Review Index:
Feedback

AMD Kaveri Mobile Preview - AMD FX-7600P Performance

Author:
Subject: Processors, Mobile
Manufacturer: AMD
Tagged: amd, FX, fx-7600p, Kaveri

Kaveri Goes Mobile

The processor market is in an interesting place today. At the high end of the market Intel continues to stand pretty much unchallenged, ranging from the Ivy Bridge-E at $1000 to the $300 Haswell parts available for DIY users. The same could really be said for the mobile market - if you want a high performance part the default choice continues to rest with Intel. But AMD has some interesting options that Intel can't match when you start to enter the world of the mainstream notebook. The APU was slow to develop but it has placed AMD in a unique position, separated from the Intel processors with a more or less reversed compute focus. While Intel dominates in the performance on the x86 side of things, the GPU in AMD's latest APUs continue to lead in gaming and compute performance.

The biggest problem for AMD is that the computing software ecosystem still has not caught up with the performance that a GPU can provide. With the exception of games, the GPU in a notebook or desktop remains under utilized. Certain software vendors are making strides - see the changes in video transcoding and image manipulation - but there is still some ground AMD needs to accelerate down.

View Full Size

Today we are looking at the mobile version of Kaveri, AMD's latest entry into the world of APUs. This processor combines the latest AMD processor architecture with a GCN-based graphics design for a pretty advanced part. When the desktop version of this processor was released, we wrote quite a bit about the architecture and the technological advancements made into, including becoming the first processor that is fully HSA compliant. I won't be diving into the architecture details here since we covered them so completely back in January just after CES.

View Full Size

The mobile version of Kaveri is basically identical in architecture with some changes for better power efficiency. The flagship part will ship with 12 Compute Cores (4 Steamroller x86 cores and 8 GCN cores) and will support all the same features of GCN graphics designs including the new Mantle API.

View Full Size

Early in the spring we heard rumors that the AMD FX brand was going to make a comeback! Immediately enthusiasts were thinking up ways AMD could compete against the desktop Core i7 parts from Intel; could it be with 12 cores? DDR4 integration?? As it turns out...not so much.

Continue reading our preview of the AMD FX-7600P Mobile Kaveri APU!!

View Full Size

Instead, AMD is using the FX brand in a place you likely least expected it: on a mobility APU. That's right, the AMD FX-7600P and FX-7500 models will carry the FX banner going forward. With what you ask? With slightly better performance than the other APUs below them in the product stack.

View Full Size

AMD is quite proud of its product placement in a few key areas. The AMD FX-7500 is a 19 watt APU that competes directly against the Intel Core i7-4500U Haswell processor, used quite often in notebooks today. In this slide AMD compares performance of the two processors in PCMark8, 3DMark and Basemark CL. It should not be a surprise then than the APU comes out ahead by large margins in latter two tests and is listed as "comparable" with PCMark8. 

Avid readers of PC Perspective may note that these tests are quite selective. PCMark8 uses OpenCL for some of its more demanding tests, a compute language that can take advantage of the GPU resources provided by the Kaveri part. 3DMark is gaming focused, so again, that's a heavy GPU focus and Basemark CL, as the name implies, is a large set of OpenCL tests.

View Full Size

Here is the breakdown of the parts launching today from AMD. Broken up into two different power levels, 35 watt and 19 watt. The 35 watt parts include the FX-7600P, A10-7400P and A8-7200P with Compute Core ranges of 12, 10 and 8. All three processors include a quad-core x86 processor and the GCN GPU cores drop by a set of 2 each iteration. Clock speeds remain high across the board with the flagship hitting a base clock of 2.7 GHz and a Turbo Boost rate of 3.6 GHz. All of these parts support high speed DDR3 memory (up to 2133 MHz on the FX-7600P) as well as a full x16 lane of PCI Express 3.0

The 19 watt parts are more limited, starting at 10 Compute Cores and including an 8 core option. All of them also include a quad-core x86 processor - important for general computing. Base clock speeds range from 2.1 GHz to 1.8 GHz and maximum GPU clock rates are noticeably lower than the 35 watt derivatives. The PCIe implementation is also lowered to an eight lane Gen2 connection in order to save the most power for systems that are unlikely to need more bandwidth than that connection speed can provide.

