Can you give us some idea of the type of clients, if any, who have expressed interest in HORUS systems? Are they mainly small server customers looking to move up, people cross shopping Superdomes, impulse buyers with a few hundred grand to spend?
Unfortunately, I cannot give you client details, sorry. But you can imagine the market for a system that supports 32 K8 sockets.
Since you can’t talk about clients, let me somewhat rephrase that. Can you tell me what type of market these systems are aimed at? Where do you hope to see uptake?
We are primarily targeting two areas: Transaction processing and high performance computing. They have some key differences. One requires low latency and the other requires high bandwidth. Both of these issues are addressed in our design. Multiple protocol engines and wide remote links are for bandwidth, and the RDC and DIR and 500MHz core are for latency reductions.
It seems like both Opteron and HORUS were designed to be multi-purpose, do you think that there are any design changes you might have made if you were going to specifically target HPC workloads? What about commercial server workloads (OLTP, batch processing, etc.)?
I can always trade one of the two (bandwidth vs. latency) and get better performance for one or the other. But we wanted to maximize both from the beginning. In the end, I believe we ended up with more bandwidth than we might need for OLTP.
There are three solutions for breaking the 8 way limit on the Opteron right now, HORUS, Cray/Red Storm and Cray/Octigabay. Can you contrast those, i.e. why are you better than an Octiga Bay setup?
I don’t know their implementation details. I don’t think they support 32 way SMP though. I thought they were more cluster based. I might be wrong. In short, I can’t compare since I don’t know their implementation.
Red Storm relies on message passing by it’s special co-processors, so it’s not really tightly coupled cc, IIRC.
Question in regards to HORUS and the Linux Kernel; this design will add another level of NUMAness so to speak. While sched_domains can map it appropriately, is anyone at Newisys working with kernel developers to ensure that we are scheduling accordingly, taking node affinity and subnode affinity into account?
We have a software group here in Newisys. I don’t know if they are actively working on Linux kernel tuning for HORUS. I believe that we have Linux modified locally for HORUS. I can put you in touch with them if you give me your contact info.
Be the first to discuss this article!