Benchmarked: How Well Does Watch Dogs Run On your PC?

Watch Dogs is one of the most anticipated games of 2014. So, we're testing it across a range of CPUs and graphics cards. By the end of today's story, you'll know what you need for playable performance. Spoiler: this game is surprisingly demanding!

Imagine Grand Theft Auto played back inside of Edward Snowden's worst nightmare. You now have a good grasp of Watch Dogs' premise. The game is set in Chicago, in a not-too-distant future where the entire city is run by a single operating system that knows everything about everyone all of the time. It's not the stereotypical dystopian vision of a heavy-handed big brother scenario; the Chicago operating system isn't sentient, nor is it malevolent. For the most part, it's a tool that provides a lot of convenience for its citizens. Lights at intersections are intelligently controlled so traffic jams are a thing of the past. People synchronize their schedules with smart software that actually makes their lives better. Understandably, criminals have a difficult time conducting business in a city that's aware of the location of all of its citizens. On the surface, such an existence sounds like it might be compelling, right?

Of course, the price everyone pays for this convenience is privacy. And no matter how altruistic the intentions of its visionary creators, the operating system offers incalculable potential for abuse. There's an obvious allegory to what's currently happening in our modern world.   

The player explores this universe through the eyes of Aiden Pierce, a cyber-thief whose mentor hacks the wrong network during a heist and draws the unwanted attention of bad guys behind the curtain. Of course, the protagonist is targeted. His niece is tragically killed in an attack meant to scare him, and Aiden subsequently becomes a vigilante.

He puts his talents to work by coding software that uses the Chicago operating system to predict violent crime based on the location and attitude of its citizens (leading us to ask the obvious question, why doesn't the game's police department do this?). Aiden has to wait until crime is actually perpetrated before stepping in. Otherwise, the criminal is scared off before doing anything wrong. Minority Report, anyone?    

But enough about the plot. How is the gameplay? It's a sandbox-style title obviously influenced by the GTA series, but with a few Deus-Ex-style twists. Aiden's smartphone is his key to controlling the Chicago OS. Among other things, it can change traffic lights, stop trains, hack other phones, control security cameras, raise bridges, engage barricades, and simply blow things up. It's definitely a cool dynamic.

In addition, the profiler app is always searching for potential crimes, and it'll direct you to different locations. There are many side missions, including a feature similar to Far Cry 3's radio towers. If you're a fan of sandbox games, I can almost guarantee you'll find something you like.

My favorite character is Chicago itself. The metropolis' digital incarnation has a great flavor, and the city-themed songs on the game's radio channels make for welcome background noise. For automotive enthusiasts like myself, you'll spot many vehicles that pay blatant homage to specific models. For instance, there's a car with an 80s Pontiac Firebird body and 70s Firebird front-end. From modern Dodge Chargers, Cadillacs, and Lamborghinis to the Volkswagen Rabbit and Honda Civic, a seemingly infinite number of different vehicles in the game were plucked from reality and given subtle changes. I haven't seen better adaptations of real-world cars since Burnout Paradise, so clearly the developer chose to create its own designs rather than pay royalties to an automaker.

What didn't I like? Well, for a vigilante fighting on behalf of the people, Aiden has surprisingly little conscience when it comes to stealing the common man's property. I suppose you could make a case that the character is complicated, but it doesn't feel plausible that a guy who cares enough about human beings to put his life at risk and save them from gun-toting thugs is also totally cool with absconding with their savings or partnering up with killers to achieve his goals. The game also makes it a challenge to engage in a heated car chase without mowing down innocent pedestrians. You'd think Aiden would be emotionally crippled by hitting a mother at a bus stop, given his original motivations. In that context, I would have appreciated if bystanders were harder to mow down. Maybe I'm just a terrible driver, but if you can finish the game without the blood of innocents on your hands, you have my respect.

I didn't play through as much of the story as I wanted, since my primary purpose was finding a taxing and consistent benchmark run. In the end, I chose a pre-planned path through the outskirts of the city, driving for 90 seconds per test. Thus, the results are quite repeatable, despite the many variables this game introduces.

We'll get to the performance results in a couple of pages. First, let's look at the game engine and its settings.

    Your comment
  • coolcole01
    Running on my system with ultra and highest settings and fxaa it is pretty steady at 60-70 fps with weird drops randomly almost perfectly to 30 then up to 60 almost like adaptive sync is on, Currently playing it withe the texture at high and hba0+ and smaa and its a pretty rock steady 60fps with vsync still with the random drops.
  • coolcole01
    definitely does not like to run up the vram
  • edwinjr
    why no core i5 3570k in the cpu benchmark section?
    the most popular gaming cpu in the world.
  • chimera201
    So a Core i5 is enough compared to Ubisoft's recommended system requirement of i7 3770
  • jonnyapps
    What speed is that 8350 tested at? Seems silly not to test OC'd as anyone on here with an 8350 will have it at at least 4.6
  • Patrick Tobin
    Most 780Ti cards come with 3GB of ram, the Titan has 6GB. This is an unfair comparison as the Titan has more than ample VRAM. Get a real 780Ti or do not label it as such. HardOCP just did the same tests and the 290X destroyed the 780 since the FSAA + Ultra textures started causing swapping since it was pushing past 3GB.
  • tomfreak
    If u dont have 780ti, 780, just show us stock Titan speed, Why would u rather show us Titan OCed speed than showing Titan stock speed & all that without showing 290X OCed speed? Infact an OCed Titan does not represent a 780Ti, because it has 6GB VRAM. Vram is a big deal in watchdog. So ur Oced titan does not look like 780ti nor a real titan.
  • AndrewJacksonZA
    Hi Don

    Please could you include tests at 4K resolution, and also please use a real 780Ti and also a 295X2? Can you not ask another lab to do it, or get one shipped to you please?

