[App] Cpu Performance Tester - Windows 8 Development and Hacking

This application simply does large amounts of mathematical operations on one, two, three, four, or eight threads (which corresponds to the number of cores used) and measures the amount of time it takes to complete. It makes it easy to see the cpu power of one computer compared to another. Enjoy
Some devices for comparison
Intel i7-3770 @ 3.4 ghz (quad core ht) -
1 Thread - 23 seconds
2 Threads - 23 seconds
4 Threads - 24 seconds
8 Threads - 27 seconds
Intel i5-3317U @ 1.7 ghz (dual core ht) -
1 Thread - 35 seconds
2 Threads - 36 seconds
4 threads - 40 seconds
8 threads - 1 minute 16 seconds
Intel Atom-Z3740D @ 1.33 ghz (bay-trail, quad core no ht)
1 Thread - 1 minute 24 seconds
2 Threads - 1 minute 24 seconds
4 Threads - 1 minute 25 seconds
8 Threads - 2 minutes 35 seconds
Source code can be found in the text file bellow.
Performance Tester:
https://dl.dropboxusercontent.com/u/46535328/Performance.exe - For X86
https://dl.dropboxusercontent.com/u/46535328/Performance-RT.exe - Windows RT edition

Cool. Looks easy to port to jailbroken RT; and chance of an ARM build? (I'd do it myself but I'm away from my dev machine at present).

Of course. I will just post it above for now.

That would be great, thanks.
I'm somewhat amused by somebody with a C# logo in their profile image posting a C++ app but that does make sense for performance testing programs.

Scores for my AMD FX-8350 (stock @ 4.0GHz):
1 Thread: 26 seconds
2 Thread: 27 Seconds
3 Thread: 27 seconds
4 Thread: 27 Seconds
8 thread: 29 seconds.
Will give it a whirl on my Q6600 machine in a sec!
Q6600 (stock 2.4GHz):
1 thread: 47 seconds
2 threads: 47 seconds
3 threads: 47 seconds
4 threads: 48 seconds
8 threads: 1 minute 26

On Windows 8 (64 bit) it gives me the following error:
The program cannot be started, because MSVCP110.dll it missing.

You need .net framework or Visual c++.
http://www.microsoft.com/en-au/download/confirmation.aspx?id=30679
Works on 7, my work machine which has a core 2 duo 4600(yup running one of these old guys!) scored 56 seconds on 2 threads.

@chazram: Cool (though not surprising) to see that the clock cycles per (arithmetic) instruction are basically the same between Intel and AMD (which allows the AMD's greater clock rate to smoke the Intel). Also interesting to see the difference between actually having eight integer cores (AMD) and having eight hardware pipelines but only four ALUs (Intel). It *is* actually a bit surprising to see how close the AMD is on 4 threads and 8, given the 8350 only has 4 floating-point ALUs and a good bit of the computation is FP. On the other hand, repeating the same operation over and over again can easily be optimized, possibly even my microcode.
Of course, real-world usage is a lot more than just crunching numbers over and over, but it's a pretty good set of ALU benchmarks.

Is the test loop doing floating point or integer calculation ?
shouldn't thread priority be high ?
Like memory consumption exceed the simple CPU Cache, how to be sure that results aren't influenced by ram quality ?

GoodDayToDie said:
given the 8350 only has 4 floating-point ALUs and a good bit of the computation is FP. On the other hand, repeating the same operation over and over again can easily be optimized, possibly even my microcode.
Click to expand...
Click to collapse
Should also be mentioned that the AMD floating point units consist of 2 128 bit FPU's which can be addressed by each half of a module independently or the 2 128 bit units can be ganged together to form a single 256 bit FPU. I think intel utilise a near identical unit. But I should think the effective 8 128 bit FPU's of the 8350 should suffice?

