AMD's Six-Core Istanbul Opterons 123
EconolineCrush writes "AMD's latest 'Istanbul' Opterons add two cores per socket, for a grand total of six. Despite the extra cores, these new chips reside within the same power envelope as existing quad-core Opterons, and they're drop-in compatible with current systems. The Tech Report has an in-depth review of the new chips, comparing their performance and power efficiency with that of Intel's Nehalem-based Xeons. Istanbul fares surprisingly well, particularly when one considers its performance-power ratio with highly parallelized workloads."
Istanbul runs your shells (Score:5, Funny)
Through shaves as tight as Dardanelles.
Use Opteron and the gallant foamy,
And thus avoid Gallipoli [wikipedia.org].
Burma Shave
Re:Istanbul runs your shells (Score:5, Funny)
Istanbul was Constantinople
Now it's Istanbul, not Constantinople
So if you were waiting for a core called Constantinople
It's been released as Istanbul.
Re: (Score:2)
Back to Byzantium (Score:2)
Uhhharg. I thought that song was hilarious the first time I heard it. But it's turned into a mind-worm that goes round and round every time I hear "Istanbul". And I hear it a lot because I'm writing documentation for an Istanbul-based server.
Just for that, I'm going to force you to watch this really dumb video [youtube.com]. You are required to drink a V8 every time you spot a geographical blooper.
Actually, a lot of Greeks find this song extremely unfunny, because the name change reflects the way Greek communities have b
Re: (Score:1)
Re: (Score:2)
But wasn't Istanbul called Constantinople?
And what do the Turks think about that?
Re: (Score:3, Funny)
Then, not so much.
Re: (Score:1)
Re: (Score:2)
Wasn't it called Constantinople? (Score:5, Funny)
Re:Wasn't it called Constantinople? (Score:5, Funny)
Re: (Score:1)
I won't be impressed until it is... (Score:2, Funny)
Over 9000!!!!1
Now where can I ... (Score:1)
get a couple of these to test? Sounds like we could get some pretty good number-crunching results.
But... (Score:1)
And 14 cores is nothing compared to 64 threads (Score:3, Interesting)
From http://www.sun.com/processors/UltraSPARC-T2/features.xml [sun.com]
"Features and Benefits
With eight cores and 64 threads on one chip, integrated 10 GbE networking, crypto, and PCI-Express expansion, you have the jump on anything else on the market. The opportunities for system consolidation and virtualization are here like never before. Consumes less power per core and thread than any processor in its class - without compromising on performance. The UltraSPARC T2 processor gives OEMs a massively threaded, multi-c
Re: (Score:2)
Hyperthreading helps you avoid the cost of context switches when multithreading, but a) the cost of context switches is remarkably lower these days due to register renaming and other tricks and b) only on Unix do you care anyway; traditionally we spawn lots of processes on Unix and lots of threads on Windows. It's not necessarily the right way to do things, and the Windows thread-heavy model is paying off now that multicore processors have brought multiprocessing to the masses.
Re: (Score:2)
Hyperthreading helps you avoid the cost of context switches when multithreading,
I had impression that it's not about context switches. In case of Sparc T2, they actually try to execute several threads in parallel. If one thread stalls on memory access or IO, CPU picks some other thread to execute.
I can't say overall, but for well optimized C/C++ programs this is a disaster. My employer did benchmarks on Sparc T2. With HT enabled system couldn't deliver stable latencies: performance figures were shattered all over the graphs. With HT disabled it performed just like on other Sparcs,
Buy it from sun.com (Score:2)
http://www.sun.com/servers/coolthreads/t5440/specs.xml [sun.com]
4 UltraSparc T2 processors with 4 processors x 8 cores per processor x 8 threads per core = 256 threars
Press "Get it". Prices start from $91,995.00 with 256 threads and $51,795.00 with 128 threads
Re: (Score:2)
Feh. And it doesn't even run Crysis.
Re: (Score:2)
What kind of application are you running? Niagara is designed for application servers with many connections. Hence the build in ethernet and crypto support and adequate memory support. It's foolish to run anything on it that does not have multiple threads yes, and it's single thread performance is low (unless you are using the cryptographic coprocessor of course). I've torched the use of Niagara processors in my company for a specific application server as well; it was used to receive large XML data sets fr
Re: (Score:2)
That's nothing compared to 14 cores.
