Intel's Best x86 Server CPU

The launch of the Nehalem-EX a year ago was pretty spectacular. For the first time in Intel's history, the high-end Xeon did not have any real weakness. Before the Nehalem-EX, the best Xeons trailed behind the best RISC chips in either RAS, memory bandwidh, or raw processing power. The Nehalem-EX chip was well received in the market. In 2010, Intel's datacenter group reportedly brought in $8.57 billion, an increase of 35% over 2009.

The RISC server vendors have lost a lot of ground to the x86 world. According to IDC's Server Tracker (Q4 2010), the RISC/mainframe market share has halved since 2002, while Intel x86 chips now command almost 60% of the market. Interestingly, AMD grew from a negligble 0.7% to a decent 5.5%.

Only one year later, Intel is upgrading the top Xeon by introducing Westmere-EX. Shrinking Intel's largest Xeon to 32nm allows it to be clocked slightly higher, get two extra cores, and add 6MB L3 cache. At the same time the chip is quite a bit smaller, which makes it cheaper to produce. Unfortunately, the customer does not really benefit from that fact, as the top Xeon became more expensive. Anyway, the Nehalem-EX was a popular chip, so it is no surprise that the improved version has persuaded 19 vendors to produce 60 different designs, ranging from two up to 256 sockets.

Of course, this isn't surprising as even mediocre chips like Intel Xeon 7100 series got a lot of system vendor support, a result of Intel's dominant position in the server market. With their latest chip, Intel promises up to 40% better performance at slightly lower power consumption. Considering that the Westmere-EX is the most expensive x86 CPU, it needs to deliver on these promises, on top of providing rich RAS features.

We were able to test Intel's newest QSSC-S4R server, with both "normal" and new "low power" Samsung DIMMs.

Some impressive numbers

The new Xeon can boast some impressive numbers. Thanks to its massive 30MB L3 cache it has even more transistors than the Intel "Tukwilla" Itanium: 2.6 billion versus 2 billion transistors. Not that such items really matter without the performance and architecture to back it up, but the numbers ably demonstrate the complexity of these server CPUs.

Processor Size and Technology Comparison
CPU transistors count (million) Process

Die Size (mm²)

Cores
Intel Westmere-EX 2600 32 nm 513 10
Intel Nehalem-EX 2300 45 nm 684 8
Intel Dunnington 1900 45 nm 503 6
Intel Nehalem 731 45 nm 265 4
IBM Power 7 1200 45 nm 567 8
AMD Magny-cours 1808 (2x 904) 45 nm 692 (2x 346) 12
AMD Shanghai 705 45 nm 263 4

 

Test Servers and Benchmark Setup
POST A COMMENT

62 Comments

View All Comments

  • Shadowmaster625 - Friday, May 20, 2011 - link

    Yeah but what you gonna do with those two extra Xeons? You cant just bolt them to the side of that two socket server. There is a huge price divide between 2 and 4 socket servers. Your numbers are totally disingenuous. You'd need to drop an extra 6 grand just to move into the 4 socket platform. You can see that right on HP's site. For $12,000 more you get 2 extra sockets, and all four chips get upgraded to 4850s. The upgrade is worth 4 grand. You also get a memory upgrade to 128GB. When you also subtract a couple grand for that 64GB of ram upgrade, you're left with 6 grand for the dual-to-quad socket upgrade.

    btw you posted the same link twice.
    Reply
  • jihadjoe - Saturday, May 21, 2011 - link

    Spec sheet, from GP's link:
    http://h18000.www1.hp.com/products/quickspecs/1366...

    It's a 4-socket server but 'as configured' comes with two processors. You put those extra xeons in the two extra sockets inside it.
    Reply
  • Shadowmaster625 - Monday, May 23, 2011 - link

    So why is the 4 core server $6000 more expensive, when factoring out the added parts? Is that what they charge just to install two processors? Is there really that much waste in the IT world? If so then it is no wonder IT is being outsourced at a breakneck pace. Any IT professional who would pay HP 6 grand just to install a couple cpu's needs to be "downsized" immediately. Reply
  • JarredWalton - Monday, May 23, 2011 - link

    The AMD server is here:
    http://h10010.www1.hp.com/wwpc/us/en/sm/WF06b/1535...

    It only supports 32 DIMMs vs. 64 DIMMs, as another disadvantage.

    As L. points out below, the Intel setup is also using the E7-4830, which is 8 core instead of 10. And then the upgraded Intel setup with 10-core CPUs also bumps up to 128GB RAM and 4 x 1200W PSUs and ends up at $26819 (with 4 x E7-4850) -- note that the AMD setup already had 4 x 1200W PSUs.

