The Next Generation of Micro Server SoCs: ECX-2000 vs Atom 2000by Johan De Gelas on October 29, 2013 5:59 AM EST
- Posted in
Our review of the Boston Viridis, one of the first Calxeda ECX-1000 based servers, was a pretty weird one. Instead of trying out different server workloads, we deliberately went for one of the few scenarios where the server might make sense: hosting light webservers. There were a few others like Content Delivery Network server or storage server, but those were about it. The quad ARM Cortex A9 inside the ECX-1000 was faster than the contemporary Atom SoCs, but missed the RAM capacity and raw performance of low power Xeons to be an alternative in most server workloads. The measured (!) 8 Watt per server node was however simply spectacular and the network fabric was one of the best in the industry. Calxeda was on the right track - they only needed more RAM and single thread performance in a server node.
Calxeda has announced its second generation server SoC yesterday, the EnergyCore ECX-2000. Based upon the more powerful ARM Cortex A15, this new SoC should be able to deliver up to twice as much performance at 1.8 GHz than the ECX-1000 at 1.4 GHz and offer four times more RAM (16 GB per node). Although we will not believe the performance claims until we have tested them ourselves, it is not impossible to speculate. Anand compared the Google Nexus 10 with the Samsung Galaxy Tab 3 8.0: the former has a Samsung Exynos 4 based upon a dual Cortex A15 at 1.7 GHz inside, the latter a very similar Samsung design, the 4212 based upong a dual Cortex A9 at 1.5 GHz.
|Benchmark||A15 vs A9|
|Sun Spider 1.0||140%|
It is impossible to estimate the performance of server SoCs by looking at browser benchmarks on tablet SoCs, but it gives us rough idea of how much extra crunching power the A15 delivers. At 7-zip.com we can compare an A15 at 1.7 GHz (Samsung Exynos 5250) with an A9 at 1.4 GHz (Samsung Exynos 4412):
|Benchmark||A9||A15||A15 vs A9|
As we posted before, the LZMA compression does have some similarities with typical server workloads. A Xeon "Sandy Bridge EP" 1.8 GHz scored 2793 with one thread, an EnergyCore ECX-1000 at 1.4 GHz scored 833 according to our own benchmarking. So we can estimate that a ECX-2000 would probably score around 1600, or similar to a modern Xeon at 1 GHz. Not earth shattering, but when you start looking at power consumption these numbers start to make sense.
While the ECX-1000 needed 5 (1.1 GHz) to 6W (1.4 GHz per SoC), according to Calxeda the ECX-2000 needs about 7 to 10W (1.8 GHz). This equates to about 2.5 W per 1.8 GHz core. The best low power Xeon, the Xeon E3-1230L V3, has 4 cores (with HT) at 1.8 GHz with a TDP of 25W, or around 6W per physical core. Even though we do not know exactly what kind of server performance the ECX-2000 at 1.8 GHz will deliver, the limited data that we have makes it very likely that the ECX-2000 is going to be very interesting from a performance/watt point of view.
Of course, the real challenge will be the newly released Intel Atom C2000. Let us compare the new Calxeda SoCs with Intel's second generation of Server SoCs.
Post Your CommentPlease log in or sign up to comment.
View All Comments
texadactyl - Friday, November 1, 2013 - linkJust be sure to stay clear of the NANO-5050 based on my experience with the Intel D2550MUD2. They're both running GMA3650 graphics ("Cedar Trail", originally developed by PowerVR). There is very limited support in Linux and Windows. In fact, Intel only provides Windows driver downloads for *32*-bit Windows (nothing for Linux). Personally, I'll be very careful before buying another Intel based or provided motherboard.
stagn - Tuesday, October 29, 2013 - linkYes, where can I buy micro-boards (smaller than mini-ITX) with these. At least 2 gigabit Ethernet and 2 cores. I don't need a Bay Trail chip since I won't be connecting it to a monitor.