You are bad at math.
EPT? (Score:2)
Re:EPT? (Score:4, Informative)
AMD has supported nested page tables since the Shanghai series processors.
Re: (Score:1)
Re: (Score:2)
Re: (Score:1)
Fuck Everything, We're Doing Six Cores (Score:5, Funny)
Here's the report from Engineering. Someone put it in the bathroom: I want to wipe my ass with it. They don't tell me what to inventI tell them. And I'm telling them to stick two more cores in there. I don't care how. Make the cores so thin they're invisible. I don't care if they have to cram the sixth blade in perpendicular to the other five, just do it!
Re: (Score:2)
Re: (Score:2)
So basically we have pentium's covering each other's asses?
Only five blades? (Score:2)
Meet Quintippio [quintippio.co.uk]: The new 15-Blade Mega Shave!
Show your beard who's boss!
Re: (Score:1, Informative)
Re: (Score:2)
You know what's crazy? I have tried those different shavers, and three blades seems to be the sweet spot - with those 4 and 5 blade cartridges, there isn;t enough room between the blades for them to get close enough to your skin, for the stubble to get in there - they just don't work as well, but I guess the standard marketing bullshit applies: "More is better*, because this is America"
*if it doesn't work better, or even as well, that is okay, because the customer's pride in knowing that they have the newes
Do i need Erlang? (Score:2)
Harnessing muli-cpu machines with these installed is going to be.... Interesting.
Re: (Score:2)
Not really.
Threaded programming has been around for many years now, and multi-process computing has been around for decades. If you can't utilise multiple cores by now you're way behind the curve.
That said, I will watch the progress of these languages designed specifically for the task, though I don't see them unseating C/C++/Java any time soon.
Re: (Score:2)
Once the operating systems handle them well, and application programmers are more aware of these issues, things will be much better in multi-core-land.
It's not the OS. (Score:2)
It's the applications.
Actually, you might have a point -- I honestly don't know how well OS kernels are implemented for this sort of thing. On the other hand, Linux has been ported to machines with more cores (and CPUs!) than that before. Worst case, the kernel-level stuff won't receive a boost -- your filesystem won't go much faster -- but how much of your CPU time is currently spent there?
No, most CPU time is spent in applications, as it should be. And that's where you have the issues you describe -- eith
Re: (Score:2)
From Wikipedia:
As its name suggests, CSP allows the description of systems in terms of component processes that operate independently, and interact with each other solely through message-passing communication.
Sounds pretty much exactly like Erlang. Your description only reinforces that:
There are plenty of CSP-style large scale projects that really are assembled from discrete single-threaded programs; this works well for distribution across loosely-coupled clusters (and even topologically distant sets of those) whereas most multi-threaded programming paradigms make strong assumptions on things like expected delays and bandwidths and their derivatives (like uniformity).
It depends on the program, of course, but Erlang itself makes no such assumptions. It includes both a simple RPC system, and robust binary processing and network libraries, making it quite easy to build loosely-coupled clusters -- but each program is already written in that style to begin with, as Erlang can also run thousands of simultaneous "processes" (actors).
By using that actor model pervasively, and thinking i
Re: (Score:2)
That said, I will watch the progress of these languages designed specifically for the task, though I don't see them unseating C/C++/Java any time soon.
I think I prefer languages matched primarily to the problem the program is solving, rather than languages matched primarily to the hardware used to run the program (primarily; some degree of the latter is necessary, for example if your hardware is a GPU or an FPGA). ;)
No. (Score:5, Insightful)
Harnessing muli-cpu machines with these installed is going to be.... Interesting.
No more interesting than existing many-core machines.
Seriously, having a couple dozen or more cores is nothing new.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
not 'tunnels'. the word you're looking for is 'tubes'.
as in the famous revelation about teh internet: "My god, it's full of tubes!"
Re: (Score:2)
Re: (Score:2)
HyperTransport is not a big truck! It's a series of tunnels!
Too bad it's not a station wagon full of tapes.
