P570 vs T5220

P570 vs T5220

Summary: There aren't many public benchmarks that allow you to compare an IBM p570 to Sun's T2 series gear - but one that does, a business applications benchmark officiated by SPEC.ORG, shows Sun to have about a 1.67 x speed advantage - while other data suggest a six times price advantage.

SHARE:
TOPICS: Processors, Hardware, IBM
13

As part of JesperFrimann's challenge with respect to the options available to a company with two aging p690s running Sybase, he suggested that:

It's 256 threads per box on the T5440 versus 16 on the p550, and furthermore the power 550 supports shared pool virtualization.. so it will allow for a much higher utilization. So the price performance for the p550 wil be far superior to the T5440 IMHO.

So today's question is: is this true?

Notice, before we get into this, that while I don't want to make excuses for the Power6 it's important to note that this machine has to carry two enormous burdens for IBM:

  1. first, it was designed to counter Itanium; not SPARC, and not x64. As a result IBM focused on cache and megahertz to the near exclusion of power efficiency and hardware assist processing (things like Sun's onboard crpytology and packet managers) because those were never going to be issues in competing with stuff from Intel; and,
  2. second, Power6 is the penultimate transition product - the next generation, the Power7, is planned to have eight cores, each offering four SMT style, switchable "threads" running at 4Ghz, and will be the last of the PowerRISC line. After that (assuming IBM can make and hold to the decision) the project Quasar CPUs, built on the 16 SPU Cell2 processor and expected to appear at 4Ghz with up to four processor sets (G6+, SAP, and 16 SPUs) on a single die, will take over.

On its own, however, the Power6 is very nice processor I never hesitate to recommend to IBM customers or as an alternative to Itanium -but the errors IBM made in the strategic assumptions underlying its development show up in odd places: including a fairly desperate scarcity of auditable benchmarks run against SPARC or higher end Opteron (and, soon, Nahelem) gear.

One important one does exist: a SPECjAppServer(R)2004 result featuring the Apache based Websphere with J2EE and DB2 under IBM AIX 5L v5.3. Here's the hardware summary:

J2EE Application Server Database Server
Model p570 (4.7 GHz, 4 core, SMT on) p550 (2.1Ghz, 4 cores, 2 chips, 2 cores/chip (SMT on))
Memory 16GB 16GB
Internal disk 4 x 146 GB 15K RPM SAS Disk Drive 73.4 GB 10,000 RPM Ultra320
External disk 2 x DS4700 Express Model 70 (2 GB Cache) on FC, each 14 x 36.4 GB/15K
Working together these two machines produced a score of 1,197.51 SPECjAppServer2004 JOPS@Standard.

The comparable Sun result, using a 32GB T5220 with one CPU (8 cores) at 1.4Ghz, running the J2EE application server against Oracle Application Server 10g on a 64GB T5120 (1 CPU, 8 cores, at 1.2Ghz) is 2,000.92 SPECjAppServer2004 JOPS@Standard - 1.67 x IBM's number.

IBM does not publish detailed P570 pricing. However, they do participate in the TPC/C benchmark Sun has considered obsolete since the mid ninties as both too unrealistic to be meaningful and too simple to game - and one result there includes cost details for the p570.

Based on the detail shown my estimate for the discounted price of just the p570 used in the J2EE test is $249K versus the $71,200 shown by Sun's on-line store for the T5220 and T5120 units combined - giving Sun a six times performance advantage per dollar if you could buy the p570 and get the p550 and disk free.

Interestingly, the details published by IBM for the TPC report include this bit:

4.7GHz POWER6 -2 Core Processor Card, 0-core active $11,500 (each, times 8) One Processor Activation for Processor Feature #7380 $23,000 (each, times 16)

suggesting that the cost of just the four licensed cores in the P570 used as the J2EE server is 1.6 times the nominal list price of the two Sun servers combined: delivered, set-up, and fully warranted for three years - and because the Sun combo is also rather a lot faster, the real bottom line here is that only a truly committed IBMer would ever buy a p550/570 combination for something a T2 will do.

Topics: Processors, Hardware, IBM

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

13 comments
Log in or register to join the discussion
  • IBM 1, Sun - all the rest

    One point in the "favor" of IBM is the way you can get "pay as you go" pricing. IBM would routinely provide DOUBLE the CPUs/Memory that you paid for - they would be in the chassis and unconfigured. If you wanted more CPU/memory, you would just call IBM and get an activation code. This makes it easy to part with your hard-earned dollars . . .

