The Hardware
From a physical standpoint, the Ryzen APU uses the same AM4 hardware platform that we saw released with the Ryzen CPU. The same motherboards and coolers will work with the Ryzen platform today will work with the Ryzen APUs with Vega graphics, giving potential buyers a wide and immediate potential install base.
With the price point of the Ryzen 5 2400G and the Ryzen 2200G where they are at, we hope consumers are smart enough to limit themselves to appropriately priced components. That means B350 motherboards and air cooling (more than likely). Overclocking is possible but spending $100+ on a high-end liquid cooler just doesn’t make sense.
For our testing AMD sent along the Gigabyte AB350N-Gaming Wi-Fi motherboard and a 16GB kit of G.Skill FlareX memory running at DDR4-3200. While a $99 motherboard is a great pairing with these processors, the memory kit is currently selling for $250 on Newegg. While I understand that rising memory prices aren’t AMD’s fault, pairing a $170 CPU with a $250 kit of memory is hard to swallow.
The problem of course is that graphics performance scales very well with memory speed, so getting it as fast as possible will help. For AMD’s goal of getting positive reviews and impressions of the hardware to reviewers, I get it. But for a consumer buying a system today, something along the lines of 2400 MHz memory is going to be more amenable at around $170. Even that is a tough pile to swallow…
For our testing, we are using that aforementioned motherboard and memory, but running at 2400 MHz because of the above. I think we will revisit the combination of memory speeds and Ryzen APU performance in the coming days in order to properly gauge the value of the memory speed increases, and then weight the cost/benefit of those upgrades. We used our Corsair H100i GTX – overkill for the 65 watt processor but an already-integrated part of our CPU test bed.
If you have an existing AM4 motherboard or are looking to pick one up, keep in mind that you should be able to boot and update the BIOS with the new Ryzen APU parts. There is a chance that an older board, stuck in a warehouse or on a shelf for a while, might not be able to – which is a big pain in the ass for a new system build. Some motherboards support a BIOS flashback capability to allow upgrades without a CPU even installed, so keep an eye out for that. Otherwise…it’s a return that will be in order.
System Setup and Testing Methodology
- 7-zip Compression
- Audacity MP3 Encode
- Blender
- Cinebench R15
- Euler 3D
- Geekbench
- Handbrake
- POV-Ray
- SiSoft Sandra
- SYSmark 2014 SE
- WebXPRT
- X264 Encode
The full testbed configuration is listed below.
| Test System Setup | |
| CPU | AMD Ryzen 5 2400G AMD Ryzen 3 2200G AMD Ryzen 5 1400 AMD Ryzen 3 1200 Intel Core i5-8400 Intel Core i3-8100 |
| Motherboard | Gigabyte AB350N-Gaming WiFi ASUS STRIX Z370-E Gaming |
| Memory | 16GB Corsair Vengeance DDR4-3200 (running at DDR-2400 on all configurations) |
| Storage | Corsair Neutron XTi 480 SSD |
| Sound Card | On-board |
| Graphics Card | NVIDIA GeForce GTX 1080 Ti 11GB |
| Graphics Drivers | NVIDIA 390.77 |
| Power Supply | Corsair RM1000x |
| Operating System | Windows 10 Pro x64 RS3 |
How hard would it have been
How hard would it have been to include 2 extra data points on each graph for the APU’s tested at 2400 and also at max RAM speed?
To not include it in the initial launch review is doing a disservice to anyone reading your review interested in the new APUs.
Yes, please do this soon! I
Yes, please do this soon! I am buying a 2200G soon and I will be pairing it with a 3200MHz kit. Can you also please look what impact the size of the frame bufer has like J2c did?
Never mind on the frame
Never mind on the frame buffer. I was basing my request on a Jayztwocents video, but it turns out he was a bit more drama than data. Hardware Unboxed did a very comprehensive review of frame buffer impacts and found it negligible.
Good review! I know that
Good review! I know that traditionally you guys don’t go through the full FCAT routine with your cpu reviews, but I would have loved to see that for the integrated graphics on this one. Minimum frame rate would be good too.
Great review thanks.
On the
Great review thanks.
On the part about chosen memory speed & using 2400MHz memory speed. First off most would at least try to get 2933Mhz memory and then try to OC it to 3200Mhz using 2400Mhz just seems like a lot of wasted performance potential. If the board & CPU/APU support it and the kit was provided at that speed then set the system up with those specs please to not gimp the system just because 10% of the buyers may opt to use slower memory to save a few bucks. Like you said it is not AMD’s fault the prices are out to lunch so why gimp their hardware to make a point. By the way all of Intel’s coffee lakes the non K versions were all tested on the highest end chipset the z370 and with memory speeds at 3200Mhz or above. Only because Intel chose to hold back the lower end chip sets for the CPU’s fully knowing the lower spec CPU’s would be tested on boards that allow for memory overclocking which also gave the lower spec CPU’s a big performance boost.
