Cellular

From a cellular perspective there really are two entirely different Galaxy Nexii, as noted before. The GSM/UMTS (maguro) version which is slightly thinner, and the CDMA/LTE (toro) version which is on sale on Verizon at present and no doubt also Sprint later on when their LTE network is fired up. To accommodate those different combinations of air interfaces, Samsung made two very different phones.

GSM/UMTS Galaxy Nexus

Galaxy Nexus (GSM/UMTS) - Network Support
GSM/EDGE Support 850 / 900 / 1800 / 1900 MHz
WCDMA Support 850 / 900 / 1700 / 1900 / 2100 MHz
HSPA Speeds HSDPA 21.1 (Cat.14) / HSUPA 5.76 (Cat.6)
Baseband Hardware Intel/Infineon X-Gold 626 w/I5712 transciever and RFMD RF6260 PA

Cellular connectivity on the GSM/UMTS Galaxy Nexus is courtesy of an Infineon/Intel X-Gold 626 (XMM6260). This is the same baseband as what shipped in the GT-I9100 SGS2, which isn’t much of a surprise at all. That means HSPA+ with HSDPA 21.1 (64QAM) and HSUPA 5.76. From a cellular perspective, the Galaxy Nexus is actually virtually identical to the GT-I9100 SGS2, and includes the exact same Infineon Smarti UE2 RF transceiver marked 5712, and the RFMD RF6260 quad-band multi mode (GSM/EDGE and WCDMA) power amp which works between 1710 and 1980 MHz, and 824–915 MHz, supplanting the need for individual power amps for each band. Of course, the 2100 MHz band has its own power amp elsewhere. The GSM/UMTS Galaxy Nexus also takes a regular size Mini SIM (not microSIM). In addition, though XMM6260 supports it, the Galaxy Nexus doesn’t include Rx diversity, and instead just locates its single cellular antenna at the very bottom of the device in one module, and the WLAN antenna off to the side.


Intel X-Gold 626 circled in black above - Image courtesy iFixit

When the Nexus One (and later Nexus S) originally came out, I was puzzled by why Google hadn’t asked its OEM to include pentaband WCDMA support. Back in the Nexus One days, Google was making its big push to try and reshape the way smartphone shoppers buy phones in the US (and make things more like how the rest of the world does it). At the same time however, the hardware launched effectively only with the WCDMA band support for T-Mobile in the US. Third time is a charm, and the GSM/UMTS Galaxy Nexus is now both the first Nexus phone with pentaband WCDMA support, and, moreover, the only Android phone with pentaband WCDMA. The result is a new level of freedom - fully carrier-unlocked and hardware-unlocked hardware that is finally completely carrier agnostic (at least in GSM/UMTS land).

We tested the GSM/UMTS Galaxy Nexus on T-Mobile’s HSPA+ network and burned through the 2 GB of data on Google’s SIM before burning through another couple of GBs on our own T-Mobile SIM. All in all we ran over 200 tests and generated the histograms that normally grace our smartphone reviews.

T-Mobile HSPA+

Downstream Stats (Mbps) Avg: 3.522; Max: 8.689; Min: 0.029, StDev: 2.022

Upstream Stats (Mbps) Avg: 1.192; Max: 3.259; Min: 0.025, StDev: 0.814

Latency Stats (ms) Avg: 196.93; Max: 500; Min: 105, StDev: 57.122

Number of Tests Run: 244

Downstream Throughput Upstream Throughput Latency

Downstream ECDF Upstream ECDF Latency ECDF

T-Mobile can be fast when it wants to, hitting 8 Mbps down at one point, and 3.2 Mbps up at another. Obviously the Galaxy Nexus is where it should be considering its HSPA+ category. I also tested the GSM/UMTS Galaxy Nexus on AT&T but didn’t run enough tests to generate the most meaningful histograms. I couldn’t get the Galaxy Nexus to go quite as fast as I’d expect it to on AT&T HSPA+ for some reason, even using the “phone” APN.

AT&T HSPA+

Downstream Stats (Mbps) Avg: 2.114; Max: 4.396; Min: 0.083, StDev: 0.979

Upstream Stats (Mbps) Avg: 1.201; Max: 1.681; Min: 0.043, StDev: 0.449

Latency Stats (ms) Avg: 280.479; Max: 951; Min: 107, StDev: 281.045

Number of Tests Run: 71

Oh, and needless to say, the Galaxy Nexus doesn’t suffer from any deathgrip or related signal propagation issues that I can identify.

