A place for everything NVIDIA, come talk about news, rumours, GPUs, the industry, show-off your build and more. This Subreddit is community run and does not represent NVIDIA in any capacity unless specified. Helpful Links. Latest Driver: 417.35 WHQL.
Guides. Other Links. Tune in to NVIDIA's 'The AI Podcast':. / /.
Rules. No Tech Support posts - Use the stickied Tech Support and QA thread for any tech support and Q&A posts. No referral links. No self-advertising. Please flair accordingly.
Low quality posts will be removed. This includes memes & fake news. News/Review/Benchmarks related posts should not have editorialized titles. For non-English submission, please use generic titles for ease of searching.
GeForce GTX 750 Ti is a performance Graphics Card based on the first variant of the Maxwell architecture. Architecture The first variant of the Maxwell Architecture succeeds the Kepler Architecture.
Be respectful, be civil, be nice, no witch hunting, personal attacks, bashing, or mudslinging. No religion & politics talk.
Do not post derogatory references or other insults. No buying/selling/trading in this subreddit. Please visit for that. Duplicate news content will be removed.
Follow Related Subreddits. I recently did a very deeply thought out response on a post about the expected performance of the 1050 Ti. The 1050 Ti should be somewhere just above the performance of a GTX 770, which turns out to be about 4-5% faster than a 960. Here's my stupidly long and hopefully pretty accurate (within 5-10%) response. '7Gbps memory from Samsung, 4GB of GDDR5 on a 128 bit bus, 768 Cuda cores, and what looks to be a boost clock that will actually turn out to be around 1500MHz (with the way GPU Boost 3.0 works).
This card seems to be a very good looking entry level gaming solution. I expect this card to be a GTX 960 replacement, and to be more powerful than a GTX 770.
To come to this conclusion I did some quick and rough math comparing each cards performance score and Cuda core count to come up with a points per core system. All of these numbers are pulled from the Nvidia website (Cuda core count) and (graphics score). I found that as a card has fewer and fewer cores it has a better per core score. For instance, the Titan X Pascal has a graphics score of 13,820, and has 3,584 cores. This creates a score of 3.86 points per core. I divided the graphics score by the core count. The 1080 (score of 12,457 with 2,560 cores) has a per core score of 4.87.
The 1070 (score of 11,533 with 1,920 cores) has a per core score of 6.0. The 1060 (score of 9,223 with 1,280 cores) has a per core score of 7.21. The 1060 3G (score of 8,901 with 1,152 cores) has a per core score of 7.73.
I did the math to find the average difference between each cards per core number and found it to be 0.9675. I decided to remove the 3GB 1060 in the end as it skewed the number set and found that a very linear efficiency curve was unfolding. The 1080 gets 1.01 more points per core than the Titan X. The 1070 gets 1.13 more points per core than the 1080. The 1060 gets 1.21 more points per core than the 1070.
So, if we follow this curve, the 1050 Ti should get about 1.3 more points per core than the 1060, making its overall graphics score 6,536. Example: 768 cores multiplied by a per core score of 8.51 gives us 6,535.68, which I rounded to 6,536. This would put the 1050 Ti comfortably on top of the 770, with a spread of 394 points between them.' Now, this math doesn't take into account memory bandwidth, but then again the 1050 Ti isn't exactly meant for 4K gaming, so the memory bandwidth becomes far less of an issue. I personally run 650 Ti BOOST SLI, and my single card performance is only about 4-5% behind a 750 Ti.
If we look at my projected score of the 1050 Ti and compare it to the 750 Ti we will see that the 1050 Ti is expected to be about 78% faster than the 750 Ti, which doesn't even take into account the further improved compression algorithm Nvidia launched with the Pascal architecture. If this is taken into account, you're most likely looking at about a 85-90% performance improvement, which is essentially the performance you would get if you were actually able to put the 750 Ti in SLI. TL;DR You will see a massive boost in performance. I would say the 1060 if you can swing it. The 1050 Ti from 660 Ti would still be a decent increase in performance, but it definitely would be proportionally less than a 750 Ti to the 1050 Ti. The math works out to a 39% performance increase over the 660 Ti, which will probably be more like 45-50%, which in my opinion is just not quite worth it.
If we look at the math between the 660 Ti and the 6GB 1060 we see a different story altogether. A massive increase of 95% makes the 1060 what I would call a very worthy upgrade from the 660 Ti, and with the improved compression algorithm and higher efficiency of Pascal overall, you're probably looking at double the performance if not slightly more. I'm actually painfully waiting to see if the 1080 Ti will be a thing before the end of January, because if it isn't, I'm going to have to cave and do 1080 SLI.
