Fairphone 3 specification speculation & leaks

[AnotherElk]

@Bert69

Guys, it’s time to face the truth

Indeed, we do already face it …

Interesting links / news articles somehow related to Fairphone

Some news on Fairphone 3: “… a listing for the Fairphone 3 has popped up on the Bluetooth SIG website. Aside from confirming the name [of the phone] and Bluetooth 5 capabilities, the listing doesn’t tell us anything else though.” https://www.androidauthority.com/fairphone-3-1014831/

Interesting links / news articles somehow related to Fairphone

FP3 should be announced by the end of August (French link) #interestingarticles https://www.lesmobiles.com/actualite/26766-fairphone-annoncera-fin-aout-un-nouveau-produit.html

Welcome to the community forum, feel free to read more .

[JeroenH]

Some of the specs are leaking here and there. I’m not a fan of (such) leaks but the information is out in the open already.

• There is a render available [3] by @evleaks from Twitter. I found it on Tweakers but the orig. source (on Twitter) is protected.

• Android 9 Pie

• 64 bit SoC

• Base frequency of 1.8 GHz

• 4 GB RAM (3.5 GB in Geekbench is an indication of 32 bit ARM kernel/userland)

• Bluetooth 5 (via Bluetooth SIG [2])

• Benchmarks via Geekbench [1]

[1] https://browser.geekbench.com/v4/cpu/14173908!

[2] https://launchstudio.bluetooth.com/ListingDetails/91763

[3] Render

[RSpliet]

@JeroenH

• 4 GB RAM (3.5 GB in Geekbench is an indication of 32 bit ARM kernel/userland)

I’m reluctant to make this assumption. The Fairphone 2 geekbench score states it has 1875MB of DRAM, even though it’s kitted out with 2GiB. Other possible explanations for this potential difference between physically available DRAM and reported DRAM are carved out DRAM sections for the modem or GPU that are invisible to user-space. I suspect we’d better just wait and see on this one.

[JeroenH]

Speculation!

So we are looking at a relatively mid-range 8-core 1.8 GHz 64-but (AArch) SoC with Bluetooth 5.

I looked around at

List of Qualcomm Snapdragon systems-on-chip

SDM630 4 + 4 cores (2.2 GHz + 1.8 GHz Cortex-A53) Bluetooth 5, NFC, 802.11ac Wi-Fi up to 433 Mbit/s, USB 3.1 from Q2 2017 (Incorrect -> see discussion hereunder -> does not contain Qcom specific chipset)

SDM636 4 + 4 cores (1.8 GHz Kryo 260 Gold – Cortex-A73 derivative + 1.6 GHz Kryo 260 Silver – Cortex-A53 derivative) Bluetooth 5, NFC, 802.11ac Wi-Fi up to 433 Mbit/s, USB 3.1 from Q3 2017.

SDM632 4 + 4 cores (1.8 GHz Kryo 250 Gold – Cortex-A73 derivative + 1.8 GHz Kryo 250 Silver – Cortex-A53 derivative) Bluetooth 5, NFC, 802.11ac Wi-Fi up to 364 Mbit/s, USB 3.0 from Q2 2018

SM6125 (665) 4 + 4 cores (2.0 GHz Kryo 260 Gold – Cortex-A73 derivative + 1.8 GHz Kryo 260 Silver – Cortex-A53 derivative) Bluetooth 5, NFC, 802.11ac Wi-Fi, USB 3.1 from Q2 2019

SM7150-AA (730) SM7150-AB (730G) 2 + 6 cores (2.2 GHz Kryo 470 Gold – Cortex-A76 derivative + 1.8 GHz Kryo 470 Silver – Cortex-A55 derivative) Bluetooth 5.0, NFC, 802.11a/b/g/n/ac/ax 2x2 (MU-MIMO) Wi-Fi up to 867 Mbit/s, GPS, GLONASS, Beidou, Galileo, QZSS, SBAS, USB 3.1 from Q2 2019

SDM845 4 + 4 cores (2.8 GHz Kryo 385 Gold+ 1.8 GHz Kryo 385 Silver) Bluetooth 5; 802.11a/b/g/n/ac/ad Wave 2(MU-MIMO); GPS, GLONASS, Beidou, Galileo, QZSS, SBAS from Q1 2018

SDM850 4 + 4 cores (2.95 GHz Kryo 385 Gold + 1.8 GHz Kryo 385 Silver) Bluetooth 5; 802.11a/b/g/n/ac/ad Wave 2(MU-MIMO); GPS, GLONASS, Beidou, Galileo, QZSS, SBAS from Q3 2018

I don’t think the benchmark is quick enough single core for a high range 800 series, and the older in the 800 series don’t have Bluetooth 5, or don’t have 1.8 GHz CPU.

