back to article IBM's new Power7+ hotness - we peek through the veil

A few weeks ago, El Reg told you that IBM was getting ready to start talking about its future Power7+ and System zNext processors at the Hot Chips conference at the end of August. Like you, I am an impatient sort when it comes to getting some insight into future processors from any vendor, and I like to poke around and see what …

COMMENTS

This topic is closed for new posts.
  1. Anonymous Coward
    Go

    Benchmarks ??

    Traditionally, POWER has been the fastest processor of all. It would be interesting to see how they currently compete against x86, Itanium and SPARC.

    Hopefully, IBM will keep doing the excellent work. I blasted IBM for the shitty aspects of some of their other products, but POWER is just a great product.

    1. Kebabbert

      Re: Benchmarks ??

      Here are some benchmarks that you asked for.

      POWER7 is 14% faster than the old Intel Westmere-EX:

      http://www.anandtech.com/print/4285

      SPARC T4 is more than 2x faster than POWER7 on TPC-H

      https://blogs.oracle.com/BestPerf/entry/20110926_sparc_t4_4_tpc

      But sure, IBM has some record benchmarks, particularly in integer arithmetic. As Larry Ellison said: "IBM is faster for integer arithmetic. If IBM think companies do a lot of arithmetic, cool. We think they access a lot of data and run a lot of Java."

      "We're better than IBM in Java, and we're going to beat them in integer arithmetic, and then there will be nothing left," he added.

      .

      I think it is funny that "IBM does not need any roadmaps". Maybe IBM had problems of keeping x86 at bay, and that is the reason? x86 is fast catching up on POWER, and the old Westmere-EX is almost as fast as POWER7 on some benchmarks. We soon have Ivy Bridge 10 core Xeons, I bet they will be faster than POWER7. And there is Haswell next year. "No need for roadmap"? That is so fanboyish blind it is funny. Normally, when IBM has something good, they bragg about it all the time. Silence is a sign of problem, and IBM tries to to not talk about their problems (for instance manufacturing problems) - as everybody does. IBM doubts the future of the POWER chips, as IBM are going to kill AIX and then have to compete with faster and cheaper x86 cpus.

      .

      SPARC T5 will also be presented at the same conference. It will be an improved T4: have double the number of cores: 16, and scale up to twice the number of cpus: 8. Which means 4x better performance than today's top T4 servers. T5 will be really brutal on database work loads, which is what really counts in the Enterprise. Oracle has officially said they will double the SPARC performance every second year, that is much more aggressive than any other cpu vendor. Not even the x86 camp is talking about such improvements.

      1. Anonymous Coward
        Anonymous Coward

        Re: Benchmarks ??

        IBM only have to keep their POWER processor road maps ahead of Intel's Itanium road maps.

        IBM won't need to match x86 (surely Itanium would be faster than x86 as it is is Intel's top-of-the-line processor and HP are always telling us how great they are...) until Intel end-of-life's the Itanium's and we get x86 versions of NonStop and OpenVMS. Then again, if you listen to Larry, maybe this isn't far off....

        As for the Sun results, I agree that a SPARC system with lots of SSD storage will be faster than a Power system with less SSD storage. Or was that not the point you were hoping to make?

      2. Anonymous Coward
        Anonymous Coward

        Re: Benchmarks ??

        Well, the issues with that are:

        1) Yes, companies do do a lot of integer work,

        2) Larry Ellison is a wanker. "Don't rely on wankers for your business needs" is a maxim that goes back to 761 BC.

        1. Ilgaz

          Re: Benchmarks ??

          I guess his company is so cool that it pays wages/ calculates them in floating point.

      3. Anonymous Coward
        Thumb Down

        Re: Benchmarks ??

        Oracle's T4 benchmarks were a fraud. They have been fined $10,000 by TPC for making false claims. Here you go, again:

        - Oracle’s T4 result needed four times the number of app nodes, twice the number of cores, almost four times the amount of memory and twice the amount of storage than the IBM POWER7 result

        - Oracle once again used 128 query streams for this benchmark compared with IBM’s 8. TPC-H has a limited number of query variations; so when you run a lot of streams, you have a high probability that the same queries will be requested more than once. Oracle is greatly increasing the probability that they will have the results of the queries stored in their cache.

