45nm Opteron Performance, Power Efficiency Tested

An anonymous reader writes "Now that Intel has unleashed its next-generation Core i7 processors, all eyes are turned to AMD and its incoming wave of 45nm CPUs. To get a feel for AMD's future competitiveness, The Tech Report has taken a pair of 2.7GHz 45nm Opterons (with 75W power envelopes) and put them through the paces against Intel Xeons and older, 65nm Opterons in an extensive suite of performance and power efficiency tests — from Cinema 4D and SPECjbb to computational fluid dynamics and a custom XML handling benchmark. The verdict: AMD's new 45nm quad-core design is a notable improvement over the 65nm iteration, and it proves to be a remarkably power-efficient competitor to Intel's Xeons. However, 45nm AMD chips likely don't have what it takes to best Intel's Core i7 and future Nehalem-based Xeons."

Comments

AMD had it going

[bb84]

...but have since really lost momentum and competitiveness. They truly awakened the sleeping giant when they were kicking Intel's ass a few years ago.

Re:AMD had it going

[speed of lightx2]

...but have since really lost momentum and competitiveness

Seven out of the top ten supercomputers in the latest top500 list have AMD in them, including the top two, so I don't really see the whole "AMD losing momentum and competitiveness.

Re:AMD had it going

[bb84]

and out of all computer owners in the world, how many of them have supercomputers? Right now, in the consumer level most people consider Intel's stuff better. No bias here, but just looking at specs and performance, Intel currently sells the best goods. That's exactly what I meant--they awoke the sleeping giant. Intel has more experience, money, employees, and resources. No, that doesn't mean they have to have the best products. However, when you take all that and combine a damaged ego when AMD first whooped 'em, they pooled their talent, money, and everything else and slammed back. I interned at Intel's fab20 in 2006. People talked about AMD and how Intel really needed to make a comeback. My impression was that they were not very amused at the ratings then, and ever since the Core 2 Duos general user preference is swinging back in their direction because they started delivering a much superior product. AMD needs to get their act together if they want to hold out against Intel in the long run.

Re:AMD had it going

[LordMyren]

Yes Intel laptop sales are better.

And I find it hilarious: Intel consistently makes better mobile CPUs definitely but everything else they do in mobile space reeks to high heaven. To this day its nearly impossible to buy a Atom netbook without a Intel GMA based chipset: thats a 2 watt cpu and a 12-25 watt chipset. If you buy a normal laptop, its probably a 45w or 35w chip, even though the Pxx00 series is 25w and almost the same price, and again it comes with an absolutely worthless video card that sucks down >10 watts.

AMD certainly doesnt have as nice a processor offering. Their power is close (31w) but the performance just isnt as good. But in my mind they more than make up for it by always having power-thrifty chipsets boasting really good graphics capabilities. Amd's gone even further by offering PowerXpress and CrossfireX, allowing users to switch between integrated and discrete video cards or to use both at once (respectively). I'll take the un-noticable cpu speed hit for a huge power savings and good integrated video boon.

The biggest thing keeping AMD down in the mobile world is the systems. OEM's tend to slap together something in a cheap case missing half the plugs you'd expect when they put together Athlon systems.

Re:AMD had it going

[Joce640k]

But ... if they're cheaper than Intel then why do they need to be faster?

PS: These days power efficiency is almost as important as speed.

Re:AMD had it going

[LordMyren]

You're perspective's demented, because you think cpu performance still matters for end users. Cpu performance has always been a rat race; the difference is that its fast enough now.

Its not the numbers of computers or supercomputers you should be counting, its the number of cores. Google runs data centers with >50,000 computers; they're working on data center #20 in the states now. Yahoo, Microsoft, Sun, Ibm, Ebay, Amazon, Pixar... they all need these colossal systems to support their business. These are huge volume sales. Ask how much CPU any of these companies wants and they'll ask how much you can give them.

The desktop on the other hand is growingly irrelevant. The square-mm of the average desktop cpu are going to shrink considerably; Atom is Intel trying to cut room for x86 in clothes of devices of a much smaller size. Consumers wont need the 6 core or 12 core cpus AMD's putting out next; most can barely use the dual core they have now. In another decade I am 100% certain most desktops will have been subsumed into phones; phones with bluetooth keyboards and some hdmi-analog. Frame buffer limitations aside, we're almost at that power level already.

In the workplace, virtualization and increasing computing power will probably lead to thin clients again. Why give everyone a $900 workstation when $250 terminals and a couple heavily virtualized servers are easier to maintain?

What me and my grandparent are saying is, if you want to build big fast machines, you need someone who has a use for those super machines. And frankly I dont see any commitment aside from dedicated gamers and the businesses for whom computing is life.

