NVIDIA Performance Benchmarks - 10.8.3 - GeForce GT 640 to GTX 680

[tonymacx86]

NVIDIA Performance Benchmarks

OS X 10.8.3
GeForce GT 640 to GTX 680​

Brand	Device	Details	OpenGL (Heaven)	OpenCL (Luxmark)
EVGA	GeForce GT 640	2GB 02G-P4-2643-KR 901MHz CUDA Cores: 384	382	148
EVGA	GeForce GT 650	2GB 02G-P4-2653-KR 1202MHz (Factory Overclock) CUDA Cores: 384	665	213
EVGA	GeForce GT 650 Ti	1GB 01G-P4-3652-KR 1071MHz CUDA Cores: 768	673	359
EVGA	GeForce GTX 660	2GB 02G-P4-2662-KR 1046MHz / 1111MHz (Factory Overclock) CUDA Cores: 960	1215	487
EVGA	GeForce GTX 660 Ti	2GB 02G-P4-3660-KR 915MHz / 980MHz CUDA Cores: 1344	1115	612
EVGA	GeForce GTX 670	2GB 02G-P4-2670-KR 915MHz / 980MHz CUDA Cores: 1344	1537	661
ASUS	GeForce GTX 680	2GB GTX680-DC2O-2GD5 1019MHz / 1084MHz (Non-Reference Model) CUDA Cores: 1536	1618	750

OS X
Version 10.8.3
Build 12D78

Systems Used
Gigabyte Z77 Motherboards
Core i5/i7 CPUs
8/16GB RAM
Mac Pro 3,1 System Definition
Chimera 2.0.1 Bootloader
GraphicsEnabler=No

Benchmarking Applications
Unigine Heaven 4.0 - download
Luxmark 2.0 - download

Testing by
tonymacx86, MacMan, Stork, Beelzebozo



Comparisons to Macs

​

Brand	Device	Details	OpenGL (Heaven)	OpenCL (Luxmark)
Apple MacBook Pro 10,2	GeForce GT 650M	1GB 900MHz CUDA Cores: 384	377	156
Apple iMac 13,2	GeForce GTX 680MX	2GB 720MHz CUDA Cores: 1536	1211	483

 

[Stork]

NVIDIA Kepler Performance Benchmarks 10.8.3: GT 640 to GTX 680

My EVGA 650 Ti benchmarks above are accomplished on my Thunderball system: GA-Z77X-UP5 TH / i5-3570 / GTX 650 Ti. Monitor is a HP ZR24W IPS at 1920x1200; I didn't run the benchmarks at a resolution of 1920x1080, but, from previous testing, there should be no difference in the results.

 

[Stork]

NVIDIA Kepler Performance Benchmarks 10.8.3: GT 640 to GTX 680

BTW, Tom's Hardware reviewed the GTX 650 Ti Boost graphics card.
GeForce GTX 650 Ti Boost Review: A Budget-Oriented GK106-Based Boss

 

[eelhead]

NVIDIA Kepler Performance Benchmarks 10.8.3: GT 640 to GTX 680

Asus GT610-SL-1GD3-L
http://www.asus.com/Graphics_Cards/GT610SL1GD3L/

Heaven Benchmark but no LuxMark as the 610 gives a runtime error

 

[eelhead]

NVIDIA Kepler Performance Benchmarks 10.8.3: GT 640 to GTX 680

Zotac GTX 670
http://www.newegg.com/Product/Product.aspx?Item=N82E16814500266

This was set to my Apple Display

This was set to my display on Ultra

And this is what most mere mortals will use, though it is bumped up a bit for gaming tastes



GE=No

 

[toleda]

NVIDIA Kepler Performance Benchmarks 10.8.3: GT 640 to GTX 680

evga gt 640/02g-p4-2645-kr

 

[tonymacx86]

There's no way we could test all models and all clocks, but hopefully the table should provide a general performance comparison between recommended GeForce models.

 

[dtrvndrn]

Isn't it strange that the 670 beats the 680 at Heaven? You hadn't sync on by any chance? That 'd explain them hoovering around 60fps.

 

[MacMan]

Isn't it strange that the 670 beats the 680 at Heaven? You hadn't sync on by any chance? That 'd explain them hoovering around 60fps.

That's because the GTX 670 is in a system with an i7 and the GTX 680 is in a system with an i5.

 

[PJALM]

Actually the 670 is known to perform better for many games over the 680. I think the 680 is better for OpenCL enabled apps.