Big-Iron to Open Up for AMD 179
vincecate writes "Traditionally the key chips that have allowed companies to
scale multiprocessors to large numbers have been proprietary.
Some examples are the
Cray SeaStar,
SGI NUMAlink,
HP sx1000,
and the
IBM X3/Hurricane.
This proprietary paradigm is about to change to a more open one.
Two companies have developed key chips for
building large Opteron multiprocessors,
and they will be
commercial off-the-shelf parts.
PathScale has
released
InfiniPath
which can be used with an
Infiniband
switch to make
a high-bandwidth low-latency interconnect for a
supercomputer cluster.
The other company is
Newisys,
which
will soon release
the
Horus chip.
This chip will make it possible to build 32 socket
(64-core) shared memory Opteron systems."
it's about time (Score:5, Funny)
Re:it's about time (Score:4, Funny)
Expect to see.... (Score:5, Funny)
Re:Expect to see.... (Score:2, Informative)
BTW, Has anyone heard of the MLX1 [extremetech.com]. Makes you wonder what would happen if you put a bunch of these on a chip with some clever caching and the mother of all memory controllers. x86 Niagra [aceshardware.com] anyone.
Re:Expect to see.... (Score:4, Funny)
64 processors will let you run a lot more spyware before your frame rate is affected.
multithreading not pervasive... yet (Score:2)
Re:Expect to see.... (Score:3, Funny)
Fortunately all computing isn't about games. With this setup you'll be able to encode those DVD rips to DIVX in seconds instead of hours!
Re:Expect to see.... (Score:2, Interesting)
E.g.
http://www.anandtech.com/printarticle.aspx?i=2410 [anandtech.com]
"Gaming performance is, currently, highly based on single-threaded performance and thus, we see no benefit from dual core. The thing to keep in mind here is that AMD's dual core solutions are closer to their fastest single core offerings in clock speed, so they end up performing more like their Athlon 64 counterparts in games - which has always been quite
Do not count out Sun (Score:2, Insightful)
-GenTimJS
Re:Do not count out Sun (Score:2)
Re:Do not count out Sun (Score:2)
Re:Do not count out Sun (Score:3, Informative)
TFA is about hardware, not software. What OS Sun runs on their servers is irrelevant to this discussion.
Quoth twiddlingbits yet again:
I don't think that you understood the word 'proprietary' in this context. Sun's technology, as well as (IBM's, SGI's, Cray's
Re:Do not count out Sun (Score:2)
Look at these pictures:
Sun V40z [sun.com]
Newisys 4300 [newisys.com]
Newisys was one of the first places to design Opteron based systems.
Indeed. (Score:2)
If I were a betting man, I'd give you my trousers.
Re:Do not count out Sun (Score:3, Interesting)
Re:Do not count out Sun (Score:3, Insightful)
Re:How about 2560 Opterons? (Score:2)
HUGE.
Re:How about 2560 Opterons? (Score:2)
-nB
Re:How about 2560 Opterons? (Score:2)
While I agree that the driver support seems to have gotten better over time, and in fact it looks as if AMD _may_ be finally coming into their own, at least as a gaming platform, I've been burnt one too many times by AMD to go ba
Re:How about 2560 Opterons? (Score:4, Interesting)
So I'm a big curious what is so vastly superior? Are you using Intel compiled codecs on AMD machines when you did your testing? Did you even do any testing? I'll admit I had some trouble getting things running smoothly with the Opteron box but the end results speak for themselves; especially when you move over to the 64bit world with 64bit capture drivers the Opteron blows away anything Intel has put out to date. Of course Intel 64bit support is slow as all hell right now so I'm sure that will change in the near future.
While you may have been burnt by AMD I will stick with them for the time being until Intel shows some signs of turning around their product offerings. I'm still curious how a processor has gone bad though. In my experience once you get back the first 90 days its smooth sailing regardless of manufacturer. Only reason I can think a chip would die later in life would be from a PSU failure or some sort of disruption. I've seen that happen, never just seen a cpu die though. Always some other component causing it.Of course this is getting off track from the article. The Opteron is very well suited for these large machines so I'll be curious how they perform in real environments like Oracle and DB2 setups. Opterons bandwidth improve the more processors you throw at it so it'll be intriguing to see the results.
Re:How about 2560 Opterons? (Score:2)
Re:How about 2560 Opterons? (Score:2)
Re:How about 2560 Opterons? (Score:2)
I don't regret the purchase, much as I don't regret the purchase of the Intel compilers or the IBM compilers for PPC either. All are infinitely preferable to GCC if you can make back what the compilers cost. They all have better ANSI compliance for example, and much better performance in the scientific apps I do.