CDMA/LTE Galaxy Nexus

When I caught wind of the fact that Verizon had also been selected as a launch carrier for the Galaxy Nexus, I speculated that Samsung would reuse the same cellular architecture it used in the Droid Charge. In that review, I talked about how the device used a combination of Samsung CMC220 for LTE, and VIA Telecom’s CBP 7.1 platform for CDMA2000 1x/EVDO. It’s a rather unique combination that also shipped in the Samsung Stratosphere.

Galaxy Nexus (CDMA/LTE) - Network Support
CDMA2000 1x/EVDO Support EVDO Rev.A (800 / 1900 MHz)
CDMA Baseband Hardware VIA Telecom CBP7.1 with FCI FC7780
LTE-FDD Support UE Category 3 - 700 MHz Band 13
LTE Baseband Hardware Samsung CMC221 with FCI FC7851

It makes sense for Samsung to reuse what work it has invested in making both Android’s Radio Interface Layer (RIL) and hardware work together for the Galaxy Nexus, and this turned out to be a pretty good educated guess. It turned out mostly correct - the CDMA/LTE Galaxy Nexus is based on a combination of Samsung CMC221 and VIA Telecom CBP 7.1.

 
Samsung CMC221 and VIA Telecom CBP7.1 - Courtesy ZDnet/TechRepublic

So what’s different between CMC220 and CMC221? No doubt, CMC221 simply is a revision of CMC220 which improves connection stability and handover performance, as one of my few complaints with CMC220 were the prevalence of paused connections in some environments.

LTE UE Categories

Beyond that, it’s hard to know exactly what’s changed, but suffice it to say CMC221 shares the properties with CMC220 that we care most about. It’s a UE Category 3 part - which is the current maximum for devices right now - it’s no doubt 3GPP Rel.8, and it supports 2x2 MIMO on the downlink. The CDMA/LTE Galaxy Nexus right now only operates on Verizon’s 10 MHz FDD LTE network, though Sprint is slated to get a CDMA/LTE Galaxy Nexus of their very own that will undoubtably be tailored to the 1900 MHz PCS block it deploys in. For now however, we’re just talking about the variant on Verizon 4G LTE.

I ran a bunch of tests on the CDMA/LTE Galaxy Nexus on Verizon 4G LTE and 3G EVDO networks. I tested 4G LTE in my Tucson, AZ market and honestly found using a UE Category 3 device quite refreshing after a long wave of UE Category 2 Motorola Wrigley. There is absolutely a difference in the kind of speeds you can see between devices, and I definitely consider the LTE Galaxy Nexus’s performance above average. I saw a maximum of 58 Mbps on the LTE Galaxy Nexus, though I didn't grab a screenshot. I did of one at 56 Mbps down, however. 

Verizon Wireless 4G LTE

Downstream Stats (Mbps) Avg: 19.715; Max: 58.294; Min: 0.089, StDev: 11.744

Upstream Stats (Mbps) Avg: 7.437; Max: 19.711; Min: 0.242, StDev: 3.699

Latency Stats (ms) Avg: 87.312; Max: 188; Min: 63, StDev: 21.285

Number of Tests Run: 342

Downstream LTE Upstream LTE Latency LTE

ECDF Downstream ECDF Upstream ECDF Latency

Again, I consider the 4G LTE performance of the Galaxy Nexus quite good, and I saw speeds in my market that I haven’t been able to achieve with other devices. The 58 Mbps test is pretty good, although I know a few other people I talk to have been able to hit over 71 Mbps on the Galaxy Nexus, and likewise consider its RF performance above average on LTE.

We also talked about the perceived LTE signal issue with the Galaxy Nexus at some length. Without revisiting the whole thing, the issue is with the way Android 4.0.2 paints signal bars as a function of signal strength. The Galaxy Nexus as of 4.0.2 does this in a way that isn’t in line with the other Verizon 4G LTE devices, and coupled with the fact that getting to LTE signal strength (and not 1x or EVDO RSSI) is difficult, misled many into thinking there was an issue when there really wasn’t one. I’ve continued comparing LTE RSRP (Reference Signal Received Power) between the Galaxy Nexus and other phones that report it, and have yet to see a significant difference. In addition, other LTE devices reported RSSI instead of RSRP - again it’s good that people are using numerics instead of bars (which are not standardized), however caution must be taken to compare the same parameters properly.