Re: (Score:2)
Well, yes you can, you just have to design your software competently. And I'm pretty sure that today's chip designs allow them to be faster per core than your old 3.5GHz P4, just by the massively improved branch prediction, faster/integrated memory bus etc.
I don't know why games tend to still be single threaded. I would think video encoding could be parallelised quite nicely too. It'll just take some work.
Actually, after a quick google - avidemux, ffmpeg and mencoder have supported threads for some time.
Re: (Score:2)
The fact that single-core speeds have now climbed back up certainly does not negate the point he was making. Scalar speed is very important, and exponential diminishing returns result from attempting threading with most common CPU-intensive tasks.
And, for the most part, you'd be wrong. In the worst case, you get about a 10% speed-up even
Re: (Score:2)
Single core performance have not just climbed back up, it has increased by a factor of ~3.
If you take a single thread task, and run in on a 3.5GHz Pentium IV, and then run the same task on a new quad core chip form Intel/Amd,
then the quad core chilp will finish the task 3 times faster. (Give or take, depending on exact instruction mix, sse usage and so on).
(And yes that is for single core task, if the task were threaded, you might gain a factor of 10).
Re: (Score:2)
As the other responder said - the speeds haven't climbed back up, they've massively surpassed the old P4. Just because the P4 had a fast clock speed doesn't mean it was faster.
It's not. Core 2 at a lower speed can far outpace it. Quit believing the old MHz thing, that's only relevant when comparing apples to apples, not apples to out of date architectures.
Re: (Score:2)
If it runs acceptably on a fast P4, then a single core of the Core2 is going to own it anyways.
Rendering is outsourced middleware (usualy from I.D., Epic, or Valve) in the majority of games, and generally *is* multi-threaded at one point or another (even the core rendering API,
Re:No. (Score:5, Informative)
You've already made this comment before, and I've already responded, so I'll keep it short and sweet.
If you're using a slow 2.2 GHz Quad core, that's not the fault of the industry, that's the fault of YOU. I have already made it clear that the top-end Core 2 Duo chips [newegg.com] would run circles around your P4, but apparently you'd prefer to pretend they don't exist. As for your dog-slow quad core, that was YOUR purchasing decision. You can purchase MUCH FASTER [newegg.com] quad cores [newegg.com] today for reasonable prices, but apparently you're still suck in the year 2006.
The reason Core 2 / Quad destroys the P4 despite having a slower clock speed: Core 2 ups the Instructions Per Clock versus the Pentium 4. The increase is between %60 and %100 more IPC. If you read my previous response to you on the subject [slashdot.org], you'd actually know that, instead of continuing to spout your ignorant bullshit.
And if you can't find a video codec with multiple core support, you're looking in the wrong place. Video decode is one of those embarrassingly-easy things to parallelize, and so your "boast" is really just outing you as a lazy bastard who can't take five seconds to search Google.
Re: (Score:2)
Core 2 also has a shorter pipeline than P4, and a better branch predictor. It spends a lot less time just throwing away results than P4.
Re: (Score:2)
He is making a point about the diminishing value of multiple cores. The fact that the industry has (finally) gotten back up to the point where a single core is clocked faster than a 10 year-old chip doesn't negate the point at all...
Re: (Score:2)
One thing that could help is fast memory. I've found dual channel DDR400 to be faster than stuff like DDR2-533, which may be the speed you'll find in a low-end newer multi-core machine. And since 3.5Ghz is not actually a speed of P4 Intel ever sold, if he's not making stuff up and is running an overclocked P4 then his bus speeds may be even faster than that. Though I don't get turning off hyperthreading - I've found that the P4 generally performed much better with it on than off, unless you're running on
Re: (Score:2)
My 3.5ghz Pentium 4 with the useless multithreading turned off kicks the crap out of HD Video rendering than anything else.
Try a Core i7 at around 3GHz and be amazed.
If you have a multi-threaded app, you get 8 really usable threads. Running six copies of LAME at the same time, I can convert a 12-track CD into MP3 in about 60 seconds on a 3.33GHz i7 920.
Video conversion is similarly speedy, although HD isn't as good without a lot of memory, too, as there's a lot more bits to move around.