AMD is also introducing a series of Pro level APUs that also fit in the 19 watt power envelope. The only outlier here is the AMD A6 PRO-7050B that drops to a dual-core x86 processor and 3 GPU cores!

Now AMD did let us test one of the new Kaveri APUs, though in a very limited one-day event in San Francisco. I did not get to keep a reference notebook for an extended period of time for a real-world analysis but the results provided by the FX-7600P are still very interesting.

June 4, 2014 | 12:48 AM - Posted by eddie (not verified)

now why is there no desktop part for this.

June 4, 2014 | 10:54 AM - Posted by ScrewySqrl (not verified)

There is. The A10-7850K is a desktop part

June 10, 2014 | 06:55 AM - Posted by g.g (not verified)

I guess the question was more like "why can't I (or won't be able to even in the future) buy a small form factor barebone with this awesome chip?" ;-)

June 25, 2014 | 12:23 PM - Posted by thatcrudedude (not verified)

That's an A series though, not an FX APU.

June 4, 2014 | 11:17 AM - Posted by Anonymous (not verified)

AMD should consider breaking the 4 core Mobile/laptop barrier with a 6 CPU core mobile part, and maybe some competition for the i5. AMD needs an enthusiast part for people who could care less about ultrabook/MacBook form factors, and need all the performance they can get in a Laptop SKU. nothing short of the complete x86 microarchitecture redesign that AMD has announced coming to completion in the next few years, is going to compete with Intel's single threaded/bandwidth/other x86 CPU core performance, that is if AMD can get some full fat x86 cores with SMP and no bandwidth bottlenecks. AMD could in the short term have made a laptop part with 6 CPU cores, and some power gating of these cores to give a 6 core laptop part with more flexibility in workloads, and better graphics by default.

I am waiting to see how these parts will perform with a Discrete AMD Mobile GPU, when these APUs get some design wins to sample.

June 4, 2014 | 12:09 PM - Posted by thet3rmun63 (not verified)

I think that AMD is on the right track. The APU is the future of computing. I think that when more software comes about that can utilize both GPU and CPU tech they will be way ahead of the curve. With GPUs being used in cars and a multitude of other items in the internet of things, AMD will leave Intel in their dust.
Now I will say that in todays world Intel is the Best money can buy and in the computer build I am working on I am going to be using an I5, But I am watching the software markets and waiting for the day that the para-dime shifts.

June 4, 2014 | 12:45 PM - Posted by Anonymous (not verified)

I do a lot of rendering, so the more cores/threads the better, if they can get ray tracing done through OpenCL on the GPU, then having as many CPU cores is not so important! Hopefully Nvidia and AMD can get some hardware Ray tracing like the PowerVR wizard has, and then I will not need a quad core i7 laptop part. I see that AMD already has some design wins with HP for the ultra low watt parts, but at that price I can get last years Intel(Haswell) core i7 laptop part on sale in an HP probook with an AMD discrete GPU! The only point that may score a Sale with me for the Kaveri APU, is if it can be paired for dual graphics, or OpenCL on the integrated GPU, and graphics and GPGPU on the discrete AMD GPU. If the makers of 3d graphics software can get ray tracing working better through openCL, then the CPU becomes a non issue, and I hope that hardware ray tracing can come to the discrete GPU market, like it has with the mobile PowerVR wizard.

June 4, 2014 | 04:30 PM - Posted by renz (not verified)

isn't that the ray tracing stuff in Power VR need dedicated hardware or some sort? like tessellation needing the specfic hardware on current gpu instead of relying on GPGPU.

http://techreport.com/news/26178/powervr-wizard-brings-ray-tracing-to-re...