[EDIT]A device released in Q3 2019 having a SoC from <= 2017 is odd to me. However the following (already mentioned) one fits exactly (AFAICT)

SDM632 4 + 4 cores (1.8 GHz Kryo 250 Gold – Cortex-A73 derivative + 1.8 GHz Kryo 250 Silver – Cortex-A53 derivative) Bluetooth 5, NFC, 802.11ac Wi-Fi up to 364 Mbit/s, USB 3.0 from Q2 2018

I found no other device which fits the scope as good as this one. Oh, btw, fingerprint scanner behind according to render and NFC[/EDIT]

[RSpliet]

The “Implementer 81” on Geekbench rules out the SDM630, as the SoC contains Qualcomm implemented cores, not a standard ARM Cortex. The variant/part/revision information seems to be uninformative.

[d2w]

just saw this: (it’s in german, but should be hard to translate)
https://winfuture.de/news,110507.html

[JeroenH]

Thanks for the heads up on the non-specific Qcom addition. I’ll strike it through.

I searched for “ARM implementer 81 architecture 8 variant 10 part 2048 revision 2” on Geekbench. Then I looked which other device had also Qualcomm Qualcomm 1804 MHz (8 cores).

Which got me this one:

https://browser.geekbench.com/v4/cpu/14195170

It has less RAM (3 GB instead) but the very same SoC information and similar benchmarks.

The specs of this device are known:

Moto G7

(See specifically the Power)

This is also the SD632 I mentioned earlier, with 1.8 GHz.

Differences are that the US version has 3 GB, and it has Bluetooth 4.2. The one with Bluetooth 5.0 has a SD636. But it is 1.8 GHz/1.6 GHz.

[JeroenH]

Interestingly the German source links to a Geekbench search query which contains a whole lot more entries:

https://browser.geekbench.com/v4/cpu/search?utf8=✓&q=fairphone+fp3

Another similar device is this one, the Xiaomi Redmi 7:

https://browser.geekbench.com/v4/cpu/14199056

However it only has 3 GB RAM, and the multicore benchmark is slower (41?? range instead of FP3’s 49?? range).

Another similar one is the Xiaomi Mi A3

https://browser.geekbench.com/v4/cpu/14197441

It has the same architecture string, little bit more RAM (but probably just 4 GB), same GHz, but its benchmarks are slightly better than FP3’s. This one has a Qualcomm Snapdragon 665. So the FP3’s is slower than that one.

Sources: https://en.wikipedia.org/wiki/List_of_Xiaomi_products

[EDIT]Another one with a similar string and performance is this one:

https://browser.geekbench.com/v4/cpu/14197042
https://www.gsmarena.com/honor_8c-9366.php

Looking around at

Kryo

Suggests either Kryo 250 or 260 based on the information above (with the SD632 / Kryo 250 being a perfect match)
[/EDIT]

[TobiasF]

The leaked rendering shows:

• The phone looks (aside from the rounded corners) quite similar to the FP2 and has nearly the same size.
• There is still a “bumper” case which has a rim around the screen.
• The phone is still modular, but the modules are different to the FP2 ones.
• The battery is much larger and is still exchangeable.
• There is a fingerprint sensor on the back.

What we don’t know is whether the rendering really comes from Fairphone or was faked.

The mentioned SoCs do not support 5G. I think we will not see 5G before FP4.

The following questions are still open:

• How large is the internal flash storage?
• Which connectors does the phone have? Does it have Micro-USB or USB-C? Does it have a headphone connector?
• Does the phone come with one or two SIM slots?
• Does it come with a Micro-SD slot?
• Does it come with a FM radio app? Does the reception differ from FP2?

[JeroenH]

The render is by @evleaks, who is a known leaker and has a reputation to keep up. None of the mentioned SoCs support 5G, but I don’t see the fuzz about that. If we assume SD632, it has QC3. All recent Snapdragons have QC3, or even QC4. The FP2 has QC2. Do devices with QC3 plus microUSB exist?

