Date: August 24, 2015
Author(s): Rob Williams
NVIDIA rounds out the entry-level part of its Maxwell-based lineup with the GeForce GTX 950. As a successor to the GTX 750 Ti, the GTX 950 boosts performance significantly while draining just a wee bit more power. And as we’ll see, the ASUS STRIX model offers its own set of perks, including possible 0dB operation.
Some 18 months after the launch of the first-gen Maxwell card, GeForce GTX 750 Ti, we’re introduced to its successor: GeForce GTX 950. While the GTX 750 Ti impressed us with its ability to deliver solid 1080p game performance without the use of a power connector, the GTX 950’s much-improved performance couldn’t afford the same design. A bummer, but one that the card manages to make up for.
As has become a theme with mid-range and lower parts, NVIDIA is touting its GTX 950 as being an ideal choice for MOBA gamers who demand excellent performance and don’t want to break the bank. Fortunately, that latter goal isn’t too hard to pull off thanks to the very nature of most MOBA games – they’re isometric, limit what you can see, don’t usually strive for realism, and thus run quite well even on older cards.
With its GTX 950 launch, though, NVIDIA wants to draw attention to the fact that a card like this one could dramatically reduce latency in MOBAs and other games. It’s not just about framerates, but also how quickly clicks will result in an action ingame. The video below explains the effect well.
For those who can’t view the video (or don’t want to), NVIDIA shows a click made with the GTX 650 taking 107ms to register the action on the screen, while the value drops to less than half for the GTX 950. While it’s obvious that a faster graphics card is going to deliver smoother and more timely frames, there’s more to it than that here.
NVIDIA’s GeForce Experience suite allows gamers to optimize their games with one click, and within that click can be many different ingame (sometimes even unofficial) settings that are tweaked. Coupled with this and the GTX 950’s architectural design, the graphics pipeline is more optimized to both increase performance and reduce latencies, ultimately giving us the result that NVIDIA shows off in the video.
Considering the article has just begun, there’s been a fair bit of MOBA talk. While NVIDIA’s targeting the genre with its GTX 950, that’s just because the card can act like a relative powerhouse when it comes to many of the titles found in it. Ultimately, the GTX 950 is designed to offer an excellent 1080p experience for the card’s $159 price point. Does it deliver? We’ll soon find out.
First, let’s take a look at the GTX 950 iteration that hit our doorstep: ASUS’ STRIX edition. It’s somewhat appropriate that this is the model we received as both the GTX 960 and GTX 970 we reviewed for launch were also STRIX cards.
This STRIX model looks similar to the others, and carries with it similar goals. At the forefront, that includes 3x quieter operation, and if loads are really light, 0dB operation. Further perks include 5-phase Super Alloy Power II, a fancy way of saying the card uses very high-quality components, and GPU Tweak II, a performance tweaker that bundles in Xsplit Gamecaster.
Before we go further, here’s a roundup of specs for NVIDIA’s latest GeForce series, along with the last-gen GTX 750 Ti as it’s in effect the predecessor to the GTX 950.
|NVIDIA GeForce Series||Cores||Core MHz||Memory||Mem MHz||Mem Bus||TDP|
|GeForce GTX TITAN X||3072||1000||12288MB||7000||384-bit||250W|
|GeForce GTX 980 Ti||2816||1000||6144MB||7000||384-bit||250W|
|GeForce GTX 980||2048||1126||4096MB||7000||256-bit||165W|
|GeForce GTX 970||1664||1050||4096MB||7000||256-bit||145W|
|GeForce GTX 960||1024||1126||2048MB||7010||128-bit||120W|
|GeForce GTX 950||768||1024||2048MB||6600||128-bit||90W|
|GeForce GTX 750 Ti||640||1020||2048MB||5400||128-bit||60W|
Being built on the second-gen Maxwell architecture, the GTX 950 stands to be much better than the GTX 750 Ti even if the specs were kept exact. But to give things a further boost, we see an increase to the cores as well as to the memory speed. Both cards come equipped with a 128-bit memory bus.
As the lowest-end card to come out 11 months after the initial second-gen Maxwell cards, it’s not too hard to surmise what it will be capable of even before we get into testing. Once again, “excellent 1080p gaming”.
ASUS’ STRIX edition GTX 950 includes a total of four ports, including two DVI, a DisplayPort, and an HDMI port. In the first shot in the slide above, you can see that unlike the GTX 750 Ti, the GTX 950 supports SLI – a nice perk, especially if you are a MOBA gamer who wants to make the jump to the “sweet spot” 1440p resolution by simply buying a second card rather than a singular higher-end one.