        1. Anonymous Coward
          Anonymous Coward

          Re: Benchmarks ??

          "- Oracle’s T4 result needed four times the number of app nodes, twice the number of cores, almost four times the amount of memory and twice the amount of storage than the IBM POWER7 result

          - Oracle once again used 128 query streams for this benchmark compared with IBM’s 8. TPC-H has a limited number of query variations; so when you run a lot of streams, you have a high probability that the same queries will be requested more than once. Oracle is greatly increasing the probability that they will have the results of the queries stored in their cache."

          The crazy part is that even after all of that gaming of the system Oracle still could not beat Power 7 on integer calculations! Sparc was just too inferior even with twice as many disk arms, twice as many processors, four times the amount of memory.

      4. Beachrider

        At least you don't have the 1999 IBM article here anymore...

        1) Anand doesn't say that Power7 is 14% faster than anything 'old'. Anand cites an SAP benchmark that compared a 14 month old result from IBM against a just-fresh Intel system benchmark-run. If you did the same thing today, using SAP run 2012015, the gap is up to 25.8% and the p460 is 65% the price of the 2010 benchmark-box. (p460 is an IBM-attempt at internal-cloud for SMB).

        They are in the same ballpark. Power is still faster. Anand's characterizations about price are, at best, under-documented (some would say more). I don't suppose that you would care to quote Anand on SPARC (he REALLY doesn't see the point of SPARC).

        2) Oracle has clearly misquoted the tpc.org page that they attempt to cite. The span of the benchmark is completely different than the results posted on tpc.org. Oracle needs to fix this before it makes sense.

        3) IBM has roadmaps. Anand has them, Register has them. Erase your rant. Use a spell-checker next time.

        4) For Java-written applications, the SPARC does do better than hardware-architects would expect.

        5) As long as you talk in the future subjunctive tense, you can say anything you want about T5. Anything COULD be true, until they bring it out...

    2. Anonymous Coward
      Mushroom

      Re: Benchmarks ??

      POWER7 absolutely crushes Itanium and SPARC already, in about every metric. x86 is actually at par performance-wise with Westmere-EX and Sandy Bridge-EP, although RAS features still have some way to go.

      1. Matt Bryant Silver badge
        Boffin

        Re: Benchmarks ??

        "POWER7 absolutely crushes Itanium and SPARC already, in about every metric...." Maybe in dragster-style "speed" benchmarks, but in real operational tests, wheer the whole solution is more important than just how fast the CPU can crunch integers, then IBM does not beat Itanium-based systems every time. The proof of this is that hp is still selling plenty of Itanium despite Oracle's FUD.

        1. Beachrider

          I never like defending Larry, but..

          On Itanium, why aren't more people going after Microsoft and RedHat. They abandoned Itanium EARLIER than Oracle. Oracle certainly has its own issues competing with Intel with Sparc, but they are trying the same thing that IBM does (i.e. sell Intel-stuff, but also market your own-stuff).

          I knew that Kebbab would be on this chain...

          1. Anonymous Coward
            Anonymous Coward

            Re: I never like defending Larry, but..

            Itanium is good and properly dead. HP's own VP of ISS had a momentary lapse of propaganda and mistakenly told the press that HP was moving to x86 - Linux away from Itanium - HP-UX and the other legacy OSs as a future strategy. Quote: "Our go forward strategy for mission-critical systems is shifting to an x86-based world," Farrand said. "It's not by coincidence that folks have de-committed from Itanium, specifically Oracle." HP had to, of course, take back the statements and say that Itanium is going to be around forever.

            http://www.serverwatch.com/server-trends/hp-moving-from-unix-itanium-to-x86-odyssey-on-linux.html

          2. Matt Bryant Silver badge
            Boffin

            Re: I never like defending Larry, but..

            People are going after Larry as he has dressed his decision up as "just good business", when the reality is the arguments he tries to aplly to Itanium are either p[atently untrue, exaggerated, or apply doubly so to Snoreacle's Slowaris servers. The sales of new licences for Oracle software on Itanium dwarf those of new ones on Slowaris and have done so for several years. In effect, this is nothing but Larry trying to carve out some Slowaris sales for his uncompetitive Slowaris kit with some classic FUD.

