Kaveri: Aiming for 1080p30 and Compute

The numerical differences between Kaveri and Richland are easy enough to rattle off – later in the review we will be discussing these in depth – but at a high level AMD is aiming for a middle ground between the desktop model (CPU + discrete graphics) and Apple’s Mac Pro dream (offloading compute onto different discrete graphics cards) by doing the dream on a single processor. At AMD’s Kaveri tech day the following graph was thrown in front of journalists worldwide:

With Intel now on board, processor graphics is a big deal. You can argue whether or not AMD should continue to use the acronym APU instead of SoC, but the fact remains that it's tough to buy a CPU without an integrated GPU.

In the absence of vertical integration, software optimization always trails hardware availability. If you look at 2011 as the crossover year when APUs/SoCs took over the market, it's not much of a surprise that we haven't seen aggressive moves by software developers to truly leverage GPU compute. Part of the problem has been programming model, which AMD hopes to address with Kaveri and HSA. Kaveri enables a full heterogeneous unified memory architecture (hUMA), such that the integrated graphics topology can access the full breadth of memory that the CPU can, putting a 32GB enabled compute device into the hands of developers.

One of the complexities of compute is also time: getting the CPU and GPU to communicate to each other without HSA and hUMA requires an amount of overhead that is not trivial. For compute, this comes in the form of allowing the CPU and GPU to work on the same data set at the same time, effectively opening up all the compute to the same task without asynchronous calls to memory copies and expensive memory checks for coherency.

The issue AMD has with their HSA ecosystem is the need for developers to jump on board. The analogy oft cited is that on Day 1, iOS had very few apps, yet today has millions. Perhaps a small equivocation fallacy comes in here – Apple is able to manage their OS and system in its entirety, whereas AMD has to compete in the same space as non-HSA enabled products and lacks the control. Nevertheless, AMD is attempting to integrate programming tools for HSA (and OpenCL 2.0) as seamlessly as possible to all modern platforms via a HSA Instruction Layer (HSAIL). The goal is for programming languages like Java, C++ and C++ AMP, as well as common acceleration API libraries and toolkits to provide these features at little or no coding cost. This is something our resident compute guru Rahul will be looking at in further detail later on in the review.

On the gaming side, 30 FPS has been a goal for AMD’s integrated graphics solutions for a couple of generations now.

Arguably we could say that any game should be able to do 30 FPS if we turn down the settings far enough, but AMD has put at least one restriction on that: resolution. 1080p is a lofty goal to hold at 30 FPS with some of the more challenging titles of today. In our testing in this review, it was clear that users had a choice – start with a high resolution and turn the settings down, or keep the settings on medium-high and adjust the resolution. Games like BF4 and Crysis 3 are going to tax any graphics card, especially when additional DirectX 11 features come in to play (ambient occlusion, depth of field, global illumination, and bilateral filtering are some that AMD mention).

Introduction and Overview The Steamroller Architecture: Counting Compute Cores and Improvements over Piledriver
Comments Locked

380 Comments

View All Comments

  • jaydee - Tuesday, January 14, 2014 - link

    I just don't "get" adding three $300 Intel CPU's to a review of 3 sub $200 AMD CPU's. We all know, or can find out how i7 SB, IB, Haswell compare to each other. I can see adding one of these CPUs to show a baseline of how AMDs top-of-the-line compares against Intels (albiet at very different price points), but having all three of them gives the impression that you want to make sure everyone knows who's boss...

    Is there going to be an update on power draw? I'm really curious to see what the 45W Kaveri draws (idle and full power) considering it is so competitive with the 95W Kaveri.
  • jaydee - Tuesday, January 14, 2014 - link

    I stand corrected, four $300+ Intel i7 CPUs in this comparison, not three. And two Intel CPU's that are actually in the price range of the AMDs. It feels as if to do this review, AT just cobbled together whatever was laying around instead of being intentional about putting together the best test bench possible to compare the review product against its real competitors.

    It's really a shame, because the commentary, the in-depth look at the architecture and the conclusions are outstanding; no website out there has tech writers as good and do as thorough of a job as Anandtech. The fact that the test benches are just after-thoughts in some of these reviews are really disappointing.
  • UtilityMax - Tuesday, January 14, 2014 - link

    I think this was mentioned somewhere in the beginning of the review. Intel likes to have the i7 reviewed instead of other parts, so the send the i7 CPUs to everyone. This is kind of like what the car makers do. Most people just buy a reasonably priced, mid-spec car model. However, the journos always get to review these ridiculously over-optioned cars, that hardly sell, like the $35 Ford Fusions, even though the base car sells for 10 grand less.
  • jaydee - Tuesday, January 14, 2014 - link

    I am well aware that when having the choice, Intel would rather you review a $340 Intel CPU against a $179 AMD CPU. But is there not any way, given the ad revenue of Anandtech, to obtain a $190-200 mid-range Intel i5 CPU (such as the i5-4440) in order to have a relevant test bench for an eagerly anticipated AMD mid-range CPU launch?
  • srkelley - Tuesday, January 14, 2014 - link

    "...do any AnandTech readers have an interest in an even higher end APU with substantially more graphics horsepower?"

    Yes, oh yes! I'm letting the APU be the core of my system build in a few months and plan to upgrade as needed. I'd like a simple solution like an APU instead of having to go with a discrete card right away. If it lets me spend more on ram and other things, keep the psu and power draw low I'm happy. The most demanding game that I play right now is the Witcher 2. Eventually I will have to go discrete for Star Citizen and the Witcher 3 but if I can get solid enough results with a high powered apu then I'll simply jump to that instead.
  • Conduit - Tuesday, January 14, 2014 - link

    This has been a long time coming thanks to Always Major Delays (AMD). Even know they can't get their sh!t together.
  • nissangtr786 - Tuesday, January 14, 2014 - link

    http://techreport.com/review/25908/amd-a8-7600-kav...
    The fpu still not improved miles still behind intel.
    http://browser.primatelabs.com/geekbench3/326781
    http://browser.primatelabs.com/geekbench3/321256
  • A5 - Tuesday, January 14, 2014 - link

    "do any AnandTech readers have an interest in an even higher end APU with substantially more graphics horsepower? "

    Maybe in the context of a Steam Machine? But for my main gaming PC, no way. Maybe something they can try out after the next die shrink if SteamOS really takes off.
  • Xajel - Tuesday, January 14, 2014 - link

    I believe the main reason for AMD to not have a new FX is SOI process.. it was okay in it's glory days, but it can't keep it up with bulk silicon...

    They want to increase the clock at least but they're trapped with 32nm SOI, and why not moving toward 28nm or lower SOI because I think that AMD is already working to convert it's entire CPU's to Bulk Silicon, so it's not logical to make a new design for still not ready yet 28nm or lower SOI while they are already designing the new core for bulk silicon.. knowing that both SOI and bulk require a complete redesign of the silicon just to convert from one to another... so it's not even logical to convert Pilediver or maybe even Steamroller to bulk silicon for only one year while they're working on the next architecture which will be Bulk Silicon...
  • jimjamjamie - Thursday, January 16, 2014 - link

    That makes sense, I was very confused as to why AMD were not going to refresh the FX line - even if HSA is the future for AMD, I presumed new FX sales throughout this year would have helped things along.

    Perhaps it wasn't worth the cost, but at the same time AMD could really do with keeping the fanboys on side.

Log in

Don't have an account? Sign up now