As for testing, we tried the ASUS Crosshair VII Hero, Gigabyte AB350 Gaming 3, and ASRock Pro 4 for overclocking the 200GE, finding that none allowed more than a BCLK offset. We ended up digging out an MSI B350 Tomahawk motherboard and installing AGESA 126.96.36.199, not presently available on ASUS or ASRock (or not disclosed on their BIOS pages), and successfully overclocked to 3.9GHz on the Athlon part. The Crosshair VII Hero was used for all stock testing. Memory is an uncontrolled variable here, as the rest of our CPU benchmarks have standardized for 3200MHz. In this test, and in the G5600 testing, we were at 2666MHz memory speed.
And again, we’re focusing entirely on the CPU today, with zero testing of the IGP.
Keep Cache in Mind
As we highlight the sometimes tremendous difference between the R3 1200 and Athlon 200GE, it’s important to keep a few core specifications in mind: Aside from the literal core specification, which is 2C/4T on the 200GE and 4C/4T on the R3, the cache is a major point of change. The 200GE eliminates much of its CCX core space in favor of an integrated Vega GPU, resulting in the reduction to 2C/4T and a halving of the cache. On the R3 1200, there’s 384KB of (the most important) level 1 cache, 2MB of level 2 cache, and 8MB of level 3 cache. On the Athlon 200GE, it’s 192KB of L1 cache, 1MB of L2 cache, and 4MB of L3 cache. The 200GE will experience more cache misses in its instruction execution and data fetching, which, you’ll see, results in overall slower processing times in some applications. Civilization VI and Blender are good instances of this.
The Athlon 200GE is also limited in memory to 2666MHz, even when using the MSI board with 188.8.131.52 AGESA. These are the primary points which establish the $45 price difference against the R3 1200 CPU.
Tested CPUs primarily include the below list:
- Intel i9-9900K
- AMD R7 2700
- AMD R7 2700X
- Intel i7-8700K
- AMD R5 2600
- AMD R5 2600 OC (2600X stand-in)
- Intel i5-8600K
- Intel i9-7900X
- Intel i9-7960X
- AMD R7 1700
- Intel i7-4790K
Motherboards used for testing are as follows:
- ASUS Maximus XI Hero Z390
- Gigabyte Gaming 9 Z370
- Gigabyte Z97X Gaming G1 BK
- ASUS Crosshair VII Hero X470
- ASUS Crosshair VI Hero X370
- Gigabyte X299 Gaming 9
Constants used are as follows:
- All desktop platforms: 2x 8GB 3200MHz Corsair Vengeance LPX 16-18-18-36 DDR4
- (Exception: Z97X uses HyperX Savage 2133MHz DDR3)
- All quad-channel HEDT platforms: 4x 8GB 3200MHz Corsair Vengeance LPX 16-18-18-36 DDR4
- Corsair AX1600i PSU
- NZXT Kraken X62 at 100% speeds
Windows 10 version 1803 was used for all testing with Windows Update disabled. PresentMon is used for framerate data logging, with custom scripts for capturing data accurately.
AMD Athlon 200GE vs. G5600, R3 1200 Benchmark – F1 2018
F1 2018 uses the CodeMasters EGO engine and, unlike a lot of other games we test, pushes extremely high framerates that rarely become GPU-bound. This is good for CPU benchmarking. At 1080p/High, F1 2018 positions the Athlon 200GE at 107FPS AVG when stock, with 64FPS for 1% lows. Overclocking the 200GE to 3.9GHz boosts it to 124FPS AVG, a climb of 16%. In terms of frametimes, that’s a move from 9.3ms average frametimes to an 8ms average frametime.
Comparatively, an AMD R3 1200 4C/4T stock CPU ends up at 144FPS AVG, with an overclocked variant at 163FPS AVG. If you’re willing to overclock the 200GE, it does start to inch towards the R3 1200 stock CPU, but manages to do so at half the price. For an ultra-budget build, this might be an OK choice; more importantly, the G5600 is Intel’s closest competition, and its price is simply too high to be worth considering at $100+. These Pentiums should be closer to $70-$80, where they’d actually make some sense.