Re: (Score:2)
Re: (Score:2)
No really, Newegg was selling E5200s for $10 extra on bundle with a Lexmark X4850, and I needed a printer. So hey, $10 CPU, how bad can it be? So I throw it in a P45 board and shake some DDR2 out of the box o random dimms, and it turns out it really likes 12x333 at stock voltage with a cheap heatsink. Since the board supports the 333/1333fsb officially, the rest of it runs at stock speeds like ddr2 1066. Prime95 and HyperPi run all day without cras
Re: (Score:2)
Nope. This is a server CPU. Things like Database servers already scale. well.
Virtualazation by definition will scale well.
Or to put it in simple terms.
You know that old four server with 8 cores total? You can now replace it with a two socket machine with 12 cores total.
Or you know that four socket 16 core server? Well you can now upgrade that to a 24 core server.
Finally (Score:2, Funny)
Re: (Score:3, Insightful)
I'll be finally able to run Crysis at a decent framerate.
Just in time to be behind the curve for Crysis 2!
Another test at anandtech.com (Score:5, Informative)
http://it.anandtech.com/IT/showdoc.aspx?i=3571 [anandtech.com]
Includes information about virtualization performance: http://it.anandtech.com/IT/showdoc.aspx?i=3571&p=9 [anandtech.com]
Conclusion:
"The six-core Opteron is not an alternative to the mighty Xeons in every application. The Xeons are more versatile thanks to the higher clockspeeds, higher IPC, Hyperthreading and higher bandwidth to memory. The Xeon 55xx series is clearly the better choice in OLTP, ERP, webserving, rendering and there is little doubt that it will continue to reign in the bandwidth intensive HPC workloads. There are two types of applications where we feel that the AMD six-core deserves your attention: decision support databases and virtualization."
Re: (Score:3, Interesting)
I believe Anandtech is showing it's bias here. I had heard great things about the Xeon 55xx series CPUs so I went and bought a couple of servers. Specifically one web server and one database server. I also had Opteron-based servers performing the same tasks. My webservers are load balanced using a hardware load balancer. During January I was under an extremely heavy load scenario. I ended up having to weight more traffic to the Opteron servers because the Xeons were choking under 100% cpu load. I barely squ
Re: (Score:1)
Wow, you mean for _just you_ your old Opterons perform better than a chip that is quite superior to it in every way including memory bandwidth?
I think _someone's_ showing their bias here...
Re: (Score:2)
Re: (Score:2)
Are they seriously touting hyperthreading as a benefit? It's a dubious-enough feature, but with 4 cores, it really stretches believability. I dare someone to find the one application that benefits from seeing 2 additional fake CPUs when there are already 4 real ones.
Re: (Score:1)
Re: (Score:3, Interesting)
If your algorithm is designed to break up the problem to exploit the ca
Re: (Score:1)
Scary Quote from the Article (Score:5, Interesting)
[...] Not only that, but it's hitting the market early. AMD had originally planned to introduce this product in the October time frame, but the first spin of Istanbul silicon came back solid, so the firm pulled the launch forward into June. Even with the accelerated schedule, of course, Istanbul comes not a moment too soon, now that Nehalem Xeons are out in the wild.
Does anyone else think that this seems a little convenient? I'm really hoping that they didn't just tone down the testing to make it to market. I'm thinking they'll go to market and then quickly release a new revision to fix the corners that they cut the first time around. I hope I'm wrong, but AMD has been slipping lately.
Any EE's out there know the process well enough to confirm or deny my suspicions?
Re: (Score:2)
I think AMD learned from their last mishap. It nearly destroyed the company.
Re: (Score:3, Interesting)
bit-tech: Has the launch of Istanbul been brought forward in response to Nehalem EX's updated launch date?
Patler: Istanbul being pulled in by five months is a result of excellent execution by our design and manufacturing teams who were about to take it from first stepping of silicon to production. Also, the fact that Istanbul is based on our existing socket infrastructure, enables our OEMs to save time on validation cycles that are normally associated with a new processor that delivers the performance Istanbul can.
Re:Scary Quote from the Article (Score:4, Insightful)
I'm in the silicon business. Not CPU, but still silicon.