Re:How about 2560 Opterons? (Score:2)
I'm curious too what is different in your setup to be able to stream 2x more data with comparable systems. Albeit the 1.8GHz Opteron has a little more horsepower than a 2.8GHz p4, but that should have little to nothing to do with video streaming. Is the Osprey in a pci-x slot on both machines? Are the memory subsystems similar?
I'm just very curious. Because 2x performance on similar machines has to be explainable beyond the processors that are almost the s
Re:How about 2560 Opterons? (Score:3, Interesting)
Re:How about 2560 Opterons? (Score:2)
This is better than a benchmark, this is real work.
I asked for you to clear things up, and you bring up more variables into the mix.
If you could, please clarify the following:
1) you mentioned dual core opterons. Was your 2x performance noticed on both the single and dual core?
Block sizes greatly effect performance. I set the Opterons to write in 8meg chunks and it seems to save
Re:How about 2560 Opterons? (Score:2)
The last Intel box I built was a BX chipset (440BX?). I have (2) AthlonXP systems still running (one is a VIA chipset the other is an nForce chipset). Plus 3 Optero
Imagine a... nah, too easy. (Score:5, Interesting)
... k, maybe not. Can't afford one anyway :-(
Re:Imagine a... nah, too easy. (Score:2)
Sure, as soon as you stop using that horrible term 'boxen'....
Re:Imagine a... nah, too easy. (Score:4, Funny)
I imagine it warms the heart of a lot of us old AMD fanboys.
Plus anything else in the room it sits in. Great solution to the heating oil crunch! These things ain't P4s, but 32 CPUs is 32CPUs.Re:Imagine a... nah, too easy. (Score:2)
Clusters vs. single servers (Score:5, Insightful)
Horus is used for building Opteron ccNUMA machines with one OS instance that can run any Linux or Windows apps. It's a very different solution for a different market.
Re:Clusters vs. single servers (Score:2)
Hasn't Beowulf been out there long enough for people to understand this?
Re:Clusters vs. single servers (Score:2)
But yeah, from a user perspective Infinipath is just like any other cluster interconnect. Perhaps a bit lower latency, but nothing dramatic (as in order(s) of magnitude lower latency).
Links (Score:4, Informative)
Oh, and the Horus link is a PDF whitepaper... please warn when a link points to a PDF.
Re:Links (Score:5, Informative)
Re:Links (Score:2)
Re:Links (Score:2)
Re:Links (Score:2)
Why? You can set up your browser to handle a pdf link any way you like (if it bothers you that much, you can redirect yourself to a google html conversion of it).
Re:Links (Score:2)
god damnit (Score:5, Funny)
Re:god damnit (Score:3, Funny)
Re:god damnit (Score:3, Funny)
Is 32 (Score:2, Interesting)
Re:Is 32 (Score:5, Funny)
For this context, 32 is plenty large. Large is relative. If you ask me how many grains of rice I ate last night, 100 would be a small number. If you ask the average slashdotter how many women he's dated, 1 is a huge number.
Re:Is 32 (Score:3, Funny)
And the date was probably huge too.
Re:Is 32 (Score:2, Funny)
Or integer
Re:Is 32 (Score:3, Funny)
heheheheheheheheheh... math humor... heheheh heheheh heheheh
Re:Is 32 (Score:2)
My poor wife.
Re:Is 32 (Score:2)
same timeframe that retail chassis are available using these chipsets, quadcore CPUs
will be on the market. That puts 128 modern CPUs in one box. Rare indeed is the whitebox
application that needs to scale beyond that. And if you cluster *those*, you're rapidly
approaching the top end of existing systems for HPC throughput.
So what? (Score:2)
This is news? We've been using an Infiniband-connected Opteron cluster for almost a year now. I got bids from half a dozen companies willing to sell us one. This is old, established tech.
Re:So what? (Score:3, Informative)
Re:So what? (Score:2)
hypertransport or pci-e|X or whatever connection you may have my friend. 32 cpu's running at 1800mhz or more are so hard to synchronize that imho this whole "megayhpe" is much talk about much nothin. yeah 32 cpu's might be a good idea, but the only occasion you really want a smp machine is to use software that uses shared memory and shared i/o access
Proprietary? (Score:2)
threat to big iron (Score:3, Interesting)
Re:threat to big iron (Score:2)
Re:threat to big iron (Score:2)
In the world of server and process consolidation, there are 4 or so 'levels'. Sun's partitioning addresses 1 of these levels. Xen/ESX/UML addresses another. (Note that the cost per hosted OS/Application goes down as you go down the levels. )
At the top level, you have things like Sun's har
Re:threat to big iron (Score:2)
While I'm excited to see the possibility, you're not going to get anyone to spend this much money on a 64-way opteron box until they have been on the market for years, have been tested, and tried, and have lots of softw
Re:threat to big iron (Score:2)
Since Sun is already in the Opteron camp, none of this news hurts them. With the cheapness of x86 hardware
Re:threat to big iron (Score:3, Insightful)
Would you buy that from Mom or Pop?
