So I flashed Skyraider today, and my quadrant score dropped 100 points from stock sense. What the hell is going on here....
The Black Droid said:
So I flashed Skyraider today, and my quadrant score dropped 100 points from stock sense. What the hell is going on here....
Click to expand...
Click to collapse
Sitting at 1032. With the stock ROM, I was getting around 1120.
Why are these scores so low???
Q scores || ROMs
Quadrant scores vary by ROM and kernel, and by different kernels in the same ROM.
Those scores are expected to vary.
Quadrant scores, don't take those to heart... It's all about feel. I've run roms where they might score low, but feel and move quick.
Sent from my ADR6300 using XDA App
I just don't understand how some of these guys have anywhere from 2700-3300 as their quad score. I would imagine their phones are running insanely fast
Maybe, they are most likely overclocking. Some phones don't like going too high. Like on mine, I can't go over 1.113, or my phone slows to a crawl and locks up/reboots.
Sent from my ADR6300 using XDA App
The Black Droid said:
I just don't understand how some of these guys have anywhere from 2700-3300 as their quad score. I would imagine their phones are running insanely fast
Click to expand...
Click to collapse
Where have you seen people post these scores for their Inc?
I'm willing to bet good money that if you are seeing people post scores that high, they are either:
A. Not running Quadrant from an Inc but a much more powerful device (like an Android tablet)
B. Faking the results in Photoshop (or just lying about the number if no screenshot is provided)
C. Running a setup that doesn't actually run through the Quadrant test properly. For example, Quadrant didn't run properly on Gingerbread roms at first, and the scores that it would produce were extremely erratic. Sometimes they would be extremely low, sometimes they would be extremely high, but they weren't accurate or repeatable.
D. Running some sort of insane setup that is only stable enough to finish a Quadrant run, and will never actually be used for anything
Frankly, even D is far-fetched.
I would be inclined to call a 100 point fluctuation in Quadrant scores insignificant.
If you really want to chase the highest benchmark scores, you'll need to overclock your CPU and run the system as lean as possible. That means uninstalling or disabling a lot of the things that make your life easier day-to-day.
Also, in case you haven't already seen it in your own testing, Quadrant scores are always lowest on the first run. If you press the back button and immediately start a new Quadrant run, you'll get a much higher score.
Like any unit of measurement, Quadrant scores do serve a useful purpose. But as is often the case when the score itself is seen by some as the end-goal, it is often misapplied.
The same can be seen in digital cameras and the megapixel arms race. Everyone wants to brag about how many megapixels their camera is capable of. Everyone wants the highest number of megapixels, assuming that more MP = better image. Few people realize what it actually means, or why it matters very little these days.
A lot of those people are overclocking to get really high scores and for all the reasons listed above (nice post!).
You really should not be looking to get that high on the incredible, you'll end up draining your battery like crazy. Around the 1,000 mark is great for playing higher-end games on the market as long as you aren't running a bunch of things in the background. Just about anything else you can think up of doing on your phone should run well, you won't have a sluggish device and you won't be killing your battery either.
If you do end up trying to overclock your phone or using a ROM or kernel combination that will give you a much higher score I don't think you'll notice any difference when doing anything on your phone, but your battery will drop much quicker.
Like other people have said, Quad scores don't matter much- take them lightly as you see them.
There is a lot that goes into that score. The highest score I could get today is 1656 but it was consistently in the Upper 1500s, I ran 5 tests.
My setup:
CM7 RC2
Incredikernel 03/06 OC to 1113Mhz/Performance Govenor
16Gb Class 2 microSD card.
If someone is using a class 4 or a class 6 card their i/o scores could be much higher than mine which would result in a much higher overall score than mine. Also keep in mind with Linux Kernels can very alot and that there are different types of task schedulers in them such as BFS or CFS which can have dramatic affect on the quadrent scores. Quadrent tends to score BFS kernels higher. So yeah I can believe people are hitting most of the score they post up. However byrong is right about it not being a setup you'd want to use on a daily basis. For me it causes random reboots, my phone gets hot and the UI becomes laggy after a little while also the battery drops like a brick.
My normal setup that I run on a daily basis is the same kernel uc to 803Mhz/smartass governor. It is extremely stable and is smooth as butter but my quadrent scores are only only in the 1100s with my high being 1244.
Its really not all about the score, if your happy with the performance who cares about the score.
