Joined
·
14,039 Posts
===================================================================================
TL;DR version:
My well-OC'd GTX670 basically matches my well OC'd SLI 470's in terms of performance, while consuming just over 1/2 as much electricity (total system power), runs 27C cooler, and does so UTTERLY SILENTLY. I am very well pleased with this card.
In the test of GTX470 SLI vs GTX670, 470SLI wins slightly in most cases, and holds all the wins that are of any significant amount (generally about 10%), but 670 wins a few test by a small margin. The setups are very close basically, but on average the 470's are probably about 7% faster.
Furthermore, the 670, SLI470, and 7970 (all OC'd), and stock 680, all share performance that is eerily close to one another. Not a lot separates any of them on any test EXCEPT Metro 2033 and Crysis/WH. In those tests, Arni's OC'd 7970 DESTROYS the nV offerings. Presumably these two games are heavily memory bandwidth restricted, and the 7970 has a huge advantage in that regard. Oh, also the 7970 crushes in Mafia 2 w/o PhysX. However that user used the Demo vs. my using the actual game, plus he's on a Sandy at 5.0GHz whereas I'm ... not. Unknown how these facts affect the scores.
After all is said and done, the final Clocks I've settled on are: 1337c/1750m at 1.162V. These are not what the tests were done at, however. See below for the actual clocks. They are a little slower than my final clocks, but oh well.
![thumb.gif]()
===================================================================================
GTX670 Unboxing Thread
This thread is intended to be a resource for folks trying to make a buying decision, in particular people who currently have Single Fermi cards who are going to be looking at the FLOOD of Fermi's I'm sure will soon be available on our Marketplace. Many I'm sure will be considering an upgrade, either to a Kepler, OR by adding a used Fermi to their existing rigs. There seems to be a lack of comparative benchmarks out there that compare SLI Fermi's to Kepler (aside from 590 results ... which actually should mirror these fairly closely)
670 (or 680) owners, please feel free to post your own MATCHING benchies to the thread. When I say matching I mean please RUN THE SAME SETTINGS, otherwise it doesn't mean anything. I've also decided to invite 7950 and 7970 owners to OC their cards as much as possible and contribute as well, so ... feel free everyone. But let there be NO FANBOY ARGUING on my thread!!!
Important Note for our AMD Contributors: Please note that not all bench runs done in the 470/670 comps below are AMD-compatible, for various reasons (due to PhysX, 32xCSAA, etc). I've marked each test that is NON-Compatible, and I'm in the process of adding a section to the end of the post here that shows my 670 scores using settings we can compare 'apples to apples' with
![thumb.gif]()
Rules to get added to OP: Post AT LEAST 4 of the benches available, at the specified settings, using normal OCN image upload procedures. Adding Spoiler tags and your system info for each (as seen below for Dimmy and Windfire's submissions below) is also greatly appreciated! And Please DO NOT CHERRY PICK which you submit based on your winning/losing
![wink.gif]()
My Current Setup:
i7 930 at 3.9GHz (195 x 20), Asus Rampage III Extreme, GTX670 at 1345c/3225m (1.162V) in PCI-Ex 2.0 (16x/16x)
NOTE: Some of the synthetics were run slightly higher (noted below) as I thought I was stable at 1350/3247 but Heaven had two artifacts pop up so I reduced it slightly for the game benches.
