Let's get as the benchmark will talk about the performance in 2021 for the GTX 960 GTA seems roughly era-appropriate for the 2015 GTX 960, so we'll start there, and then we'll get the cyberpunk 2077 we did have GTA results for this card at launch, but we've changed our testing methodology the OS has evolved into a more malware, and we change the game graphic settings, so they aren't compatible at 1080p the gtx 960m 4 gigabytes and 2 gigabyte models alike both did about 53 FPS average with our updated testing approach there's also not any meaningful difference in the 1 percent and 0.1 percent low metrics which seems to match the last time we did this in 2019.
GTA 5 Nvidia Geforce GTX 960m
We didn't notice any severe texture issues, resolution drops, or frame jobs with this game, and these settings, the GTX 970 maintains an impressively it of 52 percent, with the gtx 980 ti holding its own massively to over 100 percent beyond that one thing is clear, and it's that Nvidia has learned and in the modern era it's made sure that there's a GPU skew for every couple frames between cards given that there's also a gtx 980 that's not here it's not like the 900 series where wide clear gaps were drawn to match the clear gap some price noting that we don't have a gtx 980 on here we'll highlight some RTX 30 series cards if you want a modern comparison the rtx 3060 ti would be a good upgrade, and we'll see how the 3060 does later this week at 1440p the gtx 960m cards fall below 40 FPS average which isn't surprising and the 2 vs 4 gigabytes result don't deviate results.
The GTX 980 would still hold reasonably here, and before anyone freaks out in the comments that we have a retest of the gtx 980 lately, we'll save you the time and tell you that a 980 lands between the gtx 980 ti and a gtx 970. The gtx 980 would hold up fine at 1440p good upgrade paths include anything on this list, really, so you can pause and check it out if you want to take a closer look at the 30 or 68 series of cards, but otherwise, I use 10 series are 20 series would also be suitable.
4k is clearly a pointless task for the GTX 960 from a tenant play this game standpoint no 1 thought they were going to run 4K games on this card when it launched, but it has 1 point value the GTX 960 4 gigabytes and 2 gigabytes finally diverge with the GX 960 4 gigabyte holdings stronger 1 percent and 0.1 percent low is indicative of better overall frame time performance as a result of the framebuffer advantage the easiest way to see if the memory is insufficient as the hunt for frame time spikes.
Which is exactly what we're seeing here, so as for whether it mattered at the time, the answer is not in this game, and you could have saved some money again for this game, but it did definitely make a difference just depends on what you're tasked it further the extra memory may have been useful for other applications like 3D artists who are relying on video memory for larger projects files.
Frame time Issue on 2GB
Here's a frame time plot to help visualize things lower is better but more consistent is best neither of these as well with the 4K settings, obviously, but you can better see the spikes generated in the 2-gigabyte models data from this dataset which empirically plots the frame-to-frame performance the 4-gigabyte model averages runs between 55 milliseconds and 80 milliseconds per frame which is horrible, but it's at least fairly consistent frame the frame the GTX 960 2 gigabyte models impressive though because we haven't seen frame time deviations this bad in a long time.
Actually, it has not been a long time we saw it with the rtx 3090 running eight so-called K gaming. The GTX 960 model has a spike to almost 200 milliseconds a rapid correction after that, and this is what created those hard dips and the Low they were overweighted by the spike, and so one could argue about whether this is truly represented it here's the thing that we run a minimum of 4 tests per resolution per game, and in this instance, every single task fast had at least 1 of them if not multiple further still it's clear that the frame times are overall higher even one consistent, so GTX 960 2GB vs 4GB did matter here.
Cyberpunk 2077 is where we get into new tiles, and it's also where we have to start assigning mental asterisks to some of these results. And this game, a GTX 96 2 gigabyte card, in particular, had some serious issues with everything, but image quality is 1 of those. It wasn't necessarily always the same as the 4 gigabyte model for image quality, and so results get difficult to directly compare.
We noticed worse noise in the image and bad pop in at some locations in the game, but obviously, there are lower settings available to fly if you have to make it work. We just run the set for GPU benchmarks. Here's a graph we only tested 1080p for this one, and it'll be obvious why the GTX 960 4 gigabyte model had a meaningful, measurable difference. You're not going to be playing this game on the settings on either of these cards, so don't get us wrong, but the science is there to support that GTX 960 2GB vs 4GB does matter in the sense of these results.
The degree to which it matters is mostly irrelevant, seen as a 19 FPS average versus 16 doesn't really make it better to play. However, it still showed up in the charts, and that's cool on its own the low results also support that the 2gb model stalled out and frame times where the 4-gigabyte model held on now as for upgrades, we recommend starting your wish list at RTX 2060 and up.
