MSI N460GTX Hawk (1GB) Overclocking Results

[profJim]

OK Darkje, I'll take up your "challenge".
Clearly your GTX 285 (GDDR3) at 756/1620/1350Mhz. Core/Shader/Mem setup beats my brute force approach with my GDDR5 460 Hawk, but hopefully only for a little while longer. Let me check, your gpu o/c is 16.7% and your memory/shader o/c is 9.5%. OK, that means that I'll try ...... oh wait a minute, I think that a saw several reviews where the 285 bested the 460 in a couple of benchmarks.

Never mind.

[Darkje]

It's the 512bit bus saving the day for the old 285, but it's ageing fast these days, ie. no dx11. I've been looking to replace it with a 460perhaps, but I'm not convinced yet, maybe I'll wait one more round.

[profJim]

Hmmm..... a 460 SLI setup seems to be one of the best setups for the price.
Does your mobo have 2 pci-e slots? A new 460 for your primary card and your 285 for PhysX.
But what do I know, I just updated from a 7600GT to my 460 Hawk.
Come to think of it, I think that I flunked physx in high school, or may it was spelling. :)

[Darkje]

Yeah that was what I was thinking of... the mobo has the slots and is sli/xfire certified, but will run pci-e at 2x8 I think. I could try and pick up a 2nd hand 285 too, it will be cheap and fast for dx9&10 but still no dx11...

On the other hand, budget is tight atm., not doing much gaming lately, and seeing sandybrige on the the horizon... mmm... choices, choices...

[profJim]

Well it looks like my system will kick your system's butt when it comes to DX11 benchmarks, for a while at least....

[Darkje]

Yes in that department you will be faster for now. I kind of expected the 460 to do roughly the same as the 285, but it seems like it's a little bit slower on the stuff we can compare.

Have to add the x3tc bench just loves cpu cycles, so I guess that is where some of the difference came from.

[profJim]

OK, I'm totally humbled by graphics overclocking.
It seemed too good to be true that my first stable efforts were a bit better than many of the Hawk reviews.
Apparently overclocking and testing with DX9 is relatively easy, DX10 and DX11 are far more picky.

I won't bore you with my trials and tribulations.
Enclosed are four DX11 Heaven benchmark results.
The first three are with stock Hawk settings; 780MHz gpu and 3600MHz (QDR) memory.
The last one uses 850MHz gpu with no voltage increase and 4000MHz (QDR) with a +20mv voltage increase.
Note the different tessellation settings in each test. All other test settings are the same.
You will probably need to copy the picture to your image viewing program to easily read the information.

[Darkje]

that's like a 20% hit on the tesselation from disabled to moderate ...
but nice score on the last one with it disabled...

[profJim]

Yep, I didn't test with tessellation = EXTREME, 'cause that probably would have lowered the score to about 30 FPS with the stock settings.
When the benchmark is running with tessellation enabled, there is a tessellation button at the top of screen, that presents 3 adjustable sliding bars that can be changed. I don't even know what tessellation means or does, let alone mucking with some settings that I know even less about.

I think that it's about time that I do a little reading about shaders and other graphics stuff, even though I don't do gaming.

Do I detect some tessellation envy?

**EDIT** Come to think of it, I paid twice as much for my Hawk card as I paid for my E6300 cpu.
My 2MB of cpu L2 cache certainly isn't helping with my benchmark scores. :(

[profJim]

Here are the Heaven Benchmarks for gpu = 850MHz and memory = 4000MHz (QDR) for DX9, DX10, and DX11.

Additional benchmarks for 875, 900, and 925MHz will follow over the next few days.