The Test

For this review we’re using the latest drivers from both NVIDIA and AMD. NVIDIA’s launch drivers are 334.69, which add support for the GTX 750 series but are otherwise identical to the 334.67 drivers currently available as a public beta for existing cards. The release 334 drivers include a fairly impressive tune up of NVIDIA’s OpenCL stack, so we’re finding that OpenCL performance is significantly improved in some of our benchmarks, which helps to close the gap with AMD. On the other hand these tune-ups have not come bug free (as Ganesh has found), and while our compute benchmarks are fine, Ganesh has run into some issues with some of his OpenCL based video utilities. Other than that, we have not encountered any stability problems with these drivers.

Meanwhile for AMD’s cards we’re using the recently released Catalyst 14.1 beta v1.6 drivers.

CPU: Intel Core i7-4960X @ 4.2GHz
Motherboard: ASRock Fatal1ty X79 Professional
Power Supply: Corsair AX1200i
Hard Disk: Samsung SSD 840 EVO (750GB)
Memory: G.Skill RipjawZ DDR3-1866 4 x 8GB (9-10-9-26)
Case: NZXT Phantom 630 Windowed Edition
Monitor: Asus PQ321
Video Cards: NVIDIA GeForce GTX 750 Ti
Zotac GeForce GTX 750 Ti
Zotac GeForce GTX 750
NVIDIA GeForce GTX 660
NVIDIA GeForce GTX 650 Ti Boost
NVIDIA GeForce GTX 650 Ti
NVIDIA GeForce GTX 650
NVIDIA GeForce GTX 640
NVIIDA GeForce GTX 550 Ti
AMD Radeon R7 265
AMD Radeon R7 260X
AMD Radeon R7 250
AMD Radeon HD 7750
AMD Radeon HD 5770
Video Drivers: NVIDIA Release 334.67 Beta
NVIDIA Release 334.69 Beta (GTX 750 Series)
AMD Catalyst 14.1 Beta v1.6
OS: Windows 8.1 Pro

 

HTPC Aspects : Miscellaneous Factors Metro: Last Light
Comments Locked

177 Comments

View All Comments

  • kwrzesien - Tuesday, February 18, 2014 - link

    Seems coincidental that Apple is going to use TSMC for all production of the A8 chip with Samsung not ready yet, maybe Apple is getting priority on 20nm? Frankly what nVidia is doing with 28nm is amazing, and if the yields are great on this mature process maybe the price isn't so bad on a big die. Also keep in mind the larger the die the more surface area there is to dissipate heat, Haswell proved that moving to a very dense and small die can create even more thermal limitations.
  • DanNeely - Tuesday, February 18, 2014 - link

    Wouldn't surprise me if they are; all the fab companies other than Intel are wailing about the agonizingly high costs of new process transitions and Apple has a history of throwing huge piles of its money into accelerating the build up of supplier production lines in trade for initial access to the output.
  • dylan522p - Tuesday, February 18, 2014 - link

    Many rumors point to Apple actually making a huge deal with intel for 14nm on the A8.
  • Mondozai - Wednesday, February 19, 2014 - link

    Maybe 14 for the iPhone to get even better power consumption and 20 for the iPad? Or maybe give 14 nm to the premium models of the iPad over the mini to differentiate further and slow/reverse cannibalization.
  • Stargrazer - Tuesday, February 18, 2014 - link

    So, what about Unified Virtual Memory?
    That was supposed to be a major new feature of Maxwell, right? Is it not implemented in the 750s (yet - waiting for drivers?), or is there currently a lack of information about it?
  • A5 - Tuesday, February 18, 2014 - link

    That seems to be a CUDA-focused feature, so they probably aren't talking about it for the 750. I'm guessing it'll come up when the higher-end parts come out.
  • Ryan Smith - Thursday, February 20, 2014 - link

    Bingo. This is purely a consumer product; the roadmaps we show are from NV's professional lineup, simply because NV doesn't produce a similar roadmap for their graphics lineup (despite the shared architecture).
  • dragonsqrrl - Tuesday, February 18, 2014 - link

    "Meet The Reference GTX 750 Ti & Zotac GTX 750 Series"

    "This is the cooler style that most partners will mimic, as the 60W TDP of the GTX 650 Ti does not require a particularly large cooler"

    You mean 750 Ti right?
  • chizow - Tuesday, February 18, 2014 - link

    The performance and efficiency of this chip and Maxwell is nothing short of spectacular given this is still on 28nm. Can't wait to see the rest of the 20nm stack.

    Ryan, are you going to replace the "Architectural Analysis" at some point? Really looking forward to your deep-dive on that, or is it coming at a later date with the bigger chips?
  • dgingeri - Tuesday, February 18, 2014 - link

    In the conclusion, the writer talks about the advantages of the AMD cards, but after my experiences with my old 4870X2, I'd rather stick with Nvidia, and I know I'm not alone. Has AMD improved their driver quality to a decent level yet?

Log in

Don't have an account? Sign up now