For frametimes, remember that lower is better, but more consistent is better than lower. For every spike you see in the frame-to-frame interval, that’s a longer period of time between the previous frame and the current frame, and the interval between frames becomes noticeable to the user – particularly those 40ms spikes, which are perceived as stuttering. The 200GE overclocked to 3.9GHz does well overall, with consistently lower frametimes than the stock version. The lengthened line also illustrates more overall frames rendered. If we plot the overclocked 1200, you’ll see the consistency of the red line is overall improved versus the 200GE, showing the best experience of the three plotted devices.
At 1440p, the stack is similar. We lose some framerate off the top, but overall performance is nearly identical. That’s because we’re not changing anything in the environment that stresses the CPU, and instead focusing on imposing a GPU limitation at the top-end. These results are functionally the same as at 1080p, as is scaling versus the R3 and Pentium CPUs.
AMD Athlon 200GE & Overclock Benchmark – Assassin’s Creed: Origins
Assassin’s Creed: Origins gives us a look at a popular series that likes frequency, but also leverages core count more heavily than most other games on the market. This is made evident by positioning of the 7980XE and 7960X CPUs, which often have a frequency deficit but core advantage.
At 1080p/Medium, the Athlon 200GE also illustrates this core and cache demand at the low-end. We’re bottlenecked down to 41FPS AVG when overclocked to 3.9GHz, a 9% lead over the stock 38FPS AVG of the 200GE. The G5600 is able to lead at 52FPS AVG with its increased frequency, but still struggles when compared to other CPUs on the market. The R3 1200 stock CPU ends up at 53FPS AVG, an increase over the 200GE’s 38FPS AVG of about 8ms average frame-to-frame latency. That’s noticeable. Overclocking the R3 1200 further assists in playability, bringing the average framerate up to 59FPS AVG.
None of the lows or frametime consistency suffer disproportionately from the averages in this lineup, it’s just that the averages are low overall for the 200GE. Decreasing geometric complexity would assist, but there’s only so low you can go in the settings.
1440p doesn’t change anything for Assassin’s Creed, of course, except for the chart-topping CPUs like the 9900K and 7980XE. At the low-end, performance and scaling are identical as at 1080p. We are heavily CPU-bound by these low-end products, and playability of Assassin’s Creed does suffer with the $55 AMD Athlon 200GE CPU, even when overclocked. It could be done on the lowest settings, but doubling the budget to an overclocked R3 1200 boosts the average frametime from a dismal 27ms to an acceptable and fully playable 17.3ms average frame-to-frame interval.
AMD Athlon 200GE Benchmark – Far Cry 5
Far Cry 5 has had some odd frametime performance in some CPUs lately, something we’ve illustrated time-and-again with the 9600K test variability data we presented in our 9600K review. We’re just now starting to dig into Far Cry’s odd performance behavior, as it reminds us of the GTA V bug we discovered a few years ago, where some CPUs would hit the engine’s frame rate cap and then stutter hard as they bounced off of the framerate cap. In essence, the GTA V bug meant that higher framerates resulted in worse frametimes, but only for the quad-thread i5 CPUs of the time. We’re digging into Far Cry 5 to see why it behaves in similar ways, so for now, we’re only going to show AVG and 1% until we have a full picture of what’s going on at the low-end of frametimes.
At 1080p/Normal, the Athlon 200GE stock CPU ends up at about 62FPS AVG, which is playable and has reasonably consistent frametimes for a $55 part. Overclocking with the MSI board helps significantly, boosting us to 71FPS AVG. The R3 1200, for perspective, leverages its additional die space for cache and physical cores well, running at 81FPS AVG stock and 92FPS AVG when overclocked to 3.9GHz. The G5600 operates at 85FPS AVG, continuing to prove an overall poor value at its current variable price-point of roughly $100.