Scores from my Win8.1 device:
One Thread - 27
Two Threads - 27
Three Threads - 27
Four Threads - 28
Eight Threads - 30
Lenovo Y410P, with an Intel i7-4000MQ @ 2.40GHz
Connor Baker

Win 8.1
1 thread: 34 seconds
2 thread: 38 seconds
AMD Phenom II X2 560 @ 3.3GHz

core i7 3630 QM @ 2,3 ghz
1 threads 27 seconds
2 threads 27 seconds
3 threads 27 seconds
4 threads 28 seconds
8 threads 32 seconds

Surface RT running RT 8.0, NVidia Tegra 3 (32-bit ARM) @ 4x1.3GHz
1 threads 2:46 (166 seconds)
2 threads 2:47 (167 seconds)
3 threads 2:47 (167 seconds)
4 threads 2:44 (164 seconds)
8 threads 5:00 (300 seconds)

Intel(R) Core(TM) i3-4130T CPU @ 2.90Ghz Win8.1
1 thread 30 Seconds
2 threads 30 Seconds
3 threads 33 Seconds
4 threads 33 Seconds
8 threads 1:04

Makes me feel better that my old 2007 mac pro running Windows 8.1 64-bit was able to get the following
One Thread - 37
Two Threads - 37
Three Threads - 37
Four Threads - 37
Eight Threads - 37

Scores from Windows 8.1 x64 update 1:
One Thread - 28
Two Threads - 28
Three Threads - 28
Four Threads - 29
Eight Threads - 39
Lenovo Ideapad y510P, 4th Generation Intel Core i7-4700MQ Processor (2.40GHz 1600MHz 6MB)

Any linux edition?
Want to test it on my hp.
Отправлено с моего HTC One S через Tapatalk

I could write up a version for Linux using pthreads easily enough, but I forget whether Linux schedules pthreads on different processors or not. Different processes using fork/clone is more common, but not very efficient and not entirely in the original spirit of the test.
In other words, I *could* write it, but performance testing code for an OS should probably be written by somebody who has actually developed code for that OS in the last 5 years or so.

GoodDayToDie said:
I could write up a version for Linux using pthreads easily enough, but I forget whether Linux schedules pthreads on different processors or not. Different processes using fork/clone is more common, but not very efficient and not entirely in the original spirit of the test.
In other words, I *could* write it, but performance testing code for an OS should probably be written by somebody who has actually developed code for that OS in the last 5 years or so.
Click to expand...
Click to collapse
Any new news?
Отправлено с моего HTC One S через Tapatalk

Related

Tegra 3 Overclock..?

