IPC Increases: Double L1 Data Cache, Better Branch Prediction

One of the biggest changes in the design is the increase in the L1 data cache, doubling its size from 64 KB to 128 KB while keeping the same efficiency. This is combined with a better prefetch pipeline and branch prediction to reduce the level of cache misses in the design. The L1 data cache is also now an 8-way associative design, but with the better branch prediction when needed it will only activate the one segment required and when possible power down the rest.  This includes removing extra data from 64-bit word constructions. This reduces power consumption by up to 2x, along with better clock gating and minor adjustments. It is worth pointing out that doubling the L1 cache is not always easy – it needs to be close to the branch predictors and prefetch buffers in order to be effective, but it also requires space. By using the high density libraries this was achieved, as well as prioritizing lower level cache. Another element is the latency, which normally has to be increased when a cache increases in size, although AMD did not elaborate into how this was performed.

As listed above, the branch prediction benefits come about through a 50% increase in the BTB size. This allows the buffer to store more historic records of previous interactions, increasing the likelihood of a prefetch if similar work is in motion. If this requires floating point data, the FP port can initiate a quicker flush required to loop data back into the next command. Support for new instructions is not new, though AVX2 is something a number of high end software packages will be interested in using in the future.

These changes, according to AMD, relate to a 4-15% higher IPC for Excavator in Carrizo compared to Steamroller in Kaveri.  This is perhaps a little more what we normally would expect from a generational increase (4-8% is more normal), but AMD likes to stress that this comes in addition to lower power consumption and with a reduced die area. As a result, at the same power Carrizo can have both an IPC advantage and a frequency advantage.

As a result, AMD states that for the same power, Cinebench single threaded results will go up 40% and multithreaded results up 55%. The benefits are fewer however the further up the power band you go despite the increase, as the higher density libraries perform slightly worse at higher power than Kaveri.

Efficiency and Die Area Savings Power Saving and Power Consumption
Comments Locked

137 Comments

View All Comments

  • renegade800x - Thursday, June 4, 2015 - link

    Although viewable it's far from being "perfectly" fine. 15.6 should be FHD.
  • albert89 - Tuesday, June 23, 2015 - link

    You don't need a strong CPU since win8 because most laptops use atom, Celeron or Pentium processors. AMD APU's are the natural choice !
  • mabsark - Wednesday, June 3, 2015 - link

    AMD should make Steam Box's. They already do APUs, chipsets (which are going on die) and memory. It would be pretty simple for AMD to partner with a motherboard maker. Imagine a Steam Box about the size of a router, with a nano-ITX motherboard, a 14 nm APU with HBM, wifi, a few USB ports and an HDMI port to connect to a TV.

    An AMD/Valve partnership could potentially revolutionise the console market, providing cheap yet powerful and efficient console-type PCs.
  • Refuge - Wednesday, June 3, 2015 - link

    HBM isn't coming to APU's anytime soon.
  • Cryio - Saturday, June 6, 2015 - link

    Probably the first APU after Carrizo
  • coder111 - Wednesday, June 3, 2015 - link

    Aren't Steamboxes supposed to run Linux?

    AMD drivers for Linux are a bit weird. Catalyst is the official supported driver but it's buggy.

    Open source drivers are quite good but they are slower than Catalyst and don't support latest OpenGL spec. There is no Mantle/Vulcan/HSA/Crossfire support with Open-Source drivers either. OpenCL is in alpha stage.

    So AMD would need to man up and do the Linux drivers properly. They are working on it and making good progress but I doubt it is ready to be used at the moment as it is...

    Besides, lots of games these days get developed with Nvidia's "help" to ensure they run well on Nvidia GPUs and run like crap on AMD GPUs. And if the games are built using Intel Compiler, they'll run like crap on AMD CPUs as well. All of these tactics are anticompetitive and should be illegal IMO but who said the world is fair...

    And don't get me wrong, I love AMD, I use Linux + AMD dGPU + APU, but I don't think it's ready for the masses yet.
  • AS118 - Wednesday, June 3, 2015 - link

    I agree. I'm a double AMD Linux gamer and I've run into the exact same problems as you have, and I wish they'd be more serious about Linux. Sure they have Microsoft's support, but I feel that they should take Linux more seriously outside of the enterprise (where they do take Linux more seriously).
  • yankeeDDL - Wednesday, June 3, 2015 - link

    I disagree.
    For casual gaming on laptops, 1366x768 is just fine. You'll need a lot more horsepower to drive a fullHD screen and battery life will suffer.
    I won't say that there's no benefit gaming at fullHD vs 1366x768: obviously, the visuals are better, but if you want an "all rounder" laptop which does not weight one ton (like "real" gaming laptops) and that it is below $500, it's not bad at all.
  • BrokenCrayons - Wednesday, June 3, 2015 - link

    I personally would rather have a cheap 1366x768 panel. I don't care about color accuracy much, light bleed, panel responsiveness or much of anything else and haven't since we transitioned from passive to active matrix screens in the 486 to original Pentium era of notebook computers. In fact, I see higher resolutions as an unnecessary (because I have to scale things anyway to easily read text and interact with UI elements and because native resolution gaming on higher res screens demands more otherwise unnecessary GPU power) drain on battery life that invariably drives up the cost of the system to get otherwise identical performance. The drive for progressively smaller, higher pixel density displays is a pointless struggle to fill in comparable checkboxes between competitors to appease a consumer audience that has been swept up in the artificially fabricated frenzy over an irrelevant device specification.
  • yankeeDDL - Wednesday, June 3, 2015 - link

    I think it depends on the use, ultimately.
    For office work (i.e.: much reading/writing emails), a reasonably high resolution helps making the text sharp and easier on the eyes.
    For home use (web browsing, watching videos, casual gaming) though, I find it a lot less relevant.
    Personally, at home, I rather have a <$400 laptop always ready to be used for anything, to be moved around, even in the kitchen, than a $1000 laptop which I would need to treat with gloves for fears of damaging. Since Kaveri I also started recommending AMD again to my friends and family: much cheaper than Intel and with a decent GPU makes them a lot more versatile. Again, my opinion, based on my use. As they say: to each his own...

Log in

Don't have an account? Sign up now