I'm still alive!

While that may not be a major accomplishment for many people it is becoming a major accomplishment for me to just survive the day. My kness, back, and feet are sore, swollen, and tired. I know that's the prettiest picture, but I'm getting worn out pretty quickly. But, there are so many cool things happening at the show I just have to suck-it-up and keep going.

So without further ado, let's move on to Day 2 at SC07.

The day started off well since I didn't have early meetings. So I can to tour the floor without a lot of people and without vendors wanting to scan my badge. I saw lots of things that impressed me. I then had a round of meetings but got back to the show floor in the afternoon to tour the floor and talk to people.

• Green Computing
• Heterogeneous Computing
• Commodity CPUs are getting FAST

Green Computing

I always thought people were thinking about Green Computing. Transmeta was one of the first to develop products targeted at reducing the power consumption of systems without unduly hurting performance. This lead to companies like RLX and Orion Multisystems using Transmeta processors in HPC clusters. But, unfortunately these companies didn't survive (I like to think that these companies gave their "lives" so to speak, to advance technology for the rest of us).

Today we have SiCortex who is making systems with a very large number of low power CPUs (derived from embedded 64-bit MIPS chips). Prior to this year many companies paid lip service to lower power, better cooling, etc., but I had not seen many companies actually doing anything about it. When you wander the floor at SC07 virtually any company that makes hardware is touting that they are now "Green". I've seen signs in booths for companies where they are claiming that they are Green when in fact I know they have done nothing to make systems more energy efficient (or safer for the environment). But it is absolutely in vogue to say that you are a Green company.

But, there are companies wo are truly doing something about becoming Green. One of the trends I see is the push for blade servers. Inherently blade servers can be more energy efficient than normal rack nodes because you have fewer supplies powering more sockets that rack nodes, the cooling is more efficient, and the power supplies can be run in the most efficient zone for better power utilization. So companies like IBM, HP, Sun, and Dell, all have blade servers and are energy efficient blades into the market. In addition to being energy efficient, these companies are also working very hard to make sure that blades are cost competitive are even cheaper then rack mount nodes (in the past blades have been more expensive than rack nodes).

However, it's not easy to save large amount of power and cooling in clusters because you are fundamentally constrained by certain things. For example, pretty much all of the vendors use the same CPUs, memory, interconnects, and hard drives in their systems. So how can they save power if all of them have the same computing internals? That's a great question and one I will be asking around the show floor. But at a quick glance I think the way companies are working to save power and cooling is to focus on the pacakaging. I know this sounds a bit boring - redesigning cases - but it's actually very important. You can save 30-50% (at least) by designing the packaging better. Using better fans, fans that controlled by temperature, better airflow paths through cases, running power supplies at higher efficiencies, and other ideas, can all be combined to give signficant savings in power and cooling.

Queitly walking the floor early in the morning I can see that there are some companies, from really small to really large, that are doing something about reducing the power and cooling. I've seen some interesting ideas including the idea of using water cooling for rack mount nodes (not a new idea, but one that is coming since water cooling is more efficient than air cooling), and new chassis designs. It's hard to take apart cases to saee what makes them better than other cases, but it is easy to tell from the outside that companies are taking the design of chassis extremely seriously. I hope to have some pictures for you in a write up of SC07 next week (if I recover).

Sun brought one of their "clusters in a shipping container" solutions to the show. I didn't a chance to get into the shipping container but I peaked inside. Very cool cluster. You bring it in on an 18-wheel truck, plug in the network, the chilled water and power, and bingo - a cluster. Since they using chilled water it's a very green solution. Rackable has a similar solution.

Are we on the verge of a Green revolution. Quite possibly. I at least think that most companies are at least thinking along these lines and perhaps doing some significant research into more efficient computing. As an example of this, take a look at Green500. It's a website devoted to listing the top500 most efficient systems in the world. Dr. Wu-chun Feng, previously of Los Alamos and now at Virginia Tech has been a champion of lower power systems and started the idea of the Green500 to promote the idea. Keep an eye on this.

Heterogenous Computing

The second trend I see is somewhat opposite to Green Computing but really it is perfectly in-line with Green Computing. Since we are fundamentally limited by the same CPUs, hard drives, interconnects, and memory, the power consumption of the core of systems is about the same. Also, CPUs are not really getting faster clock speeds, so the per core performance is not improving at a really fast rate. So many people are looking for ways to dramtically improve processing power while at the same time staying very energy efficient. This trend is what people are calling Hetergeous Computing.

The idea behind heterogeneous computing is to provide some additional computing resources in systems to dramatically improve processing power. The current contenders are,

• FPGA's (Field Programmable Gate Arrays)
• Clearspeed
• GPUs (Graphic Processing Units)
• Cell processor

All of these technologies (and companies) are trying to provide massive increases in computing power in a different way. I think I've said this before, but a couple of years ago, my bet was placed on GPUs. The reason is simple - commodity pricing.

The other three technologies are very niche. But GPUs are sold by the millions every year. The crazy gamers out there who have to have the latest and greatest fastest GPU(s) so they can enjoy their games have been pushing the GPU market really hard the last few years. So this means that GPUs have become commodities. I can go into any computer store anywhere and find very fast graphics cards. Hell, I can even go into Walmart and find them! (when you in Walmart, you have arrived). So Nvidia and ATI (AMD) can spread development costs across millions of GPUs, allowing them to sell the cards for a reasonable price. But the gamers have also been pushing for beter and better performance. God bless those gamers.

The other technologies simply don't have this commodity market working for them. This means they have to spread their development costs over a much, much smaller number of products, which forces the prices way up. This is why I think that GPUs will be the winner in this accelerator contest. Also, I'm not alone.

I think everyone saw the AMD announcement about a double precision GPU card that does computations. The board has 2GB of memory (the largest that I know of with GPUs), uses 150W of power (while it sounds like a lot, it isn't too bad), and costs $2,000 (that's a bit out of the commodity range. In addition, AMD is going to finally offer a programming kit for the GPU. They will be offering a derviative of Brook called Brook+.

Nvidia is showing their Tesla GPU computing product. I haven't had a chance to see it yet, but I will.

Commodity CPUs are getting FAST

While HPC is already about performance, the other thing that I think is significant at SC07, is that Intel announced their Penryn family of chips. While normally this isn't such a big deal, the benchmarks that Doug has run benchmarks that inidicate that it an amazing improvement in performance. Doug's benchmarks were sponsored by Appro. Look at Appro's website for a copy of the paper. It's truly amazing.

I see these benchmarks as being a new trend in commodity processors. The CPU manufacturers are introducing changes into their chips that are really the HPC community. Look at the fact that SGI and Cray are using commodity processors in some of the products. Commodity CPUs are getting FAST.

Stopping Here

I'm going to stop here for today. I have had a chicken club sandwich and a package of crackers to eat in the last 2 days and my stomach is becoming louder than the slot machines in the casinos.

Enjoy - Jeff!

Jeff Layton is a cluster enthuasist who travels the world writing about clusters while trying to stamp out cluster/HPC stupidty where ever he can find it (it's definitely out there). He can be reached laytonjb@gmail.com.