            Microsoft and Red Hat simply couldn't convince enough customers to choose their OS over hp-ux or VMS. I know that because I spent a lot of time trying to convince our own board that RHEL on Itanium (including on Superdomes) was a good choice, but they prefered the comfort they got from buying hp-ux as they trusted hp. Microsoft shot themselves in the foot by attempting to protect their licensing stream by not porting certain products to Itanium, such as Exchange. Exchange on Itaniuim would have had the scale to match Domino on AIX-Power for large, centralised email solutions, but it would have cannibalised the existing Exchange base (why buy a dozen Exchange licences for x86 when you can buy two for a clustered Itanium version and do the job better?).

        2. Anonymous Coward
          Anonymous Coward

          Re: Benchmarks ??

          "The proof of this is that hp is still selling plenty of Itanium despite Oracle's FUD."

          HP's Itanium has been in free fall since the Oracle announcement. Down 30% every quarter.

  2. Anonymous Coward
    Stop

    POWERCloud ?

    Also, where is the POWER-based cloud so that I can purchase processing capability "on demand" ?

    1. Jesper Frimann
      Holmes

      Re: POWERCloud ?

      Been here for years, it's called Cloudburst, and been around for years.

      // Jesper

    2. Beachrider

      Power cloud is here...

      There are two tracks to using IBM Power in a cloud-approach:

      1) Capacity on demand with p770 or p780 (depending on scale of upper-end), they are both 2 year old boxes, though.

      2) IBM Flex p260 and p460 are more current and start much cheaper.

      ... There are several willing IBM superstars that watch that Register. They can expand for you.

  3. Anonymous Coward
    Anonymous Coward

    Random number generator for financial services or physics?

    Are we talking about hardware to generate genuine random numbers, which can't be done in software, or hardware acceleration for pseudo-random numbers, which could be done in software but less efficiently?

    If it's for financial services or physics, that suggests pseudo-random numbers. For cryptography you need real random numbers, in theory, but for physics, etc, pseudo-random numbers are good enough and actually better because you can rerun a simulation when investigating a peculiarity or bug.

    1. Anonymous Coward
      Stop

      @AC: Randomness / NOT True

      Pseudo-random number sequences should be looked at with much scepticism. The standard-issue rand() function is utter shite, but also more sophisticated things such as cryptographic stream ciphers (e.g. LFSRs, RC4) are not perfect at all. Just because there are no easily visible/testable statistical imbalances does not mean there are none.

      Physical Randomness is very valuable for many applications and I am missing cryptography here. From creating "small" keys (up to 10kByte) up to one-time-pads, this is essential. Linux currently uses a mix of strange kludges (e.g. timestamps of network packet arrivals) to generate physical random numbers (which are in turn accessible via /dev/random).

      With this feature, you can easily create CDs full of one-time-pads in short order. These are provably unbreakable if IBM has done the RNG properly ( I would trust them they have some people to do this).

      Also for simulations - use physical random numbers if your code is stable and you want best results !

    2. Anonymous Coward
      Go

      Yes, hardware/physical RNG

      They use something like the jitter of a special oscillator or the noise of a P-N junction. So yes, real, physical noise converted into numbers. Some other CPUs have this capability, too (e.g. some SPARC).

      1. Anonymous Coward
        Anonymous Coward

        Re: Yes, hardware/physical RNG

        Is that similar (in capability) to the RNG that Intel implemented in Ivy Bridge?

    3. Ilgaz

      Re: Random number generator for financial services or physics?

      Wouldn't they have a pci card solution for such jobs? Remember these are enterprise machines, IBM customers are very different profile compared to Intel, especially mainframe type wants special purpose chips that won't bug the cpu.

  4. P. Lee
    Coat

    I wonder...

    if we can get it run an old copy of OSX...

    /ducks, runs

    1. Ilgaz

      Issue is, how to try on a million dollar monster with armed guards

      There is a possibility since IBM still uses CHRP boot process they designed with Apple :)

      http://www.answers.com/topic/chrp

      Besides jokes, you can boot a apple g5 64bit Linux CD (at least via firmware commands) on these monsters, of course it would be real insane waste of cpu (and possibly your job) trying.

