AMD FX-9590 8-Core CPU Review Last Look Before Ryzen

Jump To:

Real World Benchmarks

Dolphin 5.0 x64 Emulator Benchmark: link

The long awaited Dolphin 5.0 release happened in 2016 and thanks to a major cleaning up of the codebase Dolphin has reached a new level of efficiency, powered by a revitalized dynamic recompiler. Dolphin is considered by many to be the best Nintendo Wii emulator for PC you can find. It also works for Gamecube. We are running the official Dolphin 5.0 benchmark as it offers closer mapping to real world Dolphin performance as the previous version was extremely floating point heavy. We feel this is a pretty good general CPU benchmark for real world performance as emulation workloads are something that most gamers will run at one point or another.  We benchmark the standard Wii homebrew application and run it with the speed limit set to ‘unlimited’ and the External Frame Buffer set to ‘real’ in case you wanted to run this on your personal system.

Agisoft Photoscan 1.2.6 x64 – 2D to 3D Image Manipulation Benchmark: link

Agisoft PhotoScan is a stand-alone software product that performs photogrammetric processing of 2D digital images and generates 3D spatial data to be used in GIS applications, cultural heritage documentation, and visual effects production as well as for indirect measurements of objects of various scales. We us the 50 images from the ‘Building‘ sample data download page for our benchmark. We take the total time it takes to complete four steps: Align Photos, Build Dense Cloud, Build Model, Build Texture with all the default settings for each.

KeyShot 6.3 – 3D Rendering and Animation: link

KeyShot 3D rendering and animation software is one of the fastest, easiest way to create amazing, photographic visuals of your 3D data. We installed KeyShot 6.3 to do some benchmarking and real-world stress testing using the camera_benchmark.bip scene that is included with the application. This benchmark tests a 800×554 pixel image with a continuous sample rate and shows the Frames Per Second (FPS) that the scene is being rendered from. This scene has nearly 42,000 triangles and does a good job at using all available cores to render the scene.

Blender 2.78a Open Source 3D Creation Benchmark: link

Blender is the free and open source 3D creation suite. It supports the entirety of the 3D pipeline—modeling, rigging, animation, simulation, rendering, compositing and motion tracking, even video editing and game creation. We use the BMW CPU Benchmark (CCO, 3MB) created by Mike Pan for our testing. We load bmw27_cpu.blend, hit f12 to render the animation and use the final time for the chart below. We do not use Blenchmark, so do not compare scores from that add-on to ours.

Real World Benchmark Results Summary: When it comes to our ‘real world’ benchmark tests the AMD FX-9590 performed behind the stock Intel Core i7-2700K ‘Sandy Bridge’ processor in all but one of these benchmarks. The new Intel Core i3-7350K dual-core processor was less than 40 seconds slower than the AMD FX-9590 in Photoscan, which is impressive considering the benchmark takes about 20 minutes to run on each of those processors.

Print
Jump To:
  • Timothy Isenhart

    I don’t know what memory was used on the FX 9590 but I have all of the games you tested but my average FPS are much higher than you report. I use 32 GBs of Gskill Ares DDR3 2400 RAM. (11-13-13-31 2T) ( NB 2400/HT 2200) I do notice when people test the FX 9590 the usually use 1866 but the FX 9590 and FX 9370 have an additional memory multiplier for DDR3 2400 and have A.M.P. settings that go up to DDR3 2800. And my AMD system games much smoother and fluid compared to my Intel Core i7 3770K system even when swapping video cards.

    • Yes, I’ve had the same experience with online comparisons. I think that the benchmark sites are either too lazy to optimize their config or taking money from Intel. AMD sued intel for that and won.

      BTW, do you know that if you run AMD, Intel compilers deliberately don’t use common optimizations that work on any CPU? They even admit it.

      For that reason alone, I swore I’d never buy (or recommend, which does a lot more damage) Intel processors, and I told them so.

