AMD EPYC 3101 Market Positioning
Thes chips are not released in a vacuum instead, they have competition on both the Intel and AMD sides. When you purchase a server and select a CPU, it is important to see the value of a platform versus its competitors.
AMD EPYC 3101 v. AMD Alternatives
In terms of segmentation, the AMD EPYC 3101 is one of eight SKUs that were announced in the AMD EPYC 3000 first generation. Half have two die per package, the other half has a single die. Adding that second die puts that half of the SKU stack in a different league. On the single die side the AMD EPYC 3101 is the low-end SKU with 4 cores and four threads while the high-end EPYC 3251 has 8 cores and 16 threads. TDP on these single die parts ranges from 30W to 50W. With only four SKUs covering this space, there is plenty of room for additional segmentation. We are eagerly awaiting dual-die AMD EPYC 3000 SKUs coming to channel platforms.
AMD has great segmentation between the AMD EPYC 3101 and the lower end of the 7000 series, the AMD EPYC 7251 to the point where they are very different products in just about every aspect.
Compared to the AMD Opteron X3421, we see better performance with the EPYC 3000 series. It would be interesting if HPE decided to drop the GPU and make a HPE MicroServer Gen10 using the AMD EPYC 3101, perhaps utilizing the AMD quad 10GbE NIC. That would be a game-changer in the lower-end edge market and the AMD EPYC 3101 has the same TDP as the X3421.
AMD EPYC 3101 v. Intel Alternatives
Comparing the AMD EPYC 3101 to its Intel counterparts is easier than with the eight core parts. Intel has many lines available between the Intel Atom C3000 “Denverton”, Intel Xeon D-1500, Intel Xeon D-2100, Intel Xeon E-2100/ Core series, and the Xeon Scalable SKUs, but there are few that are focused around efficient quad-core performance. Given TDP requirements and packaging size requirements, the Intel Xeon E-2100/ Core series and Intel Xeon Scalable do not currently compete in this segment. That makes the comparison group here the Intel Xeon D-1500, Xeon D-2100, and Intel Atom C3000.
The AMD EPYC 3101 is faster core-for-core than the Intel Atom C3000 series. It falls somewhere between the Intel Xeon D-1500 and Xeon D-2100 series performance. Power consumption wise the AMD EPYC 3101 below the Intel Xeon D-2100 series and closer to the Intel Xeon D-1500 series. In many ways, AMD’s entry split the difference.
AMD EPYC 3101 v. Arm Competition
The AMD EPYC 3101 is more vulnerable to the Arm competition than higher-end SKUs. As Arm cores near x86 cores in terms of general compute capability, memory bandwidth, and I/O, the easiest adjacency is the segment that AMD EPYC 3101 occupies. There are eight core Arm chips on the market that if they became SBSA compliant, and added additional memory controller and PCIe capabilities could occupy this space. The technology is there, and we expect the 7nm generation to be the one where we start to see real competitors emerge.
While those competitors are coming, the AMD EPYC 3101 has matured to the point that it works out of the box using the same ISO OS install images that are used on Intel machines. As one can also see from the one year lag between when we saw the launch and when channel partners started making adaptable products, AMD also has an advantage that it has been in the market.
Final Words
If you are building a lower-cost appliance, and 35W TDP is in your range (perhaps say 46-50W on a configured system), then the AMD EPYC 3101 is an intriguing option. In fact, if you use Intel Broadwell-DE or Skylake-D Xeon D solutions in your channel offerings or embedded servers, you should get an AMD EPYC 3000 series, perhaps an AMD EPYC 3101 series platform and try it out. That means for Supermicro X10SDV and X11SDV customers try the M11SDV as we tested here.
As we saw with Spectre and Meltdown as well as L1TF and Foreshadow the performance impacts of security flaws can be different on Intel and AMD platforms. For customers who had to choose between motherboard suppliers in the embedded space, one can now choose between SoC suppliers as well. It is important for the ecosystem that there is a viable alternative in the market and the AMD EPYC 3101 should have minimal to no new software steps needed to use it as a second source alternative.
That’s hot. I can’t wait to get a few to test.
I’d be interested in an AMD option that has a feature set similar to the X11SDV-4C-TP8F (Multiple 10GbE, extra 1GbE, lots of SATA ports). I’m hoping ya’ll find something like that to review in the near future.
This isn’t a bad option compared to the mITX Xeon D options though, I hope they also throw in Flex ATX.
Hi Eddie – we have the M11SDV-4C-LN4F in a test rig, but that does not have the features you are looking for, albeit with a faster CPU in this platform. I too hope the line expands, but the X10SDV-F line was one then a second board in the D-1540/ D-1520 days. It is now much larger. My sense is that product lines on new chips in the embedded market take time to develop.
John,
Should M11 be H11?
These will be the M11SDV series.
Well I’m not envious yet, the lack of SATA ports does not make this a good NAS option.
Also dying to know pricing.
Noob about the server type CPUs, but apart from placing those in huge clusters to cut power consumption, bla bla, is there a point for a regular consumer to invest in one of those CPUs? Will it be better or irrelevant or too expensive if this will be in a media server box, streaming music in the house or the web etc
Sorry again for he noobness
I can’t wait idle power consumption tests of this platform product line
I googled and found a heap of info on the M11SDV series and I’ll be damned if I can find any config with either SFP+ ports or more than 4 SATA, rendering the series useless for a NAS.
I want 10GB networking AND more than 4 SATA ports.