When the lower tier chip sets come out all of the data done in testing on 95% of the sites just becomes useless because these lower tier chips will now be bought with the lower end boards that do not allow over clocking of any kind. My point is AMD gave/sent you a review kit that yes will allow the kit to perform at it’s best but the hardware actually supports everything in the kit sent out for reviews at least they were not trying to fake the numbers like Intel pretty much did with non K Coffee lakes CPU’s by allowing reviews to be done on hardware that will not be put together as a kit once the non OC chip sets are released for the non OC CPU’s. The point is test the hardware sent out and at least try to set the hardware up so it shows what it’s full performance potential is not gimping it because a few may not buy the higher spec memory for it.
Hello second post for
Hello second post for this
“This clearly proves out AMD’s case that changes to the Precision Boost 2 technology can help with gaming performance, but the IPC advantages Intel holds remain the difference.”
It is not so much the IPC gain Intel has over AMD (which is only 5%-6% by the way) but more so the actual clock speed advantage Intel has over AMD right now oh & in the i5 case 2 extra cores over the AMD 4/8 setup as well. Those 2 extra cores more than make up the 8 threads the 2400G has because they are hardware cores not logical cores as not physical cores. My best guess is if you could take a AMD Ryzen APU and pit it up against a Intel part with same core count and had the clock rates setup at the same speed Intel would only behead by 5%-6% because of slightly better IPC on the Intel parts. But hey it is good to see intel’s Coffee lake 6th gen CPU’s doing so well oh wait that’s 8th gen nope wrong 6th gen because they are sky lake cores just a few more added to the mix pretty sad really when my Old Sandy Bridge i7 OC’ed to 5Ghz is faster than a 6700K stock and gets near or passes a 7700K which is 7th gen and mine is second gen core series. Pretty sad also that a second gen can clock as high or faster than a 3rd,4th,5th,6th,7th gen class CPU’s and run cooler while doing it even though it is on a lot bigger node process.
I do have to admit the 6th/8th gen Coffee lakes do seems to finally OC a bit better with some of them reaching 5.3-5.4Ghz with good cooling. My CPU only dreams of getting to 5.3GHz not alone 5.4GHz but I have had it at 5.2GHz for benching and it will run 24/7 stable at 5.1GHz so yep sad that a old 2010 or is it 2011 CPU is able to still play with the big boys and show them that Grand dad still has a few tricks up it’s sleeve.
Does anyone know if any of
Does anyone know if any of the major PC manufacturers are going to put these APU’s in the computers they sell?
Hey, just learned AMD will
Hey, just learned AMD will loan out a old CPU to do an bios upgrade that is needed on a lot of newer low cost motherboards.
I was running cpu-z or some
I was running cpu-z or some benchmark and my 2200G with the larger wraith cooler, the xfr jumped up to 4.125. A 400MHz jump ain’t bad and is up there with a 1300X.
Very interesting article. The
Very interesting article. The R5-2400g is not intended to compete with the I5-8400. The discreet tests should probably have included the R5-1600.
The moral is simple. If you need integrated graphics, then the R5-2400g is obviously far superior to Intel. If you need discrete graphics, then the R5-1600 is a great competitor to the I5-8400, at the SAME EXACT PRICE.
Hey paper question reguarding
Hey paper question reguarding down/spect intel vs amd (down/spect is was chosen by Steve Gibson among the huge pile he received )anyway from what www say , it look like I got cheated ! Every place online was saying intel was better when in fact bang for $ amd was the king of the hill ! My question ! You guys used to do article with link when you didn’t do them directly I know w going back in the past is a Nono but could you confirm performance lost from r a série , like the sandy bridge say i5 2500 k
Just the performance lost is good , it like ok like web say 30 %.but since I know you test very differently I suspect it must be a different ! Why I ask ? Tired of having excuse from intel! People paid way more for intel unit and the whole point was to get the best people can afford! Steve Gibson hinted that a lot of intel processor might loose performance or security reguardless of what google say (Microcode and all that) he was suggesting for people wait till the dust settle ! Till the dust settle can you look into down/spect intel vs amd and see the performance drop! Since I mainly talk about gaming keeping it simple is good enough
Wow Ryan, your skimping
Wow Ryan, your skimping lately, just let Josh do amd, you are not even showing its OC potential. What happen to u? 😉