Greetings all. Allow me to introduce myself and explain my noob credentials before making my main point. I am an electronic engineer by day, and I am not a software person - assembler is as high-level as I am comfortable programming. My phone hacking credentials extend only as far as rooting and unlocking one galaxy S, and installing cyanogen 7 on a ZTE Blade, another Galaxy S and my personal Defy.
The blade was the first cyanogen install I did, and I did not find it to be spectacularly fast. However, on a 600MHz processor, was I expecting too much?
Next up was my defy. Oh my word... This is absurdly fast, as we all know. My phone scores 2100 in quadrant (for whatever that is worth - we know quadrant has its flaws).
Finally, I installed CM7 on a Galaxy S, and was marginally disappointed. Yes, it was a bit slicker than touchwiz, but not much. Quadrant of 1615, despite the significantly greater 3d capability of the SGS over the Defy.
So, some questions in my mind - Why does the Defy benefit so much, and how does it end up significantly outperforming the SGS when on paper they are pretty even with a slight advantage to the SGS - both 1GHz, both 512MB RAM, SGS has SGX540 and lower res screen (fewer pixels needing fill rate can potentially improve graphics framerates).
What it comes down to, I believe, is the RAM. The SGS has a strange memory architecture for its 512MB, with 128MB as "onedram" offering ~10GB/s speeds, required by the SGX540 and apparently solely dedicated to it. The remaining 384MB is LPDDR (1.6GB/s) and is what is available to the system (with CM7 reporting 342 total in about phone).
In contrast, the defy has 512MB LPDDR2 (3.2GB/s), which is all available to the system with dynamic allocation to the GPU when required. CM7 reports 477.
Therefore, in situations where the GPU is not being heavily used (which is most of the time for general system use) the Defy has much more and much faster memory for the system.
So it seems that the key for CM7 performance is memory speed and capacity, not necessarily processor power.
Should ICS come to the SGS with GPU-accelerated UI then I think the SGS would suddenly have a big advantage over all other single-core phones.
Quadrant Score for CM7 on the Defy is absolutly not relevant.
I should not use this as point of comparison ... not at all.
Does your Defy run slicker than your GS2 ? Does apps run faster ?
Do you have same settings on the both device ?
It is an SGS (i9000), not a GS2 (i9100).
Yes. The defy does seem to run a bit slicker and more fluidly than the SGS.
Also for me, defy + cm7 is way better then my ex-dhd....
Le_Poilu said:
Quadrant Score for CM7 on the Defy is absolutly not relevant.
I should not use this as point of comparison ... not at all.
Does your Defy run slicker than your GS2 ? Does apps run faster ?
Do you have same settings on the both device ?
Click to expand...
Click to collapse
Of course it would give some insight even if it's flawed...
Sent from my MB525 using xda premium
LauncherPro and browser opera use gpu for its work. If you are right, sgs would benefit greatly by using those instead other cpu based.
Related
http://androidandme.com/2010/06/new...s-all-other-android-phones-in-gpu-benchmarks/
I found this to be interesting and a bit odd that we have all of this talk about the "Mid-Range" Cpu at 600 Mhz but no talk about the GPU. Does anyone have some information on the GPU and if they are receiving similar scores in this category?
Can we officially stop calling this a "Mid-Range" phone if this turns out to be true?
Remember the phone is only 320x480 res vs the N1, Droids, etc 480x800+. That alone will greatly improve the performance (if same GPU).
IMO a Large res display is required for a high end phone too.. though a good one its still a mid level.
Please post this is the correct section.
mesajoejoe said:
Please post this is the correct section.
Click to expand...
Click to collapse
This question is very relivant to the development of this device.
This all goes back to my mantra, lastest dont always mean the greatest. Hell, put it in this perspective...any woman would sacrifice an inch for better performance. I would much rather something do what its supposed to do and do it with blazing colors, than have something look pretty with all the bells and whistles and hiccup in comparison. Ive played with all the android phones from tmobile thus far, and the slide is by far my fav.
My reasoning behind throwing this in the development section was to hopefully find someone who could shed some more light on the GPU in this phone. I am sure that the pixel differences between this device and others are part of it but it seems as if more is going on in the background with this phone. Here are some questions this benchmark test brings up for discussion:
1.Is this GPU Open GL ES 2.0 capable?
2.Has Qualcomm opened up more possibilities with this GPU drivers compared to other phones?
3.It seems that this is an Adreno GPU, but which one? The Nexus has an Adreno also.
4.How much of this phone is hardware accelerated compared to other phones? I remember taking up the RAM from my G1's GPU to help run the main components of the phone, maybe this phone has a bit more help built in to it.
