Soldato
Latest officials. Seems inline to me, well mostly. 
-
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**
- Thread starter humbug
- Start date
Man of Honour
4 x 290Xs @1200/1500
4930k @4.6
4K
DX11
Mantle
Mantle is not a serious option for 4K, check the minimums above. Also where it uses all the VRAM (I was even running Windows basic background) it will cause the PC to freeze up very quickly. It only just managed to finish the bench. The PC was ok again after I used task manager to shut down Thief.
4930k @4.6
4K
DX11
Mantle
Mantle is not a serious option for 4K, check the minimums above. Also where it uses all the VRAM (I was even running Windows basic background) it will cause the PC to freeze up very quickly. It only just managed to finish the bench. The PC was ok again after I used task manager to shut down Thief.
Caporegime
So Mantle again is using more VRAM than DX11. Strange one that.
Caporegime
Good thing Mantle is optional so in the few rare circumstances it performs worse than DX you can just use DX instead. Best of both worlds for AMD users. 
EDIT
Kaap did you see what happens if you select the option called Automatically limit texture settings or something. Does that have any bearing on the vram usage for Mantle and does it have any noticeable effect on image quality?
EDIT
Kaap did you see what happens if you select the option called Automatically limit texture settings or something. Does that have any bearing on the vram usage for Mantle and does it have any noticeable effect on image quality?
Last edited:
Soldato
Good thing Mantle is optional so in the few rare circumstances it performs worse than DX you can just use DX instead. Best of both worlds for AMD users.
EDIT
Kaap did you see what happens if you select the option called Automatically limit texture settings or something. Does that have any bearing on the vram usage for Mantle and does it have any noticeable effect on image quality?
AMD should use that for their Mantle Marketing!
"Mantle is optional so you can just use DX instead!"
Man of Honour
Good thing Mantle is optional so in the few rare circumstances it performs worse than DX you can just use DX instead. Best of both worlds for AMD users.
EDIT
Kaap did you see what happens if you select the option called Automatically limit texture settings or something. Does that have any bearing on the vram usage for Mantle and does it have any noticeable effect on image quality?
I was just running the bench to see what happened.
Although DX11 performance is better @4K there is still something wrong with it that the game Devs need to get sorted.
There is no way you should get a CPU bottleneck with the FPS that low whether you use DX11 or Mantle. Bearing in mind the lows in the Mantle version, the DX11 version should be getting higher averages but it is not. The performance with DX11 and Mantle points to a badly written game. I don't think it is fair to blame Mantle for bad performance here when it is the game Devs and their sloppy work letting the side down.
Caporegime
Fair enough. I think there might be another patch for Thief soon.
Man of Honour
Soldato
Mantle hemorrhages memory. Developers really need to work on that. I think it's where they're leaving it to Mantle to dictate rather than doing any of the paging themselves. Extra work they don't have time for. One of the many reasons why coexisting API simply rocks
Aside from True CPU multi-threading and lower overheads Mantle also has the option of pre-loading textures.
The benefit of that is the GPU does not need to load them on the fly as your going along, visually; it cuts down on texture filling in as your moving through the map, (its already there before you see it) for performance it keeps frame rates more consistent, the performance isn't so much up and down as its not loading and unloading texture as your moving around.
The down side is a constant pool of texture preloaded increases the amount of vRam used.
Thread updated.
Caporegime
Well that's interesting. Still need to sort out a BIOS. Still crude overclocking at the moment and running ram at XMP 2133CL9 CR2 so there is plenty left:
EVGA Titan Black (WC)
Stock Bios (220+offset)
1350/2100
[email protected] (WC)
Nice score, but how are you scoring lower than Alatar with a faster gpu and higher clocks?
Soldato
My guess would be CPU speed. If you look Greg's clocks aren't far off but his CPU is a generation older.
Also look at Protoo's (2nd) entry, faster clocks than Tommy but he's below him. I'd suggest it's the AMD CPU, although would've thought Mantle would've removed that from the AMD cards.
Also look at Kaap's 290X result, faster clocks than Protoo and Tommy but he's still behind.
I don't think there's anything wrong with _Alatar_'s score.
Also, does Protoo get 2 290X scores because he's using AMD or would everyone get multiple entries if they submit slightly different clocks? Does it apply to GPU clocks too or just different CPU clocks?
Man of Honour
4770k is a better CPU for single card benching and one that will do 5.0ghz is better still.
Soldato
mmm. Yeah that's what I keep telling myself
To be fair, I think Alatar's system is quite tuned. When I clocked 98 fps in the earlier build my ram was flat out and very tight 2400 CR1 CL9, I'm just running a bit of a crude overclock at the moment.
And yes his cpu at 5Ghz will help some.
Saying that it was also on stock BIOS. If I ran it again now, I'm fairly sure I could beat it
Soldato
Gregs is an original Titan. Alatar is as as well. My clocks are boost so would have fluctuated in that test. It would be easy to beat on non boost BIOS now, not bothered though
Permabanned
I haven't done screenshots or anything I just though I'd post a couple of comparison scores for the Sapphire 265 I just got.
Mantle: 13/34.6/51.2
DX11: 23.5/33.5/48.3
The rig has a stock clocked 4790k on an Asus Z97 Hero board, 8gb's of 2133 ram and the OS is 8.1 pro.
Mantle: 13/34.6/51.2
DX11: 23.5/33.5/48.3
The rig has a stock clocked 4790k on an Asus Z97 Hero board, 8gb's of 2133 ram and the OS is 8.1 pro.
Soldato
I thought Matt said that it was just BF4 and DICE which had cocked the memory (mis)management?
Caporegime
4K - SLI Titans 1267/3758 - 3930K @ 4.6
CPU: AMD FX-8350 @ 4.6Ghz
GPU: R9 290 @ 1200/1500
GPU: R9 290 @ 1200/1500
