Factory OC graphics card worth the price premium?

[iGas]

IMHO, it doesn't.

I took my GTX 460 from 811MHz to 1001MHz and saw a measly 3.15% gain in Unigine score, minimum frame rate gone from 2.2FPS to an impressive 13.6FPS, and maximum frame rate went down to 76.5 from 76.7.

GTX 460 Hawk talon attack, Phenom 940BE @ 3.0 GHz, 4GB DDR2, Windows 7.

Unigine

Heaven Benchmark v2.1

Render: direct3d11
Mode: 1920x1200 fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 4x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
Replication: disabled
Tessellation: normal

GTX 460 811MHz factory OC

FPS: 34.0
Scores: 857
Min FPS: 2.2
Max FPS: 76.7


GTX 460 1001MHz

FPS: 35.1
Scores: 884
Min FPS: 13.6
Max FPS: 76.5

 

[happy medium]

What? Bench some games at 675 core (stock gtx460) vs 811 core , then bench from 811 core to 1000 core. I bet you framerates go through the roof.

You might want to overclock that cpu some more also.

 

[ModestGamer]

What? Bench some games at 675 core (stock gtx460) vs 811 core , then bench from 811 core to 1000 core. I bet you framerates go through the roof.

You might want to overclock that cpu some more also.

Just becuase you overclock something doesn't mean it will gain tremendous performance in some cases it may fall off.

 

[happy medium]

Just becuase you overclock something doesn't mean it will gain tremendous performance in some cases it may fall off.

Oh really? and which case is that?

The only overclock I ever seen less performence with is my current cards memory overclock because of the ecc memory, never with a core overclock.

The gtx460 will see a performance increase up to 940 core according to reviews I've seen. I see no reason why 1000 core would be any different.

 

[ModestGamer]

Oh really? and which case is that?

The only overclock I ever seen less performence with is my current cards memory overclock because of the ecc memory, never with a core overclock.

The gtx460 will see a performance increase up to 940 core according to reviews I've seen. I see no reason why 1000 core would be any different.

Instead of a canned answer, try researching it. It can and does happen. Not from the standpoint your thinking but becuase you can't keep component ratios in check and certain portions will out run one or the other cuasing a fall off.

It takes tweaking to get good performance out of a overclock.

 

[notty22]

I have run all sorts of o/c benchmarks myself on a single gtx 460, from 725mhz up to 930mhz and I saw gains equal to the o/c scale all the way up. I'm running a i5 750 /1080p.
Also there are many reviews showing this. There is something in the op's rig bottlenecking performance.

I think I can replicate some numbers. @725mhz I got 33.4
I will o/c 1 card and get a number.

 

[ModestGamer]

I have run all sorts of o/c benchmarks myself on a single gtx 460, from 725mhz up to 930mhz and I saw gains equal to the o/c scale all the way up. I'm running a i5 750 /1080p.
Also there are many reviews showing this. There is something in the op's rig bottlenecking performance.

I think I can replicate some numbers. @725mhz I got 33.4
I will o/c 1 card and get a number.

What was the performance increase from 725 to 930 ? it should be around 20%

 

[happy medium]

What was the performance increase from 725 to 930 ? it should be around 20%

No card gets perfect overclocking scaling. Can you please stop derailing this thread and help the op?

 

[Termie]

OP - what's your memory overclock for each core clock setting? You may be getting negative returns due to error correction that are cancelling out the core gains. Unigine should not be CPU-limited, so your processor isn't the problem.

 

[notty22]

Here is the results.
I ran the o/c at 900/980 stock 725/900

 

[ModestGamer]

so at best maybe a 10-12% increase in performance ? hows that work when you goto say 1ghz ?

My point is that if you need more performance, buty the right card to start with. unless your a gifted OC'er and you willing to try a zillion different configs of memory and gpu speeds, you might never get good performance from a overclock.

Not to mention the dramatic rise in power consumption that comes with higher clocks and more heat.

 

[ModestGamer]

No card gets perfect overclocking scaling. Can you please stop derailing this thread and help the op?

Should I tell him to a buy a OC product ? whats your prefernece ? Should we recomend that ?

