Mid-2015 MBP i7-4980HQ, Jetstream scores: chrome: 176, safari: 259, FF: 198. All tests done in browser with single tab open, rest of system idle. So it's more of a browser than hardware test.. it'd seem.
Then again, my expectations for Atom are low to begin with and a score of 23 would about match them. Doesn't mean that it couldn't go significantly higher if optimised. But that's a pipedream, since none will invest the effort to optimise this stuff anyways. It's doomed.
If anything, it's Intel's Core line that has more potential to run out of steam (and clearly has, as we can see with the modest performance increases over the past few generations). Atom on the other hand... there's plenty of room to make it wider.
I highly doubt the bad test result really has much to do with the cpu performance. Airmont might not be the fastest cpu core in the world, but there's no doubt it should beat Cortex-A53 cores (even if those are clocked slightly higher). It is still a "modern" out-of-order cpu design (versus the more simple in-order design of the A53). Granted it's a relatively simple and narrow OoO design, which probably should lose to the big Cortex-A cores (A72/73/A75), but it's not THAT bad. OTOH having 8 of these cores in a phone surely seems like a marketing gimmick to me, "cause everybody has 8 cores!" (intel never produced cpus with more than 4 of these cores for desktop/mobile use, albeit some server chips had more).
Strange to see an Android phone with an old 28nm chip outperforming an Atom Windows tablet with a 14nm chip. The Snapdragon 650 also shows much higher results on individual benchmarks like bigfib.cpp, gcc-loops.cpp and regexp-2010.
The SD650's A72 cores also didn't kick in during the test so I assume it was maxing out the A53s only. The 650 would get a much higher result of the high power cores had been running.
The Atom is not true test of Intel's x86 cores - I put it on performance levels of older core 2 laptops from 10 years ago or so. It's technology has lead way into some higher performance chips like the Core M and also the forthcoming 9805g in the Dell XPS 15 2in1. It also include 16 core and possible more Server based chips aim with low power in mind.
Also just including Jetstream benchmark is not to realistic comparison. This test should not be used as ARM vs x86 comparison.
Actually it's yet another good example why x86 is not suitable for anything low power. Atom has always been too slow or too power hungry compared to Arm cores.
I fear you're right, but it should be noted they're comparing an old architecture from 2013 against more modern ARM cores. A more fair comparison would be Goldmont or Goldmont+ (couldn't they find another mont?).
This was the only one that was reasonably power efficient - if this had been released 4 years ago it may not have looked so slow. Goldmont(+) is faster indeed but uses a lot more power - there was a good reason the mobile version was cancelled.
That's no secret that Intel's Atom chips are not good. They are designed to run Android, but they are NOT best in it, too. Qualcomm are much better, even its mid tier. Totally, Android is optimized for ARM, too. Intel is best in laptop and desktop, not in mobile. But they have some advantages like mobility quality of Intel is better Perhaps. On paper things are different, though. ARM SOC is so good on paper, but, in action, I see them being normally good and CAN do the job, just like the atom that is not very optimized thing. I don't mind the benchmarks. I may see, but they are okay as long as it is not called a performance bug which makes problem in action. I also want better battery life rather better performance only....
The real problem for Android devices (and Chromebooks for that matter) is the lack of optimization for x86. Atom CPU's are totally usable on Windows devices where the kernel is built specifically with Intel x86 in mind, not a recompilation retrofit lacking all sorts of optimizations like the Android kernel.
Android and Chromebook devices use the Linux kernel, which has always been developed with x86 support and optimization as the top priority. There might be some Atom-specific platform features that differ from PC architecture where Intel hasn't contributed drivers that are as good as the ARM counterparts, but its more likely that the Atom hardware is simply inferior to a good ARM SoC in those respects.
I'm not sure why Intel couldn't use its i5-7Y54 as an soc chip. Not sure power consumption but it lists TDP of 4.5. It looks like its benchmarking scores are competitive with the newest gen of SOC. Would assume they just don't want to sell the chip for $60 like all the other companies do.