Let's all run to tmo and drop one so we can see what's inside! I for one just compaired it to the mt1.2 and it blows it away! The wife loaded up her baseball flash site and flipped at how fast it rendered the stuff. The screen res ain't got nothing to do with that. Maybe they fibbed on the cpu info so we wouldn't know. Or they got themselves one smart dev for software building.
Has anyone tried linpack tests yet?
mejorguille said:
Has anyone tried linpack tests yet?
Click to expand...
Click to collapse
Linpack is 2.6+ for the one i tested at the store.I think the CPU is honestly down-clocked from what it really is, for all we know it could go up to 800Mhz or 1GHz. When linpack ran there were some b/g apps open.
Re: Mid-Range? Whatever...
If you look at the power.xml in github for Wes Garners's developments on this phone you will see the option for 998mhz as the high end for this cpu. I'd imagine overclocking it will bring at least 1.3 to the table. No info on the GPU yet.
Sent from MTS using Tapatalk
Re: Mid-Range? Whatever...
mejorguille said:
Has anyone tried linpack tests yet?
Click to expand...
Click to collapse
Yes. Mflops came out around 2.2. It needs to be tweaked severely but id assume this is due to the bloated processes and underclocking of the CPU.
Sent from MTS using Tapatalk
i dont understand why everyone is hating on this phone, my g1 is perfect. does what i need from android. and i like it over all. i never have problems with my processor. so this phone is a huge plus to me. although id love to make the sence ui on it a lil differnt, like the normal sence
mtjnkee said:
Yes. Mflops came out around 2.2. It needs to be tweaked severely but id assume this is due to the bloated processes and underclocking of the CPU.
Click to expand...
Click to collapse
Really? That's very low in my opinion when a droid, also running a 600 Mhz get much higher, and my rooted g1 with cyan kernel 2.6.34 get 2.3-2.4 on linpack.
Wikipedia says that the mytouch slide has a MSM7227 processor compared with Snapdragons which are QSD8xxx series. Some google searches found that the 7227 is for "cheap", sub $150 smartphones. Here are the specs for the processor:
Code:
The platform -- known as the MSM7227 -- is apparently designed with higher-power,
lower-cost handsets in mind, offering HSDPA / HSUPA radios, a 600MHz CPU, 320MHz
application DSP, 400MHz modem processor, hardware-accelerated 3D graphics, Bluetooth 2.1,
and GPS support integrated in a 12mm x 12mm chipset. The MSM7227 can also supposedly handle
8 megapixel camera and 30 FPS WVGA encoding / decoding with ease, and is capable of
supporting Android.
1
2
mejorguille said:
Really? That's very low in my opinion when a droid, also running a 600 Mhz get much higher, and my rooted g1 with cyan kernel 2.6.34 get 2.3-2.4 on linpack.
Wikipedia says that the mytouch slide has a MSM7227 processor compared with Snapdragons which are QSD8xxx series. Some google searches found that the 7227 is for "cheap", sub $150 smartphones. Here are the specs for the processor:
Code:
The platform -- known as the MSM7227 -- is apparently designed with higher-power,
lower-cost handsets in mind, offering HSDPA / HSUPA radios, a 600MHz CPU, 320MHz
application DSP, 400MHz modem processor, hardware-accelerated 3D graphics, Bluetooth 2.1,
and GPS support integrated in a 12mm x 12mm chipset. The MSM7227 can also supposedly handle
8 megapixel camera and 30 FPS WVGA encoding / decoding with ease, and is capable of
supporting Android.
1
2
Click to expand...
Click to collapse
Well thats because the droid has a better processor (not speed but type) and the G1 is rooted and all the bloatware cut plus probably jit enabled plus swap/comcache. Besides that score seems rather low. I have a G1 and the MTS is way faster. I dont need a benchmark test to tell me that. Its definately as fast as my friends droid may be faster.
I know its been said before but it has the same gpu as the n1 for me that makes it all the proccessor i need
http://forum.xda-developers.com/showthread.php?p=5719829
neva dinova said:
I know its been said before but it has the same gpu as the n1 for me that makes it all the proccessor i need
http://forum.xda-developers.com/showthread.php?p=5719829
Click to expand...
Click to collapse
awesome
Z51 said:
awesome
Click to expand...
Click to collapse
either way, its better than the g1 and mt3g and even the 32a mt3g's.. when it gets rooted, this device will fly even moreso than it already does. bloatware be damned, we'll have it working wonders and fixing aids!*
aids may not actually be fixed by rooted slide
I know right. This morning I was looking at the running apps and saw the following that ran even if you cleanly booted it up:
* Visual Voicemail (don't get many VMs, so it's useless to me)
* Faves (useless)
* Swype (useless to me)
* IM (why is this even running?)
