Posts by danolds 156 publicly visible posts • joined 20 Nov 2009
Did I say I had a disk drive? It wasn't all that fancy. I had to spin it myself with a foot powered petal to keep it moving. And if I spun it too slow or too fast, it would screw up the read or write and I'd have to start over. I used to dream of having a reliable disk drive or even a cassette tape drive that worked.
Of course, I didn't get much time for computing....in the morning we'd have to get up, clean the bag, and then sweep the road clean with our tongues...
Loved the Springsteen reference - know the song well - always thought of it as "Thunder Road II: reality sets in"
I am going to do some overclocking with Hydra, but not sure to what degree. The water cooling is working great - system very cool under load, and I have gear that will overclock well. So I'll dip my toe into the overclocking waters and see how it goes. Thanks for the encouragement, it's much appreciated....
What I was alluding with the Windows 7 comments in the article is that I had a full-on general purpose operating system running while I was pushing the hw with this massive benchmark. It wasn't taking up a huge portion of compute cycles - but it was taking up some of them. If I had gone whole-hog and installed the most stripped down Linux o/s I could, then it would free up more cycles for Linpack. From what I've heard from professionals in the industry, it would also give me more knobs and sliders to play with to optimize the o/s to run the benchmark.
My first computer was a Sanyo MBC-550....which was billed as 80% IBM compatible. Meaning that a program would get 80% of the way loaded until it crashed. And you had 1MB of memory?!! I only had 128k RAM - and a single floppy - couldn't afford the dual floppies...and YOU'RE complaining about long processing times? lol lol
First, if I had bought Hydra from a workstation vendor, I'm betting the all-in cost would be closer to $12,000. I'll check that out, I'm curious now. Second, I didn't actually spend that amount of money on that system. I'm very lucky in that I work in the industry and can get engineering samples and reviewer samples of some products every once in a while. For this system, a very helpful HPC vendor helped me get the Xeon processors and NVIDIA gave me two evaluation video cards. That helped defray the overall cost of the system considerably - phew....There will be more details on this when I start publishing the Hydra blogs...
Yeah, you're right, my results are only grossly comparable to 'real' Linpack run by professionals. There are many reasons, here are a few major ones: 1) I'm running an abbreviated Linpack on Windows - if I were doing this as a serious exercise, I'd be running it on as stripped down a version of Linux as possible.... 2) I'm not tuning the system or the benchmark at all. I should have run many many iterations of Linpack with different problem set and array sizes to see exactly which config gives a bigger number....3) Theoretical max on Linpack is "cores" x "frequency" x "FP operations per cycle". There are ways to tune each of those factors, none of which I did.
I think I probably got to about half of the Linpack potential on the big machine - maybe a bit better on the smaller ones.
Good point...the wife has a NVIDIA 285 (I think), which should run CUDA no problem. But it's the Hydra machine that I really want to try. It has two NVIDIA 590's and should really be able to pull a good CUDA-enabled Linpack if I can find the code. Maybe I'll reach out to NVIDIA and see if they can point me in the right direction....
In the original version of the story, there was a typo near the top saying that we needed an addition 90 pflops to hit exascale - it should have been 990 pflops. It was corrected as soon as we saw it. I think that covers at least part, if not all, of your rant.
For the rest of it, I agree, we certainly did have 'flops' before we had mega, giga, tera and petaflops....and we had 'flopm' (Floating Point Operations Per Minute) too, right? Although those were probably measurements of humans with pencils or chalk, I guess I should have factored those in to be complete.
Oh, and you misspelled 'cheque' in your last line. The correct spelling is 'check', as in 'the check is in the mail' or 'you should check your watch to make sure you're on time for your upcoming court-mandated anger management session.'
I posted a story in the Reg a few months back taunting various high profile universities around the world, trying to draw them into the competition, but to no avail. While the competition did get ore entries than ever before, I don't believe that any UK university put in an application. For that matter, the rest of Europe (with the exception of Russia, of course) was absent. Where's Germany, France, etc.? The world wonders...
...but only with an Australian friend playing it back on a DVR - so we could fast forward through a huge amount of it. Plus we were a bit drunk. But I learned all about the legend of Andy F**king Bickel and why he was a God-like figure to Aussie cricket fans. Then I lapsed into unconsciousness....
Hi, I'm Dan Olds and my company did the survey. What we asked in that "Mission critical" question was the proportion of mission critical workloads on users current commercial Unix platforms. What we wanted to see was if these platforms were primarily hosting highly important apps vs. hosting some highly important apps plus being general purpose systems.