They could make an SOC using that core if they wanted, but the existing chips wouldn't be viable in a phone platform from an idle power level standpoint because of things like the PCIe bus.
Intel stopped making Atoms with 2W TDP when it killed Willow Trail. A chip with higher TDP doesn't make sense as an SoC because it draws too much power on a tablet or smartphone. Bay Trail and Cherry Trail were amazing chips because they could meet TDP and SDP limits without throttling even at full load; a Core M is faster but it throttles hard once it hits redline.
"x86" is a such mixed bag. On the downside, the x86 instruction set is just awful to decode on the frontend. On the the upside (and when Intel cares), they have historically overcome the limitations of x86.
If you have really branchy code, or something that benefits from AVX/AVX512 (but not GPUs), then Intel is really the only game in town.
Why didn't Microsoft and Intel just make a Phone that ran full x86 Windows? It would have been a killer app. Instead they both wasted billions of dollars with x86 Android phones and ARM Windows phones. So stupid.
Nobody wants full Windows on a phone. Note here that "nobody" doesn't mean "absolutely no people", as there are some people (such as you) who believe that they want it (although if they actually had it they probably would quickly realize that they didn't actually want it). But the number of people who would trade the secure, well curated walled garden of Android/iOS for Microsoft Windows is vanishingly small. I'm so glad that my mom got a Samsung Galaxy instead of a phone running full Windows -- no way I'd want to be on the hook for fixing her phone every time it borks itself and cannot boot properly, or more likely, gets a virus, which would definitely be the experience with a full Windows stack on a phone, just as it is for the full Windows stack on any device.
Microsoft had no hope of getting their OS onto a phone, it was a big waste of money and everyone knew that going in. Intel would have done better to license the ARM instruction set and make a better implementation than everyone else. But it's a business that probably doesn't have high enough profit margins to entice Intel.
Clarification: I do know that Microsoft got their OS onto some phones, but what I meant was that "Microsoft had no hope of getting their OS onto a *successful* phone".
I'd love a Snapdragon 845 phone running full Windows, with CShell and WoA. However, the default settings should restrict software installs to Store-obtained apps, and prohibit Win32 programs from running except when docked (admittedly mostly for UI reasons on the second option). Both of those could be overridden by an advanced user, of course. I'd probably put these options on a toggle somewhere so I could lock it down or open it up whenever I wanted.
But basically there's no problem running "full" Windows as long as the average user is prevented from doing stupid things. Heck even wide open x86 Win10 installs are better in that regard than previous versions. I haven't had to fix either of my parents Win10 machines. Most of what they use is either web-based or available as an app. So they don't need to go hunting on the web trying to download something (and often running into malware in the process). With Win7 and prior, seemed like every couple months I was called in to look at something that was borked. YMMV but that's been my experience.
Oh and if PWAs take off (backed by Google and even Apple as well as MS) then they won't have a problem getting all the must-have apps for a theoretical mobile-capable Windows build. This will be a while down the road though, most existing apps aren't going to be PWA overnight, and demanding apps will need to stay native for the forseeable future.
What a naive viewpoint, you really think android is significantly more secure than windows? I know windows is patched quite often because I hear my co workers bitch every time there computers take a bit longer to shutdown. On the other hand, most android phones are lucky to even get major OS updates in a timely manor. The krack exploit is still probably in the wild on an alarming high number of android devices. Microsoft had that patched time day it was announced to the public.
Microsoft's issue was they were always on the outside of phone popularity. They never brought anything to game that the other players weren't already doing. The issues was not lack of care or technology but poor strategy and marketing.
Yes, I think that Android is significantly more secure than Windows. I wouldn't trust Microsoft to secure anything. They have proven incompetent at it too many times in the past.
krack is not much of a security hole. I mean yeah you can, if you try hard enough, maybe you can read some network traffic. Nobody's going to bother to try hard enough though because you have to have physical presence, and if you're really trying to be a bad guy then you'd just mug your target instead of hoping to gather something of value from possibly sniffing some of their network packets.
