AMD EPYC 7371 Power Consumption
The other side of the equation is power consumption. The AMD EPYC 7371 is putting up some impressive benchmark numbers, but that does have an associated cost. Here is what we saw on our PDU after a few runs:
- Idle: 83W
- 70% Load: 246W
- 100% Load: 321W
- Peak: 374W
Note these results were taken using a 208V Schneider Electric / APC PDU at 17.8C and 71% RH. Our testing window shown here had a +/- 0.3C and +/- 2% RH variance.
First off, we were told this is a 200W TDP CPU which puts its TDP above the AMD EPYC 7601. This may change as they are released next quarter. On the other hand, when one is talking about hundreds or thousands of dollars per core for a software license, a bit of extra power consumption per node ends up being a sub 1% TCO delta.
More intriguing is that we actually put this into the Gigabyte H261-Z60 server we reviewed recently both in single and dual socket configurations. Officially, that server is not specified to support to 200W TDP CPUs in its 8 sockets (2U 4-node) design. The server was able to maintain the 3.1GHz all core turbo base clock while keeping CPU temperatures acceptable. We saw slightly less time in the maximum boost frequencies in that design. For a CPU that is not officially available, and therefore not officially qualified in that dense compute platform to perform so well out of the box was extremely impressive.
Keeping a longer-term view in mind, most expect AMD EPYC 2 “Rome” CPUs to have SKUs in excess of 200W per socket in 2019 so server vendors in less dense 1U and 2U designs that 2U4N offerings already contemplated cooling chips with 200W TDP+ CPUs. Intel chips will use less power, but $15-20 per month is under $1200 over five years. Given the cost of software licenses, and the systems that use frequency optimized SKUs, this is a minor TCO delta which is why we are going to start seeing CPU and accelerator TDP jump over the next few generations.
AMD EPYC 7371 Market Positioning
AMD has a number of competitive vectors. The obvious competition is the Intel Xeon Scalable line. AMD also has P series parts that compete in this market. We wanted to explore the market positioning of the chip since it is not released in a vacuum.
AMD EPYC 7371 v. Intel Xeon
Performance wise, each AMD EPYC 7371 CPU is competitive with either dual Intel Xeon Gold 6134 or single Gold 6130 CPUs. We do not have the hard data from our lab, but the Intel Xeon Gold 6142 will trade blows with the AMD EPYC 7371. There are workloads that favor AMD. There are others that favor Intel. Still, AMD is competitive and if you are in a per-core licensing model, for example with Windows Server. Prior to the AMD EPYC 7371 this was starkly the domain of Intel Xeon. With the EPYC 7371, there is competition in this market.
We did not get firm pricing on this CPU. The AMD EPYC 7351 costs around $1250 and we expect there to be a premium for the EPYC 7371. AMD can justify a $2600 price on this SKU and be competitive with Intel head-to-head. The options are the Intel Xeon Gold 6130, that the EPYC 7371 will usually perform better than, the Intel Xeon Gold 6142 which we can extrapolate will be competitive with the EPYC 7371, or two Intel Xeon Gold 6134 CPUs.
There is a great question here regarding how do you price the parts given what is available in this range.
Driving this is a more complex topic than base clocks and turbo clocks. We narrowed our list to show what is driving many of our benchmark numbers. Not all chips will sit at turbo frequencies. Base frequencies and all core turbo frequencies become important. Here is a simplified view that will make sense given the benchmark results we have seen.
Here the reason why the dual Intel Xeon Gold 6134 setup is so much faster than the Intel Xeon Gold 6130 single CPU solution makes more sense. Base clocks are much higher and the 16-core turbo speeds (with two Gold 6134’s) are about 32% faster than the Gold 6130. That is also why the AMD EPYC 7371 is so intriguing in this space because it has the clock speed to near dual Intel Xeon Gold 6134 levels of performance both in single and multi-threaded workloads, in a single socket.
Intel still has some “exotic” parts like the Gold 6144 and Platinum 8156 that minimize cores pushing clock speeds and cache per core. On a per-socket basis, the AMD EPYC 7371 is changing the narrative for mainstream frequency optimized server parts. Whereas that has largely been an Intel domain, AMD now has an extremely competitive part. The EPYC 7371 with more cores, more memory capacity, and more PCIe lanes per socket and can go toe-to-toe with Intel’s mainstream frequency optimized parts for per core performance.
AMD EPYC 7371 v. AMD EPYC
While the AMD EPYC 7371 versus Intel Xeon is a fairly complex question, versus AMD EPYC the story is much simpler. If you need frequency optimized parts for per core licensing or simply speeding up applications, get the AMD EPYC 7371. If you want more cores AMD has 24 and 32 core options. If you want less expensive single socket solutions, get the P series parts which offer an exceptional value.
This is a chip with a specific place in the server SKU stack.
The one area we think is intriguing, is the developer workstation. The AMD Threadripper 1950X and 2950X are great chips and are better suited to workstations than AMD EPYC. At the same time, if you needed 128 PCIe lanes per socket or you needed more memory capacity with RDIMM support, the AMD EPYC 7001 series has been attractive but the clock speeds have been too low to realistically use in modern desktop workstations. Plus, four NUMA nodes in Windows is far from ideal for a desktop. At the same time, with a 3.8GHz 8-core turbo clock, this is the first AMD EPYC SKU which is in the range of the desktop parts which you might be able to get by on using it as a workstation part.
In fact, prior to writing this review, I put one of the EPYC 7371 chips into an ATX motherboard that is in our review pipeline and swapped it for my personal Threadripper 1950X to write this review. There are places where the Threadripper is superior, having fewer NUMA nodes is an advantage. At the same time, it is perfectly usable as a desktop which is something I would not have recommended with the first generation AMD EPYC SKUs.
With clock speeds approaching standalone desktop levels, Windows Server 2019, VMware, and Citrix administrators are going to have a new VDI processor option in the datacenter.
Final Words
Make no mistake, the AMD EPYC 7371 is a big deal. This will be AMD’s first publicly available frequency optimized part. AMD did not shoot for the low-end of Intel’s stack. Instead, they are delivering a part that is extremely competitive and Intel does not have a great direct answer to. This will change the server narrative as henceforth one cannot simply dismiss AMD EPYC as a high core count multi-threaded performance design. Instead, with the AMD EPYC 7371 it can compete, and win, head-to-head versus Intel Xeon Skylake-SP.
 in a market shifting manner.){kind=link}
On page 4 how do I know which terminal is for which processor?
I started reading and I was like whoa baby this is a 5 page CPU review why? Then I got into it at lunch and I know why. You’re right on the impact. That per-core performance is why we haven’t moved our Windows Hyper-V cluster to EPYC or even started to test it there.
You didn’t really mention it, but the clock speed also helps licenses in VMs. Maybe that’s obvious, but maybe to some it isn’t.
For SQL server you’re right that there’s other chips that might be better but those are extremely targeted products. It’s like EPYC’s first parts covered 75% of the market. These get them 20% more. Then there’s 5% that Intel still has better parts for.
Randy Bostrom if you look at the prompt text in those screenshots you’ll kick yourself. It’s a long review. It’s taken me 40 minutes to read.
CAN SOMEONE GET THESE GUYS GOLD 6142 CHIPS PLEASE
It can’t be that hard! Make it happen. @Intel if you don’t we’re gonna say you’re chicken.
I read STH reviews for “it is not released in a vacuum” phrase. What a peculiar feeling it triggers… Especially today, when it got paraphrased. And with a typo!
Patrick,
Why would you not have included the 7351P CPU into the “16 Core CPU Market” grid?
It is ~$400 cheaper than the 7351 CPUs, while providing (IIRC) the same performance. When talking about core-based licensing optimization it seems that this is the CPU model that is the “one to beat”.
{So much so that AMD really needs to make sure that they offer a 7371P variant, to help lock up that licensing market}
I’m with BinkyTO when is there going to be a 7371P version? I know what I want for Christmas. I’ll pay the power man.
It’s good to see AMD’s finally getting serious about frequency optimized. Maybe it’s better called performance per core optimized but that’s why we couldn’t use the EPYC that’s out there. Chips looked cheap but they’d cost too much for us to license servers for our environments. Your analysis is spot on.
6142 max all-core turbo is 3.3 GHz on 16 cores at around $2,950
Typo: Standard instead of Stanard
The base Windows Server 2019 license for Windows Server 2019 Stanard and Datacenter are 16 cores which
We we’re having this exact conversation in the office last week. We have about 40 Windows DC hosts. We’re RAM limited at 16C and 768GB. We want to EPYC but the single core is too low. This will fix
When in Q1?
That p5 SPEC CPU2017 here’s official data since I didn’t get it on the first read through
2x Gold 6130 164 https://www.spec.org/cpu2017/results/res2017q4/cpu2017-20171114-00735.html
2x EPYC 7351 165 https://www.spec.org/cpu2017/results/res2018q4/cpu2017-20180918-08912.html
2x Gold 6142 178 https://www.spec.org/cpu2017/results/res2017q4/cpu2017-20171211-01573.html
If Gold 6130 to 6142 is up to 500mhz more. EPYC 7351 to EPYC 7371 is 700MHz more, and more at the top single thread speed, then EPYC 7371 is going to do 185+. That wasn’t clear from the article on p5 but when I read the words then I looked at Cisco’s official results it make perfect sense. Maybe that’ll help someone else out
If they can do 4GHz at 225w crank it up. Intel’s been on this low power stuff for years but our data centers sell us metered power plus a cooling adder. 1 less server per rack would pay for 50-60w per server real quick.
If this comes in under $2k it’ll be a great value but you’ve still have to get people to welcome AMD again
Hopefully sth to go against the 6144 next.
I love STH for this review and analyst piece combined in one. They’re like analysts who have hands-on experience not just theoretical exposure.
For us, this is an important launch but not because we’re going to buy it. Our IT org will only buy second gen products. Rome would’ve been considered first gen for our Windows cluster. I’m going to put these in a RFP draft when they’re on the Dell site so I can get them nixed for being first gen. That’ll give me documentation to show Rome is a second gen product later in 2019. Next-level thinking here.
I was hoping for some 7nm EPYC by end of Q1/2019.
AMD has a once in a lifetime chance against Intel with Milan, they better hurry up to enjoy it as long as possible.
I buy some EPYC 7371 today for around 1550 USD. This price is absolutly fantastic and intel have nothing competive.
What cooler did you guys use? The Tyan heatsink says it doesn’t support the 200 watt processor.