  • Neacsu Andrei

    Can someone explain to me why the 5GHz FX-9590 CPU-score is significantly lower than my FX-8350 at slightly less than 5GHz?
    Even if I scale down my score (which is correct and works, in contrast to scaling up) to 4.7GHz, I’d get 8913 CPU points.

    http://www.3dmark.com/fs/6717384 on bare metal. http://www.3dmark.com/fs/6717204 on virtual machine (notice the motherboard).

    Both on mine, and on my wife’s FX-8300, I get consistent scores of about 9500 CPU points.

    I suspect something was wrong in the benchmarks, possibly the motherboard.

  • Coach

    Nice article. Thanks Nathan. I think it’s important to have data to look back on that is relevant to the upcoming benchmark data that we will see with Ryzen and the Intel chips. With the 8370 & 9590 data, we will see how much of a step forward AMD will have made in the near future.

    • Coach

      BTW, Microcenter has 9590s @$169.99—-a deal if you want to upgrade your FX platform.

  • YOUDIEMOFO

    I would love to see the comparison of some X79 platforms some time…. I’ve been rocking the Asus RIVE with a 3970X @ 4.5ghz with 32-64gb of RAM @2400mhz. Of which gets roughly 55-65gb/s read, write and copy.

    Looking at these charts makes me feel good of my purchase alsmot four years ago. As I’ve been one to go mostly all out on a motherboard and CPU.

    Just saying I’d love to get some sandybridge love on them there comparison charts.

    • Lindsey Joyce

      exactly bro, i just got a X5675 xeon hexa core unlocked, spent 190 for the P6X58D-E mobo and the cpu was 50 DOLLARS… I get 4.8ghz on all 6 cores on a 240mm arctic cooler, and i blow away 9590FX in every benchmark… and thats for 200 bucks..

  • Confever

    Ryzen gonna be suck. Mark my word for it.
    It may be cheaper, but it’s still gonna be weaker then intel in its price range.

    • tipoo2

      No one is expecting it to beat Intels best in performance. The most optimistic hope is that if it strikes ~IVB IPC, it can offer close enough per-core performance while offering more cores for the price.

      There’s a lot of space between AMD margins and Intel margins, so it could be a win win, we get lower price/performance, AMD gets higher than AMD margins while staying under Intel margins.

    • Gogodoll

      And how that went? Since your comment had sucha scientific toneI would rally like to see your numbers you based your prophecy on so I could calculate how deep insider you are in cpu industry to tell people to mark your words. Or it is some cristal ball case? Obviously broken.

  • Konstantinos

    Are you using different settings for Blenchmark than the official ones? Your times are much longer than those posted on Blenchmark. Even I get 242 seconds on my 6-core X5675@4.17GHz.

    • Nathan Kirsch

      I’m running Blender 2.78a for Windows with the BMW Benchmark (CPU) from Blender.org. I open Blender, load bmw27_cpu.blend and hit f12 to render the animation. Not sure if that is the ‘official’ way to do it, but let me know if now.

      • Konstantinos

        I am new to this benchmark so I downloaded everything and followed the instructions. The official way is found here:

        http://blenchmark.com/article/benchmark-your-cpu-or-gpu

        1. Download the add-on
        2. Unzip the download
        3. Place the map “blenchmark” into the Blender/[version]/scripts/addons folder of your blender copy
        4. Enable the add-on in File > user preferences… > system > Addons
        5. Click on “Render” next to “File” and choose “Benchmark”
        6. Choose your Render Device (CPU or GPU compute)
        7. Hit de button “Start Benchmark”
        8. Watch the BMW being Rendered
        9. Click “Show Result” to see and evaluate your results
        10. Click “Send Results” to send the results to this website!

        If you see here http://blenchmark.com/cpu-benchmarks, all render times are much lower than yours.

        It also helps (as with any benchmark) to set process priority for Blender.exe to “Above Normal” in Task Manager in order to avoid any background process from eating up CPU cycles.

        https://uploads.disquscdn.com/images/b332f656db97384e1bc5deb337e591799922acb96a3b73ec8eca3b8e00b5ec5b.png https://uploads.disquscdn.com/images/797479d3a643681ecd965eec1acfdd894e8e3a09046a7e712a02553653d5aced.png https://uploads.disquscdn.com/images/20340df74dbe1f54bf878447ccb5dde2afad1ebfbd6dc4ec8961c2f6f42ab5b4.png

        • Nathan Kirsch

          I’m not using the blenchmark add-on, but you can do it that way as well.