* AppPack (useless)
* My Account (useless)
Each of these apps take 20 MB or more.
marvinp said:
I know right. This morning I was looking at the running apps and saw the following that ran even if you cleanly booted it up:
* Visual Voicemail (don't get many VMs, so it's useless to me)
* Faves (useless)
* Swype (useless to me)
* IM (why is this even running?)
* AppPack (useless)
* My Account (useless)
Each of these apps take 20 MB or more.
Click to expand...
Click to collapse
with a clean rom, we're gonna have so much extra space we'll have no need for a2sd
I just got the G2 and I love it. I came from a vibrant which was lagging and freezing and still had 2.1. Regret buying that phone. But this phone is a beast. My friend got the mytouch thinking it was faster.
I still think the G2 is really fast, but which processor is faster.
If you can explain in detail that'd be great!
Sent from my T-Mobile G2 using XDA App
Don't they have the same processor? Except the g2 is underclocked while the MT4G isn't?
Sent from my T-Mobile G2 using XDA App
nivlac978 said:
Don't they have the same processor? Except the g2 is underclocked while the MT4G isn't?
Sent from my T-Mobile G2 using XDA App
Click to expand...
Click to collapse
that's what I thought too
Cool stated somewhere in one of his previous OC threads that they were slightly different. Ill leave it at that since explaining is beyond my scope of knowledge.
Sent from my HTC Vision using XDA App
HTC G2 uses the MSM7230 as compared to the HTC MyTouch 4G's MSM8255.
"MSM7x30, MSM8x55 and QSD8x50A - all 3 variations share the same design, differs only by the clock speeds. These 3 variations run at 800MHz, 1.0GHz and 1.3GHz respectively. Each of these features Adreno 205 GPU which runs at significantly higher performance level than Adreno 200 found in the 1st generation Snapdragon processor."
Source: http://smartphonebenchmarks.com/for...-gen-snapdragon-processors-how-fast-are-they/
joackie27 said:
HTC G2 uses the MSM7230 as compared to the HTC MyTouch 4G's MSM8255.
"MSM7x30, MSM8x55 and QSD8x50A - all 3 variations share the same design, differs only by the clock speeds. These 3 variations run at 800MHz, 1.0GHz and 1.3GHz respectively. Each of these features Adreno 205 GPU which runs at significantly higher performance level than Adreno 200 found in the 1st generation Snapdragon processor."
Source: http://smartphonebenchmarks.com/for...-gen-snapdragon-processors-how-fast-are-they/
Click to expand...
Click to collapse
More than likely this means that the silicon on each device is QC'd to a different tolerance. IE the minimum possible speed. As everyone knows each silicon chip has a range and it will meet a minimum clock speed stable. It has been shown that most G2's can be OC'd to 1.4 ish. We don't have kernel source yet for the Glacier. When we do we will see if it is better or the same.
The MT4G also has more RAM. 768MB vs G2's 512MB.
Superfrag said:
The MT4G also has more RAM. 768MB vs G2's 512MB.
Click to expand...
Click to collapse
So what you're saying is... The MT4G is faster than the G2?
jjuice525 said:
So what you're saying is... The MT4G is faster than the G2?
Click to expand...
Click to collapse
More RAM doesn't directly translate to more speed.
More RAM very nearly directly translates to more speed for 99% of usage time because there are less reads from flash memory. I don't know how noticable it would be though. Certainly not the marked improvement seen on PCs with normal hard drives.
yes but the sense ui takes more cpu/ram to run smoothly. atleast in theory. plus the mytouch hasnt been oc'd yet so we will see.
Anomaly said:
More RAM very nearly directly translates to more speed for 99% of usage time because there are less reads from flash memory. I don't know how noticable it would be though. Certainly not the marked improvement seen on PCs with normal hard drives.
Click to expand...
Click to collapse
You are wrong.
The more RAM means you can load more stuff in RAM. Example in PC, You can open more Internet Explorer window.
But the speed is the same. Example.. we kill all background apps and both G2 (1ghz) and MT4G(1Ghz) will just run quadrant alone.. they should perform the same since the speed of their RAM are equal regardless of the amount.. example 512 vs 768.
The MT4G needs more RAM because the Sense UI is composed of useless bloatwares.
Na, MT4G won't run faster, but it'll be able to run more stuff at the same time.. like people above have explained.
But I really don't think we need that much RAM, currently.