A Sun Enterprise system is a system that is supported by the vendor for its OS (Solaris) and the hardware as well as other players, as the other systems you mention.
A 32-way SMP dual-core opteron box is just a figment of your imaginati
Re:threat to big iron (Score:2)
Both the HP superdome and most IBM Power systems run Linux too.
The value of the "big-iron" unix systems is not really in their CPUs, it is in their chipsets, which Horus does compete with, but also redundancy features like ECC and chipkill, redundant power systems, etc. Once you start adding all the equivalent featurues to a horus-based opter
Re:threat to big iron (Score:3, Interesting)
Strange use of "open" (Score:3, Insightful)
Re:Strange use of "open" (Score:2)
The AMD chip is still proprietary as are the interconnect technologies. The fact that they come in packages you can assemble yourself to build your own custom solutions is really the focus of the article, but completely besides-the-point in terms of "openness". I don't mind the article or the technology but I don't see the nee
Re:Strange use of "open" (Score:2)
Re:Strange use of "open" (Score:2)
Well, the Win32 API is documented, too. Is it open?
Re:Strange use of "open" (Score:2)
Re:Strange use of "open" (Score:2)
I don't like to confuse the fact that the API/ABI is public with the notion that the product itself and all its workings are non-proprietary -- because they are proprietary.
None-the-less, I'll conc
Re:Strange use of "open" (Score:2)
Re:Strange use of "open" (Score:2)
Re:Strange use of "open" (Score:2)
Open means you can see inside. I can see inside the API, but not inside it's implementation.
By publishing books on the subject, Microsoft has made the design open, to the degree
that it is accurately documented, but not the code. OpenSolaris is open at the code level,
but not at the process level: It does not use an open development process.
I think this is a common jargon use of the term "open", and it corresponds closely
Open Processors (Score:3, Informative)
Umm, I know there's this odd phenomenon where many people tend to label any processor that's made by either Intel or AMD "non-proprietary" and any processor made by another company "proprietary", but even still this article is a little silly. SPARC processors have been in use since the late '80s, most people consider SPARC-based machines "Big Iron", and the SPARC processor architecture is fully open -- anyone who wants to can make SPARC processors. SPARCproductDIRectory [sparcprodu...ectory.com] lists a bunch of companies who currently do. In fact, there are probably just as many (if not more) SPARC manufacturers as there are X86 manufacturers.
Word from the Mac community... (Score:2, Funny)
Great AMD is quit is doing fine. (Score:5, Insightful)
Or the server which can run highly debugged application written in mainframe assembler in 60's or 70's ?
Or atleast AMD computer with SINGLE memoryspace atleast 1TB in size?
And also how many decades of uptime is for the operating system which is used with the new AMD computer?
The horus is more or less getting close to midrange server in number of processor while it won't bring it to the reliability requirements of midrange server, to get that it would have to run its own memory controllers instead of cheap ass opteron controllers which lack for example hotswappable memory.
Sure you get speed, but after taking the speed there is eventually a crash.
The big iron is all about gettin continuing to function no matter what comes.
Only problems outside of box, like earthquake or something similar could bring it down.
Yeah. AMD is doing just fine...
Its eating the cheap ass market, not the big iron.
The price is cheap and its bought where the crash proof means better than windows which is like saying saying its unsinkable since it does better in open seas than normal rowboat used in lakes.
Lets put it this way. x86 is just used in low end boxes and in clusters of lowend boxes. And those things are not for everything. They can do much but not everything. They are cost effective when you compare only the purchase price. But not so cost effective when downtime costs a lot.
There is probably order of magnitude or TWO orders of magnitude of what joe slashdotter thinks big iron and what businesses have in big iron as in price range.
Re:Great AMD is quit is doing fine. (Score:2)
I'm having cheap ass Athlon 64 computer under my desk at home.
And I think its great, but its still a very low end computer.
And I think linux is a lot better than windows, and use it but its still no where near the mainframe OSes.
Yes the mainframes have linux as one of their guest os. Running on top of a real os, but thats another story.
Re:Great AMD is quit is doing fine. (Score:2)
You don't, you buy two.
Or atleast AMD computer with SINGLE memoryspace atleast 1TB in size?
Two thoughts on this. 1. It's really, really expensive. 2. The mainframe assembler from the 60's or 70's is 24-bit or 31-bit addressing at most, 1TB or RAM won't help you.