    As for support, IBM forces everyone to call their 1-800 support line - where you have to speak to some 1st level person. We were able to get IBM to allow us to ask for a service manager - so we could get help sooner. They also priced on-premise support people separately - so they were NOT part of the "list" price for their servers.

    Sun had top-notch people on-site who you could call directly. The price of this was all part of their Gold/Platinum service - which as Murph points out, comes bundled in with the hardware purchase.

    For my money, Sun wins every time. The price/performance of Sun is always competitive (except when they take years to deploy next gen tech - like Rock and Viking before that [ SPARC10 = 40 Mhz, HP735 = 100Mhz ]). Ford had a good thing going when they pitted Sun against HP in the data center. The price wars allowed Ford to get very good price/performance (and forced HP to up its game in support - which it did).
    Roger Ramjet
  • RE: P570 vs T5220

    Hey Murph,
    The Power 570 should really be compared to Sun's T5440
    which are closer in pricing-OTHERWISE ITS REALLY UNFAIR!

    A 4 x CPU T5440 is approx $133K while a 4 x CPU Power
    570 is about $380K.

    Whats amazing is that the T5440, even at almost 1/3rd the
    price, still outperforms Power 570 with same # of CPUs! On
    some benchmarks, T5440 is over 2x faster!

    SPECjAppServer2004 SPECjApp JOPs
    Sun T5440 4 x 1.4 GHz T2 Plus 6,335
    IBM Power 570 4 x 4.7 Ghz 1,198
    Times Faster 5.29

    SAP SD 2-tier SAP Benchmark Users
    Sun T5440 4 x 1.4 GHz T2 Plus 7,520
    IBM Power 570 4 x 4.7 Ghz 4,010
    Times Faster 1.88

    SPECjbb2005 BOPS
    Sun T5440 4 x 1.4 GHz T2 Plus 692,736
    IBM Power 570 4 x 4.7 Ghz 402,923
    Times Faster 1.72

    Siebel - CRM Rel 8.0 Siebel users
    Sun T5440 4 x 1.4 GHz T2 Plus 14,000
    IBM Power 570 4 x 4.7 Ghz 7,000
    Times Faster 2.00

    SPECint_rate2006 SPECint_rate2006
    Sun T5440 4 x 1.4 GHz T2 Plus 301
    IBM Power 570 4 x 4.7 Ghz 243
    Times Faster 1.24

    SPECfp_rate2006 SPECfp_rate2006
    Sun T5440 4 x 1.4 GHz T2 Plus 230
    IBM Power 570 4 x 4.7 Ghz 216
    Times Faster 1.06
    phil_64
    • Thanks for citing those results

      :)
      murph_z
  • RE: P570 vs T5220

    Why would I need to buy the P550? The P570 can handle both tiers just fine on its own. The 570s price/perf comes through consolidation of workloads. I don't want to live in Sun-world where I need separate pieces of iron for every service tier of an application.
    civikminded
    • Nice try

      But Sun didn't set up the benchmark result I quoted for the p series combo - IBM did.

      And no, the p570 cannot handle both at benchmark levels of performance, it chokes on I/O and RAM when you add too many processors.

      I would normally put both the app and the DB on the same machine - but only if the machine could handle the load. Notice that I only priced the p570 in the comparison? Ask yourself why..
      murph_z
  • RE: P570 vs T5220

    Sounds like your IBM world is going to cost you plenty
    more. If the P570 could handle both tiers just fine as you
    say, why then would IBM need to run 3 separate p570's
    with a total of 14 x Power6 CPU's to run the Siebel8
    benchmark where it achieved 7,000 users (See here for
    details: http://www.oracle.com/apps_benchmark/doc/IBM_Siebel8_
    7000_PSPP_On_AIX_POWER6%20Final.pdf)

    While Sun's T5440, easily ran the entire benchmark
    running all tiers on one system-DB server, Web Server and
    App/Gateway server on a single 4 x CPU system and
    achieved 14,000 users! Thats 2x higher than IBM's results
    and clearly Sun did not need more iron than IBM... Time to
    re-evaluate..

    http://blogs.sun.com/mandalika/entry/siebel_8_0_on_sun
    phil_64
    • Yep. (NT)

      :)
      murph_z
      • Wow

        Do you really feel comfortable buying servers with 'Caution' statements on their own website? I'll stick to IBM - They don't have to warn you about what kind of workload to run on your servers!