Re:AMD had it going

[Henriok]

Seven out of the top ten supercomputers in the latest top500 list have AMD in them, including the top two, so I don't really see the whole "AMD losing momentum and competitiveness.

Seven out of the top ten supercomputers have Power Architecture processors in then too, including the top two, but I'd say that Power Architecture has lost its momentum, wouldn't you?

PS. For those who don't know. Roadrunner uses PowerXCell 8i processors, which are Power Architecture. All Cray XT3/4/5 supercomputers uses PowerPC 440 based communication processors called SeaStar. BlueGene uses PPC 440/450 based custom CPUs. DS.

Re:AMD had it going

[rsmith-mac]

True, but for how much longer? The reason you find Opterons in such massive servers is because HyperTransport scales up much better in 4P+ designs than Intel's ancient FSB. Now that they have QuickPath Interconnect for Nehalem/Core i7 and its derivatives, they aren't going to be held back by buses any longer. HT was AMD's one last trump card against the Core 2 generation, but they have no such card for use against the Core i7 generation.

Re:AMD had it going

[Hal_Porter]

What scares me is that AMD might decline into a purely budget CPU house like Cyrix did and then leave the market together.

Now think back to the Itanium fiasco. If AMD hadn't have been around or hadn't been making high end chips Intel could have made the high end IA64 and gradually migrate the whole market to it. So now we'd be running underpowered and overpriced IA64 chips. In a sense the thing that prevented that was that chips were dual sourced so Intel couldn't force a transition to an inferior successor like Microsoft did with XP to Vista. And IA64 was likely so patented that no one else would be ever be able to make compatible chips.

Of course with AMD around Intel was forced to adopt x64 and produce the excellent Core, Core2 and now Core i7 microarchitectures and do it very quickly. Just imagine what would have happened if they hadn't been. Recently I've heard AMD they will go fabless for example. TSMC and other commodity fabs don't have technology to match Intel, so AMD will lag behind. For low end stuff it doesn't matter much, but it really does for the high end. Mind you Intel is kicking ass in the netbook market too. It really makes you wonder how long AMD will be around. And if AMD go under so would ATI since they bought it. I actually prefer Intel and NVidia in this generation but I'm not sure they would be much good if there was no competition.

Not a very comforting thought is it?

Re:AMD had it going

[hattig]

Luckily for them Nehalem Xeons are a long time off in the computer world, especially in 4 and 8 socket variants, where AMD excels. Indeed the graphs in the review show that given another two processors, AMD would have been far more competitive. And in the server benchmarks Shanghai performed extremely well from the start, apart from the reimplementation in C# of XMLBench (instead of using the C, C++ or Java version that is well tested) that had problems.

In addition AMD have a platform that has already been tested and used by many companies. Nehalem is a major change that would require assessment before deployment.

On top of that, AMD will have a new platform out that enables HT3 and DDR3, which will improve performance, and this will be before the 4S+ Nehalem platform is out.

And maybe someone will test virtualisation in these reviews one day, where AMD will likely beat Intel into the ground (due to nested page tables and other optimisations), even on a 2S server.

Of course Intel have a far superior core and floating point. Many would argue that you should use a GPU for serious work involving the latter now of course... so would Intel, judging by their work on Larrabee.

Re:AMD had it going

[this great guy]

I wouldn't be so quick to say that AMD has lost it all.

First of all, in the 4 and 8-socket market, AMD still has no competition. The Intel Xeon MP series is still using the outdated FSB technology. This series also requires expensive and power-consuming FB-DIMM modules instead of DDR2/DDR3. Nehalem-based Xeon MPs are not going to ship before Q4 2009. Etc.

Secondly, in the 1-socket and 2-socket market, and regarding the latest 45nm AMD Shanghai and Intel Nehalem, so far there are very few benchmarks comparing directly the 2 microarchitectures; most of the hardware review sites do the mistake of comparing Shanghai against the older Intel generation, or the older AMD generation against Nehalem. But from what I have seen, clock-for-clock, for most workloads, Shanghai and Nehalem are very close, +/-10% in terms of performance, and Shanghai seems to do this in the same or a slightly lower power envelope. Some workloads do exhibit a more significant performance difference, with either Shanghai or Nehalem pulling ahead of its competitor. Now comparing clock-for-clock isn't really what matters. What matters is dollar-for-dollar comparisons. But what is interesting is that AMD has priced the Shanghai Opterons 23xx to match very closing the Nehalem Xeon 55xx series at equivalent frequencies. This tends to indicate that AMD thinks that they offer a clock-for-clock value identical or better than Intel.