Superfrag said:
Na, MT4G won't run faster, but it'll be able to run more stuff at the same time.. like people above have explained.
But I really don't think we need that much RAM, currently.
Click to expand...
Click to collapse
true. but honestly, the only moment where you might be able to notice a slight speed difference is with numerous, and i mean a hell of a lot, of apps open in the background without killing them
hey guys!
i'm kind of a noob, so forgive me if i write something stupid i've watched a few videos of other phones being benchmarked with quadrant, and at the part where the planet and it's moon shows, other mid-range/high-end phones have no problem playing it flawlessly, and with my defy i get around 10-11fps, and some textures don't even show! is that because of a poorly optimised gpu driver, or what? any ideas? do other phones with sgx530 gpu do that too, or just the defy?
Sent from my MB525 using XDA App
It's all right, i get 9 fps at stock speed (800mhz) and up to 12 depending on the OC applied.
Maybe Defy is not so brilliant in that specific test but the overall performances are great compared to other phones with similar price.
Also consider that many devices have huge score on quadrant benchmark but they're actually not so fast.
Some of them also have stagefright enabled, as a result the quadrant benchmark fails to decode h.264 and gives huge scores...
So, don't trust quadrant too much, our gpu and drivers are ok!
thx! actually i couldn't be happier with this phone, it's by far the best in it's price range (even if we exclude it's 'life-proof'-ness and great overclocking/undervolting capability). this was just something that caught my attention. anyway, thanks again!
I saw the review of the ics gs2 and at the end i saw some benchmarks... on the cpu side the nexus seems to hold its own againts the xperia s which on paper has a faster cpu snapdragon 3 1.5Ghz compare to omap 4460 1.2.. also gs2 seems to be slightly slower in the cpu benchmarks compared to Gnex.... What do you think?
http://www.gsmarena.com/samsung_galaxy_s_ii_ics_vs_gingerbread-review-737p4.php
Sent from my Galaxy Nexus using XDA
ray3andrei said:
I saw the review of the ics gs2 and at the end i saw some benchmarks... on the cpu side the nexus seems to hold its own againts the xperia s which on paper has a faster cpu snapdragon 3 1.5Ghz compare to omap 4460 1.2.. also gs2 seems to be slightly slower in the cpu benchmarks compared to Gnex.... What do you think.
http://www.gsmarena.com/samsung_galaxy_s_ii_ics_vs_gingerbread-review-737p4.php
Sent from my Galaxy Nexus using XDA
Click to expand...
Click to collapse
gs2 is surely faster on graphics (it has a more advanced gpu, though clocked slower), but it has less screen resolution.
on the cpu side they should be pretty equivalent.
i don't know about the xperia.
In real world usage my SGS2 was significantly faster than my GNEX in every department barring taking a picture.
I wouldn't trust that much on benchmarks.
What matters is how polished the stock roms are for each device.
I love how the nexus keep running smooth even after installing a ton of apps, never experienced the same with other devices.
While the MSM8260 (Xperia S) is clocked faster, it has a single channel memory interface while the TI-OMAP 4460 (Galaxy Nexus) has a dual channel interface.
Also beware of those benchmark tools. Most of them are crap. For example they use the phone's resolution to do 2D/3D benchmark which is not a fair way to compare.
biffsmash said:
In real world usage my SGS2 was significantly faster than my GNEX in every department barring taking a picture.
Click to expand...
Click to collapse
Was it significantly faster in getting updates?
Benchmarks are cool in a sense, but I've started ignoring them. My GNex has been really fast, even with Window and Transition Animations. There are exceptions. I ABUSE my Android devices with heat (nature of my work environment). When it gets hot, it does lag. At the same time, the CPU is also scaling itself back. When I get the rare lock'n'reboot it's pretty much always due to heat. Typically, flinging through settings, menus, apps, etc is really fast. Right now I have wireless tethering on and I'm streaming audio through Tune In. My battery is almost 120F. It's slower. Any device would be. If I pick up my Nook running CM7, it'll be slower right now too due to heat.
Benchmarks mean very little to me now. I mean, what's the point? What are you gonna do, brag about a superficial number? Get it engraved on your belt buckle? I think they can be an indication of certain aspects of a device, but benchmarks do not necessarily tell someone what kind of user experience that device is going to provide.
If I pick up my favorite device and I can do what I want to do quickly and smoothly then I'm fine with it. Even OCing a high end device is something I've never been able to tell a difference with. Now my Nook, yeah, an overclock worked wonders. On the GNex, pish, nah. Remember too, a lot of these 1.5Ghz devices are pushing a bloated framework too. Beyond that, ICS is going to have it's own limitations, especially when it has bugs to get worked out.