        http://blogs.sun.com/mandalika/entry/siebel_8_0_on_sun

        "CAUTION

        Although T5440 possesses a ton of great qualities, it might not be suitable for deploying workloads with heavy single-threaded dependencies. The T5440 is an excellent hardware platform for multi-threaded, and moderately single-threaded/multi-process workloads. When in doubt, it is a good idea to leverage Sun Microsystems' Try & Buy program to try the workloads on this new and shiny T5440 before making the final call."
        civikminded
        • RE: WOW

          Yep, Sun's honest, truthful and upfront that single-threaded
          workloads aren't going to run optimally on the T5440. But find me a
          single-threaded workload in todays business/web focused
          datacenter? Clearly the 7 world record industry standard benchmarks I
          listed above show atleast 2x better performance than Power 570 and if
          you run SAP/Oracle, java (SPECjbb), application servers (SPECJApp),
          Siebel (CRM), you'll have a hard time finding a better choice. You also
          won't see IBM cautioning that their Power 570 servers cost orders of
          magnitudes more than an equivalent Sun T5440. And then of course,
          as Murph said, where does IBM go from here with Power? More GHZ?
          probably not. More cores? probably. But do you think IBM will
          continue to offer 5GHz with multi-core? Probably not. We shall see.
          IBM's Power hase been dual core for last 8 years so its time for them
          to get onto the Multi-core bandwagon.
          phil_64
    • Thanks

      I'm glad you pointed me to these benchmarks that totally prove my point:

      According to these benchmarks:
      + Sun took 32 cores @ 81% util and 128G RAM (94G used) to do 14000 users
      + IBM took 14 cores @ 59% util and 104G RAM (45G used) to do 7000 users

      Additionally, you proved how you could consolidate that workload onto a single P570 instead of 3. Based on the utilization figures in the benchmarks, that workload could be collapsed onto ONE P570 with 10 cores & 64 GB RAM. Price that out and compare.

      Who needs to re-evaluate?
      civikminded
      • Not Core vs Core, but CPU vs CPU

        IBM clearly states that SUN Niagara CPU is slower than IBM Power CPU. If that is the question, then you can NOT compare core vs core. You have to compare CPU vs CPU. IBM states that the Power CPU is faster, not that the Power Core is faster.

        How many 4.7GHz Power CPU does IBM need to match one 1.4GHz Niagara CPU? Several of them. How in earth can the IBM Power CPU be faster, if you need twice as many Power PCU to match one Niagara? Something is strange, dont you think?

        Look, if the world's fastest CPU has 10.000 slow cores and IBM Power core is faster, does that mean that the IBM Power CPU is faster? No it doesnt.

        In order to determine which CPU is fastest, you have to compare CPU vs CPU. You can not compare Core to Core, and from that infer which CPU is fastest.

        Yes, IBM core is faster but the entire Power CPU is slower. That is a fact, evidenced by these benchmarks
        Orvar
  • RE: P570 vs T5220

    Why did IBM run 3 x separate p570 servers to run the
    benchmark if they could have "consolidated" onto a single
    system like you say? The benchmark shows that IBM
    needed a total of 3 x P570 systems totalling 14 x Power6
    (dual-core) CPUs. Thats 28 cores according to my math.
    And you can't run 14 x 4.7GHz CPUs in a single Power 570.
    And that still only gets you half the performance of the
    single Sun T5440 system.

    For IBM to match Sun, you'd need 28 x CPUs (56 x cores)
    to compete. I don't think we'd need to research pricing
    here to see whos better on price/performance.

    IBM's P570 7,000 User result required:
    1) Database Server
    ? 1x4way IBM p570 Server (4 x 4.7GHz POWER6 CPUs)
    2) Gateway/Application Server
    ? 1x8way IBM p570 Server (8 x 4.7GHz POWER6 CPUs)
    3) Web Server
    ? 1x2way IBM p570 Server (2 x 4.7GHz POWER6 CPUs)

    And finally, even if your estimation on system is accurate,
    giving you benefit of doubt, even a Power 570 4.7Ghz 4 x
    CPU (8 x core) system lists close to $400K, over 2x above
    the T5440 pricing!

    You can see IBM Power 570 pricing here:
    http://www.ibm.com/common/ssi/rep_ca/8/897/ENUS107
    -288/ENUS107288.PDF
    phil_64
  • Time to start sorting out the exhibitions at the museum

    A whole lot of documentation and CDs for some language called Java turned up.

    See if you can arrange them in a diorama, so the kids can see what programming used to be like.

    tonymcs1