There are lots of mission critical apps on mainframes, Linux and even windows platforms, of course. But that question was ONLY asking about the proportion of mission critical apps on commercial Unix hosts. It wasn't asking about the distribution of mission critical apps on, for example, Unix vs. mainframe vs. Linux vs. Windows...
The job market for Unix gurus is something we didn't go into with the survey. What we did see was that a decent number of shops who have commercial Unix plan to use more of it (about 45% of them said that). That doesn't necessarily mean an significant increase in employment, though, given that these systems are designed to enable fewer techies to manage more gear and workloads.
The report is just a pdf document embedded on the blog on our main site. You can also get it in our recent research section. And, no, it wasn't produced on a Unix box....lol....
What we're seeing is that commercial Unix platforms, over time, have moved into sort of a quasi-mainframe role in many enterprises. They tend to run the apps that need a high level of vertical scalability, single system high av, etc. etc.
This isn't to say that Linux (or Windows or mainframes) don't have mission critical roles too - they absolutely do. And, as you point out, infrastrucure is mission critical and Linux has become the choice platform for infrastructure.
What we were trying to find out with those particular survey questions was if customers view Linux as a direct substitute for commercial Unix and if they believe that the two are equal when it comes to playing in the current commercial Unix sweet spot.
From what I've learned in follow-up calls, they aren't prepared to release Crysis results yet, but are optimistic that the system should be able to run a highly modified version of it (ported to SPARC, additional parallelism, etc.). With luck, they're hoping to be able to host, as they put it, 'most of their Crysis krew at the same time.' I'll let you know what I hear....
Hey Lars,
The only reason I mentioned Linux in the article is because the Fujitsu box is using SPARC chips. I know that Linux is the o/s of choice for HPC and particularly Top500. But SPARC is closely tied to Solaris in a lot of reader minds so I wanted to make sure that no one jumped to the conclusion that it was running Solaris. It's not something I'd mention when the system is based on x86 processors.
...can do a great job at handling the high heat output from blades. The downside is that you have to plumb them into a chilled water source. However, the upside is that they really work well. I had a vendor demo one for me a few years ago. The ambient temp in the room (at the front of the system) was about 75 F. The temperature coming out of the back of the rear door was a delightful 68F. If I remember correctly, the input temperature of the water wasn't all that cold - maybe in the 50's F. The cost wasn't too bad either, around $15k for the hardware, plumbing extra, of course.
Yes, there are more details in NDA presentations - not just to analysts, but to big customers and some in the press as well. Oracle has finally provided some detailed roadmaps, which is a good thing, I agree with you on that. I also agree that public roadmaps are better than private roadmaps,
As for your point about IBM being 'notorious' for pushing out release dates on NDA roadmaps? I haven't seen much of that on the areas I cover. I might be a bit more forgiving on slight slips, but they were pretty much on time with POWER 6 and 7 at least as I remember it.
Itanium has had a LOT of schedule slips in the past, absolutely true. And with Poulson, they both have to deliver on schedule and on specification this time. It's certainly possible, both companies have the resources to do it and, right now at least, it looks like they will. Time will tell...
Allison, I hope things are going well with you too....we don't talk nearly enough....lol
First, I don't have to be nearly as cautious as you might think with vendors. With the research I do, there are always winners and losers and I've found that if I do the spadework to figure out what's really happening and then I 'call 'em like I see 'em' that things generally work out pretty well.
You seem to have a fair amount of information about what Hurd does or doesn't know, more than I do. He and I have never been as close as I would hope - not even one candle light dinner.
First, on the Oracle porting issue. For years and years, Oracle has solicited and received funds from vendors to defray Oracle's porting costs. Sometimes this is in the form of advance license purchases, but that's not the only way it happens. In my opinion, HP would have been happy to pay for it. Did Oracle ask them to? Or did Oracle just drop the bomb out of the blue? That would be something to ask Mark if either of us get the chance.
As for Intel investing in future Itaniums, from what I hear from them, it's a profitable business for them and there is increasingly good synergy between the IA and Xeon teams. IBM is a bit of a different deal, due to DB2, but why wouldn't Oracle make the same move with AIX at some point down the road? Particularly if they (Oracle) thinks that DB choice trumps system and o/s choice? In other words, if the latest/greatest Oracle DB only runs on Oracle gear, then, at least in Oracle's mind, that should cause customers to scamper to them.
I don't know that Fujitsu has completely abandoned SPARC - they're pushing the SPARC64-VIIIfx processor in HPC systems and this could be extended into commercial systems.
Interesting predictions on HP, btw...not sure I see it and I really doubt it could happen before the end of April.