The only area where AMD will clearly be unable to compete in the 1 and 2-socket market is the very high end: 1-socket Shanghai processors will top out at 3.0 GHz, 2-socket processors will top out at 2.8 GHz, while Intel goes all the way up to 3.2 GHz. However these expensive processors represent a very small proportion of the market share (virtually nobody buys $1000+ processors), so it shouldn't be a huge factor regarding which processor manufacturer "wins" this 45nm battle. Intel will have the bragging rights, but that's about it.

Another last point I would like to mention is that AMD will be the only one to offer low-power 1-socket 45nm Shanghai for at least the entire first half of 2009: 55W and 75W ACP Opteron 13xx, and 95W TDP Phenom II. While Intel will only offer Core i7 and Xeon 35xx processors rated at 130W TDP (!). They are planning to release lower-power 45nm Nehalems only during the second half of 2009. I find it rather stunning for Intel to not care more about power consumption... especially for their Xeon 55xx line, the server market cares about energy efficiency. We all remember that extravagant power consumption and temperature was a major factor that caused the failure of the Pentium 4 Netburst microarchitecture...

Re:AMD had it going

[hattig]

Power and PowerPC is doing great. XBox360, PS3, Wii, Toshiba TVs, supercomputers, set top boxes, ...

Just because it isn't doing well in the desktop PC market doesn't mean it is losing its momentum.

Re:AMD had it going

[wisty]

It probably doesn't help that the Megahertz race is no longer relevant, and savvy customers can no longer divide the MHz by the price, and find out which system gives you the most e-penis for your buck. Customers no longer have a clue, so everyone just shrugs and goes with the biggest brand name.

Re:AMD had it going

[wisty]

I think Steve Ballmer has a cunning plan to prevent this :D

Re:AMD had it going

[smallfries]

There is a very simple reason that the Top500 is full of Opteron systems. Until the i7 Intel did not have an integrated memory controller. Although the Core2 does more work per cycle, at lower power, and with better caching - there is a measurable difference in large memory bound workloads. The other factors were enough to make them faster on the desktop, but the lack of integrated memory controller was killing them in large-scale systems.

The i7 continues the advantages that Core2 had over the Opteron range, but adds that missing memory controller. It's not clear yet if it is good enough. The memory subsystem graphs in the article are interesting. Intel still has a faster, larger cache, but may lack raw bandwidth to main memory.

I'm not going to disagree with your comments on the impending death of the desktop (or agree with them either). But I will point out that people have been making exactly the same comments and predictions for 20 years. We still have desktop computers.

Re:AMD had it going

[Dystopian Rebel]

which system gives you the most e-penis for your buck

I believe you are referring to the Apple iPenis.

This forthcoming product from Apple is very fashionable but it does dick-all.

Re:AMD had it going

[lawaetf1]

The biggest thing keeping AMD down in the mobile world is the systems.

Speaking beyond just the mobile market, it's important to keep in mind that Intel is facing anti-trust suits around the world. And has already been found guilty in S. Korea with Europe getting increasingly annoyed at their delays. If the accusations are true, Intel's unlimited R&D budget is ill-gotten via illegal, exclusory business practices.

Frankly I'm all but blown away at how a company with a smaller market cap than either NVIDIA or Intel can continue to compete and sometimes win.

Re:AMD had it going

[LWATCDR]

You also only kind of hinted at the ease of migration. All that the OEMs need to do to introduce the Shanghai is to put it in the socket. with the I7 family it they will need to move to a new motherboard as well. For manufactures this will be a big win since for may buyers it will be seen as a nice safe evolutionary change. The one thing that worries the server market is big changes. It is all about stability.

Re:AMD had it going

[default luser]

And I find it hilarious: Intel consistently makes better mobile CPUs definitely but everything else they do in mobile space reeks to high heaven. To this day its nearly impossible to buy a Atom netbook without a Intel GMA based chipset: thats a 2 watt cpu and a 12-25 watt chipset. If you buy a normal laptop, its probably a 45w or 35w chip, even though the Pxx00 series is 25w and almost the same price, and again it comes with an absolutely worthless video card that sucks down >10 watts.

Not true at all, where have you been reading this bullshit? The Intel 945G Mobile is the standard chipset used by netbooks, and has a TDP of between 5-7w. The Mobile 945GSE used by the Asus netbooks has a TDP of only 6w. You don't really think they'd have the power in one of those tiny nettops to power a desktop chipset for 2+ hours, do you?