My Previous Setup:
i7 930 at 3.9GHz (195 x 20), Asus Rampage III Extreme, SLI GTX 470 at 800c/900m (1.087V) in PCI-Ex 2.0 (16x/16x)
Note that GTX470 cards at 800/1800 should provide performance comparable to the following: GTX590 (with a small OC), SLI 480 (with a small OC), SLI 570 (with a small OC), SLI 560ti 448 (with a small OC), SLI 560ti 384 (with a massive OC)
Benches to be Run:
3dMarkVantage (High Preset, PPU Off) *
3dMark11 v. 4.90 (Performance) *
Unigine Heaven 3.0 (All Maxed, 8xAA, Extreme Tess, etc) *
S.T.A.L.K.E.R. Call of Pripyat Benchmark (all settings maxed/High HBAO/10.1/Tessellation/CHS all on) *
Lost Planet 2 Demo Benchmark (All Maxed, 32xCSAA, Test B) *
Mafia 2 (all maxed) *
Metro2033 (All Maxed except ADoF is off cause I hate it)
Just Cause 2 (All Maxed/Everything On/32xCSAA, No Bokeh Filter, DARK TOWER benchmark)
Crysis (All Maxed, 16xQAA, Assault Harbor benchmark, 2nd run results) *
Crysis 2 (All Maxed, 4xAA, Edge AA, Central Park benchmark)
* Denotes benchmarks that are available free, either as a standalone benchmark, or part of game demo
Notes:
Where possible, I did them all them at 1920x1080 for you guys so more people can compare. All NVCP settings were left at Default.
==============================================================================
Temps and Power Usage Numbers
Below I compared temps and power usage for both setups, both at idle, and at full load. Number were observed during a Crysis bench, maxed out, for 10 loops. Power consumption for the 670 is just over HALF that of the SLI 470's. GPU temps of course are over 20C lower. Perhaps most notable is that the System (mobo temp) actually went DOWN while the bench was running, instead of increasing by almost 10C as it did with the 470's.
Idle (470 SLI)
CPU: 33C
System 31c
Watts: 231
GPU (Top): 45C
GPU (Bot): 41C
Idle (670)
CPU: 34C
System 30c
Watts: 185
GPU: 32C
Load (470 SLI)
CPU: 44C Avg
System 41C Avg
Watts: 718W (max)
GPU (Top): 86C
GPU (Bot): 83C
Load (670):
CPU: 44C Avg
System: 29C Avg
Watts: 380W (max)
GPU (Top): 63C
Screenie after the 670 Run:
I feel I should also add ... with the 470's, those were temps with the fans blasting at full. Not that loud with my coolers, but they were running full out. With the 670 ... you CANNOT HEAR the GPU running, even in the middle of a torture test. And my case is VERY quiet to begin with.
==============================================================================
3dMarkVantage (High Preset, PPU Off) *
GTX470 (800/1800)
GTX670 (1350c/3247m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
3dMark11 v. 4.90 (Performance) *
GTX470 (800/1800)
GTX670 (1350c/3247m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Unigine Heaven 3.0 (All Maxed, 8xAA, Extreme Tess) *
GTX470 (800/1800)
GTX670 (1350c/3240m) (Note: Had two artifacts at this clock, lowered clocks for rest of benches)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
S.T.A.L.K.E.R. Call of Pripyat Benchmark (all settings maxed/High HBAO/10.1/Tessellation/CHS all on) *
GTX470 (800/1800)
GTX670 (1345c/3225m) (Note: Split-Second hitch at beginning of first test)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Lost Planet 2 Demo Benchmark (All Maxed, 32xCSAA, Test B) *
NON AMD COMPATIBLE BENCH (32xAA) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800) (Woe to you, oh Earth and Sea ... for the Devil Sends the Beast With Wrath, Because He Knows That Time is Short!)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
Mafia 2 (all maxed) *
NON AMD COMPATIBLE BENCH (PhysX On) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800) (note: GPU usage is pathetic on this test ... PhysX implementation sucks)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
(this bench is from free steam demo, I don't have full game. Bret ran full game bench, so results may be incomparable).