      How did I learn about CHRP? Well, asked AIX guys if I should try free AIX beta iso on my G5. They are polite, they just talked about this crazy idea and laughed, didn't ban me.

    2. Sloppy Crapmonster
      Joke

      Re: I wonder...

      Did you see that roadmap? Power8 will obviously be running Commodore64 BASIC.

      1. Beachrider

        Yeah, but it will be FAST...

        I can't wait to fire-up my old BASIC program for 'life' on such a box!

  5. asdf

    no surprises

    IBM as usual is a full generation behind Intel x86 on process size (45 nm is so four years ago) with x86 breathing down its neck performance wise for a fraction of the cost. What saves IBM is the volume they do on gaming consoles where x86 compatibility is not desired. Still glad to have another player out there and one who realized the Cell BE SPEs were a mistake and quickly killed the idea when the market spoke.

    1. Beachrider

      Intel HAS been sprinting to process reduction faster than IBM...

      No doubt that IBM lingers longer on a given process-size than Intel. They don't sell as many chips (don't count cell chips against Power sales, they are VERY different), either. IBM does have the advantage of RISC, which they play against x64 chips. It gives them faster performance on an older process-size. Whenever it stops working, IBM's strategy will have to change (they DO sell Intel servers). In the meantime, they get more margin on the sales where Power beats Intel-servers.

      Back when Intel COULD do GHz-cranking, the Intel-fanbois made EVERYTHING about GHz. In the last decade, where the GHz crank has done very little for Intel's progress, it is getting more about overall application throughput. That seems more fair to my needs.

      Some Intel-fanbois abstract results from $80K servers onto home-motherboards and do price comparisons with large-machines. That is a vast oversimplification.

      1. Anonymous Coward
        Anonymous Coward

        Re: Intel HAS been sprinting to process reduction faster than IBM...

        That's just a load of bull (please excuse me). POWER on 45nm manages to be faster (or at least just as fast as) than x86 on 32nm because it has a vastly higher power budget and goes into much more advanced packaging and systems with better integration. RISC vs CISC is completely irrelevant, and in fact, POWER is probably one of the least orthodox RISC implementations there are.

        And if anyone is going for GHz at this juncture, it's IBM with 5GHz+ z/Arch processors and possibly 5GHz+ POWER processors at 32nm. It serves them well with regard to single-threaded performance and probably most of all for latency-sensitive applications, but power consumption shoots through the roof. Intel can't do the same trade-off at ~130W TDP.

        1. Beachrider

          Re: Intel HAS been sprinting to process reduction faster than IBM...

          Current consumption doesn't boost performance AT ALL. It is an unfortunate fallout of larger traces and higher GHz. RISC just CAN enable better/more single-cycle performance. IBM is clearly getting benefit from that.

          Maybe I was too subtle. When Intel COULD turn the GHz crank, they convinced lots a people that GHz was ALL that mattered. Now that Intel can't get much GHz action, they make it ALL about process-size. New day, same shit.

          It is all good with me. I just need fast, cheap computers.

          1. Anonymous Coward
            IT Angle

            Re: Intel HAS been sprinting to process reduction faster than IBM...

            I'm not sure what you're trying to argue. Sandy Bridge for example came with very significant improvements to performance irrespective of the process it was manufactured on (in fact it was manufactured on the same 32nm process as its predecessor, Westmere). The fact is that Sandy Bridge-EP has performance on par with the much more expensive and advanced POWER7 while sharing motherboard, chipset etc. with workstation/enthusiast systems and its core microarchitecture with consumer products that go into cheap desktops and laptops. That is the result of brilliant R&D and engineering at Intel for many years now, not of a single-minded focus on fast die shrinks. If that was the case, Sandy Bridge-EP on 32nm would not be significantly faster than Westmere-EP on 32nm.

            And yes, increased current does boost transistor performance. If you didn't know that, I'm not sure what you're even doing on these forums. That "RISC just CAN enable better/more single-cycle performance. IBM is clearly getting benefit from that." when a 2.9GHz Sandy Bridge-EP (E5-2690) offers comparable or better performance than a ~4GHz POWER7 processor (see attached SPEC numbers) is yet another completely bizarre statement. I think your rhetoric is better suited for a time when a 600MHz Alpha was actually faster than a 1GHz Pentium III, but those days are long gone.