      • anon

        Yeah, as Konstantinos clarified, you’re using the wrong way 😀
        https://s30.postimg.org/wllqorc0h/Captura.png

        • anon

          Btw, here CPU X5690@4,5Ghz getting 2:13
          but the GPU 980ti@1306Mhz gets 1:18

          I really got surprised by this fact.

        • Nathan Kirsch

          The results and the way I’m doing it are fine and it properly shows the differences between the processors. You are using the BlenchMark Addon from Blender Cycles and following their instructions. I’m simply not using the add-on. I said what I did in the article, so not sure why you are telling me I’m doing it wrong when I never wanted to use BlenchMark.

        • anon

          My bad I understood you were using blenchmark as well for the bmw test and, because you were pressing F12 instead of the blenchmark procedure, that was the cause the inconsistencies between rendering times.

        • Coach

          I ran Blender with the RyzenGraphic27.blend again while overclocked on my i7-4790k @4.9ghz and scored 70.66 seconds. It was flying! 🙂 The Ryzen was what, 35 seconds or so? Ryzen may not topple all Intel CPUs, but it will make its mark for sure.

      • Konstantinos

        Nathan I agree completely that there is nothing wrong with what you are doing since you are consistent with your settings between processors. It would just have been useful for us to be able to replicate your test on our machines in order to get a feel of the upgrade potential. I’ll try it your way and see what happens.

      • Konstantinos

        So I found the bmw27_cpu.blend on Blender.org, (https://www.blender.org/download/demo-files/), loaded it up and pressed F12. The image was different and larger than the one used in Blenchmark.
        The render took 582 seconds on a X5675@4.17GHz. Your test is perfectly reproducible, it was my mistake to assume that your referred to Blenchmark. Thanks Nathan.

  • satrain18

    “AMD board prices have also come way down as you can now purchase an ATX board like the ASRock 970A-G/3.1 for $68.99 shipped (after a $10 rebate) that fully supports the AMD FX-9590 CPU and has USB 3.1 Type-A and USB 3.1 Type-C ports as well as an M.2 PCIe Gen 2 x4 slot for an SSD. Having a 5.0 GHz 8-core processor along with a new motherboard that has USB 3.1 and M.2 drive support for under $270 does sound appealing.”

    Until you realise that the 970A-G/3.1’s second PCIe x16 slot is wired as x4, taking SLI out of the equation(hope you like Radeon and being stuck with just CrossFire).
    It also means that the SSD will be limited to 2GB/s.

    • Nathan Kirsch

      Looks like Ryzen will still have some of the same PCIe limitations. The good news is that Ryzen will have PCIe lanes on the CPU, which is something PileDriver didn’t offer.

    • Coach

      I think that is why AMD is continuing to limit PCIe Lanes in the new AM4 boards—so you buy their GPUs. These days, I’m not going to run SLI or Crossfire. Nowadays, it just doesn’t seem to be worth it unless you know your favorite game will benefit, but that is a lot of money for a few titles/apps that see “some” gain. I had success using SLI 7600gs x2 and then upgraded to 2x 9800GTs. The 2nd 9800GT never really made much of a difference in most applications, whereas my old 7600s basically doubled any benchmark or gameplay. I also crossfired a HD4850 and HD4830 with some success as well. I prefer to spend extra on a better single card.

  • Terry Perry

    Picked up a 6350 6 core Turbo 4.2 for 75$ got a 100 $ Asus M.B. with lights picked up Titan 2 went to D.L. and there it was Titan needs a TRUE 4 Core CPU. in order to work properly. Dug out the I7 4.2

    • tipoo2

      It didn’t detect the FX6000 as a 6 core processor? Weird. Cores/modules can be debated until we’re blue in the face, but most games that check for core counts just count threads, even making dual core HT i3s work in games that require “quads”