[rainer_zufall]

jfyi - the article d2w posted

@d2w

just saw this: (it’s in german, but should be hard to translate)
https://winfuture.de/news,110507.html

does speak from the assumption of a Snapdragon 660

“However, entries in the Geekbench database and the Bluetooth SIG suggest that the processor used here is the Qualcomm Snapdragon 660 Octacore-SoC with 2.2 gigahertz maximum clock speed, combined with four gigabytes of RAM.” //translated with deepl

[JeroenH]

Yes, I read, and it does not correlate with what I posted above.

1. 4 GB RAM and Bluetooth 5 fits (though as does the SD632)

2. It is not a 1.8 GHz SoC. It is a 1.84/2.2 GHz SoC.

3. The Geekbench entries with that SoC are quicker than the ones with the SD632. The Geekbench entries with the SD632 are far more accurate.

4. Another difference is SD632 supports LPDDR3 while SD660 supports LPDDR4. The memory-related benchmarks were on par with the SD632; not a device with a SD660.

5. The SD660 supports more than 1080p: WQXGA / Quad HD / 2560x1600. That’s a very good screen, like a retina MBP. The disadvantage is its more expensive to replace (which goes against the values of Fairphone). That’s of course just my take on it…

[rainer_zufall]

thanks for your explanation and the explorative investigations you are doing! I am not tech savvy enough to understand every detail - I just wanted the speculative info of the article not to be missed…

it seems the 632 is anyhow better than the 660 (https://versus.com/de/qualcomm-snapdragon-632-vs-qualcomm-snapdragon-660-vs-qualcomm-snapdragon-801-msm8974ab-v3)

[JeroenH]

Thank you! I don’t understand everything about the topic either.

As for the SD632 being faster.

AFAIK LPDDR4 (SD660) is faster than LPDDR3 (SD632).

From this benchmark comparison I would assume the RAM in FP3 isn’t fast:

https://browser.geekbench.com/v4/cpu/compare/14194833?baseline=14173908

The Xiaomi Redmi Note 7 contains a SD660. Here’s a bunch of benchmarks. Ignore the obvious outliers and you see a clear difference with the SoC in the Fairphone 3 benchmarks (1200 range single core, 4800 range multi core).
https://browser.geekbench.com/v4/cpu/search?q=Xiaomi+Redmi+Note+7&utf8=✓

Here’s more benchmarks of the SD660
https://browser.geekbench.com/v4/cpu/search?dir=desc&q=sdm660&sort=created_at
The negative outliers are single core 1300-1400. Most are at the 1500-1600 range. FP3 is consistently in the 1200 range. Not saying its impossible, just more likely it being the SD632.

[AlbertJP]

I think discrepancies in multicore can also be explained by different power settings, i.e. if the max. power is limited to preserve battery, not all cores will run at the highest speed at the same time. At least in laptops it is common nowadays that laptop manufacturers change the max. power consumption of the CPU.

[rainer_zufall]

@JeroenH

As for the SD632 being faster.

I based my “better” soley on the slightly higher overall score of the 632 compared to the 660. I am not sure if “which memory technology is supported” went into this comparison. When looking a bit closer at both, I am not so sure anymore why the overall score is better for the 632 but I simply don’t have the expertise to judge on my own

[rainer_zufall]

Five Four new photos!
( article):

Fairphone 3 taucht auf Pressebildern auf

[urs_lesse]

Seems like modularity might be slightly reduced. I somewhat expected that, maybe contact issues with the FP2 modules prove too hard to solve.

I also like that – at least so far – it seems that Fairphone is sticking to a “one model only” policy. Of course offering a catch-all party phone will have some disappointed, but I think it’s the economically soundest decision.

[JeroenH]

Thanks for the link Rainer!

Too bad we can’t read the source directly.

It is clearly a modular phone (with the iconic and unique translucent case). Also, clearly a physically larger battery. Which fits the other smartphones using similar (suspected) SoC.

What is unknown as the article says: microUSB or USB-C or 3.5 mm jack.

I’d assume USB-C because of the (suspected) SoC and QC.

[EDIT]Wait why does this picture say “charging rapidly”? There is no wire shown https://i1.wp.com/stadt-bremerhaven.de/wp-content/uploads/2019/08/fairphone-3-3.jpeg?w=1024&ssl=1

Also, what is this black spot on the left (top left of phone)? https://i1.wp.com/stadt-bremerhaven.de/wp-content/uploads/2019/08/fairphone-3-2.jpeg
[/EDIT]

[ElKrasso]

On the backsind of the display you can recognize three symbols. Two looking similar, I guess they are symbolising SIM1 and SIM2 and the other is, I guess, a microSD.