It amazes me how people are incapable of assigning severity to security issues. So many people believe every security issue is just as bad as every other one. The windows vulnerabilities that I have personally experienced are all 100x worse than the android krack vulnerability.
Well, for starters, what could be built with early to late 2000s technology was a huge limiting factor. Back then, phones were slow and RAM poor, and that meant that running a "desktop-class" operating system was out of the question (x86 or otherwise). The fact that the iPhone worked at all in 2007 is because Apple put "OSX" on a gigantic diet just to make it fit into the resources available.
LOL could you imagine the quantity of warranty claims providers would get? Literally every single computer issue that has ever existed would turn "mission critical" because their phone is down.
Atom really is at the end of the line. Intel dropped the 2W-4W chips for tablets and concentrated on Atom for cheap desktops which also didn't take off. I would love to see an 8-core Atom with the latest cut-down HD Graphics engine for $200-300 Windows tablets but that's a pipe dream now. The x5-Z38xx chips are decent, on par with a mid level C2D, but they're outclassed by the latest Qualcomm, Samsung and HiSilicon flagship chips. If only Windows Subsystems for Linux ran on ARM...
Just for perspective, here is another benchmark on a Silvermont (22nm variant of the Airmont tested here), and a Cortex-A53 (results are times, so lower is better):
Even factoring in the differences in clock rate compared to the phones in this article, I would expect the Airmont to outperform the Cortex A53 by a factor of 1.26 on this benchmark on these phones. So the JetStream benchmark seems to favor the A53 quite a bit more than the LaTeX benchmark.
Either way a 3-way OoO core being just 1.26 times faster per clock than a 2-way in-order core is extremely poor - it means Cortex-A55 should be about equal...
Old S650 scores 44.847 mid delta in with Chrome. Saying this just to remind you how how all you need are two more power OoO core's for the Web & in generally 93% of the times as a refract how really bad engendered are all the new SoC's recently.
I'm using your cpu and schedule scripts on Lineage and I'm not seeing the A72 cores kick in during the test. I don't know if it's because of the config script or if it's an issue with Chrome Beta.
Used regular chrome. In my script HPM is set to flag similar tasks as high performance (within 240ms) and skip CPU scheduler (interactive) completely pushing maximum performance almost instantly. On S652 it even gets 5~6%+ performance as the hotplug (core_ctl) is set so that core 04 & 06 are always idling one's so all L2 is utilised & with exclusive access. So I don't really know what's the problem in your case. Is hotplug working?
I reflashed the ROM and your CPU script, re-enabled hotplugging and now I get 40 on JStream. The 2x A72 cores run at full speed, A53 cores 0 and 2 run at full speed and the rest are idle. It's supposed to be that way I guess?
Anyway Intel missed the boat on heterogenous computing. The A53s save a lot of power, the A72s pack a big punch only when needed. An Atom with 4x Goldmont and 2x Core M cores would be a screamer with excellent battery life, if only those two architectures could share caches.
Well for this benchmark it is supposed to be like that as browser's (among most other things) are only SMP 2. Yes I used same typology on small core's too.
If Intel made SoC like that it would be probably priced 300+$ so no one would buy it. Also they would had to do a big R&D on heterogeneous before so it's a lost battle anyway. Who knows me by in the future we see Falkor derivat's (what ever dragon it would be called) on mainstream mobile SoC's. It's more probable so as Samsung & Apple as prime competitors are pushing it & QC will have to respond. For me ideal SoC would be an 2x A75 & 4x A55 with big juicy GPU like A630 on 22nm FD-SOI, naturally without idiotic L4 cache found on on S845. Described as it is it would cost approximately half the price of S845 & achieved 70% performance (both CPU and GPU) in real usage so it would be a real improvement & still with lower DTP. I would pay a full price for it as for S845 if it would be coupled with 4GB HMB 2 (which is about the same price as 8BG of LPXDDR 4 and 20$ extra from the SoC for interconnect) this would make it on pair in CPU side with S845 but GPU would got a wings & would be 2x the speed of current A630 found on S845. That SoC would needles to say be much better for Android and especially for mobile VR (on Android and especially self stand)...