            POWER7:

            http://www.spec.org/cpu2006/results/res2010q2/cpu2006-20100426-10753.html

            http://www.spec.org/cpu2006/results/res2010q2/cpu2006-20100426-10752.html

            Sandy Bridge-EP:

            http://www.spec.org/cpu2006/results/res2012q1/cpu2006-20120305-19593.html

            http://www.spec.org/cpu2006/results/res2012q1/cpu2006-20120305-19595.html

            1. Beachrider

              Re: Intel HAS been sprinting to process reduction faster than IBM...

              You just keep broadening the discussion. Both vendors come out with internal improvements that help progress their product.

              The benchmark references keep moving the focus, as well. We were on the SAP benchmark, now you want to switch to spec. We would just need to reload the argument to take into account the differences between the two benchmarks. Let us stay on SAP, for now.

              My goodness, all I would have to do is broaden the discussion back to you and run cost-power-analyses of ARM or core-by-core watts for SPARC. Power is faster, Power fanbois have a story to fight Intel, too.

              I get it that you think that Intel is 'brilliant'. I repeat myself in advising that 'it is all good' and that I want faster cheaper computers, just like anyone else. Power just isn't totally beaten

              FWIW your current comment is simply wrong. Increased current doesn't improve transistor performance, voltage does. No one overclocks a board by increasing current, they increase voltage.

              1. Anonymous Coward
                Anonymous Coward

                Re: Intel HAS been sprinting to process reduction faster than IBM...

                More current flowing through transistors equals faster switching speeds. And actually, the reason I used the word current was because that was the word you used in your post (the one I was responding to). Yes, you overclock a CPU by increasing the voltage, but increased voltage enables higher performance by increasing the amount of current. If this is the kind of discussion you want to have, then you're welcome to have it with yourself.

                Also, SPEC numbers are hardly irrelevant (though they may not paint a full picture) when looking at the performance of two comparable processors, and if a 2.9 GHz "CISC" processor beats a 4GHz "RISC" processor in almost every test, then that renders your "RISC enables higher single cycle performance for IBM" comment ridiculous (especially since POWER7 is at every single point of comparison a more advanced processor than Sandy Bridge-EP). At no point did we even discuss SAP ratings, you just made that up. SAP scores will also place a greater emphasis on total system performance than SPEC ratings, which is not relevant in a RISC vs CISC debate.

                By the way, I have no major doubts that POWER7+ on 32nm will blow Sandy Bridge-EP out of the water, especially for database workloads (as it really should be). But that will have nothing do with RISC vs CISC, rather with its having possibly 5GHz+ clock speeds and 80MB of L3 cache.

                1. Beachrider

                  Re: Intel HAS been sprinting to process reduction faster than IBM...

                  Have a nice day.

                  You just keep moving from topic to topic, perhaps to 'paint a full picture'. You engaged me when I advised on a Kebabbert reference to SAP benchmarks. So I didn't make anything up, you just didn't read what was written.

                  I never said that Spec was irrelevant, just superfluous. SAP is agnostic to CISC or RISC. It just cares about results. I like that.

                  The other discussion is interesting, it gets into several cool things:

                  1) Instruction architecture (you are making a case that CISC is slower than RISC, even with GHz favoring RISC)

                  2) Benchmarking architecture (SPEC is like a lie detector, sometimes it gives 'surprising' results)

                  3) Timing of measurements (measuring entries from machines that are introduced at different times of a product's cycle often give very different results).

                  4) Hardware architecture (Chip layouts can be like city-layouts, IBM has been able to shift stuff around on the core more than Intel has, although both make changes at each new product-level)

                  All of this is cool stuff. It would require a more collegial tone that we have had, so far...

                2. Jesper Frimann
                  Holmes

                  Re: Intel HAS been sprinting to process reduction faster than IBM...

                  Eh ?

                  You do understand that SPECint2006 and specfp2006 (not rate) have been broken and has been for quite some years ?

                  You also do realise that POWER and Sandy Bridge-EP are targeted at completely different types size and types of systems as POWER has it's sweet spot at 8 sockets and .. well SB-EP at 1 ?