The chipset you are thinking of is the craptastic 945GC, which is the leftovers remaining after Intel cherry-picks the mobile versions. The only reason Intel uses these on their Atom motherboards is because the boards are designed for MINIMUM COST for developing markets, not minimum power. Since the 945GC is practically free to use (leftovers), Intel can sell the whole board + Atom for $70.

There's nothing preventing other manufacturers from releasing an Atom motherboard with a mobile chipset, aside from cost...but cost is a strong factor in designs. Typically, if they go to that kind of trouble to design for low power, they design to tighter specs and sell the board to the embedded market for $300+. In the consumer DESKTOP market, most people don't care about the difference between 20w and 40w, so the ideal board for you will never be sold.

Re:Does 'Opteron' mean 'expensive'?

[thona]

No marketing talk in those names.

Not sure you would call it expensive, but the OPTERON chips per definition are only server chips. The Opteron 23xx series (45nm shanghai) is dual processor, while the 83xx series is quad processor.

The end user equivalent is the PHENOM series.

Note that this is a technical difference, not marketing talk. The Opterons use Socket F, while the Phenoms (single processor only) use the AM2+ socket. Different pin count, different number of interconnect ports (for connecting to other processory).

45nm Phenoms are IIRC supposed to appear soonish ;) Opterons start being available now - I pick up a new server on friday.

Re:Does 'Opteron' mean 'expensive'?

[lagfest]

No, he meant multiple processors each of which has four cores. And that is indeed primarily a server feature.

Re:Does 'Opteron' mean 'expensive'?

[thona]

You should read up on the difference between dual CORE and dual PROCESSOR.

In fact, both current Opterons as well as current Phenoms are quad core systems. The Phenoms single processor (1x4 cores), the Opterons dual to quad processor (2x4 cores to 4x4 cores).

Hardly any end user system (i.e. non-server) today uses more than one processor. Dual core to quad core is normal now. But always on one processor.

The difference in sockets actually is for that - the AM2(+) socket lackss the HT bus for inter-processor communication, while the Socket F has separate lanes for the processors to talk to each other.

The main problem with Intel right now is that intel has no really nice solution at all in the multi processor side - they simply (again) do not scale from the memory side, thanks to a lack of a NUMA architecture (that they change now and coming).

Re:Site's been slashdotted

[LordMyren]

I cant resist, on account of all the people who look at the Core i7 benches and think its all over; considering that the best "review" of a Nehalem EP (the dual socket variant) is a couple of guys who have a single screenshot for spec_fp, I'd say the battle's too early to call. All we've seen are single socket Nehalems-- & thats not been AMD's strong suit for some time.

Even considering that Intel's single socket game has been largely better for a while, there are some key areas AMD systems perform better. HPC, render farms, some web serving, virtualization... for all these places where people need a lot of cpus, AMD is has stayed in the runnings or maintained a lead (depends a lot on just what you're running). Unfortunately the benchmarks usually published dont factor in these kinds of workloads much at all. Cinebench is the only benchmark in the review anywhere near the above. I think if we ran some VMWare benchmarks, things would look drastically different.

But the real quesiton here is Intel: Intel is just now doing the infrastructure AMD did in early spring `03: QPI to AMD's HT, similar onboard memory setups... and thusfar aside from some spec_fp numbers, we have no idea whatsoever how well their implementation is going to work. Once Intel releases Nehalem EP for testing, we'll have an idea.

It's a shame, too

[kmike]

I find it disappointing that the test of the supposed server-oriented processors does not include web server tests - after all it's probably the largest market for such processors.

I mean, does anyone really care about Folding@Home number these processors can crunch? Or "VRAD map build benchmark"? WTF?

Re:It's a shame, too

[ThePhilips]

Also, they have used several openly pro-Intel applications: Cinebench and M$ .Net.

Cinebench never hid the fact that they optimize for Intel and if you want to have best performance you need to buy Intel CPUs.

M$ .Net XML benchmark - M$ C/C++ compiler and libraries in many parts use Intel's hand written asm code. And it always produced code optimized for Intel architectures.

AMD benchmarks with Intel optimized code?

[GNUPublicLicense]

Again, I wonder if the benchmarks used AMD optimized code (they have to use the proper GCC backend). It seems that most of the time, the benchmarks for non-Intel processors are based on Intel optimized code. I have never seen mentionned in the benchmarks if the tools were using the best machine code for the targetted processor... yeah... that smells bad.

Re:AMD benchmarks with Intel optimized code?

[howlingmadhowie]

call me a conspiracy theorist, but i'd like to see the benchmarks on *bsd and gnu/linux systems as well. i've often wondered if microsoft has a deal with intel to slow amd processors.

more the reverse

[Trepidity]

AMD didn't really destroy Itanium and then rest on their laurels. Although you have to give them some credit for coming up with reasonably good chips that the market wanted, it was more that Itanium was the reason AMD was competitive with Intel in the x86 space for a few years in the first place.