June 4, 2014 | 05:22 PM - Posted by Anonymous (not verified)

That would be the hardware ray tracing, mentioned twice in the post you replied to. Yes I would love My discrete laptop GPU, or desktop(if I owed one) GPU to have Hardware ray tracing, or any SOC/APU to have Hardware ray tracing! The reason I have to rely on overpriced Intel CPUs, is the lack of OpenCL ray tracing support in the graphics software that I use. Hardware ray tracing would be the best, as then it would be built into the GPU hardware, and faster than OpenCL accelerated ray tracing, or CPU ray tracing. If AMD can get HSAIL adopted and abstract this need for the ability to run GPGPU ray tracing in software across any CPU/GPU device, then fine, but that is just not going to happen anytime soon, so maybe someone can Fund Imagination Technologies production of a discrete GPU version of the PowerVR wizard for pro graphics workloads on Laptops. The makers of Open Source and proprietary graphics software are starting to utilize OpenCL more, but the process of adoption takes time.

Having hardware Ray Tracing on a discrete GPU, or Mobile/Laptop SOC would remove most of the software difficulty surrounding Ray Tracing, and remove any advantage the CPU has over the GPU for Ray Tracing workloads, my primary reason for using the core i7 SKUs in my laptops.

AMD needs to show any advantage the Kaveri SOCs may have in the ray tracing area, over Intel's core i7 parts. I know that Ray Tracing presents a special problem for Gaming, with the need for frames per second over other considerations, but some of my single Image graphics workloads can take 4 hours or more to render if I have all the high AA, AO, ray interactions, and other settings turned up to high, so the ability to do the Ray interactions on a GPU, with multipliable ray tracing execution units in hardware would speed the process up by many factors over a few CPU cores/processor threads. Hell, I am hoping for discrete GPUs to get their own Integrated CPUs, to go along with dedicated Ray Tracing hardware.

June 5, 2014 | 08:45 AM - Posted by renz (not verified)

so maybe you can hope the dedicated ray trace hardware to be one of the unannounced feature of DX12.

June 5, 2014 | 12:15 PM - Posted by Anonymous (not verified)

What does a Graphics API(software), have to do with magically adding dedicated ray tracing hardware that has to be built into a GPU, if the feature is not baked into the physical hardware how can DX12 support it as an unannounced feature! M$ does not manufacture GPUs, so it has no control over any hardware in the GPU, other than what the GPU maker physically creates in the GPUs Microarchitecture.

DX whatever, is only software, and any ray tracing currently implemented on PC/Laptops has to be done in software and run on the CPU, and maybe if the gaming engine, other has some support that utilizes OpenCL to run some ray tracing via OpenCL calls that can be run on the GPU, so DX12 could maybe run some very limited ray tracing on the GPU, but DX whatever is mostly for gaming and fast rendering, and using an already busy GPU with no dedicated ray tracing hardware, is not going to produce enough ray interactions to be noticeable at 10 FPS, let alone 30 FPS and above. Currently AFIK, the only support for Hardware Ray Tracing built into any GPU is on a Mobile SOC based GPU the PowerVR wizard, and DX 12 is mostly a Desktop/Laptop gaming graphics API, and the GPU maker will be the one to write the drivers, that whatever Graphics API/s used will have to call to get to any GPU features.

Most graphics software does not use DX* for non gaming graphics workloads, but uses OpenGL, other, and OpenCL for running the ray tracing algorithms on a GPU, along with the standard CPU based ray tracing algorithms, but even with this is still going to take minutes per frame(animation), if there is to be any benefits, and many single frame Ad graphics with all the fancy reflections takes hours on high end workstations with multipliable professional GPUs and server CPUs working in tandem.

Gaming use alone in not going to sell enough APUs for AMD to make any money, AMD needs to make sure that these systems can outperform Intel's CPU for Ray Tracing non gaming workloads, if they want to get enough sales to turn a profit. AMD needs to make sure that the Open Source Graphics community is provided with the Tools and support to get their software running on and benefiting from AMD version of HSA, by making sure the OpenCL, and OpenGL support is there to leverage the APUs to Outperform Intel's core i7 CPUs/GPUs in the ray tracing Benchmarks.