I'm loving my yoga 11, however at times I just feel that Windows 8 RT slows down especially when multi-tasking. Since our Tegra's are clocked at 1.3Ghz and the same Chip in android devices runs at 1.5, with overclocked kernels available to run at 1.8-2.0Ghz, what are the chances we see this type of hack/development come to windows 8 RT? Im not sure the security obstacles that would present, but haven't seen much on this to even know if someone has looked into this or actively working on method to do so.
Thanks!
I have been thinking about this as well. Im sure it can be done, but by who? thats the question. Im sure we can easily squeeze some more power out of our device. Good luck to whoever spearheads this
ej_424 said:
I'm loving my yoga 11, however at times I just feel that Windows 8 RT slows down especially when multi-tasking. Since our Tegra's are clocked at 1.3Ghz and the same Chip in android devices runs at 1.5, with overclocked kernels available to run at 1.8-2.0Ghz, what are the chances we see this type of hack/development come to windows 8 RT? Im not sure the security obstacles that would present, but haven't seen much on this to even know if someone has looked into this or actively working on method to do so.
Thanks!
Click to expand...
Click to collapse
The tegra isnt overclocked to 1.5 in android devices. There are actually 3 models of the Tegra 3 at different clock speeds. The one used in the RT is the lowest model (1.2GHz) overclocked to 1.3GHz already. I believe the other models are 1.4 and 1.6 with a few ROMs adding about 100MHz overclock as needed. 2ghz seems extreme though.
SixSixSevenSeven said:
The tegra isnt overclocked to 1.5 in android devices. There are actually 3 models of the Tegra 3 at different clock speeds. The one used in the RT is the lowest model (1.2GHz) overclocked to 1.3GHz already. I believe the other models are 1.4 and 1.6 with a few ROMs adding about 100MHz overclock as needed. 2ghz seems extreme though.
Click to expand...
Click to collapse
I've thought about this as well but have always been too scared to ask. Windows is obviously not foreign to processor scaling and power management, perhaps there's a way to make a custom power plan or something. Maybe the way to approach overlooking is not 'like' Android, but 'like' regular old windows. I have no idea and am a noob, but I thought I'd just toss that out there.
SixSixSevenSeven said:
The tegra isnt overclocked to 1.5 in android devices. There are actually 3 models of the Tegra 3 at different clock speeds. The one used in the RT is the lowest model (1.2GHz) overclocked to 1.3GHz already. I believe the other models are 1.4 and 1.6 with a few ROMs adding about 100MHz overclock as needed. 2ghz seems extreme though.
Click to expand...
Click to collapse
http://www.nvidia.com/object/tegra-3-processor.html
Its support for Windows RT is still under development. It isn't overclocked on the Surface RT/Vivo Tab but underclocked to compensate for the missing support for the fifth battery saver core.
We should expect the performance and battery to get better as they iron this out :laugh:
Actually, for those who have gotten Surface RT since launch... I bet most of you have already experience better performance after each monthly firmware update
LastBattle said:
http://www.nvidia.com/object/tegra-3-processor.html
Its support for Windows RT is still under development. It isn't overclocked on the Surface RT/Vivo Tab but underclocked to compensate for the missing support for the fifth battery saver core.
We should expect the performance and battery to get better as they iron this out :laugh:
Actually, for those who have gotten Surface RT since launch... I bet most of you have already experience better performance after each monthly firmware update
Click to expand...
Click to collapse
That's a very good news indeed and we should then probably be able to run the Tablet at 1.6Ghz Quad core instead of the actual 1.3Ghz quad core :good:
LastBattle said:
http://www.nvidia.com/object/tegra-3-processor.html
Its support for Windows RT is still under development. It isn't overclocked on the Surface RT/Vivo Tab but underclocked to compensate for the missing support for the fifth battery saver core.
We should expect the performance and battery to get better as they iron this out :laugh:
Actually, for those who have gotten Surface RT since launch... I bet most of you have already experience better performance after each monthly firmware update
Click to expand...
Click to collapse
No where in that link does it mention it being underclocked. The 1.4ghz single core/1.3 quad core is a feature of the entire tegra product line, not jsut the surface RT.
It does mention that the 5th battery saver core doesnt work on windows RT though, that will help.
Interesting: There is a "~MHz" key in regedit under local machine -> Hardware -> Description -> System -> Central processor -> 0, 1, 2, or 3. It is set to 1300, but changing it doesn't do anything and it reverts upon reboot.
