• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

AT Fermi OC/benchmark thread

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Keysplayr said:
Video of Danger Dens GTX480 water block.
http://www.dangerden.com/New-Products/gtx480-update-4.html

Furmark 51C
Unigine 46C
MSI afterburner 51C

The guy in the video stated that he was using an older radiator and a fairly weak pump, but you can get the idea. I'm a bit tempted to go water cooling, but I change out cards so frequently that I can see a big hassle for me, and the expense of a water block unique to each video card. Yikes!

Anyway, enjoy.
Click to expand...

I was reluctant to go water as well, but it rocks. Building my first loop was the most fun I've had building a PC since my first build years ago. The guys in the case/cooling sub-forum talk about it being addictive... They do not lie. It's not a cheap "habit" though...
 
nitromullet said:
Nice. Mine is a dog on the RAM as well. I guess that extra $150 for the 480 isn't for nothing.

edit: weren't you getting more out of your RAM before the volt mod? (1675 vs 1700?)
Click to expand...

To be honest I just didn't bother tweaking memory with the final OC, most of the tasty juice comes from the core anyways. I remember reading somewhere that they had major memory controller issues, which is the reason for the low memory speeds to begin with, I am just trying not to push my luck 😛

3D Mark Vantage Screenshot (almost equivalent to a stock GTX 480):

GTX&
 
Last edited:
Nice OC's. One card of mine does 785 with .962 volts. I will try to push it a bit more later. I found that my lower warranty EVGA card is sucking 30 watts more power than the good one so I am selling it and getting a lifetime warranty card. I suppose they use inferior GPU's for those even though the retail price is the same. A bit ridiculous if you ask me. Don't ever buy -TR ended EVGA cards.

As for UAC I disable it so I don't get all those annoying messages. Afterburner is great. I just got it a week ago. Too bad Precision can't do everything it can. You can set the overclock/overvoltage to apply on bootup with the 470/480 since the voltage goes down on idle anyway. 2D/3D profiles aren't necessary.
 
Alright guys, time to resurrect this thread. I re-ran my benches at 800-1600-1675. If somebody running a OCed Radeon 5870 with a 4Ghz i7 can run these built-in benches for at least Unigine 2.0, Dirt2 and Far Cry 2 it would sure make for a good comparison. I understand that the comparison of PhysX and Cuda enabled games is pointless, I can re-run these with those disabled if a Radeon bencher requests. I have a feeling that a GTX 470 at these clocks can give even an OCed Radeon 5870 a run for its money......😛

eVGA GTX 470 Clocks

Stock: 608-1215-1675 (Stock Fan Profile)
OC1: 700-1400-1700 (Fan 65% Constant)
OC2: 800-1600-1675 (Custom Fan Profile)
--------------------------------------------------------------------
System Specs

CPU: Core i7 920 4Ghz (200x20)
RAM: 12GB DDR2-1200 (8-8-8-24-1T)
Motherboard: MSI x58 Platinum SLI
Resolution: 1680x1050
PSU: PCP&C Turbocool 860W
OS: Windows 7 Ultimate 64bit
--------------------------------------------------------------------
Unigine 2.0

Settings
Render: direct3d11
Mode: 1680x1050 4xAA fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 4x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
Replication: disabled
Tessellation: extreme

Stock:
FPS: 28.5
Scores: 718
Min FPS: 16.4
Max FPS: 70.7

OC1:
FPS: 32.0
Scores: 807
Min FPS: 18.6
Max FPS: 79.0

OC2:
FPS: 35.1
Scores: 884
Min FPS: 20.7
Max FPS: 86.9

--------------------------------------------------------------------

Dirt 2 DX11 (In Built Benchmark)
8xQCSAA
Stock:
Min:57.7, Avg: 75.7
OC1:
Min:67.8, Avg: 82.7
OC2:
Min:72.8, Avg: 86.4

--------------------------------------------------------------------

Far Cry 2 (Benchmarking Tool)
(DX10 All Max 4xAA)
Stock:
* Average Framerate: 90.25
* Max. Framerate: 129.85
* Min. Framerate: 71.08
OC1:
* Average Framerate: 100.86
* Max. Framerate: 145.20
* Min. Framerate: 80.45
OC2:
* Average Framerate: 108.63
* Max. Framerate: 156.62
* Min. Framerate: 87.02