This ain't the best suitable benchmark for my script as all do it does three passes it doesn't do each test 3x in the row. Script is made mostly for real workloads and as a effort to balance power consumption & performance needs best possible way (to my knowledge). Now can I ask you kindly for a favour? Can you install App Tune-up (Qualcomm), set it to 10 min run the chrome & same benchmark in it & after it's done take screenshots of both results & Tune-up result & post it om the thread over at XDA? I know it's lot to ask for, it will take -15 minutes of your time but it would be very helpful for me. Also forgot to ask. How are you in generally satisfied with the script?
Big.little requires identical architecture state - not just sharing caches. You literally have to be able to save context from one type of core and load it on the other.
AFAIK, the software-visible architecture state of the Silvermont is a subset of modern Core CPUs, but I somehow doubt that covers everything saved in a context-switch.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
56 Comments
Back to Article
A5 - Wednesday, March 28, 2018 - link
Huh. I wonder how much of it is Atom running out of steam as an architecture on modern workloads versus thermal constraints.My i7-4600U in a work laptop scored ~110 on this benchmark, not that it is directly comparable due to OS stuff.
igavus - Wednesday, March 28, 2018 - link
Mid-2015 MBP i7-4980HQ, Jetstream scores: chrome: 176, safari: 259, FF: 198. All tests done in browser with single tab open, rest of system idle. So it's more of a browser than hardware test.. it'd seem.Then again, my expectations for Atom are low to begin with and a score of 23 would about match them. Doesn't mean that it couldn't go significantly higher if optimised. But that's a pipedream, since none will invest the effort to optimise this stuff anyways. It's doomed.
III-V - Wednesday, March 28, 2018 - link
If anything, it's Intel's Core line that has more potential to run out of steam (and clearly has, as we can see with the modest performance increases over the past few generations). Atom on the other hand... there's plenty of room to make it wider.HStewart - Wednesday, March 28, 2018 - link
This just shows Jetstream is a bad test for performance testing on processing.mczak - Wednesday, March 28, 2018 - link
I highly doubt the bad test result really has much to do with the cpu performance.Airmont might not be the fastest cpu core in the world, but there's no doubt it should beat Cortex-A53 cores (even if those are clocked slightly higher). It is still a "modern" out-of-order cpu design (versus the more simple in-order design of the A53). Granted it's a relatively simple and narrow OoO design, which probably should lose to the big Cortex-A cores (A72/73/A75), but it's not THAT bad.
OTOH having 8 of these cores in a phone surely seems like a marketing gimmick to me, "cause everybody has 8 cores!" (intel never produced cpus with more than 4 of these cores for desktop/mobile use, albeit some server chips had more).
hecksagon - Wednesday, March 28, 2018 - link
Actually performance wise it isn't really any faster than a Krait 300.ZolaIII - Friday, March 30, 2018 - link
New Atom is huge, its 3 instructions wide OoO design & it's lot bigger than A75 which is also 3 instructions wide.serendip - Thursday, March 29, 2018 - link
Atom x5-Z8500 (4x Airmont), Windows 10, Microsoft Edge: 33.4Snapdragon 650 (4x A53 + 2x A72), Android Lineage 14.1, Chrome Beta: 35.1
Strange to see an Android phone with an old 28nm chip outperforming an Atom Windows tablet with a 14nm chip. The Snapdragon 650 also shows much higher results on individual benchmarks like bigfib.cpp, gcc-loops.cpp and regexp-2010.