And, finally, many vendors have outsourced their 'tradeshow girl' vetting processes to The Reg. You simply need to send photos of yourself in tradeshow girl trim and performing the duties, along with a frank and candid letter outlining your qualifications. Keep in mind that you can't be shocking enough - they've seen it all....good luck
First, I'm an analyst - not a journalist. As an analyst, I get briefings from vendors on road maps and a fair amount of it is under NDA. However, I'm incredibly skilled when it comes to ferreting out info from vendors. I somehow managed to find a PDF document titled: "Intel Public Roadmap, Desktop, Mobile & Data Center" (http://download.intel.com/products/roadmap/roadmap.pdf) that very clearly shows the upcoming Poulson and Kittson Itanium processors. I hope you can find it, since it was the THIRD result yielded from a google search of the terms "intel itanium roadmap". It's dated 1H2011, btw, so it's current stuff.
So that's two generations....and all that I can talk about without violating NDA (I'm checking into that now, things might have changed given Oracle's actions). Is that good enough? Straight from Intel's corporate website?
For my part, I haven't written anything in these pages about SPARC roadmaps, so you can take that up with TPM. Suffice to say that roadmaps in the Sun days, particularly towards the end, were constantly morphing and hard to pin down.
Now I double dog dare you to go out and do some actual research...
Intel just explicitly disavowed Oracle's statements about Itanium being at end of life, btw. As an industry analyst who follows the RISC/Unix and x86 markets, I know that there's quite a bit of Unix and HP-UX in use in the real world. I also know that the Itanium and HP-UX business is profitable for Intel, HP and, yes, even Oracle.
I also know that I see this as Oracle pulling a power move on the industry - specifically HP and Intel. Oh, and customers too. This is partially an attempt to stampede customers away from HP-UX systems and, in their mind, towards Oracle system, but also partially an expression of contempt and dislike from Ellison/Hurd towards HP.
There are a lot of implications to this and it's going to be interesting to see how it shakes out. I'll be writing some blog entries that might make it into El Reg on this topic soon. But am interested if hearing how people who actually own HP-UX based systems are taking this news? Will there be a backlash against Oracle? Or will the need for the Oracle DB and apps trump all else?
I'm not familiar with the random names you're citing above. Are they researchers or scientists? I'd love to friend them on Facebook if they're on it, or have them on my LinkedIn list.
Technical Leggo World Series? Hmm....interesting, I'm going to look into it and learn more...could be another event that The Reg should be covering. Thanks for the tip!
It's the kind of language you find yourself using when you've seen way way too many product pitches and have been thinking way to much about marketing strategy and theory. Be glad I didn't get talking about synergy and 'leveraging audience-appropriate brand assets' or some crap like that...
My point in bringing in the hosting provider/outsourcing angle isn't that I think they are necessarily more secure than clouds or 'better'. What I meant is that with a traditional hosting/outsourcing agreement you have a negotiated contract and the ability to nail down highly specific terms and conditions. We don't seem to be seeing that level of agreement with clouds.
I fully agree that the customer/buyer HAS to do their security, avaiailability, flexibility and cost due diligence with ANY third party provider - even if it's just a website in the clouds....
...sure Linux was ported to Alpha, but it was, at least in my opinion, too little too late. Alpha was over as a Unix competitor in the late 90's. They just didn't get enough traction with ISVs and even though the systems offered amazing performance, customers couldn't get the range of apps they needed. Microsoft did deliver on Alpha, but only after a lot of delays. In my mind, it was the lack of an ecosystem that killed Alpha, rather than technical considerations.
...and there were even two sequels to the original book. I vaguely remember reading the first sequel that begins with Colossus has been driving the earth bus for a few years. It's a police state, but Forbin is well taken care of since he's the 'Creator'. Some try to rebel against the machine and are killed. Colossus does testing to destruction on humans to develop metrics. Martians start talking to Forbin's wife over the radio. Colossus not happy with this. Forbin gets special logic bomb from said Martians and uses it to take Colossus down.
I didn't read the third and final chapter to the Colossus story, but, according to Amazon reviewers, it seems that the Martians want 50% of our oxygen. Since they're quite a ways ahead of us tech-wise, they can take it. Forbin stalls them and brings up Colossus to save the day..or something along those lines....
...as I would have liked. And they couldn't tell me as much as they would have liked either. A lot of the problem is that they're a start-up and they're concerned about too much info getting out there. When they're a bit farther along, I'll look at them again. What I'm really interested in is in empirical results from using their stuff - in other words, does it really work?
I just reread the article and I don't see where I even mention the size of the racks - much less mistake the size of it. The only place I get close to that is where I touch on the 1100U figure for a POD. That was directly from slides given to me in an HP briefing. So what article were you commenting on?
