AMD Evolving Fast to Survive in the Server Market Jungle
by Johan De Gelas on June 18, 2013 6:00 AM EST- Posted in
- AMD
- CPUs
- Arm
- Enterprise
Low Power Server CPUs
When you read about micro servers, the spotlights are on very low power servers with Atom and ARM servers. The reality in the server market is however very different. But as you probably understand from the reasoning above, a server with an Atom CPU or ARM CPU is hardly a good solution for the homogenous webfarms.
In fact, with the exception of Seamicro, many micro servers have utterly failed in that market. Quite a few server vendors have offered dense Atom based solutions, but those solutions were poorly received. Simply cramming tens of Atom based servers in a small chassis is a pretty bad idea:
1. The single threaded performance of the Atom is even for webfarms too low
2. The power gains that you make by using lower power CPUs are negated by the cable management costs, the higher amount of PCBs and PHYs.
To sum it up: the performance per watt of those servers did not and will not thrill anyone. Seamicro, the pioneer of this niche market, was successful despite and not thanks to the Atom processor inside it’s micro servers. When Seamicro offered Low Power Xeons instead of Atoms inside their Borg Cube inspired server, the demand for their products really took off.
And even Calxeda, the champion of the low power servers admits that the current “ultra low power but low performance” microserver market is a small one. Calxeda has high expectations for its next generation of servers as they will be based on the more powerful Cortex A15 and A57 CPUs.
Seamicro and Calxeda succeeded where others fails as they understood that an optimized PCB and network fabric was necessary to make the concept of micro server work. Seamicro reduced the serverboard to its bare minimum (“credit card size”), turned unnecessary features off and connected all I/O via a high performance 3D torus interconnect.
Calxeda integrated several servers on one PCB and connected them together with a 2D torus network fabric. The result was a low power draw per server, not just per CPU. Once you add a CPU with good enough single threaded performance, things get very interesting.
Facebook
Twitter
RSS
80 Comments
View All Comments
Wilco1 - Tuesday, June 18, 2013 - link
Didn't you read the news that Intel appears to be delaying 14nm introduction to 2015? And where did you see that Avoton has 20 cores? At best they might have 8.Nobody has compared A57 and Avoton branch predictors yet, so the jury is out on that. Same for power efficiency.
1008anan - Tuesday, June 18, 2013 - link
A friend Ashraf wrote an article about the 14 nm "delay" not being that big of a delay:http://seekingalpha.com/article/1503982-intel-s-14...
Avoton SoC packages might include 20 cores. More than one die per package.
JDG1980 - Tuesday, June 18, 2013 - link
I know I love to get my tech news from *investment* websites. (/sarcasm)Wilco1 - Tuesday, June 18, 2013 - link
A delay is a delay. And a link to a tech site which mentions that 20 core Avoton?iwod - Wednesday, June 19, 2013 - link
14nm is only being delayed on Desktop where there are NO competition. It is on schedule for Atom and Mobile CPU.LordanSS - Tuesday, June 18, 2013 - link
I believe the road AMD has chosen has a good future for them... the only problem is that, currently, that road is very rocky and difficult to drive through.Once they manage to fulfill their "Fusion" plans, we won't need to be bothered by the anemic FPU units paired on each Piledriver/Steamroller/whatever module, mostly because that computing should be done on the GPU (which, by the time the Fusion is complete, would be an integral part of the processor itself).
Unfortunately we're still a few years away from such... and I hope they (greatly) improve integer performance on their cores until then. But I do believe it's going to happen, and it'll be great.
Think of the old "math co-processors" of the past, back in the 386 days (they got integrated on the 486 models). The only difference now tho, is that a GCN IGP unit of today has an order of magnitude more compute/FPU power than Intel parts. If you fast forward to the future, and get that improved even more and fused into the CPU....
RandSec - Tuesday, June 18, 2013 - link
"AMD can't expect third-party code to be rewritten to accomodate their processors." AMD doesn't have the market share to tell software vendors to do things their way."Not true. AMD has *more* than mere market share, it has *complete* *market* *dominance* in the new gaming consoles. *All* of gaming graphics software expertise is now focused on leveraging CPU / GPU tradeoffs in the AMD APU design. If advantages really exist, they will be found and exposed, and new software will go that way as well, if only for AMD users. For example, it would seem that the GPU has plenty of fast FP, which need not also be in the CPU, *provided* it can be accessed easily without copying back and forth.
ruiner5000 - Tuesday, June 18, 2013 - link
What is this second great depression we are talking about? I think that is a myth. I don't see any dust bowls or soup lines. Certainly not here in Austin, where AMD is. What you had to wait an extra week to pick up your iPhone 5 and were depressed, and that's the 2nd great depression? Have you ever talked to anyone who was actually alive during the great depression? Wimps.SunLord - Tuesday, June 18, 2013 - link
We're not going to see any major x86 architecture changes from AMD till at least 2015 if not 2016 thats about how long it takes to design and deploy a newly designed architecture which is hopefully what the major rehires from last year will be working on. So we're stuck with mostly minor tweaks and enhancements till then.TiredOldFart2 - Tuesday, June 18, 2013 - link
Forget low power arm servers, give me a hybrid. Give me a box with x86 and arm cpus, and allocates resources based on use scenario.Give this to the market cheap and with low power usage and carve your way back into relevance.