serendip - Thursday, March 29, 2018 - link
The SD650's A72 cores also didn't kick in during the test so I assume it was maxing out the A53s only. The 650 would get a much higher result of the high power cores had been running.ZolaIII - Friday, March 30, 2018 - link
S650=S636 regarding this test and Web usage in general.peevee - Tuesday, April 3, 2018 - link
I wonder how much of it is JIT translation of ARM code into x86 code vs natively compiled/optimized x86 code.HStewart - Wednesday, March 28, 2018 - link
The Atom is not true test of Intel's x86 cores - I put it on performance levels of older core 2 laptops from 10 years ago or so. It's technology has lead way into some higher performance chips like the Core M and also the forthcoming 9805g in the Dell XPS 15 2in1. It also include 16 core and possible more Server based chips aim with low power in mind.Also just including Jetstream benchmark is not to realistic comparison. This test should not be used as ARM vs x86 comparison.
Wilco1 - Monday, April 2, 2018 - link
Actually it's yet another good example why x86 is not suitable for anything low power. Atom has always been too slow or too power hungry compared to Arm cores.mode_13h - Monday, April 2, 2018 - link
I fear you're right, but it should be noted they're comparing an old architecture from 2013 against more modern ARM cores. A more fair comparison would be Goldmont or Goldmont+ (couldn't they find another mont?).Wilco1 - Saturday, April 7, 2018 - link
This was the only one that was reasonably power efficient - if this had been released 4 years ago it may not have looked so slow. Goldmont(+) is faster indeed but uses a lot more power - there was a good reason the mobile version was cancelled.nobodyblog - Wednesday, March 28, 2018 - link
That's no secret that Intel's Atom chips are not good. They are designed to run Android, but they are NOT best in it, too. Qualcomm are much better, even its mid tier. Totally, Android is optimized for ARM, too.Intel is best in laptop and desktop, not in mobile. But they have some advantages like mobility quality of Intel is better Perhaps. On paper things are different, though. ARM SOC is so good on paper, but, in action, I see them being normally good and CAN do the job, just like the atom that is not very optimized thing. I don't mind the benchmarks. I may see, but they are okay as long as it is not called a performance bug which makes problem in action. I also want better battery life rather better performance only....
Thanks!
Samus - Wednesday, March 28, 2018 - link
The real problem for Android devices (and Chromebooks for that matter) is the lack of optimization for x86. Atom CPU's are totally usable on Windows devices where the kernel is built specifically with Intel x86 in mind, not a recompilation retrofit lacking all sorts of optimizations like the Android kernel.Billy Tallis - Wednesday, March 28, 2018 - link
Android and Chromebook devices use the Linux kernel, which has always been developed with x86 support and optimization as the top priority. There might be some Atom-specific platform features that differ from PC architecture where Intel hasn't contributed drivers that are as good as the ARM counterparts, but its more likely that the Atom hardware is simply inferior to a good ARM SoC in those respects.peevee - Tuesday, April 3, 2018 - link
"Android and Chromebook devices use the Linux kernel, which has always been developed with x86 support and optimization as the top priority."Which has little to nothing to do with Javascript-in-some-unknown-browser tests.