                  You also realise that the increased use of 'accelerators' inside processors today make it increasingly hard to make general direct comparisons ?

                  So I think using the phrase "If you didn't know that, I'm not sure what you're even doing on these forums.", is a bit harsh.

                  // Jesper

                  1. Anonymous Coward
                    Anonymous Coward

                    Re: Intel HAS been sprinting to process reduction faster than IBM...

                    I admit that SPEC ratings never make for a perfect comparison (and I think I did in the original post) but if you look at almost every individual test (whether base or optimized), Sandy Bridge is purely and simply faster. That is unlikely to be the result of just compiler tricks for every single benchmark.

                    I don't see how accelerators impact the comparison unfairly when you are, as in SPEC ratings, looking at exactly the same general purpose code for a large variety of different tests. If Sandy Bridge has some kind of accelerator that gives it an edge over POWER in general purpose code (I guess that would be AVX and compiler auto-vectorization), then it apparently does so across the board and that advantage would be translatable to many real-world applications. libquantum does look like it's completely broken by both IBM and Intel compilers, and that is why I suggest you look at individual tests.

                    I know that you in particular think that SPEC numbers are completely worthless, but a lot of other also very knowledgeable people (without a doubt more knowledgeable than me) disagree. You could look at rate benchmarks also and see that Sandy Bridge's performance carries over rather well. The reason I didn't link to them is that I think (though I might be wrong) that the differences in memory and disk subsystems (the POWER machines had twice the amount of RAM and an array of SSDs compared to just one HDD in the System x) would give POWER an unfair advantage in a throughput scenario.

                    I would agree that SAP and TPC benchmarks give a better idea of the performance of these systems for what they would actually be used for, but that was NOT the purpose of this discussion, rather if RISC gives POWER an inherent advantage over x86. In that particular case, I believe that SPEC is in fact a more valuable benchmark.

                    1. Jesper Frimann
                      Headmaster

                      Re: Intel HAS been sprinting to process reduction faster than IBM...

                      @af Horunge

                      You start to use SPEC® CFP/INT2006 Results to compare processors, now these benchmarks are supposed to tell you about single threaded performance, which haven't directly got much to do with 'processor performance'.

                      If you want to look at 'processor performance' or throughput you really need to look at the rate numbers.

                      Now if you think that Memory capacity and SDD disks have any what so ever impact on SPECint/fp®_rate2006, you are quite wrong.

                      As for accelerators, then for example POWER7 has Decimal floating point support which SB hasn't, and SB has AES support which POWER7 hasn't.

                      // Jesper

                      1. Anonymous Coward
                        Anonymous Coward

                        Re: Intel HAS been sprinting to process reduction faster than IBM...

                        Well, even if you look at rate results with POWER7 systems and Sandy Bridge-EP systems with the same amount of cores, performance is still comparable (see attached links further below). There is definitely nothing to suggest that RISC gives POWER7 (again with substantially higher clock speed, 3.86GHz vs. 2.9GHz) "higher single clock cycle performance" than a comparable x86 challenger. In reality, POWER needs a higher clock frequency to compete with Sandy Bridge-EP (though I am not saying that its nominally RISC ISA makes POWER inefficient, I am saying that it's probably irrelevant). I don't care that POWER7's sweet spot is at more than 2 sockets, because the discussion we were having was not about systems performance, but core performance (where RISC vs. CISC could even imaginably be relevant).

                        And how exactly do AES acceleration and decimal floating point support come into play in SPEC benchmarks?

                        Links to SPEC_rate results:

                        POWER7:

                        http://www.spec.org/cpu2006/results/res2010q1/cpu2006-20100208-09586.html

                        http://www.spec.org/cpu2006/results/res2010q1/cpu2006-20100208-09582.html

                        Sandy Bridge-EP:

                        http://www.spec.org/cpu2006/results/res2012q1/cpu2006-20120305-19594.html

                        http://www.spec.org/cpu2006/results/res2012q1/cpu2006-20120305-19596.html

                        1. Jesper Frimann
                          Holmes

                          Re: Intel HAS been sprinting to process reduction faster than IBM...

                          Eh, you can read a benchmark submission can you ?

