Computer Upgrades: A Data-Based Perspective


<< PREVIOUS            NEXT >>

Analysis and Frame Times

We’ve seen what each platform can do with twice the amount of graphics power. Now let’s look at each computer “before upgrading.” i.e., with a single GPU. Yes, the original question assumes a user with an AMD platform, but to get a better idea of the two platforms’ capabilities let’s look at them in a base configuration.

I’m not going to run through all of the tests again – the direct comparison will be at the end anyway. Instead, let’s isolate two good examples of games that identify different system bottlenecks and apply it to our computer systems.

There’s actually a few conclusions you can draw from these (other than I’ve got too much time on my hands), but the following graphs are why I recommended an upgrade to an Intel platform first. Using the same (single) GTX 970 on both platforms, I ran through the same section in two different games and graphed the time each platform took to render each frame. Now, really, the only appropriate graph here is the game that YOU play – some games/engines are relatively platform agnostic, which means this article in general doesn’t really apply to you (although – if a game is platform agnostic – why not pick the platform that works across a variety of games/engines anyway?)

First up: the case for AMD. Well, sort of. This graph plots each individual frame rendered, compared to the total consecutive time used (two minutes/120000 ms for each game). Basically, the computer that reaches the top first is the “slowest.” The faster the line rises, the longer that system took to render the frames overall. The AMD system ultimately wins this race in Crysis 3 – it generates more frames in the same amount of time as the Intel system, meaning a higher frame rate overall.

crysis3 2015-10-15 23-54-55-69-Time

As we’ve seen though, frame rate isn’t necessarily the end-all-be-all of computer benchmarking. Compare the frame time plot of the AMD system above with the Intel system below.

crysis3 2015-10-15 23-36-40-55-Time

Same graphics potential, different platforms – and different result. The Intel system experiences a noticeably smoother / more consistent output. While the AMD system has a slightly higher average FPS, I’m willing to bet most users would find the occasional stutters distracting after playing on the smoother Intel platform.

2xSLI_FramesARMA3That was a best case scenario for AMD. To be fair, there are a lot of games that are relatively indifferent to the CPU used. However – and this is the point I try to make to the AMD enthusiasts out there – there are games that run poorly on AMD CPUs (interestingly enough, the same can’t be said of Intel). ARMA3, with it’s complex engine and AI is one of those games. Here, the gap between the two computer systems is even greater. The Intel system churns out almost twice the amount of frames in the same time as the AMD machine – and that’s before we look at the frame times.

arma3 2015-10-15 23-42-45-85-Time

There are still a few stutters on the Intel system. Interestingly enough, the same pattern shows on the AMD frame time plot…

arma3 2015-10-16 00-03-00-19-Time

…but the stutters are magnified, affecting the minimums accordingly. The AMD system suffers a 20 FPS loss across each of the frame time percentile categories.

When comparing the two platforms with the processor being the main difference, and removing the extra GPU power available in SLI, the differences become even more apparent. Yes, there are games that run just fine on AMD. As we’ve found, there are also games that do not. The crux of the argument then, is as follows: the same statement does not appear to apply to Intel systems.

The enthusiast, then, is faced with a choice when picking a platform: a platform that performs consistently across the board? Or one, like it’s frame times, that works well in some places and not in others?

The unfortunate part is this: there isn’t a significant price difference between these two desktop computer systems, making performance the great decisive factor. The extra cooling and stronger motherboard required to get the AMD machine to perform at this level erases the (minor) price lead it had.

Based on this conclusion alone, it’s easy to see the win in favor of Intel.

Does that continue to hold up in the context of our original question, for a user that could just add another GTX 970 to their already-existing AMD system?


<< PREVIOUS            NEXT >>


  1. Caring1

    So it’s not a data based upgrade discussion, it’s an AMD V’s Intel debate again.
    Can we stick to facts and leave bias out of it please!
    This article implies “enthusiasts” don’t use AMD platforms!
    The true cost of switching platforms should be looked at, there is more than the cost of a Motherboard and Processor for enthusiasts.

    1. Tom Jaskulka

      I take it you haven’t read the complete article? I felt there was more than enough data to form the conclusion that I made, and I stand behind my results (they should be reproducible for anyone, if you’d like to gather more data of your own). I’m not saying enthusiasts don’t use AMD platforms (they do, and I’m one of them), I’m using benchmarking to attempt to analyze where an enthusiast would get the best performance for their money.

      The fact – supported by data – remains: there are games and situations where switching to Intel CPUs will net a greater performance advantage for the same or less money than using an AMD FX CPU. The worst case scenario I’ve stumbled across so far is ARMA3 (see article for examples).

      Analyzing the cost of the platforms would have been a little out of the scope of what I was trying to do, but you have a point – perhaps I should include the total cost of the systems in the article (at the time I had purchased them, as the cost will change from month to month).

      For purposes of discussion, the Intel system was approximately $1000, the AMD (with one GTX 970) was the same ~$1000 (for the components required to generate the performance result – CPU, cooling, etc. – storage wasn’t included, although it should be added to the cost as well). I could add a cost breakdown to the systems if that would help.

      Let me ask this: given that $1000, and the option to buy one of the two platforms shown here to play games on, which one would you buy if you wanted the best gaming experience?

  2. David Gilmore

    Thank you, Tom, for this very interesting article and for your honest conclusions. I understand that you are not anybody’s “fanboy”, just someone who is tired of empty promises, exaggerated claims, and unproved “truisms” (like AMD is best bang for buck). Someone just looking for the best perceived gaming experience, like the rest of us, (even Mr. Caring 1, although his obvious bias towards AMD clouds his perception). It’s just sad that an honest effort for a real answer still somehow brings out the misguided emotions. I also gave AMD platforms a try for various builds, and my first two video cards were AMD. For a while I thought that all discreet GPUs were more trouble than they’re worth, because mine were constantly crashing my PC, or doing weird things while gaming, now I know why that was happening (AMD drivers were either poorly written or not validated with enough different hardware). Since Sandy Bridge I’ve only built with Intel and nVidia, and you know what? It is a more satisfying experience, and this article helps to explain why.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>