Based on what Intel disclosed today, here’s a snapshot of Calxeda EnergyCore 1000 vs. Intel’s new S1200 chip:
| ECX1000 | Intel S1200 | |
| Watts | 3.8 | 6.1 |
| Cores | 4 | 2 |
| Cache (MB) | 4 Shared | 2 x .5 MB |
| PCI-E | 16 lanes | 8 lanes |
| ECC | Yes | Yes |
| SATA | Yes | No |
| Ethernet | Yes | No |
| Management | Yes | No |
| OOO Execution | Yes | No |
| Fabric Switch | 80 Gb | NA |
| Fabric ports | 5 | NA |
| Address Size | 32 bits | 64 bits |
| Memory Size | 4 GB | 8 GB |
So, while the Centerton announcement indicates that Intel takes “microservers” seriously after all, it falls short of the ARM competition. It DOES have 64-bits and Intel ISA compatibility, however. Most workloads targeting ARM are interpreted code (PHP, LAMP, Java, etc), so this is not as big a deal as some would have you believe!Intel did not specify the additional chips required to deliver a real “Server Class” solution like Calxeda’s, but our analysis indicates this could add 10 additional watts PLUS the cost. That would imply the real comparison is between ECX and S1200 is ~3.8 vs ~16 watts. So roughly 3-4 times more power for Intel’s new S1200, again, comparing 2 cores to 4. Internal Calxeda benchmarks indicate that Calxeda’s four cores and larger cache delivery 50% more performance compared to the 2 hyper-threaded Atom cores. This translates to a Calxeda advantage of 4.5 to 6 times better performance per watt, depending on the nature of the application.
Many have asked for benchmarks; if Intel publishes benchmarks for Centerton, we will do the same for ECX1000.
Um, I made a mistake on the PCI-E. I meant to say we support PCI-2 x8. I said we support 8 lanes, obviously not the same thing! There are up to 16 lanes of PCIe Gen 2 that can be used as follows:
* Two PCIe x8 lanes OR
* Four PCIe x4 lanes (which could support x1, x2 devices as well)
Sorry about that! At least we are BETTER than reported, not worse!
NO BS, when will calxeda based budget NAS come out? OR can consumers buy calxeda bga from retail shop? How much is that?
No BS: Penguin Computing has a storage box available today. We don’t sell through retail (yet); you need to purchase a complete system.
Not a fan of the S1200s faster clock and greater number of HW thread contexts per core? Can you tell us how the NEON SIMD extensions are implemented in Calxeda’s ECX1000? Which toolchain can I use for the ECX1000? Do you have SPEC2000 or SPEC2006 FP or INT performance numbers for the ECX1000?
We like frequency so long as it doesn’t create too much leakage
Our 1.4 Ghz part delivers about the same perf/watt as our 1.1 (+/- 3%). Our internal benchmarks show we beat the 2 hyper threaded cores by up to 50% for integer codes, while the Phoronix benchmarks (http://www.phoronix.com/scan.php?page=article&item=calxeda_ecx1000_atom&num=2) show that Atom has better floating point at 1.6 than our 1.1, but our 1.4 outperforms their 1.6 by about 10%. Hope this helps!