Qualcomm has Snapdragon 8cx beating Core i5-8250U in PCMark 10

[DrMrLordX]

No WiFi built in?

That would be insane, no cell coverage is perfect, even in towns and cities buildings can interfere with the signal - I used to live on a street 10-15 minutes from the city centre of Swansea, and it had terrible reception for 4G.

Welllllll

https://www.anandtech.com/show/1442...5g-laptop-a-lenovo-with-qualcomms-8cx-and-x55

I think the x55 actually supports 802.11ac so I might be a bit strident here. The underlying problem is that the only way to buy one of these in the US may be from a company like Verizon that will insist on selling you one with a data plan. Whether or not 802.11ac will be enabled or not in firmware is anyone's guess. I HOPE that it is. It would not surprise me if Verizon would want it disabled.

 

[AkulaMD]

I'm still really looking forward to Rockchip's A76 SoC. All those 8cx machines look like "5G always-on" machines that don't even have wifi built-in. What a crock.

Yeah I'm looking forward to the RK3588. It should be much cheaper too.

As for the 8cx notebook. It would be a suicide if it really comes with no Wi-Fi.

 

[Tup3x]

I'm still really looking forward to Rockchip's A76 SoC. All those 8cx machines look like "5G always-on" machines that don't even have wifi built-in. What a crock.

There's no way that these wouldn't have WiFi. 8cx has supports the following
Wi-Fi standards: 802.11ad, 802.11ac Wave 2, 802.11a/b/g, 802.11n
Wi-Fi Spectral Bands: 2.4 GHz, 5 GHz, 60 GHz
MIMO Configuration: 2x2 (2-stream)

 

[DrMrLordX]

@AkulaMD , @Tup3x

See my above post. Qualcomm is relying on the x55 for everything, and it does support 802.11ac. As you stated. So it certainly has the hardware for it. But good luck buying one of these without a 5G plan.

 

[soresu]

Yeah I'm looking forward to the RK3588. It should be much cheaper too.

As for the 8cx notebook. It would be a suicide if it really comes with no Wi-Fi.

Yeah, historically speaking RK's stuff ends up in alot of sub £100 products (android tv boxes) as well as stuff like Chromebooks and SBC's.

With any luck we might even get another Samsung ARM Chromebook based on it, but I wouldn't bet on it.

 

[Thala]

@AkulaMD , @Tup3x

See my above post. Qualcomm is relying on the x55 for everything, and it does support 802.11ac. As you stated. So it certainly has the hardware for it. But good luck buying one of these without a 5G plan.

The Samsung Galaxy Book S does have an 8cx but no x55 modem.

 

[Lodix]

Qualcomm's SoC usually have all the connectivity integrated inside them.

 

[Tup3x]

But good luck buying one of these without a 5G plan.

That's going to be a non-issue here in Finland at least. Also it's up to laptop makers to choose whether or not they want to use the 5G chip.

 

[Tup3x]

The Samsung Galaxy Book S does have an 8cx but no x55 modem.

It has LTE connectivity (from 8cx) and obviously WiFi too. It's going to be sold here in Finland like any other laptop.

 

[DrMrLordX]

The Samsung Galaxy Book S does have an 8cx but no x55 modem.

Hmm okay. It does claim, from the product page, though:

• Wi-Fi 802.11 a/b/g/n/ac (2.4/5GHz), VHT80 MU-MIMO

*May differ by market and mobile operator.

It's that asterisk that has me concerned.

@Tup3x

Looks like there may be variances based on the market.

 

[soresu]

Hmm okay. It does claim, from the product page, though:



It's that asterisk that has me concerned.

@Tup3x

Looks like there may be variances based on the market.

I think that there may be areas where certain wifi spectrum is not quite as approved by local governments as it is in the west, every time a new cellular modem generation comes out it comes up, but seldom if ever about WiFi, makes me wonder just how contested it is.

 

[Tup3x]

Yeah, it definitely refers to different WiFi channels.

 

[dark zero]

Yeah I'm looking forward to the RK3588. It should be much cheaper too.

As for the 8cx notebook. It would be a suicide if it really comes with no Wi-Fi.

Sadly the GPU is Mali G52 and not G77 in order to be decent...

 

[soresu]

Sadly the GPU is Mali G52 and not G77 in order to be decent...

I'm pretty sure that G52 spec is for RK3530.

The press release for RK3588 mentioned both chips, but was very ambiguous about RK3588 GPU.

 

[soresu]

You must comment on your post. No pics or links only, without comment


esquared
Anandtech Forum Director

 

[Tup3x]

The custom chip (SQ1) on Surface Pro X is rather interesting. 7W TDP and 2 teraflops of GPU power which is far from weak compared to full below AMD APU even.