It sounds as if AMD budgeted time for another pass at the design, and turned out not to need it. The amount of time they pulled out of the schedule looks more like a silicon pass than short-cutting testing and validation. Adding that extra pass, and making sure it was scheduled is probably a result of having been so badly burned last time, but that's good. You can always be a hero by doing better than plan.
Re: (Score:1)
Re: (Score:2, Interesting)
Actually, testing was increased for 6Core... As our 4Core tests no longer stressed a system with 50% more cores the same way.
What changed was Process. 6Core uses all of the 'good' tech from Shanghai, then implements a few things differently (rev upgrades, etc). The reason 6Core launched soo quickly, is we learned all of our lessons on the initial quad core fiasco. We did things 'right' this time, and the result is... a launch date that is nearly 12mos ahead of the initial schedule (which was set 2yrs ago
Yeah, but... (Score:2, Funny)
Yeah, but will it run a hackintosh?
Re: (Score:2)
How's this offtopic? It's a legitimate question. I have an older AMD that will not run the hackintosh software. I like AMD products - they _seem_ to be faster - but I'm not spending money on this, as nice as it may be, if it won't run what I want.
Re: (Score:2)
This is a (very expensive) server CPU... I don't think you're going to spend money on this to run hackintosh either way.
Underutilized (Score:1)
How many of your favorite app already re-written to take advantage of the additional cores?
How many of your favorite compiler already re-designed to generate codes that uses additional cores?
How many of your favorite boss already re-wired to fuss about the additional cores?
Re: (Score:2)
My favourite apps are written in Fortran, so it only takes a nice compiler to generate multiprocessor code from it. The first time I did something like that was in 2001, so the compilers have certainly been around for a while.
Speaking for Intel, spokesperson Nigel Tufnel said (Score:1, Redundant)
When asked by the reporters, as to why Eleven was chosen as the target number of cores, Nigel said
It's six louder than AMD! I mean faster...
In other news ... (Score:1)
it's interesting, but not becase of 6c (Score:5, Informative)
the real news here is not the extra couple cores, but coherency snooping. this feature will make 4/8s machines far more attractive; it doesn't hurt that with 48 cores and 32 ddr3/1333 dimms, you have quite a monster. _and_ incidentally something that Intel can't currently answer.
there's no question that nehalem has put a serious dent in the market, but Intel's going quite slow in rolling out higher-end products. yes, a nehalem socket delivers about 50% more bandwidth than a current opteron socket, but show me the 8s nehalem machines. nehalem-ex is coming, but how soon and at what price?
one thing I haven't seen is any attempt to measure real SMP performance on new-gen chips. I don't mean something like Stream or VMs, where there is no real sharing inherent to the workload. how long does it take to exchange a _contended_ lock between cores (in the same socket vs remote)?
finally, the real question is whether there is actual demand for more-core chips. I'm in HPC, and we always want more, and throw good money. but it has to be smart more - the 6-core core2, for instance, was just asinine because even 2c core2 is drastically memory-bandwidth-starved. nehalem-ex seems quite promising, but if it's cheaper to cluster dual-socket machines rather than pay the premium for 4s's, the 4s market will be stunted and less successful in a self-fulfilling way...
Re: (Score:2)
the real news here is not the extra couple cores, but coherency snooping. this feature will make 4/8s machines far more attractive; it doesn't hurt that with 48 cores and 32 ddr3/1333 dimms, you have quite a monster. _and_ incidentally something that Intel can't currently answer.
That's actually 16 channels of DDR2/800, according to page 1 of TFA. I think it's supposed to be what comes out after this one that goes to 4xDDR3 per socket.
Re: (Score:2)
Scaling vertically hasn't been a good idea for a long time unless you're app has trouble scaling horizontally. I'm in the process of creating a proposal with a back-end database cluster considering of 4-6 nodes. Now I could achieve the same horsepower by buying an 8s or a 4s server and not have to buy as many machines but 4s servers seem to be 3 times more expensive than 2s socket servers so I can just buy more dual processor servers and scale out to achieve the target number of connections served.
