though.. I bet a game that was released close to the 290 would now perform better on latest driver vs what driver was out at that time.

like you said also its great to see they still patching this game though.

Oh yeah, definitely an issue too. However Nvidia/AMD didn't make the benchmark and so have no reason to want it to perform consistently.
It'd almost be nice if benchmarks would let you select the version of the game to be benchmarked. Then you could compare the improvements made in new versions, but for threads like this you could offer some sort of consistency, driver improvements aside.
 
Last edited:
Caporegime
Joined
24 Sep 2008
Posts
38,322
Location
Essex innit!
His first score (see the quoted post right after his entry from scotty) was 99 fps, then edited and suddenly it was 95fps.

From the horses mouth.

Are you being serious? A guy realises he is using his tweaks that HWBot allow and then corrects his post to follow the rules of OcUK and he is getting picked up for cheating?

Come on Matt, you are better than that.
 
Soldato
Joined
25 Nov 2011
Posts
20,639
Location
The KOP
He didn't say it was cheating, you branded it as that. He said tweaks and 'shady stuff' which could mean anything but is apparently acceptable for HWBot submissions.




Oh yeah, definitely an issue too. However Nvidia/AMD didn't make the benchmark and so have no reason to want it to perform consistently.
It'd almost be nice if benchmarks would let you select the version of the game to be benchmarked. Then you could compare the improvements made in new versions, but for threads like this you could offer some sort of consistency, driver improvements aside.

Thief does allow you to down grade the patch version. Am not sure if all games allow this though.

Image
Screenshot%20%28143%29.png
 
Last edited:
Soldato
Joined
29 Aug 2010
Posts
7,850
Location
Cornwall
Are you being serious? A guy realises he is using his tweaks that HWBot allow and then corrects his post to follow the rules of OcUK and he is getting picked up for cheating?

Come on Matt, you are better than that.

Maybe a news site should pick up on the fact that an AMD official has said that HWBot allows cheats to be used in their benchmarks. Is this AMD stating that HWBot benchmarks and records are worthless?
 
Man of Honour
Joined
20 Sep 2006
Posts
34,050
That's because it using more vram than you have S2KIP.

Just beat Alataars score at 1080P. Even 5+fps ahead of the one he posted prior to that using cheats.

Yeah I wasn't monitoring VRAM usage to be honest. Is it just Thief that will use it 'up'?
 
Back
Top Bottom