Overclock.net › Forums › Graphics Cards › NVIDIA › Fermi Tessellation Performance Comparison
New Posts  All Forums:Forum Nav:

Fermi Tessellation Performance Comparison - Page 11

post #101 of 105
Thread Starter 
Quote:
Originally Posted by BallaTheFeared View Post
Thread Necro!!!!


Looks like the 460s lost 43.2% while another 560 ti rolled out a 40.7%.

Looks like this confirms a direct correlation to tess units and performance hit.
That is a bit surprising. The 460 and 560 Ti have the same ratio of shaders to tess units, but the 460 seems to be taking a slightly larger performance hit according to the samples so far.

I'm too tired to try and crunch all these numbers together and come up with some kind of conclusion atm

Interesting results though. I'd KILL for some 550/450 results to fill in the lower end.
post #102 of 105
Thread Starter 
So... for kicks I decided to test this at my native res.

The results were interesting to say the least.

Attachment 224485

Using stock clocks again as a baseline I got 54 fps with no tess, 38.8 fps with extreme tess... for a total performance loss of ~28%.

It would seem then that the performance loss shrinks as the resolution increases. If it was proportional, it should have stayed the same, shouldn't it?
post #103 of 105
I've got science for any occasion.
275.33, stock 752, and the ubik 4.0
2560x1440 = 46.5 Extreme Tess, 60 no Tess = 22.5% drop(ing science...)

Attachment 224515


@1080 = 70.3 Extreme Tess and 101.7 no Tess, so inline with the others results.

Attachment 224517


Edited by Draygonn - 8/15/11 at 6:47pm
VR machine
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 4790k  ASUS Z97I-Plus mITX ASUS GTX 1080 Ti Turbo 8G EVGA 1600 C9 
Hard DriveCoolingOSMonitor
Samsung Spinpoint 1TB Corsair H60 Windows 8.1 64 bit Sony HW45ES Projector w 120" Acoustically Trans... 
MonitorKeyboardPowerCase
VR- HTC VIVE  Razer Turret EVGA Supernova 550 G2 Corsair 250D 
AudioAudioAudio
Yamaha RX-V781 Emotiva A-700 and Airmotiv Speakers Dual SVS SB1000 Subs 
  hide details  
Reply
VR machine
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 4790k  ASUS Z97I-Plus mITX ASUS GTX 1080 Ti Turbo 8G EVGA 1600 C9 
Hard DriveCoolingOSMonitor
Samsung Spinpoint 1TB Corsair H60 Windows 8.1 64 bit Sony HW45ES Projector w 120" Acoustically Trans... 
MonitorKeyboardPowerCase
VR- HTC VIVE  Razer Turret EVGA Supernova 550 G2 Corsair 250D 
AudioAudioAudio
Yamaha RX-V781 Emotiva A-700 and Airmotiv Speakers Dual SVS SB1000 Subs 
  hide details  
Reply
post #104 of 105
Thread Starter 
Quote:
Originally Posted by Draygonn View Post
I've got science for any occasion.
275.33, stock 752, and the ubik 4.0
2560x1440 = 46.5 Extreme Tess, 60 no Tess = 22.5% drop(ing science...)
Awesome. So the % performance hit significantly decreases relative to resolution. This is a triumph!
post #105 of 105
EDIT: this is what happens when you dont read the last few pages of posts lol, kinda repeated whats been said but never-the-less ive still got some results and im waiting for booty warrior or another 560 ti.

Just had a quick idea that might help:

If two people with very similar systems (processor, speed, also ram)

One person with a 460 and one with a 560 ti, or else one 570 and one 580

Because these cards are identical bar 1 more tessalator, clock speed and shaders

Set unigine engine benchmark to no tessalation

Declock the 560 or 580 to make them match the average framerates or the 460 or 570

Now we are in a positon to see exactly what one more tessalator does for performance by just twitching tessaltion on and compare the scores.

Just an idea. We could then draw more conclusions.

Another thought: after this scale back performance all the way down to match a 560 from the 580, tess off, then flick it on and see the differences between the two for another comparison. Slightly larger amount of tessalators to compare, ut since we would now know the difference 1 extra makes using the 460 and 560, we could just work that into the new difference and gain more info.