At 1440p, the same sort of scaling appears. We have some fluctuations in framerate, but given that only the chart-toppers hit a GPU bottleneck, there’s not much movement at the low-end of the chart.
AMD Athlon 200GE Turn Time Benchmark – CIV VI
Civilization VI offers a look at entirely CPU-bound computational workloads for real gaming scenarios, like turn-based strategy games and Grand Campaign games where AI players calculate thousands of possible actions. For CIV VI, the AMD Athlon 200GE really suffers: Its average turn resolution time is 24.6 seconds, meaning that a game with 5 AI players will, upon clicking your “end turn” button, take 123 seconds – two full minutes – until you can play again. For an extreme example, the 7980XE at 4.6GHz would take 56.5 seconds, or roughly half the time, to process the same 5 turns with its 11.3 second-per-turn average.
Overclocking the 200GE fully demonstrates why we say CIV VI still prefers frequency for its largely single- and dual-thread-dependent crunching, where our performance climbs to a 21-second turn time average. This is a turn time requirement reduction of 13.4% per turn, which is absolutely noticeable. Moving to the R3 1200 stock CPU doesn’t provide much value against the overclocked 200GE, but does provide a significant turn time reduction of 17% against the stock 200GE. Overclocking the R3 1200 to 3.9GHz propels it to an 18-second turn time. For these two CPUs, we’re looking at Zen 1 architecture on both the 1200 and the 200GE, with the differences at 3.9GHz being a higher core count of 4C/4T on the 1200, alongside a massive difference in cache. The R3 1200 has more die space allocated to the CPU components, as there’s no IGP, and that allows for the extra physical cores and the 384KB of level 1 Cache, alongside 2MB of level 2 cache. Level 1 is the most important, as it’s the SRAM closest to the cores, which is why the 200GE’s 192KB L1 cache is such a punisher in these types of titles. The 200GE has half the cache across all three levels as compared to the 1200, not to mention a more limited memory controller.
AMD Athlon 200GE – Blender Benchmark
Blender is 100% unintended for the Athlon 200GE processors, clearly, but we ran it through the workload anyway, just for fun. We don’t have the other low-end CPUs here, as we do generally draw a cut-off when it becomes unusable, but this provides good perspective. The 200GE as a mixed workload part that may occasionally be used for 3D rendering is, as expected, completely unrecommended. It took 145 minutes to finish our GN Logo single-frame render, although overclocking aided significantly against that time horizon to bring down render time to 119 minutes. That’s an 18% time reduction, but the R5 2600 destroys the 200GE’s performance at 39 minutes to do the same work. It’s more expensive, but if you’re doing this type of work, it really is worth trying to skip a few meals to afford the better part. It’ll be worth it.
Conclusion: Athlon 200GE CPU + Cheap GPU
Finding these performance figures, and keeping in mind the $55 price-point of the Athlon 200GE, we find this CPU to be an acceptable solution for an ultra-budget PC. The goal would be to combine the 200GE with a low-end dGPU, something like a GTX 1050 or RX 550/560. The 200GE would be well-paired with one of these GPUs and performs well overall. The graphics capabilities of such a combination would exceed those of the R3 2200G and 2400G, although the 2200G comes in unbeatably cheap and was an award recipient for GN’s Best Budget Gaming CPU award. We acknowledge the 2200G primarily for its endeavor to enable reasonably high framerate gaming in “eSports-style” titles, e.g. Overwatch and Rocket League, while being the cheapest reasonable option. The 200GE fills a slightly different role with its price, freeing-up more budget for a dGPU on the cheap.
We think it does well in its market position and would feel comfortable recommending the 200GE for someone who genuinely can’t afford something like an R3 1200 or better. It’s an extra $50, but $50 is a lot when working with a budget of, say, $400.
AMD has done well to exploit Intel’s present weakness, which is a 14nm shortage minimizing availability of low-margin product, like the G5600 series.
Editorial, Testing: Steve Burke
Testing: Patrick Lathan
Video: Andrew Coleman, Keegan Gallick