STRIX emphasizes effective cooling, and shots above highlight that well. Looking at the card top-down, it’s easy to see that ASUS tried to give this card as much breathing-room as possible. It has two mammoth heatpipes coming out the top, and fins running along the entire surface. Also seen in these shots is the singular 6-pin power connector the card requires.
Remember how GPUs of this price point used to look? Yeah – I’m sure glad the times have changed that. This STRIX card looks great and will enhance any PC it’s installed into.
At Techgage, we strive to make sure our results are as accurate as possible. Our testing is rigorous and time-consuming, but we feel the effort is worth it. In an attempt to leave no question unanswered, this page contains not only our test-bed specifications, but also a detailed look at how we conduct our testing.
The below table lists our testing machine’s hardware, which remains unchanged throughout all GPU testing, minus the GPU. Each card used for comparison is also listed here, along with the driver version used.
|Graphics Card Test System|
|Processors||Intel Core i7-4960X – Six-Core @ 4.50GHz|
|Motherboard||ASUS P9X79-E WS|
|Memory||Kingston HyperX Beast 32GB (4x8GB) – DDR3-2133 11-12-11|
|Graphics||AMD Radeon R9 280X 2GB – Catalyst 13.12
AMD Radeon R9 285 2GB (MSI Twin Frozr IV) – Catalyst 14.30
AMD Radeon R9 290X 4GB – Catalyst 13.12
NVIDIA GeForce GTX 750 Ti 2GB – GeForce 334.89
NVIDIA GeForce GTX 770 2GB – GeForce 340.52
NVIDIA GeForce GTX 780 3GB – GeForce 331.93
NVIDIA GeForce GTX 780 Ti 3GB – GeForce 331.93
NVIDIA GeForce GTX 950 2GB (ASUS STRIX) – GeForce 355.65
NVIDIA GeForce GTX 960 2GB (EVGA SuperSC) – GeForce 350.12
NVIDIA GeForce GTX 970 4GB (ASUS STRIX) – GeForce 344.11
NVIDIA GeForce GTX 980 4GB – GeForce 344.07
|Storage||Kingston HyperX 240GB SSD|
|Power Supply||Cooler Master Silent Pro Hybrid 1300W|
|Chassis||Cooler Master Storm Trooper Full-Tower|
|Cooling||Thermaltake WATER3.0 Extreme Liquid Cooler|
|Displays||ASUS PB278Q 27″ 2560×1440
Dell P2210H 22″ 1920×1080 x 3
|Et cetera||Windows 7 Professional 64-bit|
Important Note: EVGA’s GTX 960 SuperSC was tested using a different motherboard (ASUS X99-DELUXE) and processor (Intel Core i7-5960X, overclocked to 4GHz). We don’t ordinarily change test platforms without retesting everything, but this is a special case, and overall, performance is extremely similar between the two platforms (based on 3DMark results). Once Windows 10 is released, and with stable graphics drivers, we will be performing a GPU test suite overhaul and retest all cards.
Notes About Our High-end System
The goal of our performance content is to show you as accurately as possible how one product compares to another – after all, you’re coming to us for advice, so we want to make sure we’re giving you the best possible information. Typically, one major step we take in ensuring that our performance results are accurate is to make sure that our test systems are void of all possible bottlenecks, so for that, high-end components must be used.
In the case of our graphics card test system, the processor chosen has six-cores and is overclocked far beyond reference clocks. Most games nowadays are not heavily CPU-bound, but by using such a chip, we feel that we completely rule it out as a potential bottleneck. The same can be said for the use of an SSD (as opposed to latency-ridden mechanical storage), and even our memory, which is clocked at the comfortable speed of DDR3-2133.
Why this matters to you: Our test PC is high-end, and it’s very likely that you’d encounter a bottleneck quicker than us. Our goals are to rid all possible bottlenecks, whereas yours is to build the PC you need. In our case, we need to go overboard to attain as accurate a representation of a graphic card’s performance as possible.
If your PC has at least a modern (~2-years-old) quad-core or better processor, and at least 8GB of fast memory (DDR3-1866+), that chances of you running into a bottleneck with today’s hottest game is admittedly low. If you’re using lower-end gear, you can absolutely expect that the rest of your system could be a bottleneck. It should be noted, though, that if you’re seeking out a lower-end graphics card, the importance of a bottleneck would of course be lessened.