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
(all maxed) - my CPU should be at 3.88GHz but somehow it is recognized as 3.21GHz
==============================================================================
Metro2033 (All Maxed except ADoF is off cause I hate it)
NON AMD COMPATIBLE BENCH (Advanced PhysX On) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
==============================================================================
Just Cause 2 (All Maxed/Everything On/32xCSAA, No Bokeh Filter, DARK TOWER benchmark)
NON AMD COMPATIBLE BENCH (32xAA, GPU H2O simulation on) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800) (note: for some reason the top card would only do about 85% usage, while the bottom stayed maxed ... probably hurting the score here)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
Crysis (All Maxed, 16xQAA, Assault Harbor benchmark, 2nd run results)
NON AMD COMPATIBLE BENCH (16QxAA not supported on AMD) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
Crysis 2 (All Maxed, 4xAA, Edge AA, Central Park benchmark)
GTX470 (800/1800)
GTX670 (1345c/3225m)
Dimmy's Results 1 (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Dimmy's Results 2 ((2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1245/3386 memory))
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
======================================================================================
======================================================================================
AMD-Compatible Benchmarks
*Section IN PROGRESS*
==============================================================================
Lost Planet 2 Demo Benchmark (All Maxed, 4xAA, Test B) *
GTX670 (1345c/3225m)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Mafia 2 (all maxed, Apex PhysX OFF) *
GTX670 (1335c/3500m) (used my final OC for this one)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Metro2033 (All Maxed except ADoF, Advanced PhysX both OFF)
GTX670 (1345c/3225m)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Just Cause 2 (All Maxed/4xAA, Except No Bokeh Filter, No GPU H2O Simulation, DARK TOWER benchmark)
GTX670 (1345c/3225m)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Crysis Warhead (All Maxed, 4xAA, Ambush, Frost, Avalanche Flythrough benchmarks (using WH bench tool made by Framebuffer) , use 2nd run results)
***Instead of Crysis I'm using WH cause the Crysis BM tool is totally unreliable, plus Crysis BN's on my chip ***
GTX670 (1325c/3600m) (sacrificed some core to crank the memory bandwidth)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1245/3375 memory))
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
TL;DR version:
My well-OC'd GTX670 basically matches my well OC'd SLI 470's in terms of performance, while consuming just over 1/2 as much electricity (total system power), runs 27C cooler, and does so UTTERLY SILENTLY. I am very well pleased with this card.
In the test of GTX470 SLI vs GTX670, 470SLI wins slightly in most cases, and holds all the wins that are of any significant amount (generally about 10%), but 670 wins a few test by a small margin. The setups are very close basically, but on average the 470's are probably about 7% faster.
Furthermore, the 670, SLI470, and 7970 (all OC'd), and stock 680, all share performance that is eerily close to one another. Not a lot separates any of them on any test EXCEPT Metro 2033 and Crysis/WH. In those tests, Arni's OC'd 7970 DESTROYS the nV offerings. Presumably these two games are heavily memory bandwidth restricted, and the 7970 has a huge advantage in that regard. Oh, also the 7970 crushes in Mafia 2 w/o PhysX. However that user used the Demo vs. my using the actual game, plus he's on a Sandy at 5.0GHz whereas I'm ... not. Unknown how these facts affect the scores.
After all is said and done, the final Clocks I've settled on are: 1337c/1750m at 1.162V. These are not what the tests were done at, however. See below for the actual clocks. They are a little slower than my final clocks, but oh well.
===================================================================================
GTX670 Unboxing Thread
This thread is intended to be a resource for folks trying to make a buying decision, in particular people who currently have Single Fermi cards who are going to be looking at the FLOOD of Fermi's I'm sure will soon be available on our Marketplace. Many I'm sure will be considering an upgrade, either to a Kepler, OR by adding a used Fermi to their existing rigs. There seems to be a lack of comparative benchmarks out there that compare SLI Fermi's to Kepler (aside from 590 results ... which actually should mirror these fairly closely)
670 (or 680) owners, please feel free to post your own MATCHING benchies to the thread. When I say matching I mean please RUN THE SAME SETTINGS, otherwise it doesn't mean anything. I've also decided to invite 7950 and 7970 owners to OC their cards as much as possible and contribute as well, so ... feel free everyone. But let there be NO FANBOY ARGUING on my thread!!!