ShieTar - Tuesday, October 29, 2013 - linkNope. Out of curiosity, what is you design logic for wanting less than mini-itx, but gual GigE ports? I can think of a few applications, but basically all of them seem to be served as well by a router with an open source firmware.
pjkenned - Tuesday, October 29, 2013 - linkHi Johan,
If you want me to run tests on the C2750 let me know. Published these a while back: http://www.servethehome.com/Server-detail/intel-at...
In some synthetic benchmarks the 8-core C2750 is more than 10x faster than the S1260. My general characterization is that it feels more like half of a Intel Xeon E3-1220 V3 in terms of performance. Power consumption for a mITX C2750 platform including 32GB (4x8GB @ 1.35v) cooling and a SSD is 33w with everything including the 4x Intel i354 Gigabit Ethernet ports running.
About the same both on the Supermicro and ASRock units we have tested thus far. Happy to run some benches if you want. Should be getting the fourth platform in the next two weeks so can spare.
Other cool news is that the C2750 runs ESXi fine. If Intel included VT-d it would be a category killer with passthrough. I think you are right that the C2550 or C2730 may be better NAS chips because even on the 12 disk ASRock C2750D4I you can run RAID-Z2 and have plenty of CPU power left over.
JohanAnandtech - Wednesday, October 30, 2013 - linkInteresting comparison. However, I have a few nits to pick. "There is a major myth that the Intel Atom C2750 is many times slower in single threaded performance", but most of your "closer to realworld" benches tell exactly the opposite :-). See the crafty bench for example.
milli - Wednesday, October 30, 2013 - linkJohan that's not really a fair thing to say. The Xeon E3 and Opteron turbo to 3.7 and 3.6Ghz respectively. The Atom will turbo to 2.6Ghz. Compared to the Xeon L5520 (2.26Ghz), it's only around 20% behind. That's quite a feat for a cpu core that uses around 2W.
With the clocks normalized, it's almost 5x faster than the S1260 it replaces, only 35% behind the Opteron and 2x slower than the E3.
He concludes the following: "performance generally around half of the Intel Xeon E3-1200 V3 series". Seems right to me.
JohanAnandtech - Wednesday, October 30, 2013 - link1. The clockspeed is part of the µarchitecture trade-offs. So you can not simply normalize clocks.
2. The point I was making was about "single threaded performance". It is not even close to being half that of a Xeon... more like one fourth. Not bad, but it is *not* a myth that Atoms are "several times" slower single threaded.
Not saying the Atom C2000 is a dog (it is pretty amazing in fact), just pointing out that the C2000 is still "several times" slower per thread. The myth has not been debunked yet...
pjkenned - Thursday, October 31, 2013 - linkEven there 1/3 of the E3 V3 performance at < 1/4 the platform power consumption on a single threaded application. The difference is that the C2750 is not 1/10th or less the speed like the S1260 was.
geekfoo - Friday, November 1, 2013 - linkPatrick, how come you dont run x264 and ffmpeg tests given these are quite popular things to do on servers both home and industrial ?
something along the lines of
FOR %%A in ("*.mkv") DO (
ffmpeg.exe -threads 0 -i "%%A" -f mp4 -vcodec libx264 -crf 16 -minrate 900k -maxrate 2100k -bufsize 2100k -refs 3 -preset slow -vprofile high -strict -2 -acodec aac -ac 2 -ab 192k -vf yadif,hqdn3d,unsharp=5:5:0.5,gradfun "%%~nA.MP4"
for .bat and easy to edit slightly for bash
for f in *.mkv
name=`echo "$f" | sed -e "s/.mp4$//g"`
./ffmpeg -threads 0 -i "$f" -f mp4 -vcodec libx264 -crf 16 -minrate 800k -maxrate 1800k -bufsize 1800k -refs 3 -preset slow -vprofile high -strict -2 -acodec aac -ac 2 -ab 192k -vf yadif,gradfun,hqdn3d,unsharp=5:5:0.5 "$name.MP4"
pjkenned - Sunday, November 3, 2013 - link@geekfoo - will look into adding the bash one. The new development version of the script adds a redis benchmark and sysbench CPU tests. It also now works on CentOS as well as Ubuntu from a clean install:
Feel free to give it a try to compare. Will add the above as an open item once I get back to the US this week.