Of course, that’s really not all there is to talk about - what about 1x and EVDO performance on the VIA CBP7.1? My initial investigation only looked at 4G LTE, and since then I’ve had countless people email or otherwise notify me that EVDO also needs a good looking at.

I believe I’ve nailed down why there’s a perception that something is wrong with the Galaxy Nexus’ EVDO connectivity. Interestingly enough, the reason is again a nuance in the way that the device reports signal, but on EVDO. In particular, I noticed something in common with the Droid Charge that is no doubt unique to CBP 7.1. Namely, the baseband reports signal level in a quantized fashion at a few different levels in dBm. Ordinarily I see signal level quantized to just integer values, however on the CBP 7.1 based devices I’ve seen (like the Droid Charge and CDMA Galaxy Nexus) that quantization takes it to a few different levels. That means the values below are the only ones shown when connected to EVDO. You could be in an area with -51 dBm EVDO RSRP, but the maximum you’ll ever see reported is -75. There’s some hysteresis as well and the device doesn’t switch between levels that often.

-75
-83
-93
-100
-113
-120 (0 signal)

I have a sneaking suspicion that these values originally were designed to correspond to some decent cutoffs for signal bars (eg a 5 bar system with -120 being 0 bars), though I’m not entirely certain. The result of this quantization again is that you can be misled into thinking signal is worse or better than it really is depending on conditions, and I wager that this is what leads many to report that things are different than shown on other devices. I also compared EVDO signal strength between the CDMA/LTE Galaxy Nexus and a Droid Bionic, and quantization issues notwithstanding I saw very similar measures.

The caveat is that no doubt CBP 7.1 + FC7780 has different performance than the MDM6600 that has been ubiquitous for CDMA connectivity, so the fact that phones might perform differently in the same location doesn’t surprise me. That said, I’ve yet to see anything dramatic at all.

Nevertheless, I ran a ton of tests on EVDO (eHRPD is EVDO anchored through the PDN-Gateway) to test performance.

Verizon Wireless 3G EVDO Rev.A

Downstream Stats (Mbps) Avg: 0.891; Max: 2.525; Min: 0.07, StDev: 0.560

Upstream Stats (Mbps) Avg: 0.851; Max: 1.534; Min: 0.149, StDev: 0.254

Latency Stats (ms) Avg: 145.036; Max: 800; Min: 103, StDev: 97.1341

Number of Tests Run: 139

Downstream EVDO Upstream EVDO Latency EVDO EVDO ECDF Upstream ECDF Latency ECDF

Honestly I can find nothing wrong with EVDO performance on the Galaxy Nexus. In fact, if you go back to the Droid 3 review, you’ll see that the Galaxy Nexus numbers are slightly higher in the upstream department and pretty similar (though differently grouped) on downstream.

Call Quality

I don’t usually include call quality measures in this section, but it also bears talking about since we’re dealing with some reports of things varying from what they’ve seen before on the CDMA/LTE Galaxy Nexus. Call quality on the GSM/UMTS Galaxy Nexus on T-Mobile is pretty par for T-Mobile 3G voice (AMR).

GSM/UMTS Samsung Galaxy Nexus on T-Mobile by AnandTech

CDMA voice on the Galaxy Nexus does sound a bit different from other devices, however. Compare to the RAZR below which is again based on MDM6600 and its own vocoder.

CDMA/LTE Samsung Galaxy Nexus on Verizon by AnandTech
Motorola DROID RAZR on Verizon by AnandTech

It isn’t a huge difference, but enough to account for why many people have reported that voice sounds different on the CDMA/LTE Galaxy Nexus. Again, there are bound to be subtle differences between implementations, and I wager Qualcomm’s CDMA market domination in the US accounts for why the difference is something so many have talked about. It’s interesting to me that none of these nuances attracted as much attention with the Droid Charge, but then again I doubt as many people made a switch at that point.

That said, I’ve been using the CDMA/LTE Galaxy Nexus for almost all my calls, and did encounter one circumstance where voice continually cut in and out inexplicably. I haven’t experienced that behavior since.

When it comes to ambient noise suppression, both the CDMA/LTE Galaxy Nexus and GSM/UMTS Galaxy Nexus have a second microphone for common mode noise rejection. The microphone is right near where the battery cover makes contact with the top, slightly offset to the right. It’s a small port but easy enough to pick out.

