Sign in with
Sign up | Sign in

Game Benchmark Results

Radeon R7 250X Review: Reprising Radeon HD 7770 At $100
By

We begin with Battlefield 4, tested using the game's Medium preset at 1920x1080. Click the screenshot for a full-size representation of how this title looks as we cruise through our  benchmark.

Click for 1080p pop-upClick for 1080p pop-up

As you can see, Battlefield still looks good at this detail setting, and AMD's Radeon R7 250X maintains at least 54 FPS. Let's move on to Metro: Last Light.

Click for 1080p pop-upClick for 1080p pop-up

Although a DirectX 10 code path and the Low detail preset don't sound particularly cutting-edge, bear in mind that at its most demanding settings, Metro: Last Light brings almost any graphics configuration to its knees. As you can see in our screenshot, the game still looks great at our compromised settings. You still get little extras like shadows, even.

The Radeon R7 250X never falls below 33 FPS, and it maintains an average of 54 FPS. Not bad at all.

Click for 1080p pop-upClick for 1080p pop-up

In contrast, the Radeon R7 250X can handle BioShock Infinite's High preset, which looks quite nice (as the screenshot shows). AMD's "latest" keeps its nose above 33 FPS and averages 49 FPS.

Click for 1080p pop-upClick for 1080p pop-up

Assassin's Creed IV: Black Flag has to run at its lowest detail settings on this class of card, but we are able to raise the texture quality setting to Normal and the environment quality from Very Low to Low. As with Metro: Last Light, that doesn't sound like an attractive combination. However, the output still looks pretty darned good. What's more, the Radeon R7 250X doesn't drop below 42 FPS.

Click for 1080p pop-upClick for 1080p pop-up

Finally, let's look at Grid 2, which isn't particularly graphics-bound and allows us to utilize the High preset in addition to 4x MSAA. AMD's Radeon R7 250X delivers minimum frame rates above 59 FPS with an average in the 70s.

Ask a Category Expert

Create a new thread in the Reviews comments forum about this subject

Example: Notebook, Android, SSD hard drive

Display all 67 comments.
This thread is closed for comments
Top Comments
  • 21 Hide
    cleeve , February 10, 2014 1:53 PM
    Quote:
    Why do the reviewers keep using a 2500K?


    Why wouldn't we? What's wrong with an overclocked 2500K?
  • 14 Hide
    cleeve , February 10, 2014 1:55 PM
    Quote:
    why would you get this card? why not get the 7770 that's now in the same price range and has 256 more shaders instead. I mean the r250 is the same as the 7700 in every way except the 7700 has 640 more shaders compared to the r250's 384. so it wouldn't make sense to buy an r250 esp now since the 7700 is in the same price range.


    If you read this review you would see that the R7 250X is *exactly* the same as the 7770.

    It does not have 384 shaders, it has 640. You're thinking of the R7 250, not the new R7 250X. ;) 

  • 12 Hide
    cleeve , February 10, 2014 5:56 PM
    Quote:
    I really dislike the way that the performance numbers for these low-end cards are only shown compared to other low-end cards. To see the big picture, we really need to see a much wider array of boards.


    We'll have to agree to disagree on that. We use detail settings that make sense for the boards we test.

    Seeing a 290X get 200 FPS at low settings doesn't provide much insight, nor does it make sense to use high resolutions and details playable on the 290X that deliver 8 FPS on low-end cards.

    Quote:
    Also, what's with the exaggerated Newegg hints? Yeah, we get it.... they sponsor you. But tailoring your articles to drop their name? Geez.


    That's not how it works, Achoo. I mention Newegg because I use them as a price indicator, not because we have any directive to do so.

    The content management system will automatically target any keywords that pay and highlight them as links. Welcome to the future...
Other Comments
  • 0 Hide
    rolli59 , February 10, 2014 12:35 PM
    Renaming is a trend nowadays so no surprises!
  • 0 Hide
    icraft , February 10, 2014 12:38 PM
    Eh. I used the HD 7770, and while it is a good budget card, it had its issues. They should have put much faster memory on there, especially for 2GB cards, and bumped core to 1100 or better. My card saw better performance with a 200Mhz memory bump than with a 100Mhz core bump, so memory bandwidth is an issue here.
  • 21 Hide
    cleeve , February 10, 2014 1:53 PM
    Quote:
    Why do the reviewers keep using a 2500K?


    Why wouldn't we? What's wrong with an overclocked 2500K?
  • 14 Hide
    cleeve , February 10, 2014 1:55 PM
    Quote:
    why would you get this card? why not get the 7770 that's now in the same price range and has 256 more shaders instead. I mean the r250 is the same as the 7700 in every way except the 7700 has 640 more shaders compared to the r250's 384. so it wouldn't make sense to buy an r250 esp now since the 7700 is in the same price range.


    If you read this review you would see that the R7 250X is *exactly* the same as the 7770.