Of cours
I'd rather have faster disk I/O (Score:3, Informative)
We run a lot of commerical OCR (as in millions of images), which is extremely processor-intensive, disk-intensive, memory-intensive, you name it. Our current main OCR server is a dual quad-core Xeon X5355 box with 16 GB of RAM. Our OCR software multithreads and the processor is no longer the bottleneck -- it's now disk I/O. While current drives continue to increase in size, their read / write speed is what keeps us from getting work done faster. It now takes several orders of magnitude longer to build,
Re: (Score:2)
Maybe I'm missing something here... but if you process data in 2GB chunks shouldn't your software just keep it all in memory. Once processing is complete writing it out to one of those SSD arrays should take 10 seconds (which is nothing for 2 hours worth of processing time!!!). If you don't have access to the source code, a quick fix is to just mount a RAM drive.
Furthermore, OCR is stupidly easy to parallelize. The results of each page do not depend on previous pages. You can process each page independ
computers are more confusing than ever (Score:2)
Tried pricing up a decent box for some heavy-lifting, there's just so much complexity out there! It's hard to figure out where the bleeding edge is and where the most effective bang for the buck zone is behind all the blood. 286, 386, 486, a man used to be able to tell where computers sat! And then all that Pentium bullshit started. I don't know what the fuck I'm looking at. I'm crossing my fingers and going with a Tom's Hardware recommended build list.
video? (Score:2)
Anyone have any clue how Nehalem and these multicore AMD beasts would compare for video editing or render farm applications?
Re: (Score:1)
Mmmm...yeah, but wasn't Constantinople sacked by the Turks, thereby causing the fall of the Eastern Roman Empire?
Oh, wait, I think I get it now... ;)
Re: (Score:2)
yeah, but wasn't Constantinople sacked by the Turks
see what happens when you don't have enough stock to satisfy consumer demand :)
dont bullshit (Score:1, Offtopic)
egyptian civilization started well around 4000 BC and lasted until 400 BC. thats around 3600 years.
eastern roman empire is from ad 250 ish to ad 1453.
EVEN if you add entire roman empire history unto that, which makes from 500 BC to 1453 AD, it still makes 2000 years. doesnt come anywhere near egypt.
Re: (Score:2)
He said Egyptian empire, which is somewhat different from just Egypt, just as the Roman empire is different from just Rome.
Re: (Score:2)
There never was an Egyptian empire, there was the state of the pharaos, starting about ~2700 B.C. and falling to Alexandre the Great in the 4. century BC, followed by the greek-ptolemaean Egyptian kingdom, which was coming to an end in 30 BC with the suicide of Cleopatra.
PS: We are completely offtopic, because Byzanz/Constantinople/Istanbul never was a part of Egypt anyway.
'so what' is that : (Score:3, Informative)
ancient egypt is THE source of many of your philosophies and sciences. from 1000 BC and onwards, early greeks were coming to egypt for education. egypt had 2 schools - school of life, and school of death. school of life was teaching stuff related to this world, ie, medicine, land registry, writing, government, and school of death taught stuff pertaining to abstract world. not to mention that many of the professions people identify themselves today originated in egypt.
even before knossos was known, medicine
Re: (Score:2)
yeah, but there would be no Egypt without the global warming trend that made the Sahara desert...
Heh, now we are REALLY off topic!
Anyone got a 6-pack of Fascists! to finish off the party?
Re: (Score:2)
Huh? If you take the founding of the Eastern Empire to be Constantine's moving the capital to Byzantium (renaming it Constantinople), that's 330AD to its fall in 1453. Dynastic Egypt traditionally dates back to somewhere around 3100BC, and the end is usually marked as the fall of the Ptolemaic Dynasty in 30BC. So t
Re: (Score:2)
The Sea People
The Nubians (Kush)
The Assyrians
The Persians
Alexander
The Ptolemaic era was the last time a culture with any kind of connection to the Old Kingdom ruled Egypt but it makes no sense to talk about the time from the Old Kingdom to the Ptolemys as the duration of the "Egyptian Empire"
Actually, the only time when Egypt really had an empire (a small one) was the New Kingdom. That period was 1550-1070BC and considerably shorter t
Re: (Score:2)
If that's the case, then you'd better include the Latin Empire into your counting of years for the Byzantine Empire. That was a major interruption of the political continuity of the Empire (and pretty much lead to its final decline).