    +1 also on what @Patrick Tobin said.

    I can appreciate that you might've spent a lot of time on this review, and we'd really appreciate you doing the final bit of this review. I know that not a lot of gamers currently game at 4K, but I am definitely interested in it please.

    Thank you!
  • Lee Yong Quan
    why doesnt you have the high detail setting? and would a 7790 1gb perform the same as 260x 2gb in medium texture? if not which is better
  • chimera201
    We need more variety of CPUs
  • icepick314
    anyone know if Watch Dogs have SLI profile?

    does the game utilize SLI or Crossfire setup on PC?
  • That_Guy88
    I usually use toms as my definitive sit for performance benchmarks but the lack of variety in both cpus and gpus here is really disappointing, especially for this being the first "next gen" game
  • rolli59
    This needs to be redone to see if it actually needs 8 thread capable CPU so I5 and I7 on a lga 1150/1155 should have been included!
    Techspot did include those and no difference between I5 and I7 not even lga2011 hexa core!
  • ubercake
    2GB or more VRAM is required when running at MSAA x8 (this requirement appears when you turn on the MSAA).
    The game looks beautiful.
  • wtfxxxgp
    You guys are being a bit unnecessary regarding the inclusion of the Titan OC'd to simulate the 780Ti - he simply used what he HAD. I think the choice to use medium textures renders the 6GB VRAM vs 3GB VRAM mostly moot. This was just to give us an indication, why do people have to get so darned technical all the time? You guys should really try to wrap your head around the various scenarios to be tested and the time it takes to be done before you give the Authors grief about "limited this and limited that". The game looks good, thanks for the brief review
  • Empyah
    First you put the R9 290X(CATA 14.6?) without OC against the Titan with OC,
    and then the FX-8350 against a freaking i7-3960X and NO OTHER intel CPU. [edited for language]
    For freak sakes i am really trying to follow you as a serios tech-site without bias,
    please do not make it any freaking harder for me.
  • Memnarchon
    Why no core i5 4670K nor i7 4770K at benchmarks? :O

    edit: Actually someone did some nice tests for CPUS:

    CPU performance with GTX 780
    CPU performance with R9 290X
  • TheMentalist
    I wanted to know how a 770/760 4gb edition performs on this, the 760 already performs great on 1080p though
  • Onus
    I would have liked to have seen more CPUs tested, in particular three that are widely discussed and recommended in the forums, the i5-4670K (or i5-3570K), FX-6300, and 760K.
    I hope there is a followup article, focusing on some specific details. These include VRAM limitations, and more tweaking to see which settings changes most affect not only raw FPS but also smoothness. It looks like some settings lead to a very distracting experience, and it would be nice to know what those are.
    Edit: Thanks, Don, for adding the FX-6300 and i5-3550; those are useful numbers to have. Here is one title where the FX clearly beats the i3, so core count must matter.
  • somebodyspecial
    Take a look at the links the OP gave with 780 vs. 290x. 290x lost. Not sure what all the whining is about. 290x has more ram than a 780 right? Who cares above this res when only 2% are using over 1080p?

    Claiming something wins where 98% of us NEVER play is ridiculous. You want to know who wins in 98% of users cases. Those fps are too low for me anyway, as barely breaking 30fps min is not enough. You will see dips even on AMD while playing. They're only showing a snapshot here. They dropped textures to high at hardocp (the 2nd test) and NV won. So yeah if you want to push things to where we probably wouldn't enjoy it, AMD wins. Yay. But if you play at 1080p, the links above show NV winning. I think FAR more people are worried about 1080p. Having said that, this game would laugh at my PC...ROFL.
  • InvalidError
    Interesting how Don recommends a minimum of i5 or FX6300 but did not include those in the CPU scaling benchmarks. There should be a LGA115x i7 in there too for a smooth progression from 2C4T to 6C12T - the Extreme CPU has over 3X the i3's raw performance but only manages twice the score; it would have been interesting to see a smooth progression on how much benefit it gets out of extra threads vs extra cores.
  • Onus
    I must say, I much prefer Tom's video game reviews to what HardOCP does. As much as I enjoy HardOCP's PSU reviews and believe they are well-done, their video game reviews seem devoted strictly to [near] top of the line hardware running at UltraMaxOhWOW! settings that are absolutely irrelevant to the average gamer. I want to see a lot more data points than that. Yes, more would be nice, but at least here we do get enough data for some interpolation/extrapolation to alternate hardware.
  • neon neophyte
    how are 4 core intel processors, both with and without hyperthreading NOT on the cpu benchmarks. it has been claimed that hyperthreading is required for ultra but this wasnt even tested? for all we know the high end intel processor you did test performs no better than an i5.

    considering this game is cpu bound, how is there not a more comprehensive cpu benchmarking being done? what a waste.
  • maddogfargo
    I'm on AMD, but there are a LOT of gaming rigs and gamers out there using i5 3570k/4670k and i7 3770k/4770k. I was surprised not to see these tested. That would have been some very useful information for a lot of people. :(

    Is there any way the benchmark charts could be updated to show results for these parts?