Wilco1 - Saturday, April 7, 2018 - link
x86 has really good JavaScript engines. Atom is simply really slow compared to modern Arm CPUs.newblar - Wednesday, March 28, 2018 - link
I'm not sure why Intel couldn't use its i5-7Y54 as an soc chip. Not sure power consumption but it lists TDP of 4.5. It looks like its benchmarking scores are competitive with the newest gen of SOC. Would assume they just don't want to sell the chip for $60 like all the other companies do.III-V - Wednesday, March 28, 2018 - link
Well... yeah. It's way more expensive to make. And there's no integrated modemDanNeely - Wednesday, March 28, 2018 - link
They could make an SOC using that core if they wanted, but the existing chips wouldn't be viable in a phone platform from an idle power level standpoint because of things like the PCIe bus.skavi - Wednesday, March 28, 2018 - link
It's huge. It would hardly fit. It's also extremely expensive.hecksagon - Wednesday, March 28, 2018 - link
It's real life TDP is way higher than 4.5w. I have an M3 7Y30 and it routinely tops 10w when the CPU and GPU are loaded.serendip - Friday, March 30, 2018 - link
Intel stopped making Atoms with 2W TDP when it killed Willow Trail. A chip with higher TDP doesn't make sense as an SoC because it draws too much power on a tablet or smartphone. Bay Trail and Cherry Trail were amazing chips because they could meet TDP and SDP limits without throttling even at full load; a Core M is faster but it throttles hard once it hits redline.superkev72 - Wednesday, March 28, 2018 - link
Heh, my i7-7700k scored 312.4 on the jetstream 1.1 test. Still - portable devices have genuinely become quite impressive.Elstar - Wednesday, March 28, 2018 - link
"x86" is a such mixed bag. On the downside, the x86 instruction set is just awful to decode on the frontend. On the the upside (and when Intel cares), they have historically overcome the limitations of x86.If you have really branchy code, or something that benefits from AVX/AVX512 (but not GPUs), then Intel is really the only game in town.
takeshi7 - Wednesday, March 28, 2018 - link
Why didn't Microsoft and Intel just make a Phone that ran full x86 Windows? It would have been a killer app. Instead they both wasted billions of dollars with x86 Android phones and ARM Windows phones. So stupid.bji - Wednesday, March 28, 2018 - link
Nobody wants full Windows on a phone. Note here that "nobody" doesn't mean "absolutely no people", as there are some people (such as you) who believe that they want it (although if they actually had it they probably would quickly realize that they didn't actually want it). But the number of people who would trade the secure, well curated walled garden of Android/iOS for Microsoft Windows is vanishingly small. I'm so glad that my mom got a Samsung Galaxy instead of a phone running full Windows -- no way I'd want to be on the hook for fixing her phone every time it borks itself and cannot boot properly, or more likely, gets a virus, which would definitely be the experience with a full Windows stack on a phone, just as it is for the full Windows stack on any device.Microsoft had no hope of getting their OS onto a phone, it was a big waste of money and everyone knew that going in. Intel would have done better to license the ARM instruction set and make a better implementation than everyone else. But it's a business that probably doesn't have high enough profit margins to entice Intel.
bji - Wednesday, March 28, 2018 - link
Clarification: I do know that Microsoft got their OS onto some phones, but what I meant was that "Microsoft had no hope of getting their OS onto a *successful* phone".Alexvrb - Wednesday, March 28, 2018 - link
I'd love a Snapdragon 845 phone running full Windows, with CShell and WoA. However, the default settings should restrict software installs to Store-obtained apps, and prohibit Win32 programs from running except when docked (admittedly mostly for UI reasons on the second option). Both of those could be overridden by an advanced user, of course. I'd probably put these options on a toggle somewhere so I could lock it down or open it up whenever I wanted.But basically there's no problem running "full" Windows as long as the average user is prevented from doing stupid things. Heck even wide open x86 Win10 installs are better in that regard than previous versions. I haven't had to fix either of my parents Win10 machines. Most of what they use is either web-based or available as an app. So they don't need to go hunting on the web trying to download something (and often running into malware in the process). With Win7 and prior, seemed like every couple months I was called in to look at something that was borked. YMMV but that's been my experience.
Alexvrb - Wednesday, March 28, 2018 - link
Oh and if PWAs take off (backed by Google and even Apple as well as MS) then they won't have a problem getting all the must-have apps for a theoretical mobile-capable Windows build. This will be a while down the road though, most existing apps aren't going to be PWA overnight, and demanding apps will need to stay native for the forseeable future.Zeratul56 - Wednesday, March 28, 2018 - link
What a naive viewpoint, you really think android is significantly more secure than windows? I know windows is patched quite often because I hear my co workers bitch every time there computers take a bit longer to shutdown. On the other hand, most android phones are lucky to even get major OS updates in a timely manor. The krack exploit is still probably in the wild on an alarming high number of android devices. Microsoft had that patched time day it was announced to the public.Microsoft's issue was they were always on the outside of phone popularity. They never brought anything to game that the other players weren't already doing. The issues was not lack of care or technology but poor strategy and marketing.