 

[Thala]

The custom chip (SQ1) on Surface Pro X is rather interesting. 7W TDP and 2 teraflops of GPU power which is far from weak compared to full below AMD APU even.

Thats interesting, as Qualcomm already put a larger GPU in the 8cx compared to the Snapdragon 855. Apparently Microsoft wanted to get even something bigger. (its called Adreno 685 vs Adreno 680 (8cx)). And yes 2 teraflops are very impressive in a 7W TDP device.

 

[AkulaMD]

The custom chip (SQ1) on Surface Pro X is rather interesting. 7W TDP and 2 teraflops of GPU power which is far from weak compared to full below AMD APU even.

Seems a bit better than the 8cx. Hope this SQ1 will be available on notebooks with more conventional form factor.

 

[DrMrLordX]

I finally got my Snapdragon 855+ running old Java code. Doing so was not necessarily easy, but it works, and it's . . . faster than I thought it was. I'm using the Liberica JDK/JRE for Java 13.0.1 (not touching Oracle's binaries).

Not too long ago, I wrote a Java implementation of Dr. Cutress' 3DPM benchmark. My version only features the trig model. You can find it in this thread here:

3DPM: Can we fix it? Yes we can!

Cue up Bob the Builder intro music. Some of you may remember this thread in which Ian Cutress showed some code which is allegedly representative of what is in his 3DPM benchmark. Subsequent posts by bronxzv and JoeRambo go into some detail as to why the code is perhaps less-than-great. We...

forums.anandtech.com

Anyway this 855+ in "x-mode" (which means it's staying at 2.96 GHz for as long as my phone's cooling system permits) maxes out at somewhere around 190-198M movements/second. Which is faster than my old A10-7700k @ 4.5 GHz. It seems like running the benchmark for more than 1 minute consecutively finally overloads the ROG Phone II's cooling system (and external fan) and makes it throttle a bit. That lowers performance to ~180M movements/second.

In contrast, my "stock" 3900x turns in a time of over 1100M movements/second. My old A10-7700k @ 4.5 GHz managed 141M movements/second. Consider yourself pwned, Kaveri.

 

[DrMrLordX]

@Thala

Are you able to run a version of the JRE on your SQ1 machine?

 

[Thala]

@Thala

Are you able to run a version of the JRE on your SQ1 machine?

Indeed i did test it. I am having issues with getting the clock frequency up when running the benchmark. So as workaround i am running 7-zip benchmark on one core to get the clock frequency up and then the JRE in parallel under WSL2. Getting scores between 197M and 216M steps per second. I assume that is still with 7-zip occupying at least one core partially.

ps. With little further investigation it looks like as soon as i start JRE under WSL2 it moves 7-zip to the small cores and when JRE stops 7-zip goes back to the big cores. No throttling observable and machine stays cool - assume with just the 8 cores it is impossible for the machine running into the 7W power limit. Extrapolating what if i had all cores would give results around 230M steps per second.

 

[DrMrLordX]

@Thala

Interesting! I need to see if I can get the core clocks higher on my end and get scores up there. Regardless, SQ1 is performing quite well. If I could stop being so lazy, I could use the code pasted by Dr. Cutress in the old 3DPMRedux thread to add the other 5 testing algorithms. Then we'd have a somewhat-faithful multiplatform version of 3DPM v2

 

[moinmoin]

If I could stop being so lazy, I could use the code pasted by Dr. Cutress in the old 3DPMRedux thread to add the other 5 testing algorithms. Then we'd have a somewhat-faithful multiplatform version of 3DPM v2

Shouldn't @Andrei. be interested in this?

 

[DrMrLordX]

@moinmoin

One would hope so. Ultimately, it would be better to get a native-compiled version of 3DPM v2.1 with NEON support. I think? I'm not sure if the x86 version of 3DPM v2.1 that's available for download to the public has AVX2/AVX512 support (I'm guessing not), but I know AT has used that version in CPU reviews recently, such as the 10980XE review:

It’s a Cascade of 14nm CPUs: AnandTech’s Intel Core i9-10980XE Review

www.anandtech.com

In the meantime, it's pretty easy to compile Java software on an x86 machine and run it on an ARM machine, assuming the OS allows a full JRE (which was a bit of a trick on Android, but it is possible).

 

[Shivansps]

I think It Is a matter of price than anything else.

For example im looking for an arm board with better gpu perf than the rpi 4. They are just too expensive. To the point that im better off getting a 2200g combo.