                          "Intel Turbo Boost Technology up to 3.80 GHz" versus"Intelligent Energy Optimization enabled,

                          up to 3.94 GHz" for POWER7.

                          So the processors run at practically the same clock frequency.

                          With regards to design points.

                          Then sure it matters if the processor is designed for 32 sockets or 2 sockets. POWER7 chip has 3 Intra node busses, and 2 internode busses, compared to the 2xQPI links that the SB-EP uses to connect two processors to each other. Again scalability comes at a price.

                          With regards to an imaginary comparison of cores, then sure you can do that, but it will always be in the context of two multicore chips, that are designed for different types of servers and on different production technologies. If you really wanted to compare core performance on a theoretical levels as you are suggesting, then you basically would need to 1 core chips manufactured and designed for the same types of servers. A theoretical adventure.

                          Furthermore when you compare SB to POWER7 you have to take into account that one is sparkling new, and the other is a 2 1/2 year old product. Which kind of makes your theoretical comparison.. well shaky.

                          Furthermore when it comes to accelerators, then on a benchmark like 462.lipquantum, Xeon processor have managed to increase the per core score with a factor of 7 going from pre AES Xeons to the current SB-EP.

                          So where I might agree that SB-ep is a great processor, then I think that the trees don't grow into the heavens.

                          // Jesper

                          1. Anonymous Coward
                            Anonymous Coward

                            Re: Intel HAS been sprinting to process reduction faster than IBM...

                            First of all, I doubt that either processor is running near its turbo limit in the rate benchmarks (actually I don't think they are running with turbo at all). Second of all, if you had actually paid attention to what I have been writing, you would know that I have not stated that Sandy Bridge-EP is an equal or superior processor for an 8+-socket database server, nor that it is fair to make such a comparison between a processor that is brand new and one that is 2½ years old. I fully agree that POWER7 and Sandy Bridge-EP compete in different classes and that the valuable metric for the purpose of making a server purchase is not their per-core performance but their performance as complete systems. I have also said that POWER7+ is likely to be hands down, significantly faster than Sandy Bridge-EP and it will naturally be much closer in time of release.

                            The only thing that I have been contesting is the quote "RISC just CAN enable better/more single-cycle performance", and for that reason I have compared SPEC results and clock frequencies with equal amounts of cores (16 cores in this case) and reached the conclusion that POWER7 does not in fact do more work per clock cycle than Sandy Bridge-EP. The age of the processors in question and how far their system interfaces really scale etc. is not relevant, as we are not comparing systems but how fast their cores run and whether or not RISC vs CISC has any impact on their performance. Now if you disagree with that quote like I do, then we are pretty much in agreement. If you agree with that statement, then I have to seriously doubt how much you really understand about these matters. It's not anywhere near informed or correct territory.

                            1. Jesper Frimann
                              Headmaster

                              Re: Intel HAS been sprinting to process reduction faster than IBM...

                              @af Horunge

                              I disagreed with your arguments and only the arguments, not your attack on the validity of "RISC just CAN enable better/more single-cycle performance" in the case of SB-EP versus POWER7. Basically POWER was never the pure RISC processor and Intel x86 processors aren't really a CICS processor, and haven't been for years.

                              And IMHO you are still wrong with many of your arguments.

                              "and how far their system interfaces really scale etc. is not relevant, as we are not comparing systems but how fast their cores run"

                              Sure it is, 'interfaces' take up chip real estate, adds to memory latency etc. etc. hence it has an impact on performance, also for single core throughput.

                              And if you look at Turbo boosts for both processors then running one core they will surely run at peak speed, both of them.

                              // Jesper

  6. Anonymous Coward
    Anonymous Coward

    Sparc fanbois are hilarious

    Sun fired saled itself. Oracle EVPs are IMing about it being a "dog" and "dead dead dead"... yet there are still Sparc people out there who carry on... despite Sun not being around, despite it clearly not being a priority for Oracle... Larry's ego is the only reason it is still around, despite sales of Sparc machines nose diving for about a decade.

    1. asdf

      Re: Sparc fanbois are hilarious

      Not so sure. Yes there are still some Solaris fanbois around but I haven't heard anyone outside of Oracle use the term SPARC without dead, dog shit or dog slow in quite some time.

This topic is closed for new posts.

Other stories you might like