I haven't heard of many people buying empty PODs yet - from any vendor, but I can see where it might be an interesting offering from one to the data center design firms.
It's been a while since I've renewed my passport photo, but I do remember that 'no smiling' rule. However, is this the case for every country? Couldn't there be a govt. out there that realizes that their country is such a stinking hole that any rational citizen should be grinning like an idiot at the possibility of leaving it, if only for a short trip?
First of all, this was a blog entry, not a research report. It was intended to highlight an interesting interview and session at the GPU Conference. NICTA's facial recognition technology seems to be quite a bit ahead of other technology that I've seen out there.
I'm not an expert in this field and do not present myself as such, but I do have a smattering of technical background. I believe that what they presented at the show fairly represented what they can do now and is a preview of what the technology will be able to do in the future.
It seems that you have a bit of a problem with their approach and results, which is fine. But I'm not overly pleased with your saying that my article "misled" or was filled with "wild claims.” This isn't the kind of behavior that I'd associate with an Esquire, and it saddens me. Here in America, we expect more from guys who append ‘Esq.’ to their names.
It's probably relevant for our readers to know that you have your own agenda here. You have some sort of proposal dealing with some identity system based on mobile phones or some crap… I found it online, but lost interest pretty quickly.
I'd suggest that you take up your concerns with NICTA directly rather than use open letters posted in online publications. They seemed like a pretty good bunch of guys, and they'll probably respond if you approach them in the right way. You know, like someone who actually wants to learn something – not like a blowhard gadfly.
You promised that you wouldn't log onto The Reg and rip me anymore. I'm going to show this to mom....and a real man would suck it up and risk the damage by watching me stumble and mutter through the entire video. Yeah, I'm bad in the first 30 seconds, but you have no idea how much worse I get later on. If you weren't such a baby quitter whiner, you'd know that and would have much more fodder to nail me on.
I'm the analyst who did the filming and wrote the short blog piece - not a Reg reporter...just wanted to say nice work, Adam 73.....you explained the challenge of SMP scaling in a clear and concise way. Meaning I'm going to steal it and start using it as my own right away. The factors that you write about above are also what makes it so hard to scale things like Oracle RAC or other applications using distributed systems - there's just too much locking and internode communication flying around.
The question you bring up about the speed/latency of IBM's expansion stuff - their scalability port - is one that I've had as well. I've been assured that it's fast enough these days to ensure flat memory access - but we won't know without testing with benchmarks and real world workloads.
As you point out, the Unix guys control their own hardware and os's and thus have the ability to optimize for SMP performance. Hmm...Open VMS was your favorite? For me, it's the original Cray/Sun E10k with the cross bar....true flat SMP that scaled like a howler monkey climbing a tree. As we used to say back then "NUMA is a bug" that need to be stomped out. However, tech marches on and there are now network connections like Infiniband that top that crossbar in terms of bandwidth, but aren't quite there in terms of latency (yet). When they get the latency number down, we might see true scaling using distributed systems - assuming the operating systems are designed to take advantage of it.
Again, I'm the guy who did the original story. And I'm not a reporter hired by El Reg, I guess I'd say that we're partners. I have my own small analyst firm and I sort of consult with the Reg as their tame HPC Industry Analyst.
I went to the HW vendors to see what they were bringing to the table in terms of big iron. Aside from my own weakness for systems that get big and go fast, I believe that large shared memory spaces and lots of cores (plus I/O) are the most efficient way to host lots of virtualized workloads. Having more system resources to parcel out among workloads can yield higher overall efficiency since you can host more apps and spread the virtualization overhead across a larger number of guests (up to a point).
I also think that having larger systems enables admins to build more customized system images to optimize performance for apps that might become memory or CPU starved in smaller boxes under peak load. I think we're going to see a greater awareness of the need for 'lumpy systems' rather than the past 'balanced system' mantra that the vendors used to chant. Lumpy systems (with greater or lesser amounts of memory, CPU, or I/O shares) fit system resource shares to the workload rather than simply throwing it an equal share of horsepower. This will lead to higher overall performance and, with informed placement of workloads on the virtualized boxes, will also lead to higher system utilization. [-end rant-]
They are connecting the memory drawer to the host system via Intel QPI and what IBM used to call their Scalability Port (they might still call it that). This port is the same port they used to use to pin 4-socket chassis together to build out 16-socket SMP systems. I don't think they ever published full specs on this port, but it was fast and wide enough to give them reasonable SMP scalability on commercial apps.