Even if we can't overclock this thing, is there a way to resurrect the "High Performance" power plan that disappeared in RT? One that would set the CPU to 100% by default, all the time?
Any update or more info on this?
bigsnack said:
Any update or more info on this?
Click to expand...
Click to collapse
+1
hope to see a 'high performance' feature on the pwr mgnment as well, especially when we are hooking up RT onto the power line and battery life is not so much of an issue in this case.
Rogerngks said:
hope to see a 'high performance' feature on the pwr mgnment as well, especially when we are hooking up RT onto the power line and battery life is not so much of an issue in this case.
Click to expand...
Click to collapse
iirc, you can still set your cpu states through powercfg in the command line. I might be wrong though.
Is the 5th power saving core just disabled or not present on our hardware?
bigsnack said:
Is the 5th power saving core just disabled or not present on our hardware?
Click to expand...
Click to collapse
According to NVidia's website, Tegra 3 for RT is "still under development." (http://www.nvidia.com/object/tegra-3-processor.html) It also lists it as only being quad-core on Windows 8 devices.
I had personally reeealy hoped that one of the highlights for RT 8.1 was going to be reworked support for the 5th core, bringing performance and battery life improvements. Alas, it was not to be.
jtg007 said:
According to NVidia's website, Tegra 3 for RT is "still under development." (http://www.nvidia.com/object/tegra-3-processor.html) It also lists it as only being quad-core on Windows 8 devices.
I had personally reeealy hoped that one of the highlights for RT 8.1 was going to be reworked support for the 5th core, bringing performance and battery life improvements. Alas, it was not to be.
Click to expand...
Click to collapse
I cant see how the 5th core would bring a performance improvement. The system cannot use the 5th core as an actual 5th core, it shuts most of the other cores down to sleep when it needs the 5th which is also an incredibly low performance core, its just for power saving really, or simply hopping around the UI and checking your email, NVidia claim that android can also play video while running purely on the 5th core although this never happened on my Nexus 7 without any other apps running, it carried on running using 1 of the main cores for that.
Would definitely boost the battery life though and thats not something to be ignored. But there are few times where that 5th core really comes into its own, perhaps it just wasn't worth the time for MS to add companion core support to windows RT 8.1 when not all RT tablets use the tegra.
SixSixSevenSeven said:
I cant see how the 5th core would bring a performance improvement. The system cannot use the 5th core as an actual 5th core, it shuts most of the other cores down to sleep when it needs the 5th which is also an incredibly low performance core, its just for power saving really, or simply hopping around the UI and checking your email, NVidia claim that android can also play video while running purely on the 5th core although this never happened on my Nexus 7 without any other apps running, it carried on running using 1 of the main cores for that.
Would definitely boost the battery life though and thats not something to be ignored. But there are few times where that 5th core really comes into its own, perhaps it just wasn't worth the time for MS to add companion core support to windows RT 8.1 when not all RT tablets use the tegra.
Click to expand...
Click to collapse
I always thought that the 5th core could run simultaneously with the other 4 to manage background tasks, etc, thus leaving less side work for the others. I could be wrong though. Also, I know of only one RT tab to NOT use Tegra (Dell), and it was the first to drop price and flop.
Anyways, the exciting thing about kexec/Linux prospects is that if we were to get in, there are a lot of Android and Linux versions that run on Tegra 3, which hopefully means we wouldn't have too tough of a time getting at that 5th core working then.
Sent from my SCH-I535 using xda app-developers app
Well the Samsung Ativ Tab RT is also using the S4 cpu, but that device had a limited release from what it seems like in North America. I too was under the assumption that th3 5th core could be used at the same time with the other cores, which could free up power for other things. Like the 5th core would be used for the low power task, while at the same the the other 4 cores are being used for a more process heavy task.
It would be interesting to have Android or Linux running in a dual boot situation on our RT devices, or if even possible do what Samsung is doing, and have it emulated in Windows so you can run apps side by side.