Intel has orders of magnitude more R&D budget and especially capital for fab construction than AMD does. So AMD is perpetually at least a half-generation behind Intel on the tech curve: they keep coming up with chips that could beat Intel... if they had come out a year ago. Now when Intel effectively skips a generation, as they did when they sunk all their resources into Itanium and mostly ignored x86 for a year or two, this is enough to give AMD the lead. But once Intel shifted fully back into x86, they crushed AMD again.

Re:Does 'Opteron' mean 'expensive'?

[this great guy]

Well to be pedantic:

• The Opteron 1xxx series is using the same AM2+ socket as the Phenom processors, and are in fact rebranded Phenoms (no technical difference). But you are correct in that Opteron 2xxx and 8xxx are completely different animals.
• The Opteron 8xxx series is for systems with 4 or more sockets (not restricted to 4). This is made possible because each of the 3 HT links per processor is running the cache coherency protocol (whereas only 1 out of the 3 HT link of an Opteron 2xxx runs the protocol).

Re:Site's been slashdotted

[this great guy]

You are so true about virtualization. Just 2 or 3 weeks ago I was benchmarking Linux and Windows VMs compiling Java code under Qemu/KVM-75 on an 2-socket 8-core 2.0GHz Opteron 2350 (non-Shanghai) system, and on a 1-socket 4-core 2.4GHz Core 2 Q6600 system. The VMs were configured with 1, 2, or 4 virtual processors and not more, to not give an unfair advantage to the AMD system which had twice the number of cores. Despite the lower CPU frequency as well as lower memory throughput and latency (registered DDR2-667 vs. unbuffered DDR2-800 for Intel), the AMD system was kicking the ass of the Intel system in every case by as much as 10-30%. Most likely this was because of the integrated memory controller and support of nested paging (aka "Rapid Virtualization Index"). Now Intel has cloned these 2 features in their Nehalem microarchitecture. I am very impatient to see how they perform.

NUMA

[DrYak]

i've often wondered if microsoft has a deal with intel to slow amd processors.

Yes, sort of.
It's called NUMA - Non Uniform Memory Architecture.

Up until recently Intel platforms had the memory controlled by the northbridge, with all CPUs and all cores having the same access to the memory.

Newest Intel platform and all 64bits AMD had the memory controller on the processor package. In a multi-socket configuration, each processor controls it's own chunk of memory, so for some range, the access will be faster because the processor is directly accessing it, and for other the latency will be increased because the processor has to ask its neighbour over HyperTransport / QuickPath.

To be able to function in a such configuration, an OS should pay some attention when scheduling process and threads to cores : it should be best that all threads from some process are all scheduled to cores having all direct access to the resources used by said process. (While at the same time scheduling two threads at a physical core and it's corresponding hyperthreading virtual core if there's a another physical core sitting idle)

Windows has always deeply sucked at this. Opensource OS, on the other hand, have much more work applied to them for that. (That's why they are much more popular on super computers).

This also introduces technical difficulties (like keeping the cache coherent). That's also why heavily multi-socketed (4 and up) motherboard won't be coming during the first year of Core i7's life. They probably have to fix all the fine details before that. As usual expect a change in socket format and a new iteration of Core i7 not quite exactly compatible with the previous one.

On AMD's side, currently sold Opteron are already adapted for 4 and more sockets configuration. (As explained by other /.ers, the 8000 series has a coherency protocol running on 3 HT interconnects, which should be enough to help on 4 and more sockets).

Re:AMD had it going-939

[Enter the Shoggoth]

Absolutely, I was happy to support the underdog until I was the beneficiary of a '939 shafting; AMD promised AMD-V support on Socket 939 and then pulled it, and although not AMD's fault (I'm looking at you nVidia & ASUS) the chipset/motherboard performance, driver support and hardware reliability left me with a very sour taste.

Now that there is no longer any viable non-x86 solution, I've gone intel 100%. not just the CPU but I'll now only buy Intel motherboards, turns out they are _very_ reliable, stable, and the BIOS and drivers are QA'd properly. Who would've thunk it?

Re:it's a shame

[Enter the Shoggoth]

I love bagging the itanic as much as the next guy, but to be fair it does perform well under certain loads; give it an online transaction processing load and it will shine, which by an amazing coincidence is where HP's target market for the platform is.

Re:Does 'Opteron' mean 'expensive'?

[MikeBabcock]

For the people who like pictures, feel free to check out pictures of one of our socket F server boards before it got closed up and installed.