 

[notty22]

I think thats about 20%, gain. In my example for 24% o/c
You see the OP gained in Min FPS, big time.
I would say its a combination of cpu speed / memory speed, thats bottlenecking fps gains.

another example:

The overclocks of our card are 850 MHz core (17% overclock) and 1055 MHz Memory (17% overclock). Very nice results here. We have seen core overclocks well over 800 MHz from all the GTX 460 cards that we reviewed today.
The actual 3D performance gained from overclocking is 15.3%.
http://www.techpowerup.com/reviews/MSI/GeForce_GTX_460_Cyclone_OC_1_GB/34.html

 

[iGas]

OP - what's your memory overclock for each core clock setting? You may be getting negative returns due to error correction that are cancelling out the core gains. Unigine should not be CPU-limited, so your processor isn't the problem.

1950MHz factory default memory clock.

CPU at 3.0GHz & GTX 460 at 675MHz core clock:

FPS: 30.1
Scores: 759
Min FPS: 14.0
Max FPS: 65.8


CPU at 3.4GHz & GTX 460 at 811MHz core clock:

FPS: 35.2
Scores: 887
Min FPS: 13.5
Max FPS: 76.6

It look like Unigine bench is some what CPU dependence, because 3.4GHz 940BE & 811MHz GTX 460 equal performance of 3.0GHz 940BE & 1001MHz GTX 460.

It could very well be that my CPU is under power for the OC GTX 460.

 

[ModestGamer]

I think thats about 20%, gain. In my example for 24% o/c
You see the OP gained in Min FPS, big time.
I would say its a combination of cpu speed / memory speed, thats bottlenecking fps gains.

15% is stretching it.

 

[happy medium]

I think thats about 20%, gain. In my example for 24% o/c
You see the OP gained in Min FPS, big time.
I would say its a combination of cpu speed / memory speed, thats bottlenecking fps gains.

Max fps not rising is usually a good indicator of your cpu not being fast enough.

@ modestgamer
Yes, I do recommend a factory overclocked card if the price is right or if your going for maximum overclocks.

My point was games are a good indication of performance gains not synthetic benchmarks.

With the gtx460 a 20% overclock usually nets your 16% more performance.

 

[apoppin]

IMHO, it doesn't.

I took my GTX 460 from 811MHz to 1001MHz and saw a measly 3.15% gain in Unigine score, minimum frame rate gone from 2.2FPS to an impressive 13.6FPS, and maximum frame rate went down to 76.5 from 76.7.

GTX 460 Hawk talon attack, Phenom 940BE @ 3.0 GHz, 4GB DDR2, Windows 7.

Unigine

Heaven Benchmark v2.1

Render: direct3d11
Mode: 1920x1200 fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 4x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
Replication: disabled
Tessellation: normal

GTX 460 811MHz factory OC

FPS: 34.0
Scores: 857
Min FPS: 2.2
Max FPS: 76.7


GTX 460 1001MHz

FPS: 35.1
Scores: 884
Min FPS: 13.6
Max FPS: 76.5

Something is way wrong. Perhaps your OC is not stable. it is pretty hard to get 1000MHz on the core. Your CPU is also pretty slow. Increase the details and add AA to give your video card more of a workout.

For my benching with GeForce 262.99, benching Heaven 2.1 at 1920x1200 with all setting at maximum including 2xAA/16xAF, shaders high and tessellation extreme, with PC in sig and GTX 460 (Galaxy OC and EVGA FTW)

GTX 460 (675/1800MHz) - 23.0
GTX 460 (810/2000MHz) - 27.2
GTX 460 (850/2000MHz) - 28.6

 

[happy medium]

1950MHz factory default memory clock.

CPU at 3.0GHz & GTX 460 at 675MHz core clock:

FPS: 30.1
Scores: 759
Min FPS: 14.0
Max FPS: 65.8


CPU at 3.4GHz & GTX 460 at 811MHz core clock:

FPS: 35.2
Scores: 887
Min FPS: 13.5
Max FPS: 76.6

Compared to your original benchmmarks, it looks like this benchmark is very inconsistant.