No, the 5th core is not an actual 5th core. The idea is you have 4 full blown cores at 1.2, 1.4 or 1.6ghz depending on the tegra model (and then the tegra can overclock automatically to 1.3, 1.5 or 1.7), thats quite power hungry really. But as CPU usage falls the tegra shuts a few cores off, if the system cant benefit from all 4 cores being active it will drop to 3, then 2 and then 1. Sometimes even that 1 core running at 1.2ghz is compartively power hungry, so the tegra shuts the final core down and fires up the companion core which I think runs around the 700MHz range, its slow at any rate, its also built optimised purely for power consumption over performance. Idea is you can go from a full quad core chip when you need the performance but then when the device is idling you can switch over to the companion core and shut the main 4 all off and save alot of power.
NVidia claim that the companion core combined with the hardware video acceleration of the tegra should be able to play HD videos on its own. That doesnt really seem to happen outside of the lab. But when you lock the screen on your android device it often jumps into companion core mode, you can browse around the android home screen and use a few lightweight apps on the companion core no problem, and when it does begin to struggle the tegra just has to skip over to its main core and gradually bring the other 3 main cores online as it needs them.
It never has the companion and main cores on in a state able to be used by the operating system simultaneously though.
Samsungs so called octa-core chips also do the same. They arent really octa core chips, in reality they are a quad core cortex A15 chip and a quad core lower clock speed cortex A9 chip (possibly even A7) on the same piece of silicon, when CPU load is high it runs as a quad core A15, when it doesnt need so much performance it shuts down the A15 and swaps for the A9, the 2 CPU's are near separate and at any one time the chip is only running as a single quad core processor not an octacore. Similar to the companion core design this can lead to a massive boost in battery life. In both A15 and A9 modes the processor is capable of shutting down individual cores as need be.
Tegra may well be the chip in all main tablets, but when microsoft first started working on windows RT there were meant to be qualcomm snapdragon, NVidia tegra and texas instruments OMAP devices all coming to market so of course microsoft at the time needed RT to run on all 3. The original plan was that there would be56 3rd party manufacturers manufacturing RT tablets, 2 per chip vendor except TI. Originally qualcomm partnered with HP and Samsung, NVidia with Lenovo and Asus and Toshiba with TI In the end TI dropped out and shortly after downscaled OMAP production (I think it has completely stopped with the exception of existing contracts now, or at least chips intended for tablet usage have been, they had a few industrial chips under the OMAP branding that might still be available, their ARM based microcontroller and DSP lines are still going fine), TI took Toshiba with them. Of course by the time TI dropped out there were already running builds of RT. HP dropped out and were replaced by dell. Acer were slated to be joining the program but didn't, when MS unveiled the surface that killed it for acer.
Another limitation is that Windows RT is essentially just an ARM port of windows 8, windows 8 and the NT kernel in general didnt already have support for the companion core or similar tech, it would be pointless adding it to the base NT kernel as hardly any devices use it and it would probably lead to issues introducing it only for tegra.
Surely Microsoft can see that getting the maximum out of the CPUs in their own devices is a good thing? I get that they have to support a few ARM architectures, but there's no reason why Windows RT can't be optimised with a specific update for the Surface?
bydandie said:
Surely Microsoft can see that getting the maximum out of the CPUs in their own devices is a good thing? I get that they have to support a few ARM architectures, but there's no reason why Windows RT can't be optimised with a specific update for the Surface?
Click to expand...
Click to collapse
It would be a maintenance nightmare. You know the way everyone *****es and moans about the non existent android fragmentation (or at the very least hugely over exaggerated)? Now apply that to windows RT, its already a struggling platform. You don't want more ammo for the opposition, the extra effort probably isn't worth it. Under sleep mode or single core mode (non companion, RT will scale back to single core non companion happily) the battery life is good enough, companion would be nice, but non essential. Companion core would need to be supported at a kernel level. It would be a nightmare to keep one version of the kernel (if you don't know what a kernel is, consider it the chassis of a car or the foundations of a house, its the very core of the operating system) for each tablet.

