Intel proves that it has what it takes when it comes to LTE

Signals Research Group (SRG) recently completed its eighth collaborative effort with Spirent Communications and its sixteenth “Chips and Salsa” report on cellular chipsets. In the most recent collaboration, we brought together LTE baseband chipsets from eight different suppliers (Altair Semiconductor, GCT, Intel, NVIDIA, Qualcomm, Renesas Mobile, Samsung, and Sequans) to determine who has the best performing chipset, based on a series of 32 test scenarios that we derived from industry accepted 3GPP test specifications. SRG facilitated the benchmark study and was responsible for reviewing and analyzing the results. Spirent provided engineering support, and most importantly, the use of its 8100 test system to conduct the automated and highly repeatable tests on each chipset.

The most recent study marked our second benchmark study of LTE chipsets. Previous studies with Spirent have included HSPA+, HSDPA, UMTS call reliability and A-GNSS. To date, we are still recognized as the only independent provider of baseband chipset performance benchmark studies in the industry. And as a testament to our long-standing relationship, the companies that participated in the most recent round are already clamoring for the next round to take place. The companies that came out on top want to prove that they are not a one trick pony and the companies that came out toward the bottom want redemption. The few companies that were not ready to participate in the last study are also ready to enter the competition. There was a reason that we titled the report, “Sweet 16 and never been benchmarked” since some of these companies have been noticeably absent from prior studies due to the uncertain viability of their chipsets.

The results from the most recent round are interesting, to say the least. First, Spirent and SRG were able to bring together numerous pre-commercial and commercial chipsets. I imagine that most people were surprised that Intel actually had a working LTE chipset, let alone find out that it was the best performing chipset (more on this facet in a bit). Additionally, the list included pre-commercial solutions from Sequans, Renesas Mobile and NVIDIA. It would be virtually impossible for any organization to assemble such a line-up!

As I hinted in the title, Intel came out on top – beating the likes of perennial favorite and San Diego native, Qualcomm. To be fair, the results were incredibly close with only a few percentage points separating the two companies, but Intel’s results were better and close only counts in horseshoes and hand grenades. We could add another activity to the list, but this blog is intended to be family friendly. And if you are assuming that Qualcomm came in second place then you might want to rethink your assumption – nothing we wrote in this blog suggests that they did.

In hindsight, Intel’s results should not be all that surprising since it highly leverages the Infineon 3G platform and stellar RF performance that has since evolved to support LTE under the Intel moniker. Infineon, I note, was always a strong performer in our HSPA+/HSDPA chipset studies and it was in the original 3G iPhone until Qualcomm won the slot, in part due to its ability to support the requirements of a certain North American operator whose name rhymes with Horizon Direless. Intel may have lost the ARM war, but you can’t throw the baby out with the bath water.

Separate from the overall results, I once again saw some pretty big performance differences among all of the chipsets, in particular for the more challenging fading scenarios. As a side note, in addition to the more basic static channel conditions, our 32 test scenarios included various simulated fading channels (EVA5, EPA5, ETU70, and ETU300), SNR values, and MIMO correlation factors to create a range of challenging, albeit realistic, scenarios. In many cases the variance between the top-performing and bottom-performing LTE baseband chipset exceeded twenty percentage points. Even for the top-performing LTE baseband chipsets, it was clearly evident in the results that some chipsets did better in some scenarios than in other scenarios.

Now that we’ve set the bar for how chipsets should perform, I expect to witness material improvements in our next round, which we have planned for later this year. Just to keep everyone honest, I plan to change the test scenarios for the next round. In the interim, Spirent and SRG are investigating some additional benchmark studies that we can do together. These studies could include the industry’s first independent over-the-air (OTA) testing of leading platforms in commercial devices (imagine Samsung S III versus Apple iPhone 5) as well as our second round of A-GNSS testing.

If you are interested in the published report, please feel free to visit our website at www.signalsresearch.com where you can download a report preview.

comments powered by Disqus
× Spirent.com uses cookies to enhance and streamline your experience. By continuing to browse our site, you are agreeing to the use of cookies.