If AMD's APUs can Not beat last years core i7 in ray tracing, then buying last years core i7 based laptop with AMDs or Nvidia's discrete mobile GPUs will be the more affordable solution for rendering. I can always get a last years' core i7 based new HP on sale at a big box electronic(Not best Buy) store, for the price of any new model APU AMD based, or Intel based, HP/other laptop of comparable performance and price, not that AMD can yet compete with Intel's x86 CPU performance. If AMD or Nvidia can get ray tracing hardware built into their Discrete GPUs, then for rendering(Ray tracing) the CPU would become a non issue for graphics rendering.

June 11, 2014 | 12:20 PM - Posted by Gadgety

"Gaming use alone in not going to sell enough APUs for AMD to make any money, AMD needs to make sure that these systems can outperform Intel's CPU for Ray Tracing non gaming workloads, if they want to get enough sales to turn a profit.AMD needs to make sure that the Open Source Graphics community is provided with the Tools and support to get their software running on and benefiting from AMD version of HSA, by making sure the OpenCL, and OpenGL support is there to leverage the APUs to Outperform Intel's core i7 CPUs/GPUs in the ray tracing Benchmarks."

What kind of market size, value wise, would the Open Source Graphics community represent?

June 16, 2014 | 12:13 PM - Posted by Rob (not verified)

> I think that AMD is on the right track.
Yes, but they drive like my Grandmother.

> The APU is the future of computing.
Part of, the beginning of, the Future.

It will be great for Tablets, which are popular for people who click more than they type since one Chip provides much of what is needed.

For people who actually do type then a Tablet is (obviously) not great, and a Laptop is an insult to the Finger. A real Keyboard is needed for typing, just as a "real" Computer is needed for Computing.

An APU is not the "real" Computer to which I refer.

What will be the Future is an HSA CPU (not an APU) where you buy a CPU that will accept Graphics Cards of the User's choosing to 'build' your own "APU".

The HSA CPU will run the OS and do the old X86 CPU duties (unless the HSA CPU is ARM based) serving the 'Compute Commands' to the Graphic Cards.

> AMD will leave Intel in their dust.
Yes, once they get SeaMicro Fabric with ARM/X86 HSA CPUs on a Server Motherboard that can hold a half dozen of AMD's powerful Graphic Cards -- then AMD will have SuperComputeServers.

Not built for your average User (that is what Tablets, and Cell Phones) are for) but I want one !

June 4, 2014 | 01:31 PM - Posted by Anonymous (not verified)

In More detail, those streaming capabilities are impressive, read and learn! Very thorough the Blighty way, they write the best computer books, hands down! Great review, and HSA explanation.

http://www.theregister.co.uk/2014/06/04/amd_releases_new_kaveri_mobile_a...

June 5, 2014 | 02:54 AM - Posted by danwat1234 (not verified)

It's only 2.7/3.6GHZ quad core with less work done per clock cycle than Sandy Bridge+, and 35W TDP. AMD needs to grow a pair and release some 47W and 57W APUs like Intel has done so it can have more room for performance

June 19, 2014 | 05:21 PM - Posted by Rob (not verified)

This Link points back at this Page.

http://www.pcper.com/reviews/Processors/AMD-Kaveri-Mobile-Preview-AMD-FX...

"... As it turns out...not so much.

(Link)_Continue reading our preview of the AMD FX-7600P Mobile Kaveri APU!!_

Instead, AMD is using the FX brand in a place you likely least expected it:..."

It seems to be the tail of the "Intro" which is erroneously included in the midst of the Article proper. Another way of explaining the prior sentence is go to the bottom of http://www.pcper.com/category/tags/fx-7600p and see the 'Link Sentence' in it's correct location.

(PS: Feel free to delete this Comment after fixing).

Thanks.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.