I have my 4790K paired with an H105, 2133 ram, and an 850 PRO 512, but when I built my computer I just rolled the old graphics cards and HDD into it and was waiting to save up enough money to upgrade the graphics cards last, then I got let go from my seasonal job, and my dream of a graphics card upgrade was put on hold. Currently in the job market so I expect to have enough money for a 1080 Ti if they release it by the time they are proposed to.
The GTX 1060 has a 574 MHz higher core clock speed and 40 more Texture Mapping Units than the GTX 750 Ti. This results in the GTX 1060 providing 86.7 GTexel/s better texturing performance.
This still holds weight but shader performance is generally more relevant, particularly since both of these GPUs support at least DirectX 10. The GTX 1060 has a 574 MHz higher core clock speed and 32 more Render Output Units than the GTX 750 Ti. This results in the GTX 1060 providing 60.2 GPixel/s better pixeling performance. However, both GPUs support DirectX 9 or above, and pixeling performance is only really relevant when comparing older cards. The GTX 1060 was released over a year more recently than the GTX 750 Ti, and so the GTX 1060 is likely to have better driver support, meaning it will be more optimized for running the latest games when compared to the GTX 750 Ti.
The GeForce GTX 750 Ti has 640 Shader Processing Units and the GeForce GTX 1060 Gigabyte GTX G1 Gaming 6GB has 1280. However, the actual shader performance of the GTX 750 Ti is 937 and the actual shader performance of the GTX 1060 is 3126. The GTX 1060 having 2189 better shader performance and an altogether better performance when taking into account other relevant data means that the GTX 1060 delivers a massively smoother and more efficient experience when processing graphical data than the GTX 750 Ti. The GTX 1060 transistor size technology is 12 nm (nanometers) smaller than the GTX 750 Ti. This means that the GTX 1060 is expected to run slightly cooler and achieve higher clock frequencies than the GTX 750 Ti.
GPU Mini Review Mini Review Overview GeForce GTX 750 Ti is a performance Graphics Card based on the first variant of the Maxwell architecture. Architecture The first variant of the Maxwell Architecture succeeds the Kepler Architecture. Despite being also manufactured with a 28nm technology, it has an extremely large L2 Cache which allows NVIDIA to produce Graphics Cards with relatively small memory data transfer rates, without causing too much impact on the overall performance. Furthermore, the Shaders have been redesigned and are both more powerful and energy efficient. GPU It equips a GPU codenamed GM107-400-A2 which has 5 SM activated and thus offers 640 Shader Processing Units, 40 TMUs and 16 ROPs, on a 128-bit memory interface of fast GDDR5. The central unit runs at 1020MHz and goes up to 1085MHz, in Turbo Mode, while the memory clock operates at 1350MHz. Features DirectX 11.2 Support (11.0 Hardware Default) and support for GameStream, G-SYNC, GPU Boost 2.0, GeForce Experience, PhysX and other technologies.
Cooling Solution The Cooling Solution consists of a Single-Fan and is identical to the one seen on GeForce GTX 650. The Card is known to run very cool and has a maximum operating temperature of 95ÂșC.
Power Consumption With a rated board TDP of 60W, it requires at least a 300W PSU and relies entirely on the PCI Slot for power. Release Price Pricing is identical to the one seen on its predecessors - $149. Performance Benchmarks indicate its performance is largely better than its predecessor's but Radeon HD 7850 and GeForce GTX 650 Ti Boost Edition still offer better performance, due to their high memory bandwidth transfer rates. System Suggestions We recommend a strong processor (Intel Core i5 Quad Core/AMD FX Six Core) and 8GB of RAM for a system with GeForce GTX 750 Ti.
Overview The GeForce GTX 1060 Gigabyte GTX G1 Gaming is a Performance Graphics Card based on the Pascal Architecture. Architecture It equips a GPU Codenamed P106-400-A1 which is a First Gen Pascal GPU amd has 10 SM activated, offering 1280 Shader Processing Units, 80 TMUs and 48 ROPs. GPU The central unit runs at 1594MHz and goes up to 1809MHz, in Turbo Mode.
Memory The GPU accesses a 6GB frame buffer of Fast GDDR5, through a 192-bit memory interface, while the Memory Clock Operates at 2002MHz. Power Consumption With a rated board TDP of 120W, it requires at least a 400W PSU with one available 6-pin connector. Performance The GeForce GTX 1060 Gigabyte GTX G1 Gaming is a high level graphics card that can competently handle triple A titles from 2016. In full HD 1080p resolution it can confidently maintain 60+ frames per second on very high graphics settings. With the graphics adjusted down to High/Medium it can also handle between 50-60 fps on 1440p. System Suggestions GeForce GTX 1060 Gigabyte GTX G1 Gaming is best suited for resolutions up to and including 2560x1440.
We recommend a High-End Processor and 8GB of RAM for Optimal Performance. Recommended CPU Possible GPU Upgrades GPU Variants.