Important Note for our AMD Contributors: Please note that not all bench runs done in the 470/670 comps below are AMD-compatible, for various reasons (due to PhysX, 32xCSAA, etc). I've marked each test that is NON-Compatible, and I'm in the process of adding a section to the end of the post here that shows my 670 scores using settings we can compare 'apples to apples' with
Rules to get added to OP: Post AT LEAST 4 of the benches available, at the specified settings, using normal OCN image upload procedures. Adding Spoiler tags and your system info for each (as seen below for Dimmy and Windfire's submissions below) is also greatly appreciated! And Please DO NOT CHERRY PICK which you submit based on your winning/losing
My Current Setup:
i7 930 at 3.9GHz (195 x 20), Asus Rampage III Extreme, GTX670 at 1345c/3225m (1.162V) in PCI-Ex 2.0 (16x/16x)
NOTE: Some of the synthetics were run slightly higher (noted below) as I thought I was stable at 1350/3247 but Heaven had two artifacts pop up so I reduced it slightly for the game benches.
My Previous Setup:
i7 930 at 3.9GHz (195 x 20), Asus Rampage III Extreme, SLI GTX 470 at 800c/900m (1.087V) in PCI-Ex 2.0 (16x/16x)
Note that GTX470 cards at 800/1800 should provide performance comparable to the following: GTX590 (with a small OC), SLI 480 (with a small OC), SLI 570 (with a small OC), SLI 560ti 448 (with a small OC), SLI 560ti 384 (with a massive OC)
Benches to be Run:
3dMarkVantage (High Preset, PPU Off) *
3dMark11 v. 4.90 (Performance) *
Unigine Heaven 3.0 (All Maxed, 8xAA, Extreme Tess, etc) *
S.T.A.L.K.E.R. Call of Pripyat Benchmark (all settings maxed/High HBAO/10.1/Tessellation/CHS all on) *
Lost Planet 2 Demo Benchmark (All Maxed, 32xCSAA, Test B) *
Mafia 2 (all maxed) *
Metro2033 (All Maxed except ADoF is off cause I hate it)
Just Cause 2 (All Maxed/Everything On/32xCSAA, No Bokeh Filter, DARK TOWER benchmark)
Crysis (All Maxed, 16xQAA, Assault Harbor benchmark, 2nd run results) *
Crysis 2 (All Maxed, 4xAA, Edge AA, Central Park benchmark)
* Denotes benchmarks that are available free, either as a standalone benchmark, or part of game demo
Notes:
Where possible, I did them all them at 1920x1080 for you guys so more people can compare. All NVCP settings were left at Default.
==============================================================================
Temps and Power Usage Numbers
Below I compared temps and power usage for both setups, both at idle, and at full load. Number were observed during a Crysis bench, maxed out, for 10 loops. Power consumption for the 670 is just over HALF that of the SLI 470's. GPU temps of course are over 20C lower. Perhaps most notable is that the System (mobo temp) actually went DOWN while the bench was running, instead of increasing by almost 10C as it did with the 470's.
Idle (470 SLI)
CPU: 33C
System 31c
Watts: 231
GPU (Top): 45C
GPU (Bot): 41C
Idle (670)
CPU: 34C
System 30c
Watts: 185
GPU: 32C
Load (470 SLI)
CPU: 44C Avg
System 41C Avg
Watts: 718W (max)
GPU (Top): 86C
GPU (Bot): 83C
Load (670):
CPU: 44C Avg
System: 29C Avg
Watts: 380W (max)
GPU (Top): 63C
Screenie after the 670 Run:
I feel I should also add ... with the 470's, those were temps with the fans blasting at full. Not that loud with my coolers, but they were running full out. With the 670 ... you CANNOT HEAR the GPU running, even in the middle of a torture test. And my case is VERY quiet to begin with.