bji - Wednesday, March 28, 2018 - link
Yes, I think that Android is significantly more secure than Windows. I wouldn't trust Microsoft to secure anything. They have proven incompetent at it too many times in the past.krack is not much of a security hole. I mean yeah you can, if you try hard enough, maybe you can read some network traffic. Nobody's going to bother to try hard enough though because you have to have physical presence, and if you're really trying to be a bad guy then you'd just mug your target instead of hoping to gather something of value from possibly sniffing some of their network packets.
It amazes me how people are incapable of assigning severity to security issues. So many people believe every security issue is just as bad as every other one. The windows vulnerabilities that I have personally experienced are all 100x worse than the android krack vulnerability.
damianrobertjones - Thursday, March 29, 2018 - link
...and I bet that EVERY single Windows user, you know, runs their machine(s) with full Admin access all day long. No wonder they get infected.bji - Thursday, March 29, 2018 - link
You'd be wrong.Hurr Durr - Thursday, March 29, 2018 - link
>android>well curated
>muh terrible viruses on windoze
Elstar - Wednesday, March 28, 2018 - link
Well, for starters, what could be built with early to late 2000s technology was a huge limiting factor. Back then, phones were slow and RAM poor, and that meant that running a "desktop-class" operating system was out of the question (x86 or otherwise). The fact that the iPhone worked at all in 2007 is because Apple put "OSX" on a gigantic diet just to make it fit into the resources available.hecksagon - Wednesday, March 28, 2018 - link
LOL could you imagine the quantity of warranty claims providers would get? Literally every single computer issue that has ever existed would turn "mission critical" because their phone is down.karthik.hegde - Wednesday, March 28, 2018 - link
No one os going to bat an eye on ARM GPU in Intel SoC?Alexvrb - Wednesday, March 28, 2018 - link
We didn't bat an eye with PowerVR GPUs in Intel SoCs for many years. Going back even to integrated desktop graphics.serendip - Thursday, March 29, 2018 - link
Atom really is at the end of the line. Intel dropped the 2W-4W chips for tablets and concentrated on Atom for cheap desktops which also didn't take off. I would love to see an 8-core Atom with the latest cut-down HD Graphics engine for $200-300 Windows tablets but that's a pipe dream now. The x5-Z38xx chips are decent, on par with a mid level C2D, but they're outclassed by the latest Qualcomm, Samsung and HiSilicon flagship chips. If only Windows Subsystems for Linux ran on ARM...AntonErtl - Thursday, March 29, 2018 - link
Just for perspective, here is another benchmark on a Silvermont (22nm variant of the Airmont tested here), and a Cortex-A53 (results are times, so lower is better):From <https://www.complang.tuwien.ac.at/franz/latex-benc...
1.052s Celeron J1900 (Silvermont) 2416MHz (Shuttle XS35V4) Ubuntu16.10 64-bit
2.32s Odroid C2 (1536MHz Cortex A53) Ubuntu 16.04 64-bit
Even factoring in the differences in clock rate compared to the phones in this article, I would expect the Airmont to outperform the Cortex A53 by a factor of 1.26 on this benchmark on these phones. So the JetStream benchmark seems to favor the A53 quite a bit more than the LaTeX benchmark.
Wilco1 - Monday, April 2, 2018 - link
Either way a 3-way OoO core being just 1.26 times faster per clock than a 2-way in-order core is extremely poor - it means Cortex-A55 should be about equal...ZolaIII - Friday, March 30, 2018 - link
Old S650 scores 44.847 mid delta in with Chrome. Saying this just to remind you how how all you need are two more power OoO core's for the Web & in generally 93% of the times as a refract how really bad engendered are all the new SoC's recently.serendip - Friday, March 30, 2018 - link
How did you get those results? On MIUI?I'm using your cpu and schedule scripts on Lineage and I'm not seeing the A72 cores kick in during the test. I don't know if it's because of the config script or if it's an issue with Chrome Beta.