CDMA-LTE Verizon Wireless Galaxy Nexus Noise Cancellation by AnandTech
GSM-UMTS T-Mobile Galaxy Nexus Noise Cancellation by AnandTech

I don’t know what solution is at play inside the Galaxy Nexus, but it honestly does a mediocre job suppressing noise, we’ve seen other phones definitely do better.

Display - 720p Super AMOLED HD WiFi, GPS, Speakerphone, Audio Quality
POST A COMMENT

185 Comments

View All Comments

  • zorxd - Friday, January 20, 2012 - link

    They can have some differences (cache size, memory bandwidth, neon instructions) but the A9 is not an ISA. ARMv7 is.

    Given that it has the same configuration, an Apple A5 behave the same as a TI OMAP4 or a Samsung Exynos of the same clock speed. I beleive nVidia tegra2 lacks the neon instructions so can be slower in some cases. There is an article on Anandtech about this.

    Given that the iPhone 4S is only 800 MHz it is the slowest A9 CPU by far.
    Reply
  • pSupaNova - Friday, January 20, 2012 - link

    The GPU's on the IPhone uses Tiling so in most GPU rendering tasks it will be a lot faster, However spit lots of Triangles at it and then see how fast it really it is. Reply
  • StormyParis - Wednesday, January 18, 2012 - link

    It's not all about performance, at least if you don't do FPS games. The screen on the Nexus is much bigger than on the 4S for example. For me, it's not about performance at all. I went for the GN for its even bigger screen, and that criteria alone was 95% of my decision, the remain 5% being "... and the rest don't suck", and "has xda-dev support'. Reply
  • humancyborg - Wednesday, January 18, 2012 - link

    Once you start accelerating the entire interface, performance becomes much more significant than just FPS games. There's a reason Apple uses such a gigantic and powerful GPU in their devices, and it's definitely not only for FPS gamers.

    Agree with you on the rest, there are other good reasons to buy this phone, just a shame that they skimped here. I have the 4S, GN and Lumia 800 currently and constantly switch around between them.
    Reply
  • metafor - Wednesday, January 18, 2012 - link

    It doesn't really take a whole lot of resources to render a 2D interface. Just about any ol' GPU with OpenGL ES 2.0 support will do it.

    About the only thing where the GPU is the limiting factor is rendering 3D games. And even then, most if not the vast majority of games on the market will continue to be written for this level of hardware for at least the coming year.

    Honestly, people take benchmarks way too seriously.
    Reply
  • doobydoo - Thursday, January 19, 2012 - link

    Actually, you're absolutely wrong.

    In fact, the GPU slowness is cited in this very article for causing slowdowns in situations where no 3D gaming is being done.

    Remember, the operating system as a whole is hardware accelerated, so every thing you do - animations, transitions, task switching, etc are carried out by the GPU. With the higher screen, the speed of the GPU becomes even more relevant.

    The combination of a high resolution screen and a low powered GPU is a bad combination and materially affects the performance of everything you do on the phone.
    Reply
  • zorxd - Thursday, January 19, 2012 - link

    Do you remember the iPhone 4? Who complained that the GPU was slow? It was much slower than the SGX540 in the Galaxy S. Reply
  • metafor - Thursday, January 19, 2012 - link

    Speculation in an article isn't exactly proof of concept.

    Alpha blending, panning, compositing are very light tasks for a GPU pipeline; it's only a problem when a GPU is TMU-limited. And if it's TMU-limited, it would be obvious all the time.

    I don't think you quite grasp exactly what parts of UI rendering are handled -- or could be -- by the GPU and just how trivial it is compared to rendering a 3D game.
    Reply
  • trob6969 - Wednesday, January 18, 2012 - link

    What i don't understand is why would samsung give the gn 1gig of ddr2 ram then give it an inferior GPU? But to be fair, Apple is no better. Why give iphone 4s a powerful GPU then give it only 512 mb of ram?! My old-ass og moto droid from over 2yrs. ago had that much! Reply
  • doobydoo - Thursday, January 19, 2012 - link

    As alluded to by numerous posters, including one in this comments section, iOS handles memory usage more efficiently than Android so it doesn't suffer any performance penalty as a result of having less RAM. Reply

Log in

Don't have an account? Sign up now