Also like you say having the 550 or 450 would be good also to do the same things. Im up for participation in this idea, if anyone thinks its even valid. My i5 750 should perform the same as an i7 1st generation with ht off, same clock and ram speed.

edit 2: refering to the unexpected performance hit looking at the 460 and 560, dont forget the shading power is greater on the 560 regardless of ratio, so performance hit is going to be greater on the 460 for that reason too. We also dont know if tessalators scale with clock speed but lets leave that variable out for this idea im having. Again the only way is make both cards equal in average frames, to assme the reduced clockspeed is making the shader power equal

edit3: just realized the cpu etc shouldnt matter as long as its clocked highly to eliminate any bottleneck, and i was just playing abut and got a nice 4GH oc with all powersavings and dynamic vcore. So im gona run tests here comparing them to your opening post

Quote:
I used a single, stock clocked 560 Ti @1080p as a reference w/ForceWare 275.50 (the 560 Ti has 8 dedicated tess units).

Tess Off: Average FPS = 50.1
Tess Normal: Average FPS = 39.6
Tess Extreme: Average FPS = 29.6

1 - (39.6/50.1) = 0.2096

Meaning going from no tessellation to "normal" incurred a ~21% performance penalty.

1 - (29.6/39.6) = 0.2525

Further, going from normal to extreme incurred a relative 25% performance penalty.

1 - (29.6/50.1) = 0.4091

And the overall performance penalty from no tessellation to extreme was ~40.9%.
So for a single 460 Hawk at 811mhz (kind of close in shader performance)(top card doesnt oc well through no spacing so this will have to do)

Tess Off: Average FPS = 42.5
Tess Normal: Average FPS = 33.6 - 20.94% difference from off
Tess Extreme: Average FPS = 24.9 - 41.41% difference from off - A further 20.47% difference

so similar result differences, because the ratio of shaders and tessalators are the same

now to compare results

tess off = 15.17% difference in performance
tess normal = 15.15%
tess extreme = 15.88%

updated:
so basically your card was 15.4% faster no matter what. If you can just lower your clock speed and run your tests again to initially match 42.5 FPS tessalation off (start around 700core maybe?), we can make a further analysis, there should be a difference in the favour of the 560 ti with tessalation, and that difference is what 1 extra tessalation unit does for performance. However if not and the tessalators seem to scale with clock speed its gona be impossilbe to work out! And the lower clock speed will cancel out having the extra unit.
Edited by Juliancahillane - 8/30/11 at 7:48am
    
CPUMotherboardGraphicsRAM
i5750 @ 3.8 (19x200)1.31vcore /1.21vtt  GA-P55M-UD4 EVGA GTX 480 @ 875/1800/2000 1.075vcore 8GB Corsair Dominator @ 2000mhz 9-10-9-27 1T 
Hard DriveHard DriveCoolingCooling
1TB Western Digital Black 64mb OS Boot 3x 60GB Vertex2E Raid 0 : 700mb/s  Prolimatech Megahalems Rev.B Zalman VF3000F 
OSMonitorPowerCase
Windows 7 Ultimate 64 42" Panasonic TXP42G20 1080p Plasma XFX 850w Pro XXX Edition (Silver) Fractal Design Arc Mini 
MouseMouse PadAudioOther
G9x Razer Goliathus Yamaha RXV671 reciever / Q Acoustic 2020i speakers Belkin n52e Keypad 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5750 @ 3.8 (19x200)1.31vcore /1.21vtt  GA-P55M-UD4 EVGA GTX 480 @ 875/1800/2000 1.075vcore 8GB Corsair Dominator @ 2000mhz 9-10-9-27 1T 
Hard DriveHard DriveCoolingCooling
1TB Western Digital Black 64mb OS Boot 3x 60GB Vertex2E Raid 0 : 700mb/s  Prolimatech Megahalems Rev.B Zalman VF3000F 
OSMonitorPowerCase
Windows 7 Ultimate 64 42" Panasonic TXP42G20 1080p Plasma XFX 850w Pro XXX Edition (Silver) Fractal Design Arc Mini 
MouseMouse PadAudioOther
G9x Razer Goliathus Yamaha RXV671 reciever / Q Acoustic 2020i speakers Belkin n52e Keypad 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › Fermi Tessellation Performance Comparison