It can kinda be like cramming 16GB's of Ram into a PC running 32 bit Windows 7. The hardware addition is irrelevant and while you can brag about 16 gigs of ram to all of your friends, it's a pointless feature of your system.
one thing for sure is that software optimization means much more than hardware differences between those 3 phones.
thus, it depends on what you're going to run: stock, aosp, cyanogen, miui? that's the question.
zapek666 said:
While the MSM8260 (Xperia S) is clocked faster, it has a single channel memory interface while the TI-OMAP 4460 (Galaxy Nexus) has a dual channel interface.
Also beware of those benchmark tools. Most of them are crap. For example they use the phone's resolution to do 2D/3D benchmark which is not a fair way to compare.
Click to expand...
Click to collapse
This. I think one of my favorite aspects of the nexus is its dual channel memory. So important for performance.
biffsmash said:
In real world usage my SGS2 was significantly faster than my GNEX in every department barring taking a picture.
Click to expand...
Click to collapse
To me it seems exactly the opossite.. my gnex is faster than my gs2.... except in gaming.. the animations on the gs2 are different, than on the stock ics...
Sent from my Gnex
My nexus is much faster at browsing than my wires gs2, and my work iphone4s..what? I didn't get to choose!
Sent from my Galaxy Nexus using xda premium
I've used both the tmobile s2 and now the nexus....I'm a big time xda guy and root user so as the developers went to the nexus, I went to the nexus! Lol! If the s2 gets stock ics, it'll be just as good but that'll be a long time! We nexus users have had ics for months already!
Sent from my Nexus Prime
Simer03 said:
I've used both the tmobile s2 and now the nexus....I'm a big time xda guy and root user so as the developers went to the nexus, I went to the nexus! Lol! If the s2 gets stock ics, it'll be just as good but that'll be a long time! We nexus users have had ics for months already!
Sent from my Nexus Prime
Click to expand...
Click to collapse
We've had a pretty stable CM9 for the i9100 for quite a while now. But it still can't compare to the Galaxy Nexus....can't wait for mine to come
Just swapped over from the t989 gs2. The nexus is much smoother. We were always telling ourselves that our phones were better because of the supposedly better hardware... There were many threads Lol. Now that I have it I can see the experience doesn't even compare. Never buying anything touchwiz again
I am a big fan of benchmarks but are there are really important
Please say what Kenel are you using.
No.
-----
I would love to help you, but help yourself first: ask a better question
http://www.catb.org/~esr/faqs/smart-questions.html
It really depends what you find important and what's useless. Its just for the certain few enthusiasts who like tweaking their devices and testing out different values to achieve a higher benchmark score and brag about it. I'm one of them although for the Note 2 I see little or no difference in performance when comparing between an over clocked and a stock setting.
Sent from the Rabbit Hole
Some people are disagree with overclock some not, benchmark are high only with overclock but if they are not important what are the reason?
I personally think that having a benchmark with a phones overclock settings turned on is the best way to see the true strength of a phone.
Some phones being overclocked do not even come close to other phones with out it being overclocked
Sent from my SPH-L900 using xda app-developers app
So overclock or no overclock?
What you prefer?
When it comes to mobile phones, be it a ultra top end smartphone, absolutely not.
Swyped from my GT-N7100
for me, benchmark is only for fun and push it to the best performance of Note 2.
but for daily i still overclocked, 1.8GHz
it was stable and response very well..
Good way to compare devices. Although I do not overclock for my daily usage.
They are not for everyday thing but they tell their stories. For example I am a GPU comparison addict. I believe the GPUs are the first aging parts of the chipsets and if the GPU is not enough no matter what CPU or phone it is, it will fade away quickly. So when I am choosing a product I take one of the Nexus devices as a reference. For my Note 2, my referance was Nexus 7. SÄ°nce it will be getting at least 1.5 year support (the worst scenerio) Note will run any game for that time period.
So GLbenchmark comes in. Then it lets us to compare Nexus 4 and Note 2. What we can learn from it is, for example, Mali 400 is better for higher resolutions since 1080p and 720p results are same. Mali 400s quad core processors are so powerfull that they can not be effected by MSAAx4 or higher resolutions where Adreno 320 tenst to slow down however it has a bottle neck on the pixel processor side and it gets stuck on the Egypt 2.5 test no matter how much we OC it. So newer games will be a problem for Note 2 we can say. For the more older-coded games which are still close to the Egypt 2.1 side Adreno barely has %5-7 advantage over the Mali 400 so no problem for a long time.
Thanks for reply:good: