Sizing Up Servers: Intel's Skylake-SP Xeon versus AMD's EPYC 7000 - The Server CPU Battle of the Decade?
by Johan De Gelas & Ian Cutress on July 11, 2017 12:15 PM EST- Posted in
- CPUs
- AMD
- Intel
- Xeon
- Enterprise
- Skylake
- Zen
- Naples
- Skylake-SP
- EPYC
Multi-core SPEC CPU2006
For the record, we do not believe that the SPEC CPU "Rate" metric has much value for estimating server CPU performance. Most applications do not run lots of completely separate processes in parallel; there is at least some interaction between the threads. But since the benchmark below caused so much discussion, we wanted to satisfy the curiosity of our readers.
Does the EPYC7601 really have 47% more raw integer power? Let us find out. Though please note that you are looking at officially invalid base SPEC rate runs, as we still have to figure out how to tell the SPEC software that our "invalid" flag "-Ofast" is not invalid at all. We did the required 3 iterations though.
Subtest | Application type | Xeon E5-2699 v4 @ 2.8 |
Xeon 8176 @ 2.8 |
EPYC 7601 @2.7 |
EPYC Vs Broadwell EP |
EPYC vs Skylake SP |
400.perlbench | Spam filter | 1470 | 1980 | 2020 | +37% | +2% |
401.bzip2 | Compression | 860 | 1120 | 1280 | +49% | +14% |
403.gcc | Compiling | 960 | 1300 | 1400 | +46% | +8% |
429.mcf | Vehicle scheduling | 752 | 927 | 837 | +11% | -10% |
445.gobmk | Game AI | 1220 | 1500 | 1780 | +46% | +19% |
456.hmmer | Protein seq. analyses | 1220 | 1580 | 1700 | +39% | +8% |
458.sjeng | Chess | 1290 | 1570 | 1820 | +41% | +16% |
462.libquantum | Quantum sim | 545 | 870 | 1060 | +94% | +22% |
464.h264ref | Video encoding | 1790 | 2670 | 2680 | +50% | -0% |
471.omnetpp | Network sim | 625 | 756 | 705 (*) | +13% | -7% |
473.astar | Pathfinding | 749 | 976 | 1080 | +44% | +11% |
483.xalancbmk | XML processing | 1120 | 1310 | 1240 | +11% | -5% |
(*) We had to run 471.omnetpp with 64 threads on EPYC: when running at 128 threads, it gave errors. Once solved, we expect performance to be 10-20% higher.
Ok, first a disclaimer. The SPECint rate test is likely unrealistic. If you start up 88 to 128 instances, you create a massive bandwidth bottleneck and a consistent CPU load of 100%, neither of which are very realistic in most integer applications. You have no synchronization going on, so this is really the ideal case for a processor such as the AMD EPYC 7601. The rate test estimates more or less the peak integer crunching power available, ignoring many subtle scaling problems that most integer applications have.
Nevertheless, AMD's claim was not farfetched. On average, and using a "neutral" compiler with reasonable compiler settings, the AMD 7601 has about 40% (42% if you take into account that our Omnetpp score will be higher once we fixed the 128 instances issue) more "raw" integer processing power than the Xeon E5-2699 v4, and is even about 6% faster than the Xeon 8176. Don't expect those numbers to be reached in most real integer applications though. But it shows how much progress AMD has made nevertheless...
219 Comments
View All Comments
oldlaptop - Thursday, July 13, 2017 - link
Why on earth is gcc -Ofast being used to mimic "real-world", non-"aggressively optimized"(!) conditions? This is in fact the *most* aggressive optimization setting available; it is very sensitive to the exact program being compiled at best, and generates bloated (low priority on code size) and/or buggy code at worst (possibly even harming performance if the generated code is so big as to harm cache coherency). Most real-world software will be built with -O2 or possibly -Os. I can't help but wonder why questions weren't asked when SPEC complained about this unwisely aggressive optimization setting...peevee - Thursday, July 13, 2017 - link
"added a second full-blown 512 bit AVX-512 unit. "Do you mean "added second 256 ALU, which in combination with the first one implements full 512-bit AVX-512 unit"?
peevee - Thursday, July 13, 2017 - link
"getting data from the right top node to the bottom left node – should demand around 13 cycles. And before you get too concerned with that number, keep in mind that it compares very favorably with any off die communication that has to happen between different dies in (AMD's) Multi Chip Module (MCM), with the Skylake-SP's latency being around one-tenth of EPYC's."1/10th? Asking data from L3 on the chip next to it will take 130 (or even 65 if they are talking about averages) cycles? Does not sound realistic, you can request data from RAM at similar latencies already.
AmericasCup - Friday, July 14, 2017 - link
'For enterprises with a small infrastructure crew and server hardware on premise, spending time on hardware tuning is not an option most of the time.'Conversely, our small crew shop has been tuning AMD (selected for scalar floating point operations performance) for years. The experience and familiarity makes switching less attractive.
Also, you did all this in one week for AMD and two weeks for Intel? Did you ever sleep? KUDOS!
JohanAnandtech - Friday, July 21, 2017 - link
Thanks for appreciating the effort. Luckily, I got some help from Ian on Tuesday. :-)AntonErtl - Friday, July 14, 2017 - link
According to http://www.anandtech.com/show/10158/the-intel-xeon... if you execute just one AVX256 instruction on one core, this slows down the clocks of all E5v4 cores on the same socket for at least 1ms. Somewhere I read that newer Xeons only slow down the core that executes the AVX256 instruction. I expect that it works the same way for AVX512, and yes, this means that if you don't have a load with a heavy proportion of SIMD instructions, you are better off with AVX128 or SSE. The AMD variant of having only 128-bit FPUs and no clock slowdown looks better balanced to me. It might not win Linpack benchmark competitions, but for that one uses GPUs anyway these days.wagoo - Sunday, July 16, 2017 - link
Typo on the CLOSING THOUGHTS page: "dual Silver Xeon solutions" (dual socket)Great read though, thanks! Can finally replace my dual socket shanghai opteron home server soon :)
Chaser - Sunday, July 16, 2017 - link
AMD's CPU future is looking very promising!bongey - Tuesday, July 18, 2017 - link
EPYC power consumption is just wrong. Somehow you are 50W over what everyone else is getting at idle. https://www.servethehome.com/amd-epyc-7601-dual-so...Nenad - Thursday, July 20, 2017 - link
Interesting SPECint2006 results:- Intel in their slide #9 claims that Intel 8160 is 2% faster than EPYC 7601
- Anandtech in article tests that EPYC 7601 is 42% faster than Intel 8176
Those two are quite different, even if we ignore that 8176 should be faster than 8160. In other words, those Intel test results look very suspicious.