Hitman 2 Nvidia Geforce GTX 960m
Hitman 2 came out three years after the GTX 960, but it's still 1 of the older titles that we have on this bench so run through that as well. At 1080p, the GTX 960 4gb does manage a higher average frame rate and average 1 percent point 1 percent low than we've seen that comparatively in some of the others, average FPS is a bit misleading here though, and for the same reasons we've shown in the past the GTX 960 2 gigabytes in the purest technical sense does perform relatively close to the 960 4gb and frame the frame intervals. Still, the high spikes on the gtx 960 2 gigabytes from occasionally running out of memory are what start to drag down the average.
So, in other words, it's fine until it isn't. It's similar to the 4-gigabyte card in a lot of frames. Still, the frames that it's dissimilar and are significant to the experience and that's dragged down the average anything would be a good upgrade use the 900 series at the higher end, or ten series cards would be great options if you can find them your local markets at this point might be a better place to look than the global online market we're competing with a minor.
From the 20 series, the 2060k or similar might be sensible while also unlocking 1440p gaming. The RX 6800 would be worth considering if your budget has doubled since your GTX 960 purchase. At 1440p PVC, similar behaviors but everything sort of loses median since performances so low overall you can reduce the graphics settings quality to try and accommodate the gtx 960 but clearly, this card is mostly stuck at 1080p. That's not a surprise. It was mostly marketed at 108p back when it launched despite being technically able to handle a few lightweight gains are simpler indie titles at 1440p.
Red Redemption 2
Red dead redemption 2 is up now tested first with the Dx12 mostly because the 2 gigabyte GTX 960 hates Vulcan in all tests that we tried to use it or rather maybe Vulcan hates it. That's okay though the good news is that DX 12 hates the nvidia geforce gtx 960m just as much as broken that we kept getting these error messages with red dead 2.
Eventually, we got it to work sort of the nvidia geforce gtx 960m 4gb ram poorly at 26 FPS average and with lows at 12 to 19 for the 1080p bands of red dead 2. The gtx 960 2gb was technically worse, but again that technicality is only interested in as an academic exercise of the differences in the two models otherwise, they're both useless here with these settings anyway.
I use GTX 1080 would be a good starting point for an upgrade with newer stuff jaunting far enough ahead, but you should probably be on the lookout for a higher resolution monitor as well. We technically also have 1440p results, but they're more useful for seeing what to buy rather than how the 960 does today. These settings are just too much for these cards to handle to the extent that any difference between them is irrelevant in the face of an otherwise unplayable frame rate for both. We'll skip Vulcan as well because the nvidia geforce gtx 960m 2gb just refuses to launch it, at least for the settings we were trying to use.
Rainbow Six is pretty interesting, and that's what's up next and this one, we found the nvidia geforce gtx 960m SSC 4 gigabyte card held about 93 FPS average with lows at 79 and 77 that's overall consistent and what we call good comparatively it's far far behind modern cards at over 100 FPS deficit versus the 1080 and 2060. Still, only that's not obvious the frames laws will complain about 93 fps on a 2015 GPU.
We assume that these people have evolved to lose their sense of sight and instead play games purely through the electromagnetic radiation pulses emanated from the monitor and from the noises made by their opponents as they grief them in the game.
The gtx 960m 2gb model is interestingly behind with a significant deficit 0.1 percent lower repeated across every single task fast and also a slight deficit an average FPS. This is from a few hard spikes that drag down the average, but those spikes are what matter most in a competitive game like rainbow six sieges. Still, the 2-gigabyte card is holding on well for the frame buffer that it has to work with.
Horizon Zero Dawn
Horizon zero dawn was rough for the gtx 960m. The average FPS came out better than some previous tasks, but it warned us pre-launch that the gtx 960m 2gb was below 3 gigabytes of capacity thanks to horizon zero dawn that was not obvious. The game ran choppy, and we're not convinced that the image quality was always like for like comparison, but it is hard to tell for this particular title used ten series and a low end 20 series cards would be suitable upgrades and will highlight a few of those here. The 1660 ti and superclass cards are also worth considering, although not shown here, but you can check our reviews from on those want to see where they stand relative to the others.
The strange brigade didn't run on the nvidia geforce gtx 960m 2gb card when using Vulcan, so we'll just present the DX 12 results the gtx 960m 2gb and gtx 4gb card are roughly the same here with only minor differences the 2060 runs multiples of an average of FPS over the gtx 960m. What the rtx 3060 ti far enough ahead now over 200 fps average in this game that's in a different resolution or graphics class altogether.
We'll stop on this one because we've established repeating patterns for a shadow of the tomb raider. The gtx 960m 2gb card refused to run, so we just ran the 4-gigabyte model that shows the same story. Nearly anything would be an upgrade, so whether you got lucky and grab a new card or you're getting ripped off on eBay for something old, you got reason, though, that it's likely to improve the overall experience.