==============================================================================
3dMarkVantage (High Preset, PPU Off) *
GTX470 (800/1800)
GTX670 (1350c/3247m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
3dMark11 v. 4.90 (Performance) *
GTX470 (800/1800)
GTX670 (1350c/3247m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Unigine Heaven 3.0 (All Maxed, 8xAA, Extreme Tess) *
GTX470 (800/1800)
GTX670 (1350c/3240m) (Note: Had two artifacts at this clock, lowered clocks for rest of benches)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
S.T.A.L.K.E.R. Call of Pripyat Benchmark (all settings maxed/High HBAO/10.1/Tessellation/CHS all on) *
GTX470 (800/1800)
GTX670 (1345c/3225m) (Note: Split-Second hitch at beginning of first test)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
==============================================================================
Lost Planet 2 Demo Benchmark (All Maxed, 32xCSAA, Test B) *
NON AMD COMPATIBLE BENCH (32xAA) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800) (Woe to you, oh Earth and Sea ... for the Devil Sends the Beast With Wrath, Because He Knows That Time is Short!)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
Mafia 2 (all maxed) *
NON AMD COMPATIBLE BENCH (PhysX On) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800) (note: GPU usage is pathetic on this test ... PhysX implementation sucks)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
(this bench is from free steam demo, I don't have full game. Bret ran full game bench, so results may be incomparable).
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
(all maxed) - my CPU should be at 3.88GHz but somehow it is recognized as 3.21GHz
==============================================================================
Metro2033 (All Maxed except ADoF is off cause I hate it)
NON AMD COMPATIBLE BENCH (Advanced PhysX On) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Windfire's Results (i7 [email protected], GTX570 SLI (799/1000), Vista 64-bit)
==============================================================================
Just Cause 2 (All Maxed/Everything On/32xCSAA, No Bokeh Filter, DARK TOWER benchmark)
NON AMD COMPATIBLE BENCH (32xAA, GPU H2O simulation on) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800) (note: for some reason the top card would only do about 85% usage, while the bottom stayed maxed ... probably hurting the score here)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
Crysis (All Maxed, 16xQAA, Assault Harbor benchmark, 2nd run results)
NON AMD COMPATIBLE BENCH (16QxAA not supported on AMD) - See the AMD Compatibility Section at Bottom.
GTX470 (800/1800)
GTX670 (1345c/3225m)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
==============================================================================
Crysis 2 (All Maxed, 4xAA, Edge AA, Central Park benchmark)
GTX470 (800/1800)
GTX670 (1345c/3225m)
Dimmy's Results 1 (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1137Mhz / 3005Mhz memory))
Dimmy's Results 2 ((2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1245/3386 memory))
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
======================================================================================
======================================================================================
AMD-Compatible Benchmarks
*Section IN PROGRESS*
==============================================================================
Lost Planet 2 Demo Benchmark (All Maxed, 4xAA, Test B) *
GTX670 (1345c/3225m)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
Lost Planet 2:
==============================================================================
Mafia 2 (all maxed, Apex PhysX OFF) *
GTX670 (1335c/3500m) (used my final OC for this one)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
Metro2033 (All Maxed except ADoF, Advanced PhysX both OFF)
GTX670 (1345c/3225m)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
Metro 2033:
==============================================================================
Just Cause 2 (All Maxed/4xAA, Except No Bokeh Filter, No GPU H2O Simulation, DARK TOWER benchmark)
GTX670 (1345c/3225m)
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
Just Cause 2:
==============================================================================
Crysis Warhead (All Maxed, 4xAA, Ambush, Frost, Avalanche Flythrough benchmarks (using WH bench tool made by Framebuffer) , use 2nd run results)
***Instead of Crysis I'm using WH cause the Crysis BM tool is totally unreliable, plus Crysis BN's on my chip ***
GTX670 (1325c/3600m) (sacrificed some core to crank the memory bandwidth)
Dimmy's Results (2500K @ 4.6Ghz, GTX 680 stock (max GPU boost 1245/3375 memory))
Arni's Results - 2600K @ 5.0GHz, HD7970 @1250/1700