Only problems outside of box, like earthquake or something similar could bring it down.
Now you put the second one 1000km away.
BTW - I've replaced IBM mainframes with clusters of 4-way Opte
Re:Big Iron? Uhhh... (Score:5, Informative)
Re:Big Iron? Uhhh... (Score:2)
Re:Big Iron? Uhhh... (Score:5, Insightful)
Re:Big Iron? Uhhh... (Score:2)
Re:Big Iron? Uhhh... (Score:2)
I think somebody (ie, you) is confusing what big iron *is* for what it *was*. Things have changed in the days of the Beowulf cluster.
Re:Big Iron? Uhhh... (Score:2)
No-one seriously wants to run a bank on a Beowulf cluster.
Re:Big Iron? Uhhh... (Score:2)
Nor do they want to spend 100x the cost of a Beowulf cluster on a single big iron machine that has less processing power. Which is why companies like those in the story are bridging the gap by speeding up the latency of multiprocessor computations. So you use commodity chips, get great performance, lower prices, and don't deal with clusters.
Despite your assertions, the days of classic big iron are over. DEC got bought and reb
Re:Big Iron? Uhhh... (Score:2, Interesting)
I've worked for Banks, Airlines too
The all had BIG iron in the server rooms
These people are not doing lots of hard calculations
They are moving large amounts of data...
Clock speed doesn't matter that much in these cases.
I/O bandwidth is king for these applications
The fact that these machines are ULTRA reliable
Is a really big deal for these companies.
You can't just reboot you're entire banking platform
to add disk or fix broken hardware.
Processing power just isn't the point in these
DEC, SGI, and Cray never made Big Iron. (Score:2)
Those of us who still work in the airline, banking, or insurance industries know that most of the current mainframe systems aren't going anywhere. They run on B
Re:DEC, SGI, and Cray never made Big Iron. (Score:2)
After my latest experience with EMC equipment I have concluded th
Re:DEC, SGI, and Cray never made Big Iron. (Score:2)
But, since the whole point of this thread is sales, how many of those systems are being replaced? And how quickly? And is that enough to prevent the general demise of so-called "Big Iron" from a sales perspective?
My guesses are not many, slowly, and no, respectively. There will always be niches for everything (even COBOL, G
Re:Big Iron? Uhhh... (Score:5, Informative)
There is a Cray XT3 [top500.org] that runs at 15 Teraflops at Sandia and made out of 2ghz opterons and is currently the 10th fastest computer in the world. There is a similar machine [top500.org] over at Oak Ridge National Labs that runs at 14 Teraflops and is the 11th fastest computer in the world.
In fact, those lowly AMD kids seem to also have their chips on the fastest machine at the Pittsburgh supercomputing center [top500.org] (ranked 33rd fastest computer in the world) and the US Army Research Laboratory (ranked 39th fastest) [top500.org]. The latter was actually being built by IBM for ARL, you know those guys who coined the term "big iron".
Re:Big Iron? Uhhh... (Score:2)
Re:Big Iron? Uhhh... (Score:2)
Re:Big Iron? Uhhh... (Score:2)
a) was I talking to you? a: NO!
b) did I say it was a manufacturer thing? a: NO!
c) Kid? Don't call me kid you fucking loser!
From +5 Insightful to 0 Flamebait in 1 hour (Score:2)
Compiler technology (Score:2)
The benifit of Sun's mature sparc compilers might let you squeeze more performance out of a sparc box. Although of course the opterons have a higher clock speed in the first place.
Re:Compiler technology (Score:2, Interesting)
I didn't realise that the compiler was involved in optimising code for multiple processors. I thought it was the programmer that had to do that. I must have been misinformed.
The benifit of Sun's mature sparc compilers might let you squeeze more performance out of a sparc box.
So are you comparing the Sun compiler and GCC on Sparc or Opteron because you don't make that clear? One woul
Re:Compiler technology (Score:2)
Re:Compiler technology (Score:2)
I meet way too many comp sci people of today, quite a few of them extreme open source advocates, who can only work with GCC, and are very poor when it comes to writing ANSI-compliant code. They are too used to using GCC and the non-standard extensions. Quite ironic, given their pretentious drivel about "standards compliance"
Re:SMP memory model? (Score:3, Informative)
Re:SMP memory model? (Score:2)
Re:about time and huh? (Score:2, Informative)
And what makes you qualified to state this? Opterons were designed with EXACTLY this in mind, right off the drawing board - i'd dig up some old articles about it but I'm at work. Research Non-Uniform Memory Access (NUMA), supercomputing, and the Opteron's architecture as it related to those two. AMD knew what they were doing when they designed the Opteron - Intel has been completely out-engineered.
Re:Not much to say about AMD, (Score:2)