Skip to main content

GeForce GTX 650 Ti Boost Review: A Budget-Oriented GK106-Based Boss

Results: OpenGL

Why OpenGL?

Recently, we've had a number of readers request OpenGL benchmarks to complement our DirectX-based games. So, today we're adding them to the test suite. In order to get an idea of how these cards handle 2D and 3D in a professional environment, we're subjecting them to the Cadalyst 2012 metric using a fully-patched and current version of AutoCAD 2013. Then, we enlist the help of various Unigine benchmarks running in OpenGL mode to gauge gaming performance.

AutoCAD 2013

AMD‘s cards do surprisingly poorly in the 3D AutoCAD test, considering their strong DirectX performance in our gaming benchmarks. Up and down the line-up, they're soundly beaten by Nvidia's cards. The new GeForce GTX 650 Ti Boost falls right where we'd expect it.

The results aren't quite as polarizing in our 2D run, though Nvidia's GeForce GTX 660 still tops the Radeon HD 7850. Performance from the Radeon HD 7790 is particularly bad, though, hampered by a driver issue, perhaps.

Synthetic OpenGL Benchmarks

Unigine Heaven 4.0 requires no introduction, so we’ll skip right to the analysis. This time around, the cards finish in the order we would have expected based on the DirectX results. It also appears that the Radeons have an easier time handling the tessellation load (normal) than the GeForce boards.

The Sanctuary benchmark emphasizes other areas, and with tessellation out of the picture, the AMD boards fall behind in OpenGL performance. Here is a short rundown of this benchmark’s features:

  • Five dynamic lights
  • HDR rendering
  • Parallax occlusion mapping
  • Ambient occlusion mapping
  • Translucence
  • Volumetric light and fog
  • Particle systems
  • Postprocessing
  • Interactive experience with fly-through mode
  • Stereo 3D modes

Unigine Tropics paints a similar picture as Sanctuary, although it, too, focuses on other aspects:

  • Dynamic sky with light scattering
  • Live water with a surf zone and caustics
  • Special materials for vegetation
  • HDR rendering
  • Parallel split shadow map
  • Depth of field
  • Real-time ambient occlusion
  • Up to 2M polygons per frame
  • Simulation of changing light conditions
  • Interactive experience with fly-through mode

It looks like OpenGL performance is a sore spot for AMD‘s current Radeon cards, at least in Windows. Aside from the 2D test, where driver optimizations look like they're needed, the Radeon HD 7790 falls exactly where we'd expect it to within AMD's product stack.

  • barto
    Let the battle begin!
    Reply
  • Souv_90
    anandtech review is out for 650 ti boost...just faster than 7790 and slower than 7850 in every games tested only except bf3 and Shogun

    http://www.anandtech.com/show/6838/nvidia-geforce-gtx-650-ti-boost-review-

    (wonder why tom's review is swaying more in favour of kepler everytime I see)
    Reply
  • EzioAs
    This card pricing seems a little too good that it kinda makes me regret getting the GTX660 just a while ago. Damn it, I hate the Kepler series and their pricing :/

    Although I can't say the same about the power consumption and some part in me actually hoped and expected the card would come with a game bundle.
    Reply
  • Memnarchon
    Souv_90anandtech review is out for 650 ti boost...just faster than 7790 and slower than 7850 in every games tested only except bf3 and Shogunhttp://www.anandtech.com/show/6838 st-review-(wonder why tom's review is swaying more in favour of kepler everytime I see)Techpowerup said also that 650ti boost is faster than 7850.
    edit: hardwarecanucks / PCPerspective. A lot of sites agree with Tom's, 7850 and 650ti boost are about the same in performance.

    But the point is that is similar to 7850 while its cheaper. Now that's what I wanted to see when I was saying we need better performance for $$$. Now lets see AMD dropping its prices too so we can have a sweat war between them that profits our wallet :D.
    Reply
  • tomfreak
    It looks to me form the chart except the 7900. AMD GCN is are actually far more efficient here. Better compute + better performance/watt. So Nvidia taunting about their 'most efficient GPU' seems got defeated again. I wonder what if AMD make 7900 a 1.5x of 7800.

    I still do not like the idea of Odd memory configuration. 1GB/2GB @ 192bit. The last 512MB from 2GB model is only getting 64bit memory bandwidth. Why cant they offer a a proper 1.5GB version @ middle price $159?

    Tomshardware fail to bench 2GB 7850. I like to see that vs 2GB 650Ti Boost.
    Reply
  • AMDRadeonHD
    it beats the 7790, 192 bit+2gb, so awesome...
    Reply
  • pcperson7
    Watch out gtx 660!
    Reply
  • Henrik Olsson
    from reviews around the web the GTX 650 ti boost is trading blows with the 7850 very well. i just gotta say it took nvidia awhile to get back into the game after AMD lowerd the pricing of there cards. good to see some competion again insted of just going "get the AMD card they are cheaper"
    Reply
  • de5_Roy
    darn it. i was enjoying reading the review considering i've been waiting to se how this card performs ever since i heard of it. 192bit interface with a cut-down gk106 under $170 and sli - makes gtx 650ti irrelevant and puts a good deal of hurt on the 128bit 7790 in terms of config. i doubt gtx650tiboost can outperform a 7790 in compute tasks but this is a gaming gfx card above all. this seems like a good proposition for budget 1080p gaming, so far.

    now, if amd brings back 7850 1gb, it'd be awesome (i don't buy the 'lack of gddr5 chips' excuse for a second).

    will there be gtx650tiboost sli performance analysis vs 7790 cfx and 7850 1gb cfx? hopefully nvidia has good drivers available because the sub $200 gfx card price war just got heated up.

    edit: imho nvidia is late to the sub $200 gfx card price war. this card shoulda come out instead of gtx 650ti (128bit). amd should be updating with new gcn cards soon while kepler 2.0 seems nowhere in sight. better late(!) than never, i guess..
    Reply
  • CrisisCauser
    The 7850 has a 256-bit bus, not 128-bit. Please fix your chart on page 1.
    Reply