How is powervr g6430 rogue when campared to top class gpu's like adreno 405 etc?

Is power vr g6430 any good when campared to adreno gpu's?
http://www.gsmarena.com/apple_iphone_5s_vs_lg_g2_vs_nokia_lumia_1020-review-997p5.php
The same GPU used on iPhone 5s. Based on this benchmark, it's better than Adreno 330 I think.
Adreno 405 isn't top class GPU. According to GFLOPS numbers, 405 is better than 1st gen Adreno 320 (S4 Pro, S4 Prime) and weaker 2nd gen.
But all about benchmarks, the most important is user experience and last but not least is optimization
GrandpaaOvekill said:
Is power vr g6430 any good when campared to adreno gpu's?
Click to expand...
Click to collapse
Adreno 405 is only half as power of powervgr g6430
Adreno 405 is middle range gpu
While powervgr g6430, adreno 320, 330, 420 are last year and current flagship gpu
Gpu mostly rated by gflops
http://kyokojap.myweb.hinet.net/gpu_gflops/
And adreno each generation have basic, mid, high power gpu..
Adreno 405 is 4th generation (05 means basic) and can match 3rd Gen mid
Adreno 420 is 4th generation (20 is mid) and can match 3rd Gen high gpu
See gflops of each in that above link
And yes optimization is the most for gaming
The PowerVR G6430 in Zenfone 2 is clocked higher than iphone 5s but lower than ipads and Atom 3570. Its performance is between the Adreno 330 and 430 which is excellent given that it was designed in 2012 and released in 2013. Reclocking it at 640Mhz like its 3570 brother should give a nice run for its price, still technically, it won't be as fast as Adreno 430. However, in real world usage and coupled with a more powerful Intel cpu, it should match it as the CPU is able to extract more GPU power.
If you are really looking at the most powerful mobile GPU, the Nvidia Tegra X1 is at the top, close to twice the performance of the top Qualcomm 810 GPU, Adreno 430. In Antutu, it only scores 75K because the CPU is slower than others like Intel. 75K is still unbreakeable for the moment. Surely, Nvidia and ATI have much more experience in the GPU domain so its not surprising that they are the fastest.
Now, only if ATI partner with Intel to provide us with 14nm goodies :angel:
p.s: To have a broader picture, the Tegra X1 chip is close to twice the performance of a PS3 which is astonishing considering its small size and 2W max power consumption.
Nvidia Shield TV based on Tegra X1 has active cooling system.
So, how it can be compared to phone SoCs?
My bad, I though it was found in the Nvidia Shield tablet. Its its brother the Kepler K1 that is currently used but still at 365 GFlops on nvidia website, it competes with the adreno 430. Note that the PS3 was 192 GFlops.
Interesting fact is that the Tegra X1 actually draws much Less power at idle and slightly less power (1w less than Kepler) at load. Kepler would peak at 11w. Thanks to the new 20nm tech in Maxwell cores efficiency. The Nvidia TV Shield has much more and larger components to power, its also for sure clocked higher.
''According to Nvidia, the power consumption in a tablet powered by Tegra X1 will be on par with Tegra K1. In fact, idle power consumption will be even lower thanks to the various architecture improvements. Tegra K1 was designed to operate at around 5-8 watts, with infrequent peaks up to 11 watts when running stressful benchmarks, so the X1 will be well within the realm of tablet power requirements.'' Source: greenbot.com
Heres this too: http://www.pcper.com/reviews/Processors/NVIDIA-Announces-Tegra-X1-Maxwell-Hits-Ultra-Low-Power
I really like the fact that PC manifacturers enter the mobile market, after all, they were building computer components for ages. This will open the door to more powerfull and cheaper SoCs especially because they have the ability to mass produce and develop the latest tech with many factory plants worldwide.
aziz07 said:
My bad, I though it was found in the Nvidia Shield tablet. Its its brother the Kepler K1 that is currently used but still at 365 GFlops on nvidia website, it competes with the adreno 430. Note that the PS3 was 192 GFlops.
Interesting fact is that the Tegra X1 actually draws much Less power at idle and slightly less power (1w less than Kepler) at load. Kepler would peak at 11w. Thanks to the new 20nm tech in Maxwell cores efficiency. The Nvidia TV Shield has much more and larger components to power, its also for sure clocked higher.
''According to Nvidia, the power consumption in a tablet powered by Tegra X1 will be on par with Tegra K1. In fact, idle power consumption will be even lower thanks to the various architecture improvements. Tegra K1 was designed to operate at around 5-8 watts, with infrequent peaks up to 11 watts when running stressful benchmarks, so the X1 will be well within the realm of tablet power requirements.'' Source: greenbot.com
Heres this too: http://www.pcper.com/reviews/Processors/NVIDIA-Announces-Tegra-X1-Maxwell-Hits-Ultra-Low-Power
I really like the fact that PC manifacturers enter the mobile market, after all, they were building computer components for ages. This will open the door to more powerfull and cheaper SoCs especially because they have the ability to mass produce and develop the latest tech with many factory plants worldwide.
Click to expand...
Click to collapse
Maxwell can very power hungry when you clock it all the way up, and X1 has more CUDA cores than K1. X1 has 2 SMM with 256 total while K1 only has 1 SMX with 192.
also, pc manufacturers have always been in the mobile market, or you could even say they started the mobile market. for instance, Apple was a pc manufacturer, steve jobs dedicated 70% of his life to PC rather than phones. samsung makes everything and they have a lot of experience too in making notebooks. so the two most powerful (or most successful) players in the mobile sector are also pc manufacturers, what do you mean by pc manufacturers entering the mobile market?
Its getting off topic but Intel or Apple weren't the first one to build a cell phone. Intel was the first company to build a CPU though. Motorola built the 1st cellphone.
On a sidenote, Apple never really built anything except for aesthetics, it started with IBM building for them after non-success with Synertek for a couple of months. Btw, Samsung does not manifacture PC CPUs or GPUs. Only CPU they build is the Exynos for mobile. I think you misinterpreted the fact the they sell laptops, yes they do, but they are not the one building its major components, its Intel and AMD. They may build its memory components but not CPU or GPU.
You are seeing technology the other way around. If we take, let say, a 2 years old gpu and a new one. The new one can have double the transitor and components count yet still consume less power. Its about architechture efficiency and transistor nm. e.g. the Intel in our Zenfone 2 is built with 3D 22nm transistor which is more power efficient. That's how tech flow.
Anyway, apple is slowly declining, Intel is building their PC segment, replacing IBM, and Samsung is building their next iphone and taking care of the mobile segment. We can already see whats next.
I have been building PCs for over 15 years, its my hobby.
@ mods There should be a ''resolved'' button just like other forums so threads don't get cluttered lol
GrandpaaOvekill said:
Is power vr g6430 any good when campared to adreno gpu's?
Click to expand...
Click to collapse
I know benchmarks aren't everything, but GFX gives a good idea of the performance difference between the two. Basically, the PowerVR G6430 is much more powerful than the Adreno 405.
PowerVR G6430:
https://gfxbench.com/result.jsp?ben...VR Rogue G6430&base=device&ff-check-desktop=0
Adreno 405:
https://gfxbench.com/result.jsp?ben...ter=Adreno 405&base=device&ff-check-desktop=0
Here's some videos of a Zenfone 2 with a phone that utilizes the SD 615/Adreno 405 combo
https://www.youtube.com/watch?v=N3DcRHXrTHg
https://www.youtube.com/watch?v=TYZr53U2Tfk
Hope this helps.

