AMD Takes A Swing At Intel's SysMark Benchmarks, Misses Completely

It always fun when major corporations decide to publicly slag the other one off. So to help round off your Friday with a smile, let's enjoy AMD having a good old whinge about benchmarks and bias towards Intel.

The video starts out with AMD saying "why the Federal Trade Commission has required Intel to use a disclaimer" on the SYSmark 2014 benchmark. "Choosing the wrong benchmark could result in overpaying for technology or getting a less performing PC," AMD's director of computer client products, John Hampton, starts out by saying.

Hampton and AMD engineering manager Tony Salinas then compare the biased SYSmark benchmark with others they think is fairer, namely the work accelerated test in the PCMark 8 benchmark and a "set of scripts that walked through the office suite [of programs]".

AMD's contention is that the difference between performance in the AMD laptop versus the Intel one shown is only between 6-7%, compared to the 300-plus margin with which the Intel laptop wallops its AMD competitor in SYSmark.

For reference — and you can see this in the video for yourself — the AMD laptop is a FX-8800P with an AMD Radeon R7 GPU, while the Intel laptop is an i5-5200U at 2.2GHz with an HD Graphics 5500 integrated GPU. The Intel laptop also packs 8GM of RAM, according to the visible information in PCMark 8, while the total memory for the AMD laptop is oddly reported as "0MB".

AMD does have a point here, though. Manufacturers cherry-pick things that make their products look the best all the time, although insinuating a link between Intel, SYSmark and Volkswagen's recent scandal (as Hampton did at the start of the video) is a stretch and a half.

But AMD are cherry-picking facts too. If you're going to thoroughly benchmark a system, there's nothing wrong with a tool like SYSmark that AMD says more or less exclusively tests the CPU. The trick is to use a suite of benchmarks to give you a complete picture of the system's capacity, and that includes theoretical tests alongside ones that prioritise real-world applications and programs.

And it's not like AMD's comparison is hugely fair either. PCMark 8 tests the GPU and the CPU, and the FX-8800P is AMD's top of the line APU for laptops. Intel's i5-5200U is based on the older Broadwell micro-architecture, not the newer Skylake architecture that was released later this year. On top of that, the i5-5200U didn't even have the most powerful integrated GPU among other Broadwell Core i5 mobile CPUs, let alone the latest Broadwell versions that were released last June.

Despite all of that, the Intel laptop still beat its AMD counterpart in every single test AMD showed in the video, including two of their fairer benchmarks. It just makes AMD look petty. The video starts with the tone of a current affair-style expose, only for AMD to end up deploying the same cherry-picking tactics they're accusing Intel of.

Their complaints against SYSmark aren't that helpful. Its benchmarks can be a useful yardstick for gamers looking for the highest performance possible, particularly given we're still in a world where many games are still yet to fully take advantage of multiple CPU cores. (Although with DirectX 12 around the corner, that should start to change very rapidly.)

So at the end of the day, AMD hasn't said or shown anything technically inaccurate. But their portrait of real-world performance has just as many caveats as Intel's.


Comments

    Suprise suprise, AMD CPUs are shite. Intel is better. In other news water is wet

    AMD should stick to what they do best. Mid range gaming chips and cards that are good for 90% of people, yet blow intel/Nvidia out of the water on price. Before i had money, it was AMD/ATI for me, because the performance/price ratio was higher, but im green camp now.

    I missed when AMD was better than Intel... but since i7 out... they just never been... in the front foot.

      You miss when AMD was better than intel? THAT EVEN OCCURED?

        Yes that did occur. But intel offered "price discounts" aka to pretty much all computer makers as incentives to use Intel chips which shunned AMD out of the market when AMD had the better product. It's not the first time Intel trys to maintain monopoly. Intel is is a pretty unethical corporation go read the history books

        Yep AMD were way ahead of intel when the first P4 chips shipped. The athalon 2200XP was the first and only AMD cpu I ever had and it mopped the floor with anything intel had at the time. Once intel started to ship the core2duo cpu's thats when AMD lost it and never caught up.

        Plus AMD was the one release 64 bit CPU ahead from Intel.

          That's not quite true. CPUs with HP/Intel's IA64 architecture was out two years beforehand. AMD took cues from IA64 and adapted it to work in a backwards-compatible arrangement with x86, calling it AMD64 (now mostly known as just x64 or x86-64). It was a relatively iterative change, all things considered, but its strength was that it provided emulation-free backwards compatibility with existing software, which made it much easier for the market to take up.

          The first fully 64-bit processor was probably the MIPS R4000 in 1991. Before that there were a lot of partially 64-bit processors.

    I agree that there's nothing wrong with a purely CPU based benchmark and that benchmarks should be chosen to give an overall picture of a system's performance.

    The problem with SYSmark is that it doesn't advertise itself as a CPU benchmark, instead it claims to give an overall picture of system performance from a business users perspective:

    "SYSmark® 2014 ver 1.5 is an application-based benchmark that reflects usage patterns of business users in the areas of office productivity, media creation, and data/financial analysis."

    Based on the benchmark definition I'd say AMD's point is valid.

    Yup.

    http://techreport.com/review/8616/amd-athlon-64-x2-3800-processor
    http://www.legitreviews.com/the-amd-athlon-64-x2-3800-processor-review_226

    Good ol' days of the mighty X2. Retired my 3800+ only 5 years ago when upgrading to PileDriver.

    FX-8800P is equal in performance to i5-5300U, even though most i5-5300U laptops cost more than FX-8800P.

    FX-8800P also has better graphics than any Broadwell part and still better graphics than Skylake. For higher resolutions and detailed graphics, the more expensive core i7 will have faster frame rates as CPU performance becomes the limiting factor.

    Power consumption between Carrizo and Broadwell is nearly the same, even though Carrizo is still on 28nm and Broadwell is on 14nm. This suggests Carrizo on 14nm will not only have better power efficiency than Broadwell, but Skylake too.

    Of the three Carrizo APUs, FX-8800P is the highest performance, but is much lower price than Intel's newest Skylake CPUs, so the fair comparison is still Carrizo versus Broadwell.

    Wow this is petty. I agree with their message, there is no one score that can describe a whole system. But it's not like PCMark is much more realistic. You use 3 benchmarks and 3 applications to measure a PC AFAIC. For a gaming rig use 3DMark, Cinebench and one of your choice. Then follow up with 3 games you intend to play that require the most resources.

    Well to be fair, you could take a five to six year old CPU from either Intel or AMD and not be able to see a performance difference in the real world for 95% of software that one uses day in and day out. At least in a business environment where people are doing something productive and not just playing. Assuming that you are using a desktop and had no concerns about battery life or always have your laptop plugged in.

    You forgot that the Intel-based laptop has 8GM of RAM. That's a pretty odd amount, don't you think?

    Lot of pro-AMD guest accounts came out of the woodwork for this article. Interesting how that happens.

    I still use AMD GPUs but made the change back over to intel a couple years ago the [email protected] just couldn't handle my crossfire set up.

Join the discussion!

Trending Stories Right Now