Unfortunately, we’re not able to test a single card on multiple PC configurations; each single card we test takes at least 3 hours to test, with another 2 hours added on for each additional resolution, and at least another 1~2 hours for our Best Playable results (for up to 11 hours of mostly hands-on testing for a high-end model).
Please bear all of this in mind. If you’re unsure if your PC could prove to be a bottleneck, our comments section exists for such questions.
When preparing our test-beds for any type of performance testing, we follow these guidelines:
To aid with the goal of keeping accurate and repeatable results, we alter certain services in Windows 7 from starting up at boot. This is due to the fact that these services have the tendency to start up in the background without notice, potentially causing inaccurate test results. For example, disabling “Windows Search” turns off the OS’ indexing which can at times utilize the hard drive and memory more than we’d like.
The services we disable are:
For further fine-tuning, we also use Windows’ “Classic” desktop theme, which gets rid of the transparency that can sometimes utilize a GPU in the background.
Sometimes, either AMD or NVIDIA will work with a game studio to help their development process along. As history has proven, this often results in a game that is tuned better for one vendor over the other, although sometimes the tides can change over time, resulting in the competing vendor offering the better experience.
One of our goals is to provide as neutral a benchmarking suite as possible, so while it’s impossible to avoid games sponsored by either of these companies, we can at least make an effort to achieve a blended list. As it stands, our current game list and their partners are:
(AMD) – Battlefield 4
(AMD) – Crysis 3
(AMD) – Sleeping Dogs
(NVIDIA) – Assassin’s Creed IV: Black Flag
(NVIDIA) – Metro: Last Light
(NVIDIA) – Splinter Cell Blacklist
(Neutral) – GRID 2
(Neutral) – Total War: SHOGUN 2
With that, let’s move on to a quick look at the game settings we use in our testing:
Note: The “High” preset is used for multi-monitor configurations.
Note: The “Medium” preset is used for multi-monitor configurations.
Given the sheer number of titles in the Assassin’s Creed series, it’s a little hard to believe that the first game came out a mere seven years ago. You could definitely say that Ubisoft hit the ball out of the park with this one. To date, we’ve never considered an AC game for benchmarking, but given the number of graphical goodies featured in the PC version of Black Flag, that trend now ends.
Manual Run-through: The saved game starts us not far from the beginning of the game under a small church which can be climbed to synchronize with the environment. To kick things off, I scale this church and rotate the camera around once, making sure to take in the beautiful landscape; then, I climb back down and run all the way to the water (the top of this small church and the water can be seen in the above screenshot).
Note: For some reason, Ubisoft decided to cap the framerate to 60 FPS in Black Flag even if Vsync is turned off. For most games, this would ruin the chance of it appearing in our benchmarking, but because the game is graphically intensive, I’ve chosen to stick with it, as at higher resolutions, reaching 60 FPS is a perk that will belong only to high-end graphics cards.
As soon as I’m able, I’ll be overhauling our test suite to include fresher games, and Black Flag is a good example of why. Unfortunately, this game would still be suitable had Ubisoft not implemented an asinine frame-limiter. But I digress. The takeaway with this title is that the GTX 950 can handle a game that still looks really good even today at 1080p at very high detail levels. 60 FPS is ideal, but the GTX 950 is no slouch for not dipping below 52 FPS on the low-end. It’s also starkly faster than the GTX 750 Ti it replaces.
Thanks to the fact that DICE cares more about PC gaming than a lot of developers, the Battlefield series tends to give us titles that are well-worth benchmarking. Battlefield 3 offered incredible graphics and became a de facto benchmark immediately, so it’s no surprise, then, that BF4 follows right in its footsteps.
Manual Run-through: The Singapore level is the target here, with the saved game starting us on an airboat that must be driven to shore, where a massive battle is set to take place. I stop recording the framerate once the tank makes its way to the end of this small patch of beach; in all, the run takes about 3 minutes.
The solid performance continues with Battlefield 4. 37 min / 50 avg isn’t what I’d consider ideal, but this is Ultra settings we’re dealing with. A disabling of AO and AA would improve the situation tremendously.