HTC 10: Averages 49 FPS During Computational Test

Hi all,
A user recently shared his Mobile FPS Test that gauges a phone's FPS through various animation and computation challenges. In the Android Subreddit thread, users reported that their flagship, mid-range, or older flag ship devices were consistently hitting 59-60 FPS. There weren't any HTC 10 reviews that I could find.
Disappointingly, the HTC 10 averaged 49 FPS across the tests. I nearly turned everything off in the background, returned to the Sense Home, etc., but my results were consistently the same. The other option is a factory reset.
For comparison, the Nexus 5 didn't struggle with 59 FPS.
I'd appreciate another HTC 10 user to compare the test and see if this is my unit.
Thanks for your help.
I ran the tests and mine averaged around 60. I used Firefox though as I've heard that chrome has some problems with using device native FPS. I am not allowed to post screenshots though so I guess I can just give you the average values:
Baseline: 60.0
DOM mutations: 59.6
Canvas animations: 78.7
CSS matrix transforms: 64.2
Touch events: 60.0
Hi,
Thanks for the response. You're right, I used Chrome originally with both the Nexus 5 and the 10. I switched to Firefox and my numbers were nearly identical to yours. The question is why would using Chrome have a 10 FPS loss?
jhadsfi said:
I ran the tests and mine averaged around 60. I used Firefox though as I've heard that chrome has some problems with using device native FPS. I am not allowed to post screenshots though so I guess I can just give you the average values:
Baseline: 60.0
DOM mutations: 59.6
Canvas animations: 78.7
CSS matrix transforms: 64.2
Touch events: 60.0
Click to expand...
Click to collapse
Probably some flags set in chrome? You could try going to about:flags and changing some of them to check if they lead to better results.

P600 octa-core showing only four cores

Over the years I have switched rom's many times.
Now I am back on stock 5.1.1
I use app2sd pro to help improve performance and it was in the app I noticed that it shows the device only has four cores.
So I downloaded cpu id and antutu and they all show four cores.
I am wondering why. Did I install the wrong stock rom even though it was labelled as Australia XSA?
Something else?
Strange.....
Edit: the p600 should have an octa-core
So how can one verify that it is octa-core?
meekspace said:
Over the years I have switched rom's many times.
Now I am back on stock 5.1.1
I use app2sd pro to help improve performance and it was in the app I noticed that it shows the device only has four cores.
So I downloaded cpu id and antutu and they all show four cores.
I am wondering why. Did I install the wrong stock rom even though it was labelled as Australia XSA?
Something else?
Strange.....
Click to expand...
Click to collapse
For me in Aida64 and CPU-Z it says:
Cores: 8
4x ARM Cortex-A15 @1.90GHz
4x ARM Cortex-A7 @1.90GHz
jeffrey268 said:
For me in Aida64 and CPU-Z it says:
Cores: 8
4x ARM Cortex-A15 @1.90GHz
4x ARM Cortex-A7 @1.90GHz
Click to expand...
Click to collapse
Ok.. I used those apps and the data is the same.
But it only shows four cpu/cores. Why?
Kernel adiutor shows four cpu's. How do you get access over all eight?
meekspace said:
Ok.. I used those apps and the data is the same.
But it only shows four cpu/cores. Why?
Kernel adiutor shows four cpu's. How do you get access over all eight?
Click to expand...
Click to collapse
You can't , it is first gen big.little SoC, meaning that you can only use 4 at a time. It is the SoC immediately after this that allowed for all 8 to be used synchronously.
There was a guy some time back that tried to enable all 8 cores working in synergy. Eventually he gave up IIRC , I just can't recall the reason.
IMO the reason that Samsung did not allow for it was that the SoC would throttle due to high temps if you were to enable all 8 at once leading to actually lower performance. So they simply disabled it. After a node shrink they finally managed to do it without thermal throttling, so the first phone that had it in a useful manner was the Note 4 (we use Note 3 tech). IIRC S5 had the feature as well however it throttle to hell and back, so it was not very useful there (hence why Note 4 was the first to use 8 cores syncronously in a useful manner)
Stevethegreat said:
You can't , it is first gen big.little SoC, meaning that you can only use 4 at a time. It is the SoC immediately after this that allowed for all 8 to be used synchronously.
There was a guy some time back that tried to enable all 8 cores working in synergy. Eventually he gave up IIRC , I just can't recall the reason.
IMO the reason that Samsung did not allow for it was that the SoC would throttle due to high temps if you were to enable all 8 at once leading to actually lower performance. So they simply disabled it. After a node shrink they finally managed to do it without thermal throttling, so the first phone that had it in a useful manner was the Note 4 (we use Note 3 tech). IIRC S5 had the feature as well however it throttle to hell and back, so it was not very useful there (hence why Note 4 was the first to use 8 cores syncronously in a useful manner)
Click to expand...
Click to collapse
So when you throttle the cpu, does it work on the slower or faster quad core?