...it really depends on the terms and conditions attached to the deal, right? Along with the price of the memory drawer too. Right now, it's hard to pry those details out of them without buying one myself. While I'm committed to providing as much info as I can, I'm not going to take one for the team to that extent...lol
I do seem to remember IBM and other vendors talking about how they've dropped their memory prices in past years. I'm sure that those prices aren't competitive with the cheapest memory out there - but are the street prices from the vendors competitive with what you'd get from a name brand memory house?
Yeah, that's EXACTLY what I did....I not only beat on my chest, but I also puffed it out, pulled myself up to my imposing 5'10" height and screamed threats and obscenities at them as I was dragged away. I then dissolved in tears when my exalted status didn't end up in me getting my way....it was quite a display.
...and who says I'm a professional anyway? What do you think you're reading, The Economist? And, to close, there was no threat, thinly veiled or otherwise. I don't do that....and if you persist in saying this, I'll make sure you never work in the industry again.
...when I wrote it, there wasn't much, if any, Intel/McAfee PR-ish fluff out there yet....maybe I was channeling them or the victim of some mind control beam from Intel? I'm located not far from their Hillsboro, Oregon mega-outpost, so it's certainly possible that they could have taken control of me for at least a while. I need to put another layer of aluminum foil around the west facing windows to block the beams again.
On the 'bleedin' obvious' point...yeah, it's bleeding obvious to me too - but I wasn't seeing any of that stuff in either the press or in other discussions. Hell, I'm not the sharpest knife in the drawer, so I'll point out the obvious all day long if I'm not seeing it in other places.
..you're right, security is of course SW based...I should have positioned the hardware component as a hardware assist via specialized hardware and not implied it's a totally hardware based solution. There's always going to be a big chunk of SW in any security solution that has to change to adapt to changing threats.
I also agree with your cloud point. It doesn't matter who certifies what, with clouds, you're still putting your data out beyond your own firewall and data center. Even if you encrypt the data in during transmission, it's still vulnerable when it's at rest in the cloud in the vast majority of cases. There isn't anything that's going to take away that "what if..." thought in your mind. This is why I don't see cloud as a enterprise computing panecea or game changer. This puts me out of step with many in the industry, but the potential risks of cloud outweigh the benefits by a large margin in my mind. However, having a 'certified safe' cloud will persuade some folks to put stuff out there - maybe not important stuff and maybe just for overflow processing - but the vendor who seems to have the best security story will probably do better than the rest. And that means $$
On the appliance thing - yeah, it's going to be a chipset on a board - not in a single Intel chip. I think that Intel's economies of scale in manufacturing, distribution and marketing will give them an advantage in this area.
I think the 'why' question comes down to security (already important) rising in importance and Intel looking at a way to get a piece of that pie.
I wrote a blog in the Reg on this story and checked out the numbers. I agree that the 5 degree delta seems too low, but that was the best info I could get at the time. I was expecting to see a higher output temperature given the amount of gear being cooled and the operating temps. So, as you surmise, we must be talking about a very high flow rate. I also believe that perhaps the numbers are a bit off in the materials that I was able to find (primarily a press release with a few supporting, but non-technical docs.)
While I think that more BI background would help IT pros in the UK (and elsewhere too), I think that they'd be better served by learning more about predictive analytics. By this, I mean learning more about modeling, data analysis, and predictive methods - lots of statistics. But this knowledge needs to be coupled with business expertise and curiosity. The most valuable folks will be those who can apply the analytical techniques to the unique business situation their organization faces. Machines can do the math and crunch the data, but it takes smart people to figure out the right questions to ask, the right conditions to test, and the right data to use to get the answers. In my mind, these skills are the most valuable and will be very difficult to outsource or offshore.
If you look around, AC, you'll see our stuff on a variety of vendor web sites (IBM, VMware, and others, and also on our own site). We're not Gartner or IDC, but we do OK. However, your comments have me rethinking our entire business model. Hmm....we do independent, non-sponsored research (surveys and stuff) and then write about the results… but maybe we should GIVE the results and associated analysis away FOR FREE - just like all of the other analyst firms. That might work - and it might take away some of your objections.
While we're at it, we need to make sure we bury any survey result where customers say they prefer one vendor over another on a particular item. We need to be impartial and evenhanded, despite what we hear from real customers who live in the real world. So we’ll cut everything right down the middle: "All children are attractive, well behaved, and of above average intelligence.." Yeah, that's the ticket; we want to make sure we don't offend anyone. Then we’ll look a little more like the big guys, won’t we?
It's a source of constant amusement to me how some people regard us as Capitalist Pig Whore Sellouts when the research results don't agree with their perceptions. But when the results back up what they believe, then we're merely reporting and analyzing industry facts that are obvious to all.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