When the original Crysis dropped in late 2007, it took no time at all for pundits to coin the phrase, “Can it run Crysis?“, almost to the point of self-parody. At the time, the game couldn’t have its graphics detail maxed-out on even top-of-the-line PCs, and in reality, that’s a great thing. I’d imagine few are opposed to knowing that a game could actually look better down the road as our PCs grow into them. As the series continued, Crytek knew it had a legend to live up to, and fortunately, Crysis 3 (our review) lives up to the original’s legacy.
Manual Run-through: There’s no particular level in Crysis 3 that I could establish was “better” for benchmarking than another, but I settled on “Red Star Rising” based on the fact that I could perform a run-through with no chance of dying (a great thing in a challenging game like this one). The level starts us in a derelict building, where I traverse a broken pipe to make it over to one rooftop and then another. I eventually hit the ground after taking advantage of a zipline, and make my way down to a river, where I scurry past a number of enemies to the end spot beneath a building.
At High detail, Crysis 3 manages to run better than Battlefield 4 on Ultra – although while the latter is at max detail, the former is not. Still, we’re continuing to see proof that this lowly $160 card can muster quite a bit of performance at this resolution.
For those who appreciate racing games that are neither too realistic nor too arcade-like, there’s GRID. In GRID 2 (review), the ultimate goal is to build a racing empire, starting from square one. Unlike most racing titles that have some sort of career, the goal here isn’t to earn cash, but fans. Whether you’re racing around Abu Dhabi’s Yas Marina or tearing through a gorgeous Cote d’Azur coastline, your goal is simple: To impress.
Manual Run-through: The track chosen for my benchmarking is Miami (Ocean Drive). It’s a simple track overall, which is one of the reasons I chose it, and also the reason I choose to do just a single lap (I crash, often, and that affects both the results and my patience). Unlike most games in the suite which I test twice over (save for an oddity in the results), I race this one lap three times over.
Unlike the other games up to this point, at max detail, GRID 2 runs super-smooth on the GTX 950.
Crysis has become infamous for punishing even top-end systems, but let’s be fair: The Metro series matches, if not exceeds its requirement for graphical horsepower. That was proven by the fact that we used Metro 2033 in our testing for a staggering three years – only to be replaced by its sequel, Last Light. I’m not particularly a fan of this series, but I am in awe of its graphics even at modest settings.
Manual Run-through: Because this game is a real challenge to benchmark with for both the reasons of variability in the results and the raw challenge, I choose to use the built-in benchmark here but rely on Fraps to give me more accurate results.
Note: Metro Last Light‘s built-in benchmark is not representative of the entire game; some levels will punish a GPU much worse than this benchmark will (namely, “The Chase”, which has lots of smoke and explosions). What this means is that while these settings might suffice for much of the game, there might be instances where the performance degrades enough during a certain chapter or portion of a chapter to force a graphics setting tweak.
It’s no longer a secret that the GTX 950 is much faster than the GTX 750 Ti – it’s a total domination.
Many have called Sleeping Dogs (our review) the “Asian Grand Theft Auto“, but the game does a lot of things differently that helps it stand out of the crowd. For example, in lieu of supplying the player with a gazillion guns, Sleeping Dogs focuses heavily on hand-to-hand combat. There are also many collectibles that can be found to help upgrade your character and unlock special fighting abilities – and if you happen to enjoy an Asian atmosphere, this game should fit the bill.
Manual Run-through: The run here takes place during the chapter “Amanda”, on a dark, dank night. The saved game begins us at the first apartment in the game (in North Point), though that’s not where I begin capturing the framerate. Instead, I first request our motorcycle from the garage. Once set, I begin recording the framerate and drive along a specific path all the way to Aberdeen, taking about two minutes.
Sleeping Dogs, despite having released in 2013, is still a great-looking game, and on the GTX 950, it manages to come close to 60 FPS on the minimum at 1080p at max detail. Console what?
Tom Clancy is responsible for a countless number of video games, but his Splinter Cell series has become something special, with each game released having been considered “great” overall. The latest in the series, Blacklist, is no exception, and thankfully for us, its graphics are fantastic, and not to mention intensive. For those who love a stealth element in their games, this is one that shouldn’t be skipped.
RIP, Tom Clancy.
Manual Run-through: From the start of the ‘Safehouse’ level in Benghazi, Libya, we progress through until we reach an apartment building that must be entered – this is where we end the FPS recording.
Blacklist is yet another game that looks fantastic but still runs great on the lowly GTX 950. We’re talking total hoverage around the 60 FPS mark at max detail (which includes AA).