Best AMD processors for performance

The past couple of years has seen AMD gain a better grip on the CPU market with its Ryzen series. While the Ryzen 3000 series of processors competed strongly against Intel last year, the latest generation has become a favorable choice of many thanks to the excellent performance. Gamers, PC building enthusiasts, and even professionals prefer going for the Ryzen 5000 series instead of Intel. One of the reasons for that is AMD’s Zen 3 architecture based on the 7nm node, whereas Intel is still stuck on its 14nm architecture for the past six years.
Let’s check out the best AMD CPUs for performance
AMD Ryzen 9 5950X
AMD continues to offer high-end desktop (HEDT) class processors to mainstream users with the Ryzen 9 5950X. Featuring 16-cores and 32-threads, it is one of the most powerful processors from the company. It isn’t affordable by any means especially when you look at the $799 price tag, but compared to other competitive HEDT processors, this is actually a really good price. If you don’t want to jump over to the Threadripper series, then this is your best bet.
Clock speeds: 3.4GHz – 4.9GHz
16-Cores, 32 Threads
64MB L3 Cache
PCIe 4.0
105W TDP
~$920
Buy from Amazon
AMD Ryzen 9 5900X
Sitting below the 5950X is the 12-core Ryzen 9 5900X that gives Intel’s latest Core i9-11900K a run for its money. It’s an incredibly powerful processor for gaming and creative workloads, at the same time it manages power and thermals more efficiently thanks to the 7nm process. The processor delivers more performance per watt consumed, compared to the 8-core 11900K. The only issue is that the Ryzen 9 5900X is difficult to get hold of and is currently selling more expensive than AMD’s suggested price.
Clock speeds: 3.7GHz – 4.8GHz
12-Cores, 24 Threads
64MB L3 Cache
PCIe 4.0
105W TDP
~$680
Buy from Amazon
AMD Ryzen 7 5800X
It is neck-to-neck when comparing the Ryzen 7 5800X with Intel’s Core i7-11700K. While it is slightly more expensive than the Intel counterpart, it's worth it paying extra as it offers faster gaming performance and almost the same performance when it comes to core-CPU-based tasks. There is also the additional benefit of the 5800X’s lower power consumption, which means it can reach its full performance potential even on less expensive motherboards.
Clock speeds: 3.8GHz – 4.7GHz
8-Cores, 16 Threads
32MB L3 Cache
PCIe 4.0
105W TDP
~$400
Buy from Amazon
AMD Ryzen 9 5980HX
The newly launched AMD Ryzen 9 5980HX laptop CPU is part of AMD's 5000 series 'Cezanne' generation. It is targeted towards high-performance laptops. The octa-core processor comes with a base clock speed of 3.3GHz and a boost clock of 4.8GHz. The TDP is rated at 45W which is quite impressive for a powerful processor like this. According to AMD, thanks to the Zen 3 architecture, the new 5000 series has made significant leaps in IPC compared to the previous generation with an average IPC gain of 19-percent.
Clock speeds: 3.3GHz – 4.8GHz
8-Cores, 16 Threads
16MB L3 Cache
PCIe 4.0
35-45W TDP
Beast cpu still in 2022

Categories

Resources