    So once again, we're back to comparing apples and pears -- similar in many ways, but certainly not identical. And for that very reason, you can't even begin to make statements like L.'s "AMD wins on perf/watt/dollar" because we don't have any figures for how much power such an AMD setup actually consumes compared to the Intel setup. It might be more power than our review servers, or it might be less, but it will almost certainly be different.

    My main point is that we're not even remotely close to paying 2x as much for an Intel server vs. AMD server. If you want to compare the cheapest quad-Opteron 6174 to a higher quality quad-Xeon, yes, the pricing will be vastly different; that's like pointing out that a luxury sedan costs more than the cheapest 4-door midsize sedan.
    Reply
  • L. - Monday, May 23, 2011 - link

    I think I saw that comment quite a few times .. but I only just realized there was a big problem with your numbers :
    The processors listed here are e7-4830.

    Those processors are 8 cores (not 10 like 4870) and 2.13Ghz (not 2.4Ghz like 4870).

    Assuming linear scaling (although this is absolutely not the case) you would get the same perf/watt as the above model, and a total sap score of 52,518 vs 47,420 for the AMD 6174 (10% more).

    And the price is 18% more .. looks like perf/watt/dollar crown goes to AMD again.

    Other tests are clearly impossible to guesstimate and clearly the SAP test was where the e7 was getting a better advantage compared to vAPUs mark II test.

    So yes, the argument stands that even though Intel has higher-priced extreme components, anything they have in AMD-performance-range is more expensive than AMD's option, quite logical with AMD as the underdog so far.

    BUT, as we're showing benchmark of the flagship vs flagship, there tends to be misconceptions about the rest of the product line, just like here "20k more expensive" or "so much better cpu from intel" or other random bullcrap.
    Reply
  • spanky_mcsoreass - Thursday, May 19, 2011 - link

    What software does the average business need to run on a ridiculous number of cores? The only common application that comes to mind is internet facing Linux/Apache servers, and Linux/Apache are free(anyone dumb enough to pay for and use Windows/IIS deserves what they get).

    Most businesses just need a lot of VMs running their various low intensity apps, and dedicated NAS or SAN devices. Magny-Cours Opteron devices do either just as well or better than Xeon, and using VMs to run licensed apps won't result in a penalty for software licensing.
    Reply
  • ggathagan - Thursday, May 19, 2011 - link

    Then this really isn't for "most businesses", is it?

    This is Intel's new flagship.
    As such I do not expect it to be a good fit for many businesses out there.

    That does not make it any less interesting to review, since there are businesses that *can* make use of it.
    Further, the same technology will end up in the lower tier CPU's as well.
    Reply
  • dominique_straws_con - Thursday, May 19, 2011 - link

    No, this is just another CPU that *could* be for most businesses, you could use it in the "lots of non-intensive apps running in VMs" scenario(or any other scenario), it's just an exceptionally poor value, and will probably not out-perform Magny-Cours or Bulldozer for most people's real world use. This CPU excels at benchmarking, and that's just about it.

    AMD got server CPUs right, it's all about how many cores you can fit on a rack.
    Reply
  • L. - Friday, May 20, 2011 - link

    How many cores you can fit on a rack, with what TDP ;)

    When you have to WC your rack, you have a problem - most of the time anyway.
    Reply
  • Casper42 - Thursday, May 19, 2011 - link

    You know nothing about Enterprise level IT.

    Your Example of Internet Facing Apps and Linux Apache is the EXACT opposite design methodology of how things work in the real world.

    In the real world, internet apps run on the cheapest of the cheap servers and companies just use a ton of them behind a Load Balancer.

    Now the Database serving those web servers in the background, running Oracle RAC or MS SQL or even MySQL on the other hand will make use of all these cores and memory assuming you have a large database.

    The examples given RIGHT IN THE ARTICLE about things like SAP are probably the most common thing run on these Big Iron type boxes.

    If it helps prove my point any further, over at HP on the Sales side of things, the guys that have been selling RISC based machines under the HP Integrity/Superdome name for something like a decade, are now also being paid commission when they sell the DL580/DL980 G7 servers. Those 2 models use the Nehalem EX and will soon be using the Westmere EX.
    So the type of Apps running on these CPUs are often the same things Fortune 100 companies used to run on Integrity/Sun/IBM Power/etc
    Reply

Log in

Don't have an account? Sign up now