[HH] cores: AMD vs Intel

[Tom Metro]

Federico Lucifredi wrote:
> Any nice comparison of per-core performance between AMD and Intel?

Do you literally want a comparison that looks at the performance of an
individual core, in isolation?

The comparisons I've seen uses the usual benchmarks, which attempt to
emulate real-world usage, and thus they're impacted by the imperfect way
tasks are distributed to multiple cores, and memory bandwidth, etc.

What's your application?


> Because, looking at the FX processors six and eight core prices, they
> are nowhere near what a 6 core Intel chip costs… so i'd like to educate
> myself on the why.

The short answer is they're cheaper because they perform considerably
worse. At least that's what I've read.

I don't recall a specific comparison, but at some point I read one at
Anandtech or Tom's Hardware.

Here's one:
http://www.anandtech.com/show/6396/the-vishera-review-amd-fx8350-fx8320-fx6300-and-fx4300-tested

  Last year's launch of AMD's FX processors was honestly disappointing.
  The Bulldozer CPU cores that were bundled into each Zambezi chip were
  hardly power efficient and in many areas couldn't significantly
  outperform AMD's previous generation platform. Look beyond the direct
  AMD comparison and the situation looked even worse. In our conclusion
  to last year's FX-8150 review I wrote the following:

    "Single threaded performance is my biggest concern, and compared to
    Sandy Bridge there's a good 40-50% advantage the i5 2500K enjoys
    over the FX-8150. ..."

Here's a thread linking to reviews comparing the FX-6300 to the i3:
http://www.tomshardware.com/forum/id-1623327/6300-gaming.html


The problem with these sites is that they review with desktop or gaming
performance in mind. What I wanted to know about was VM performance.
Can, for example, you attain better performance per dollar running 8 VMs
on an AMD 8-core CPU than you can on a 4-core i7?


In other CPU news, I see...

Calxeda's ARM server tested
http://www.anandtech.com/show/6757/calxedas-arm-server-tested

  At first sight, the relatively low performance per core of ARM CPUs
  seems like a bad match for servers. The dominant CPU in the server
  market is without doubt Intel's Xeon. The success of the Xeon family
  is largely rooted in its excellent single-threaded (or per core)
  performance at moderate power levels (70-95W). Combine this
  exceptional single-threaded performance with a decent core count and
  you get good performance in almost any kind of application.
  [...]
  Having four 32-bit Cortex A9 cores, each with 32 KB instruction and 32
  KB data L1 per-core caches, the processor block is somewhat similar to
  what we find inside modern smarphones. One difference is that this SoC
  contains a 4MB ECC enabled L2 cache, while most smartphone SoCs have a
  1MB L2 cache.
  [...]
  A dual Xeon E5 or Opteron 6300 server has much more processing power
  than most of us need to run one server application. That is the reason
  why it is not uncommon to see 10, 20 or even more virtual machines
  running on top of them.
  [...]
  Each server node has one quad-core Cortex A9 with 4MB of L2 cache and
  4GB of RAM. With that being the case, the question "what can this
  server node cope with?" is a lot more relevant.
  [...]
  The ARM based server is a pretty bad choice right now for memory
  intensive workloads. Even with four cores and DDR3-1333, the useable
  bandwidth is less than one sixth of what one Xeon core can sustain.

  In a similar vein, the ECX-1000 is not capable of providing more
  bandwidth than an Atom system equipped with DDR2-667. However, both
  the Atom and ARM cores are pretty bad when it comes to bandwidth.
  [...]
  Clock for clock, the out-of-order Cortex-A9 inside the Calxeda
  EXC-1000 beats the in-order Atom core. A single A9 has no trouble
  beating the older Atoms while likewise coming close to the much higher
  clocked N2800. The N2800 and ECX-1000 perform similarly.
  [...]
  Looking at both decompression and compression, it looks like a quad
  ARM A9 is about as fast as one Xeon core (without Hyper-Threading) at
  the same clock. We need about six A9 cores to match the Xeon core with
  Hyper-Threading enabled. The quad-core ECX-1000 1.4GHz is also close
  to the dual-core, four-threaded Atom at 1.86GHz. This bodes well for
  Calxeda as the 6.1W S1240 only runs at 1.6GHz.
  [...]
  We created 24 virtual machines on top of the Xeon server. ...
  [Calxeda's] server gets the same workload, but instead of using
  virtual machines, we used the 24 physical server nodes.
  [...]
  At the low concurrencies, the Intel machine leverages turboboost and
  its exceptionally high per core performance. At the higher web loads,
  the total throughput of the 96 (24x quad-core SoCs) ARM A9 cores is up
  to 50% higher than the low power 32 thread/16 core (2x Octal core)
  Xeons. Even the mighty 2660 cannot beat the herd of ARM SoCs.
  [...]
  ...each server needs about 8.3W (200W/24), measured at the wall. That
  is exactly what Calxeda promised: about 6W (at 1.4GHz) per server node
  (measured internally), up to 8.5W measured at the wall (again at
  1.4GHz). That is nothing short of amazing if you consider the
  performance numbers.
  [...]
  Let's be clear: most applications still run better on the Xeon E5. Our
  CPU benchmarks clearly indicate that any application that accesses the
  memory frequently or that needs high per thread integer processing
  power will run better on the Xeon E5. Compiling and installing
  software simply feels so much faster on the Xeon E5, there is no need
  to benchmark.

(See the article for lots more details.)

The 8-core AMD parts may be useful in a similar way to the ARM parts,
except the AMD parts tend to get dinged for power consumption, even
compared to the higher performing Intel parts. So even if they do make
sense in this context, it won't be long before the ARM servers take over
that space.

 -Tom