    It does not have 384 shaders, it has 640. You're thinking of the R7 250, not the new R7 250X. ;) 

  • 1 Hide
    The_Trutherizer , February 10, 2014 2:07 PM
    Some might say it's unfair to NVidia and Intel, but R7s HAVE to be paired with a Kevari in reviews for people to really get a sense of the options provided by AMD these days. It's just soo relevant. And with games supporting Mantle as well. ASAP. Battlefield already does... Isn't the time more than ripe for people to get a feel for what the future may hold? Personally I've ran the swarm stress test Mantle demo and I know there is a massive difference between D3D and Mantle in both performance and quality. Fair enough... I have a 280x and an Intel 4670 cpu - I'm pretty much agnostic about brands. But I'd sincerely love to see what AMD has actually been cooking up on the entry level.
  • 1 Hide
    tourist , February 10, 2014 2:37 PM
    Why use dx10 for metro test ? And no it is not explained in the review.
  • 7 Hide
    InvalidError , February 10, 2014 3:59 PM
    Quote:
    Why do the reviewers keep using a 2500K?

    Because:
    1) it works
    2) for most games and low/mid-range GPUs, it is also just about as fast as the fastest current CPUs so there is no actual urge to use the highest-powered, newest and latest CPUs available.
  • -6 Hide
    Immaculate , February 10, 2014 4:03 PM
    @Cleeve , Nothing is really wrong with it. Just saying I know you guys got 4670Ks to use.
    You can't even buy a 2500K new. Use something newer that a new build with this card might actually have in it or be able to buy new.
  • 5 Hide
    Brian Redelings , February 10, 2014 5:14 PM
    Meh. I literally JUST bought a Saphire Radeon HD 7770 GHz Edition for $99 on newegg.
  • 0 Hide
    Damon Palovaara , February 10, 2014 5:15 PM
    I think they made a good moveIt should have Mantle support since it's part of the new r7 lineup which will benefit those who have quad-core processors. Also there is a price drop so it's pretty attractive
  • 12 Hide
    cleeve , February 10, 2014 5:19 PM
    Quote:
    cleeve I did read it, maybe you should actually read what I wrote I was talking about the r250 not r250x. like I said it doesn't make sense to buy a r250 when you can find 7770's as cheap or even an r250x rebrand for the same price an an r250.


    You're kidding, right?

    You responded to a thread titled:

    "Radeon R7 250X Review: Reprising Radeon HD 7770 At $100"

    with the words:

    "why would you get this card? why not get the 7770 that's now in the same price range and has 256 more shaders instead."

    ...I mean, come on. That doesn't make a lot of sense in context.
  • 11 Hide
    cleeve , February 10, 2014 5:20 PM
    Quote:
    @Cleeve , Nothing is really wrong with it. Just saying I know you guys got 4670Ks to use.
    You can't even buy a 2500K new. Use something newer that a new build with this card might actually have in it or be able to buy new.


    An overclocked 2500K is worlds faster than a stock 4650K. Stock they're incredibly close when it comes to gaming.

    Really, Intel hasn't improved IPC much since Sandy Bridge. Ivy and Haswell have been all about graphics improvements.
  • 1 Hide
    InvalidError , February 10, 2014 5:34 PM
    Quote:
    @Cleeve , Nothing is really wrong with it. Just saying I know you guys got 4670Ks to use.

    They may have some i5-4670k rigs but that does not mean all reviewers have access to it: individual reviewers have their own permanent rigs based on their long-term testing requirements (act as a reference CPU so all future benchmarks for the foreseeable future remain valid comparisons against the original rig) and other parts may get shuffled around between reviewers located in different cities, states or even countries. Using "one of their 4670k" is easier said than done if the reviewer lives 500km from the nearest other THG reviewer who happens to have one to spare.
  • 12 Hide
    cleeve , February 10, 2014 5:56 PM
    Quote:
    I really dislike the way that the performance numbers for these low-end cards are only shown compared to other low-end cards. To see the big picture, we really need to see a much wider array of boards.


    We'll have to agree to disagree on that. We use detail settings that make sense for the boards we test.

    Seeing a 290X get 200 FPS at low settings doesn't provide much insight, nor does it make sense to use high resolutions and details playable on the 290X that deliver 8 FPS on low-end cards.

    Quote:
    Also, what's with the exaggerated Newegg hints? Yeah, we get it.... they sponsor you. But tailoring your articles to drop their name? Geez.


    That's not how it works, Achoo. I mention Newegg because I use them as a price indicator, not because we have any directive to do so.

    The content management system will automatically target any keywords that pay and highlight them as links. Welcome to the future...
  • 3 Hide
    InvalidError , February 10, 2014 8:36 PM
    Quote:
    To see the big picture, we really need to see a much wider array of boards.

    Most people shopping for a new graphics card already know either the price range or performance level they are interested in and ignore anything that is more than about a rung up/down from that since this is all they usually need to confirm that prices seem to line up with expectations.

    Having more models for a full-blown roundup/chart, sure. But for a review more intended to pin down for whom this model may make sense, comparing it to its nearest equivalents and next models up/down the food chain is fair enough IMO.
Display more comments