--------------------------------------------------------------------

Batman Arkham Asylum (In Built Benchmark)
(All Max 4xAA PhysX High)
Stock:
* Average Framerate: 65
* Max. Framerate: 97
* Min. Framerate: 35
OC1:
* Average Framerate: 71
* Max. Framerate: 104
* Min. Framerate: 39
OC2:
* Average Framerate: 77
* Max. Framerate: 119
* Min. Framerate: 40

--------------------------------------------------------------------

Just Cause 2
(DX10 All Options Maxed incl CUDA options 4xAA & 16xAF)

Stock:
The Dark Tower: 40.14
Desert Sunrise: 53.41
Concrete Jungle: 32.68

OC1:
The Dark Tower: 44.87
Desert Sunrise: 59.96
Concrete Jungle: 37.18

OC2:
The Dark Tower: 49.84
Desert Sunrise: 65.92
Concrete Jungle: 40.70
 
Last edited:
I've got my pair of GTX 480 at 870/2150. One of the cards could go to 2250 on the memory. This is on air slotted right next to each other but with extra case fans. Temps and noise are alright...I have a set of open headphones (audio technica ADH-AD700) which don't block a lot of sound and the noise is only particularly noticeable during very quiet parts of a game. That's as high as these cards are going...there's more room tempwise but they're at 1.11 volts (considered "safe" by the manufacturers) and 1.125 is the max afterburner allows. That's fine though since they were starting to level off with extra voltage giving diminishing returns. They destroy everything I throw at them and I can even keep them stock speeds for most games. The ability to really crank them is really nice for GTA 4 and Crysis though.
 
There doesn't seem to be much point going more than 3 notches higher than the VID. (.037-.038v). 10-20Mhz isn't worth the heat and stress.

Just Cause 2 is DX10 not DX11 and turning off either GPU water or SSAO boosts framerates quite a bit. I prefer the GPU water myself. Turning them both off gives a 30% higher framerate.
 
SHAQ said:
There doesn't seem to be much point going more than 3 notches higher than the VID. (.037-.038v). 10-20Mhz isn't worth the heat and stress.

Just Cause 2 is DX10 not DX11 and turning off either GPU water or SSAO boosts framerates quite a bit. I prefer the GPU water myself. Turning them both off gives a 30% higher framerate.
Click to expand...

Fixed, for some reason I thought it was DX11.
 
Glad to see this thread is still alive.

I finally got my EK water block all the way from Slovenia, but I need some new barbs which I can't get locally. I should have an update in a few days after I get the barbs and do some leak testing prior to install.
 
SHAQ said:
There doesn't seem to be much point going more than 3 notches higher than the VID. (.037-.038v). 10-20Mhz isn't worth the heat and stress.

Just Cause 2 is DX10 not DX11 and turning off either GPU water or SSAO boosts framerates quite a bit. I prefer the GPU water myself. Turning them both off gives a 30% higher framerate.
Click to expand...

I prefer to keep those and turn shadows down one notch while leaving the high res shadow option turned on. Unless you sit around and closely scrutinize the shadows, they arguably look better that way because they are a lot more defined.

I found that neither a pair of overclocked 480s nor a 5870+5970 can handle running through trees/dense grass while staying above 60fps (2560 4xaa) if the shadows are maxed...it's just too much for some reason.
 
Finally got the water block installed.

GTX 470 @ 815/1630/1700:

Crysis Warhead:

DirectX 10 ENTHUSIAST 3X @ Map: avalanche @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 16.21 Max: 40.18 Avg: 28.38 ]

DirectX 10 ENTHUSIAST 3X @ Map: frost @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 20.43 Max: 31.83 Avg: 27.63 ]

DirectX 10 ENTHUSIAST 3X @ Map: ambush @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 19.93 Max: 33.72 Avg: 28.04 ]

max gpu temps 47C with just the comforting hum of 9 yate loons spinning at 1300rpm.
 
Last edited:
nitromullet said:
Finally got the water block installed.

