Precision Boost 2

Exact per-core turbo timings for the new processors will be determined by AMD’s voltage-frequency scaling functionality through Precision Boost 2. This feature, which we covered extensively in our Ryzen 7 2700X review, relies on available power and current to determine frequency, rather than a discrete look-up-table for voltage and frequency based on loading. Depending on the system default capabilities, the frequency and voltage will dynamically shift in order to use more of the power budget available at any point in the processor loading.

The idea is that the processor can use more of the power budget available to it than a fixed look up table that has to be consistent between all SKUs that are stamped with that number.

Precision Boost 2 also works in conjunction with XFR2 (eXtreme Frequency Range) which reacts to additional thermal headroom. If there is additional thermal budget, driven by a top-line cooler, then the processor is enabled to use more power up to the thermal limit and get additional frequency. AMD claims that a good cooler in a low ambient situation can compute >10% better in selected tests as a result of XFR2.

Ultimately this makes testing Threadripper 2 somewhat difficult. With a turbo table, performance is fixed between the different performance characteristics of each bit of silicon, making power the only differentiator. With PB2 and XF2, no two processors will perform the same. AMD has also hit a bit of a snag with these features, choosing to launch Threadripper 2 during the middle of a heatwave in Europe. Europe is famed for its lack of air conditioning everywhere, and when the ambient temperature is going above 30ºC, this will limit additional performance gains. It means that a review from a Nordic publication might see better results than one from the tropics, quite substantially.

Luckily for us we tested most of our benchmarks while in an air conditioned hotel thanks to Intel’s Data-Centric Innovation Summit which was the week before launch.

Precision Boost Overdrive

The new processors also support a feature called Precision Boost Overdrive, which looks at three key areas for power, thermal design current, and electrical design current. If any of these three areas has additional headroom, then the system will attempt to raise both the frequency and the voltage for increased performance. PBO is a mix of ‘standard’ overclocking, giving an all core boost, but gives a single core frequency uplift along with the support to still keep Precision Boost trying to raise frequency in middle-sized workloads, which is typically lost with a standard overclock. PBO also allows for idle power saving with a standard performance. PBO is enabled through Ryzen Master.

The three key areas are defined by AMD as follows:

  • Package (CPU) Power, or PPT: Allowed socket power consumption permitted across the voltage rails supplying the socket
  • Thermal Design Current, or TDC: The maximum current that can be delivered by the motherboard voltage regulator after warming to a steady-state temperature
  • Electrical Design Current, or EDC: The maximum current that can be delivered by the motherboard voltage regulator in a peak/spike condition

By extending these limits, PBO gives rise for PB2 to have more headroom, letting PB2 push the system harder and further. PBO is quoted by AMD as supplying up to +16% performance beyond the standard.

AMD also clarifies that PBO is pushing the processor beyond the rated specifications and is an overclock: and thus any damage incurred will not be protected by warranty

StoreMI

Also available with the new Ryzen Threadripper 2 processors is StoreMI, AMD’s solution to caching by offering configurable tiered storage for users that want to mix DRAM, SSD, and HDD storage into a single unified platform. The software implementation dynamically adjusts data between up to 2GB of DRAM, up to 256 GB of SSD (NVMe or SATA), and a spinning hard drive to afford the best reading and writing experience when there isn’t enough fast storage.

AMD initially offered this software as a $20 add-on to the Ryzen APU platform, then it became free (up to a 256GB SSD) for the Ryzen 2000-series processors. That offer now extends to Threadripper. AMD’s best case scenario is citing a 90% improvement in loading times.

Core to Core to Core: Design Trade Offs Feed Me: Infinity Fabric Requires More Power
Comments Locked

171 Comments

View All Comments

  • just4U - Monday, August 13, 2018 - link

    Ian, were you testing this with the CM Wraith Cooler? If not is it something you plan to review?
  • Ian Cutress - Monday, August 13, 2018 - link

    Most of the testing data is with the Liqtech 240 liquid cooler, rated at 500W. I do have data taken with the Wraith Ripper, and I'll be putting some of that data out when this is wrapped up.
  • IGTrading - Monday, August 13, 2018 - link

    To be honest, with the top of the line 32core model, it is interesting to identify as many positive effect cases as possible, to see if that entire set of applications that truly benefit of the added cores will persuade power users to purchase it.

    Like you've said, it is a niche of a niche and seeing it be X% faster of Y% slower is not as interesting as seeing what it can actually do when it is used efficiently and if this this makes a compelling argument for power users.
  • PixyMisa - Tuesday, August 14, 2018 - link

    Phoronix found that a few tests ran much faster on Linux - for 7zip compression in particular, 140% faster (as in, 2.4x). Some of these benchmarks could improve a lot with some tweaking to the Windows scheduler.
  • phoenix_rizzen - Wednesday, August 15, 2018 - link

    It'd be interesting to redo these tests on a monthly basis after Windows/BIOS updates are done, to see how performance changes over time as the Windows side of things is tweaked to support the new NUMA setup for TR2.

    At the very least, a follow-up benchmark run in 6 months would be nice.
  • Kevin G - Monday, August 13, 2018 - link

    Chiplets!

    The power consumption figures are interesting but TR does have to manage one thing that the high end desktop chips from Intel don't: off-die traffic. The amount of power to move data off die is significantly higher than moving it around on-die. Even in that context, TR's energy consumption for just the fabric seems high. When only threads are loaded, they should only be with dies with the memory controllers leaving two dies idle. It doesn't appear that the fabric is powering down while those remote dies are also powering down. Any means of watching cores enter/exit sleep states in real time?

    I'd also be fun to see with Windows Server what happens when all the cores on a die are unplugged from the system. Consdiering the AMD puts the home agent on the memory controller on each die, even without cores or memory attached, chances are that the home agent is still alive consuming power. It'd be interesting to see what happens on Sky Lake-SP as well if the home agents on the grid eventually power themselves down when there is nothing directly connected to them. It'd be worth comparing to the power consumption when a core is disabled in BIOS/EFI.

    I also feel that this would be a good introduction for what is coming down the road with server chips and may reach the high end consumer products: chiplets. This would permit the removal of the off-die Infinity Links for something that is effectively on-die throughout the cluster of dies. That alone will save AMD several watts. The other thing about chiplets is that it would greatly simplify Thread Ripper: only two memory controller chiplets would be to be in the package vs. four as we have now. That should save AMD lots of power. (And for those reading this comment, yes, Intel has chiplet plans as well.). The other thing AMD could do is address how their cache coherency protocols work. AMD has hinted at some caching changes for Zen 2 but lacks specificity.
  • gagegfg - Monday, August 13, 2018 - link

    do not seem to exist more than once the 16 additional core of the 2990wx compared to the 2950x
  • Ian Cutress - Monday, August 13, 2018 - link

    https://www.anandtech.com/bench/product/2133?vs=21...
  • Chaitanya - Monday, August 13, 2018 - link

    Built for scientific workload.
  • woozle341 - Monday, August 13, 2018 - link

    Do you think the lack of AVX512 is an issue? I might build a workstation soon for data processing with R and Python for some Fortran models and post-processing. Skylake-X looks pretty good wit its quad memory channels despite its high price.

Log in

Don't have an account? Sign up now