The $60 CPU Question: AMD Athlon 200GE or Intel Pentium Gold G5400? A Review
by Ian Cutress on January 14, 2019 8:00 AM ESTOverclocking the Athlon 200GE
In recent weeks, motherboard manufacturers have been releasing BIOS firmware that enables overcooking on the Athlon 200GE. It appears that this has come through an oversight in one of the base AMD firmware revisions that motherboard vendors are now incorporating into their firmware bundles. This is obviously not what AMD expected; the Athlon is the solitary consumer desktop chip on AMD's AM4 platform that is not overclockable. Since MSI first starting going public with new firmware revisions, others have followed suit, including ASRock and GIGABYTE. There is no word if this change will be permanent: AMD might patch it in future revisions it sends out to the motherboard vendors, or those vendors will continue to patch around it. As it stands, however, a good number motherboards can now offer this functionality.
The question does arise if there is even a point to overclocking these chips. They are very cheap, they usually go into cheap motherboards that might not even allow overclocking, and they are usually paired with cheaper coolers. The extra money spent on either an overclocking enabled motherboard or even spending $20 on a cooler might as well be put into upgrading the CPU to a Ryzen 3 2200G, with four cores and better integrated graphics, which comes with a better stock cooler and stomps all over Intel's Pentium line, and is also overclockable without special firmware. The standard response to 'why overclock' is 'because we can', which if you've lived in that part of the industry is more than enough justification.
Given that our resident motherboard editor, Gavin, has been on a crusade through 2018 looking at the scaling performance of the AMD APUs, I asked if he could do a few overclocking tests for us.
Overclocking the 200GE
Enabling our MSI motherboard with the latest overclocking BIOS was no different to any other BIOS flash, and with it, the multiplier options opened up for the chip. Even though AMD's chips can go in quarter multiplier steps, we could only push this processor in full multiplier jumps of 100 MHz, but with a little bit of voltage using our usual overclocking methodology, we managed to get 3.9 GHz without any trouble.
To be fair, we are using a good cooler here, but to be honest, the thermals were not much of a problem. Our practical limit was the voltage frequency response of the chip at the end of the day, and our 3.9 GHz matches what other people have seen. The base frequency is locked, so there is little room for fine adjustments on that front.
At each stage of the overclock, we ran our Blender test. The gains went up almost linearly, leading to a 20% performance throughput increase from the stock frequency to the best frequency.
Thoughts
A 21 percent performance increase across the range of benchmarks would put the 200GE either on par with Intel on most tests or even further ahead on the tests it already wins. This now changes our conclusion somewhat, as explained on the next page.
If you want to see a full suite test at the overclocked speed, leave a comment below and we'll set something up in January.
95 Comments
View All Comments
brakdoo - Monday, January 14, 2019 - link
The reason why I bought the 200GE last month: The 5400 is 75€+ and was above 80 when I bought my parts in my country. The Intel shortage makes it easy to decide but I that it'll be over soon.BTW: Does graphics performance have an influence on WebGL stuff like google maps or isn't that challenging enough?
blu42 - Monday, January 14, 2019 - link
Shouldn't 'measured under heavy load' suggest a heavy load for the entire package, not the CPU cores alone?Robotire - Monday, January 14, 2019 - link
This. I’m surprised a pure CPU (I think?) load was used. I would be interesting to know power usage while gaming.Also I’m more likely to buy a component that uses less power… but I guess some SUV owners might prefer it the other way.
Otherwise it’s a very interesting article, thanks! I don’t care at all about overpriced products, but reviews like this one are useful.
SaturnusDK - Monday, January 14, 2019 - link
It's important to note that even if you could get a G5400 at the $64 MSRP, it would be 16% more expensive than the $55 Athlon. However, you can't get the G5400 for $64. The lowest price I have ever seen it retail at is $80 which makes the choice for a absolutely bottom bin bargain buyer easy.If you're still even contemplating the G5400 at it's actual $80 price tag then it would be wise to consider if you can go the extra $20 and get the vastly superior 2200G instead.
Drumsticks - Monday, January 14, 2019 - link
As of right now, the Pentium G5400 is $129.99 on Newegg (??????) and $101 on Amazon from an alternate seller. OTOH, the Athlon 200GE is readily available from both for $60. That's a pretty stark difference.ikjadoon - Monday, January 14, 2019 - link
The pricing is out of whack at Newegg: G5400 for $129, G5500 (3.8GHz w/ UHD630) for $114, and G5600 (3.9GHz w/ UDH630) for $113.You save $16...while buying a higher-end part.
B&H Photo has it for slightly-less absurd 24% markup over MSRP ($80): https://www.bhphotovideo.com/c/product/1393125-REG...
khanikun - Tuesday, January 15, 2019 - link
Ya, the G5400 makes no sense with it's pricing. An i3 8100 is a quad core 3.6 ghz proc with UHD630 for $119. Unless power usage is a main concern, I wouldn't bother with any of those G procs currently. Until prices drops, doesn't make much sense to buy them.Zim - Saturday, January 19, 2019 - link
$65 at Fry's https://www.frys.com/product/9499961StevoLincolnite - Tuesday, January 15, 2019 - link
Even in Australia... The G5400 is $129. - Doesn't make sense to grab that when you can get the Ryzen 3 2200G for $155... You could probably find a motherboard that reduces that $26 pricing gap.Otherwise the 200GE is $79... $50 cheaper.
The G4920 is more inline with the 200GE's price at $89 currently... And at that point, the 200GE is still the decided victor.
PVG - Monday, January 14, 2019 - link
I think TDP has to cover simultaneous heavy usage of both the CPU and IGP. POV-Ray only really stresses the CPU side.I see how one can disregard the IGP contribution on higher end chips, but on models like this, where there's a big chance the IGP will actually be put to use, I fell it should be taken into account, on the power measuring front.