GTX 470 @ 815/1630/1700:

Crysis Warhead:

DirectX 10 ENTHUSIAST 3X @ Map: avalanche @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 16.21 Max: 40.18 Avg: 28.38 ]

DirectX 10 ENTHUSIAST 3X @ Map: frost @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 20.43 Max: 31.83 Avg: 27.63 ]

max gpu temps 47C with just the comforting hum of 9 yate loons spinning at 1300rpm.
Click to expand...

25% overclock? nice! Should be at 5870 level of performance, mabe more.
 
Man you guys are getting screwed on your ram clocks =p. Though I'm not sure if there's much of a point in clocking the ram very high, I haven't benchmarked it.

Grabbed a 470 last week to try out. I have not actually tried to find the max stable clocks, I see no reason to at this point lol.

In afterburner:
1. Voltage 1.05
2. Shader Clock: 800/1600
3. RAM 2000

4. Fan speed with a custom profile that ramps it up slightly more aggressively once it gets toward 90C. Card is usually right at 80C in gaming. Fine temp imho.

I couldn't believe how well it overclocks, lol. Like I said, I haven't even attempted to try to find what it's max clocks are.

The weird thing is that I swear that when I immediately got it and used it at stock voltages and clocks the temps would get up into the mid 90's easily. Even with the fan at 70%! Now the fan is still right around 70%, the clocks and voltages are higher, and temps are a lot lower. I think the thermal paste on at least some of these cards really needs to sit under heat and pressure for a little bit and "cure" before it works well.
 
ashishmishra said:
--------------------------------------------------------------------
Unigine 2.0

Settings
Render: direct3d11
Mode: 1680x1050 4xAA fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 4x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
Replication: disabled
Tessellation: extreme
Click to expand...

I ran this, but it's v1.0... when I have time I'll run some others for comparrison.

FPS: 37.9
Scores: 955
1680x1050 4xAA
Shaders: High
Textures: High
Filter: Trilinear
Anisotroph\y: 4x
Occulsion: Envabled
Refraction: Enabled
Volumetric: Enabled

This is with a 950MHz 5870 and a 3.6GHz Phenom II. Not sure how this compares, I'll get v2 later and try again.

Downloadig now... 145KB a second, going to be a bit. 🙂
 
Last edited:
SlowSpyder said:
I ran this, but it's v1.0... when I have time I'll run some others for comparrison.

FPS: 37.9
Scores: 955
1680x1050 4xAA
Shaders: High
Textures: High
Filter: Trilinear
Anisotroph\y: 4x
Occulsion: Envabled
Refraction: Enabled
Volumetric: Enabled

This is with a 950MHz 5870 and a 3.6GHz Phenom II. Not sure how this compares, I'll get v2 later and try again.

Downloadig now... 145KB a second, going to be a bit. 🙂
Click to expand...

Not bad, but IMO the V2.0 hits the 58xx harder on the extreme tessellation setting, probably won't be a fair comparison. However I would be interested in the rest of your scores when you get the time 🙂
 
nitromullet said:
Finally got the water block installed.

GTX 470 @ 815/1630/1700:

Crysis Warhead:

DirectX 10 ENTHUSIAST 3X @ Map: avalanche @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 16.21 Max: 40.18 Avg: 28.38 ]

DirectX 10 ENTHUSIAST 3X @ Map: frost @ 0 1920 x 1200 AA 4x
==> Framerate [ Min: 20.43 Max: 31.83 Avg: 27.63 ]

max gpu temps 47C with just the comforting hum of 9 yate loons spinning at 1300rpm.
Click to expand...

Nice clocks and nice temps, tempted......but say no to water 😀 Too much effort.
 
ashishmishra said:
Not bad, but IMO the V2.0 hits the 58xx harder on the extreme tessellation setting, probably won't be a fair comparison. However I would be interested in the rest of your scores when you get the time 🙂
Click to expand...

Ok, same thing ran on v2.0. I got 21.3FPS with extreme tesselation.

I honestly don't believe that games will use that level of tesselation, it's pushed really high as this is a synthetic bench. But, the numbers are what the numbers are, 21.3FPS in this bench with my rig. It gives us an idea how the different GPU's scale.
 
extra said:
Man you guys are getting screwed on your ram clocks =p.
Click to expand...