Strategy games are well-known for pushing the limits of any system, and few others do this as well as Total War: SHOGUN 2. It fully supports DX11, has huge battlefields to oversee with hundreds or thousands of units, and a ton of graphics options to adjust. It’s quite simply a beast of a game.
Manual Run-through: SHOGUN 2 is one of the few games in our suite where the built-in benchmark is opted for. Strategy games in particular are very difficult to benchmark, so this is where I become thankful to have the option of using a built-in benchmark.
Rounding out our apples-to-apples benchmarks, SHOGUN 2 puts the GTX 950 where it’s been most of this review – well ahead of the GTX 750 Ti and behind the R9 285.
We don’t make it a point to seek out automated gaming benchmarks, but we do like to get a couple in that anyone reading this can run themselves. Of these, Futuremark’s name leads the pack, as its benchmarks have become synonymous with the activity. Plus, it does help that the company’s benchmarks stress PCs to their limit – and beyond.
While Futuremark’s latest GPU test suite is 3DMark, I’m also including results from 3DMark 11 as it’s still a common choice among benchmarkers.
3DMark is backing up what we’ve seen with the other results so far: the GTX 950 is much better than the GTX 750 Ti (a staggering 50% faster according to 3DMark (2013)), and falls short of catching the R9 285.
Unigine might not have as established a name as Futuremark, but its products are nothing short of “awesome”. The company’s main focus is its game engine, but a by-product of that is its benchmarks, which are used to both give benchmarkers another great tool to take advantage of, and also to show-off what its engine is capable of. It’s a win-win all-around.
The biggest reason that the company’s “Heaven” benchmark is so relied-upon by benchmarkers is that both AMD and NVIDIA promote it for its heavy use of tessellation. Like 3DMark, the benchmark here is overkill by design, so results are not going to directly correlate with real gameplay. Rather, they showcase which card models can better handle both DX11 and its GPU-bogging features.
Yet again, the GTX 950 falls short of the R9 285, but delivers about 33% better performance over the GTX 750 Ti (don’t let the minimum FPS fool you.)
To test graphics cards for both their power consumption and temperature at load, we utilize a couple of different tools. On the hardware side, we use a trusty Kill-a-Watt power monitor which our GPU test machine plugs into directly. For software, we use Futuremark’s 3DMark to stress-test the card, and AIDA64 to monitor and record the temperatures.
To test, the general area around the chassis is checked with a temperature gun, with the average temperature recorded. Once that’s established, the PC is turned on and left to site idle for ten minutes. At this point, we open AIDA64 along with 3DMark. We then kick-off a full suite run, and pay attention to the Kill-a-Watt when the test reaches its most intensive interval (GT 1) to get the load wattage.
The GTX 950 STRIX is the lowest-end card I’ve tested in our updated test platform (which will result in a full-blown GPU test suite overhaul in the near-future), so it’s no surprise that it reigns supreme in both the temperature and power consumption tests. While the GTX 950 falls a bit short of the R9 285 in performance, the efficient Maxwell architecture shaves 50W off at load – an impressive feat.
Based on the model name alone, it’s not hard to sum the GTX 950 up. It’s of course slower than the GTX 960, which gives us the idea that it could handle 1080p gaming with good detail levels – and that’s just what we get.
In all of our testing, the GTX 950 soared past the GTX 750 Ti, but that’s of no surprise. It’s clear why NVIDIA wasn’t able to design this card without a power connector – it’s just too powerful, and it’s hard to complain about that. In some cases, it could be upwards of 50% faster; in most, it’s closer to 33%.
In all cases, AMD’s Radeon R9 285 was faster, and as of the time of writing, it’s actually $5 cheaper if you want to take advantage of a mail-in rebate. Typically, though, the card is closer to $180 after mail-in rebate. The R9 285’s successor, the R9 380, retails for as low as $200 – in that particular match-up, the GTX 950 is highly desirable. That’s especially the case when you realize it will draw up to 50W less while gaming.
Again, the GTX 950 is just what I expected, and that’s far from being a bad thing. We’re dealing with a $160 GPU that can blow away the kind of gaming you’ll see on consoles; which is to say 1080p and 60 FPS in every single game you want to play – and probably with better quality, too.
In case you’re wondering where all of our AMD 300-series reviews are – don’t fret, we’ll be cranking some out soon.
ASUS GeForce GTX 950 STRIX
Copyright © 2005-2021 Techgage Networks Inc. - All Rights Reserved.