Forgot your password?
typodupeerror
Oracle Hardware

Oracle To Halve Core Count In Next Sparc Processor 200

Posted by timothy
from the reverse-course-full-steam dept.
angry tapir writes "Oracle will halve the number of cores in its next Sparc processor and instead improve its single-thread performance, a weak area for the chip but one that's important for running large databases and back-end applications. The next Sparc chip on Oracle's roadmap, the T4, will have eight cores on each chip, down from 16 in the current Sparc T3."
This discussion has been archived. No new comments can be posted.

Oracle To Halve Core Count In Next Sparc Processor

Comments Filter:
  • by multipartmixed (163409) on Tuesday December 07, 2010 @08:16PM (#34481692) Homepage

    Does it maybe mean more register windows?

    Because that would certainly help things like Java, and presumably oracle.

    Anybody know how often a large query spills registers?

    • by Surt (22457)

      I assume they're talking about improving their multiple dispatch, so that they can go from 3 - 4 (or is it 4-5) ops in parallel on a single core. And probably bring up the clock speed. 8 cores at 2ghz beats 16 cores at 1.5 ghz for a lot of applications.

    • by greg1104 (461138)

      There are all sorts of common database paths where slow cores are troublesome. Acquiring locks, access to shared memory, writes to redo logs; these are all examples of things that can end up serializing more than is optimal if individual cores are slow. Because of this, half as many cores that run at twice the speed is not the same net speed; it's probably faster, because each process is introducing less contention. Reducing the time things hold onto shared resources is really important for database work

  • I'm pretty sure this was on Suns roadmap. Higher throughput per thread. Higher clock speeds. So have Oracle deviated from the plan Sun had?

  • I don't think the author had any understanding of the history of SPARC or Oracle (Sun)'s product linup. Here is an informative interview from the useful Sun hardware oriented blog on the subject http://www.c0t0d0s0.org/ [c0t0d0s0.org] http://www.oracle.com/us/corporate/innovation/innovator-hetherington-191304.html [oracle.com]

  • Sparc (Score:5, Informative)

    by TopSpin (753) on Tuesday December 07, 2010 @08:23PM (#34481742) Journal

    The reduction in cores from 16 to 8 was part of the Sparc road-map [channelregister.co.uk] before Sun was acquired by Oracle. Despite a lot of speculation it appears Oracle is following through with the plans they bought from Sun.

    ... Sun was going to cut back the number of cores to eight and crank the clocks to 2.5 GHz ...

  • by Doc Ruby (173196) on Tuesday December 07, 2010 @08:29PM (#34481796) Homepage Journal

    Reducing the core count lets Oracle make each core bigger, to add features making each faster. But can't Oracle keep the same core count, and instead of increasing the core count in the next generation the way most other CPU makers will, just add circuits to each existing core? Is it really necessary to reduce the count? Process size will probably also be shrinking in that generation, and new tricks developed, as usual. Can't Oracle just make a bigger chip, and also keep the benefits of the high core count Sun already achieved?

    • by GWBasic (900357)

      Reducing the core count lets Oracle make each core bigger, to add features making each faster. But can't Oracle keep the same core count, and instead of increasing the core count in the next generation the way most other CPU makers will, just add circuits to each existing core? Is it really necessary to reduce the count? Process size will probably also be shrinking in that generation, and new tricks developed, as usual. Can't Oracle just make a bigger chip, and also keep the benefits of the high core count Sun already achieved?

      From what it sounds like, Oracle could be devoting the extra space to cache. A large cache can go a long way in CPU-bound operations; or help make a very fast database.

      Making a bigger chip isn't as easy as it sounds. As the die size increases, the probability of a defect within the die increases. (Imagine that you have 5 specs of dust on a wafer, if the die size is larger then the ratio of good to bad is worse.) Large die sizes will also have problems with heat distribution, or could limit total clock spee

      • by dgatwood (11270)

        This is why you build each group of cores and the corresponding cache on a separate die, test and bin each die independently, then wire them together [wikipedia.org] inside the package. Sure, there's the added potential for interconnect failure, but so long as you test the integrated module before you epoxy the lid on, you should be able to salvage those parts.

      • by mlts (1038732) *

        Oracle could have always gone the route IBM did with the POWER7 chips and have the best of both worlds. With Power7, you can turn half the cores off. The remaining cores will use the cache on the counterparts that are off, and the clock speed gets a decent bump.

        This is what Oracle should have done -- if someone is doing a task that is easily split up into parallel parts, or using a lot of domains/VMs, allow for this. If they need more oomph per core, have half the cores flip off, and the others use their

    • by Surt (22457)

      An increasing number of cores tends to be a challenge to keep clocked at a high speed. Every CPU developer struggles with this, and they all market higher clocked lower core count parts. Choosing to go for a lower core count in your design phase makes a lot of sense if you are single thread bound.

  • but that doesnt really matter now does it? We know your application only supports 2 and scalability isn't an option.
  • Sanity, at last! (Score:3, Insightful)

    by kanto (1851816) on Tuesday December 07, 2010 @09:36PM (#34482306)

    When will people realize that not everything runs better on more cores, especially stuff that's highly dynamic say like a database query which is effectively a long sequence of conditionals. You talk to people and the first thing they ask is "yeah, but how many cores does it have"... it's like multithreading didn't exist until dualcore cpus.

    A cpu has a limited amount of processing power; some things you can only do in sequence ergo you can't do them in parallel ergo you're limited by the core-speed ergo you're fucked with 16 core 1GHz machine against a 1 core 2GHz machine.

    • by Surt (22457)

      Not everything runs better on more cores, but so many things scale close to linearly with cores that it has become what the majority rightly want. Basically every business function that has to support N users can be partitioned over up to N cores, the more cores you pack per chip, the less chips and sockets and boxes you have to buy.

  • I said "Can HAVE cores plz"!!!!

Chemist who falls in acid is absorbed in work.

Working...