Yes, we are... The RAM on my card is crap.

ashishmishra said:
Nice clocks and nice temps, tempted......but say no to water 😀 Too much effort.
Click to expand...

Yeah, it was a massive pain too. There is an issue with the block I'm using that shorts out the card when you use the included reinforcement bracket... The solution is just to run the block without the bracket, but you only find that out after you install, fill, and bleed the system and pull your hair out trying to figure out what you did wrong.

It works like a champ once you get it properly installed though.
 
SlowSpyder said:
Ok, same thing ran on v2.0. I got 21.3FPS with extreme tesselation.

I honestly don't believe that games will use that level of tesselation, it's pushed really high as this is a synthetic bench. But, the numbers are what the numbers are, 21.3FPS in this bench with my rig. It gives us an idea how the different GPU's scale.
Click to expand...

Lets level the field then, I ran 2.0 1680x1050 all maxed 4xAA 4xAF with No Tessellation and here the are scores:

GTX 470 @ 800-1600-1675
FPS: 58.9
Scores: 1485
Min FPS: 31.2
Max FPS: 118.4

I am interested to see how your 5870 does.

Edit:

GTX 470 @ 800-1600-1700
FPS: 59.3
Scores: 1493
Min FPS: 31.4
Max FPS: 117.9
 
Last edited:
Wow thats one bad ass gtx 470 you have there. (800-1600-1675-1.087V). Wow would love to see what it does with 1.25 volts. Is'nt that the maximum?
 
happy medium said:
Wow thats one bad ass gtx 470 you have there. (800-1600-1675-1.087V). Wow would love to see what it does with 1.25 volts. Is'nt that the maximum?
Click to expand...

Actually MSI Afterburner maxes out at 1.087V. On air I think that 1.087V itself is the safe max, on water 1.25V would be certainly nice.

I would have done a step-up to a GTX 480 but at these clocks I am as fast as a GTX 480, yet my idle power consumption is lower and I paid $150 less. Unless they launch a 512 Shader (GTX 485?) in the next 7 weeks, I am all set, especially for my resolution.
 
ashishmishra said:
Lets level the field then, I ran 2.0 1680x1050 all maxed 4xAA 4xAF with No Tessellation and here the are scores:

GTX 470 @ 800-1600-1675
FPS: 58.9
Scores: 1485
Min FPS: 31.2
Max FPS: 118.4

I am interested to see how your 5870 does.

Edit:

GTX 470 @ 800-1600-1700
FPS: 59.3
Scores: 1493
Min FPS: 31.4
Max FPS: 117.9
Click to expand...

Same setting as you, 63.3FPS average. Score: 1720. Min FPS: 33.0 Max FPS: 141.0

Clearly it would seem that the new GeForce parts are more robust as far as tesselation goes. How that will translate to real world performance waits to be seen, it could be a big deal in some games, the Radeons may be 'good enough' for how tesselation is implemented in games. I guess we'll have to wait and see.

Can you do higher than 1680x1050 res? I am able to run 1920x1200. Maybe some 8xAA runs?

I ran it at 1680x1050 8xAA/16xAF normal tesselation. Some how my print screen didn't work, but the score was 23.3FPS if I recall correctly.

Just one more data point. Also, I don't know how much CPU factors into this test, it at all. But my 3.6GHz Phenom II is obviously a bit behind what you have
 
ashishmishra said:
Actually MSI Afterburner maxes out at 1.087V. On air I think that 1.087V itself is the safe max, on water 1.25V would be certainly nice.

I would have done a step-up to a GTX 480 but at these clocks I am as fast as a GTX 480, yet my idle power consumption is lower and I paid $150 less. Unless they launch a 512 Shader (GTX 485?) in the next 7 weeks, I am all set, especially for my resolution.
Click to expand...

I'm somewhat suprised that they weren't able to release a 5123SP part... that is what they stated the specs were back around January. They obviously don't mind going well past 'normal' power use specs, so I'm somewhat suprised that we don't have a 512SP part. I guess you could call it 'unmanufacturable'. 😉 j/k
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top