ZolaIII - Friday, March 30, 2018 - link
Used regular chrome. In my script HPM is set to flag similar tasks as high performance (within 240ms) and skip CPU scheduler (interactive) completely pushing maximum performance almost instantly. On S652 it even gets 5~6%+ performance as the hotplug (core_ctl) is set so that core 04 & 06 are always idling one's so all L2 is utilised & with exclusive access.So I don't really know what's the problem in your case. Is hotplug working?
serendip - Friday, March 30, 2018 - link
I reflashed the ROM and your CPU script, re-enabled hotplugging and now I get 40 on JStream. The 2x A72 cores run at full speed, A53 cores 0 and 2 run at full speed and the rest are idle. It's supposed to be that way I guess?Anyway Intel missed the boat on heterogenous computing. The A53s save a lot of power, the A72s pack a big punch only when needed. An Atom with 4x Goldmont and 2x Core M cores would be a screamer with excellent battery life, if only those two architectures could share caches.
ZolaIII - Friday, March 30, 2018 - link
Well for this benchmark it is supposed to be like that as browser's (among most other things) are only SMP 2. Yes I used same typology on small core's too.If Intel made SoC like that it would be probably priced 300+$ so no one would buy it. Also they would had to do a big R&D on heterogeneous before so it's a lost battle anyway.
Who knows me by in the future we see Falkor derivat's (what ever dragon it would be called) on mainstream mobile SoC's. It's more probable so as Samsung & Apple as prime competitors are pushing it & QC will have to respond.
For me ideal SoC would be an 2x A75 & 4x A55 with big juicy GPU like A630 on 22nm FD-SOI, naturally without idiotic L4 cache found on on S845. Described as it is it would cost approximately half the price of S845 & achieved 70% performance (both CPU and GPU) in real usage so it would be a real improvement & still with lower DTP. I would pay a full price for it as for S845 if it would be coupled with 4GB HMB 2 (which is about the same price as 8BG of LPXDDR 4 and 20$ extra from the SoC for interconnect) this would make it on pair in CPU side with S845 but GPU would got a wings & would be 2x the speed of current A630 found on S845. That SoC would needles to say be much better for Android and especially for mobile VR (on Android and especially self stand)...
Who knows... We will see both eventually.
mode_13h - Monday, April 2, 2018 - link
I'm with you on HBM2 for mobile. I've been saying this for a long time...ZolaIII - Friday, March 30, 2018 - link
This ain't the best suitable benchmark for my script as all do it does three passes it doesn't do each test 3x in the row. Script is made mostly for real workloads and as a effort to balance power consumption & performance needs best possible way (to my knowledge).Now can I ask you kindly for a favour?
Can you install App Tune-up (Qualcomm), set it to 10 min run the chrome & same benchmark in it & after it's done take screenshots of both results & Tune-up result & post it om the thread over at XDA? I know it's lot to ask for, it will take -15 minutes of your time but it would be very helpful for me. Also forgot to ask. How are you in generally satisfied with the script?
serendip - Friday, March 30, 2018 - link
Doing it now, let's hop over to XDA :) Thanks for all your hard work on the script.mode_13h - Monday, April 2, 2018 - link
Big.little requires identical architecture state - not just sharing caches. You literally have to be able to save context from one type of core and load it on the other.AFAIK, the software-visible architecture state of the Silvermont is a subset of modern Core CPUs, but I somehow doubt that covers everything saved in a context-switch.
Jiang - Saturday, June 2, 2018 - link
https://youtu.be/_XaRaSdSwdMJiang - Saturday, June 2, 2018 - link
https://youtu.be/R5B4o9ePvcIIssue overheating SoC Intel