Related
Look at this (from 1:44 on):
It's a quadrant benchmark run on a android port on the HD2. Graphics are really bad, but in the end it has approximately the same score as the benchmarking score of the Galaxy with the original firmware. I mean what is in a score? If I look at the beginning of the movie, the UI is very slow and not as responsive as the Galaxy
(BTW i got 55.7 FPS with the neocore benchmark on JM2)
This is not to say that I don't have deep respect for what the HD2-android development team is doing. Really amazing job. I just can't wait to get my HD2 back from repair.
appelflap said:
Look at this (from 1:44 on):
It's a quadrant benchmark run on a android port on the HD2. Graphics are really bad, but in the end it has approximately the same score as the benchmarking score of the Galaxy with the original firmware. I mean what is in a score? If I look at the beginning of the movie, the UI is very slow and not as responsive as the Galaxy
(BTW i got 55.7 FPS with the neocore benchmark on JM2)
This is not to say that I don't have deep respect for what the HD2-android development team is doing. Really amazing job. I just can't wait to get my HD2 back from repair.
Click to expand...
Click to collapse
Quadrant scores have been criticized for their non-descript breakdowns, at least on their free suite. Also, the fact that they chose the weighting of the scores, so should they chose 2D is equal to 3D weight, I don't know their formula (and for all I know, they give equal weighting to all or they give equal weighting to all test where the CPU has 12 tests and the 3D graphics has 4), but the fact that we, as users don't have access to their formula on their website is a bit unnerving.
Add to that the fact that many reviews and videos rely on it so heavily leaves users a bit misinformed. In reality, and thorough review should definitely run a custom test suite to give individual scores to:
CPU
Memory
I/O
2D graphics
3D graphics
That way users can compare what's important to them. The Galaxy S suffers from terrible I/O and the hacks that have given the fixes typically boost Galaxy scores to nearly double their rates, and it's majorly attributed to improving a bunk I/O score.
Totally agree. In addition, it would be really nice to know which benchmarked factors are responsible for which functions. For example it is really interesting to see how the hd2 performs before the user is running the tests. When the user is scrolling through the setting menu there is a very noticible lag. Given the fact that the total score is nearly the same as the scrore for the SGS, and thar the graphic score of the hd2 is bad in comparisson to the SGS, I would conclude that graphic performance is very important for the way the ui responds.
Sent from my GT-I9000 using XDA App
appelflap said:
Totally agree. In addition, it would be really nice to know which benchmarked factors are responsible for which functions. For example it is really interesting to see how the hd2 performs before the user is running the tests. When the user is scrolling through the setting menu there is a very noticible lag. Given the fact that the total score is nearly the same as the scrore for the SGS, and thar the graphic score of the hd2 is bad in comparisson to the SGS, I would conclude that graphic performance is very important for the way the ui responds.
Sent from my GT-I9000 using XDA App
Click to expand...
Click to collapse
From what I can tell, the HD2 got a decent score 'cos it was running Froyo. When we get bumped up to an official froyo build with JIT fully optimized, We should be top of the pile.
don't forget, android isn't working 100% on the HD2.
I personally think it's pointless comparing to a not complete port.
woops dbl post
alovell83 said:
Quadrant scores have been criticized for their non-descript breakdowns, at least on their free suite. Also, the fact that they chose the weighting of the scores, so should they chose 2D is equal to 3D weight, I don't know their formula (and for all I know, they give equal weighting to all or they give equal weighting to all test where the CPU has 12 tests and the 3D graphics has 4), but the fact that we, as users don't have access to their formula on their website is a bit unnerving.
Add to that the fact that many reviews and videos rely on it so heavily leaves users a bit misinformed. In reality, and thorough review should definitely run a custom test suite to give individual scores to:
CPU
Memory
I/O
2D graphics
3D graphics
That way users can compare what's important to them. The Galaxy S suffers from terrible I/O and the hacks that have given the fixes typically boost Galaxy scores to nearly double their rates, and it's majorly attributed to improving a bunk I/O score.
Click to expand...
Click to collapse
Even then though, it's possible to write a benchmark which wins constantly for any phone.
In regards to "terrible I/O", that might even be due to a bug in the FAT32 drivers. Yes you can benchmark it, but it wont mean much. The best way is to actually TEST the applications you need, rather than select a phone based on benchmarks. However, you are possibly best off looking at the component specs, because they ignore software bugs.
scrizz said:
don't forget, android isn't working 100% on the HD2.
I personally think it's pointless comparing to a not complete port.
Click to expand...
Click to collapse
But the topic is about "what's in a score". Maybe one can generally say that is pointless to compare devices this way. I think that such benchmark scores are only (a bit) relevant at the two poles of the benchmark score spectrum. Everything in between can be neglected due to the uninformed way sub-scores are evaluated.
You got 55.7 FPS on Neocore as the sgs has vertical sync enabled, the refresh rate on the sgs'es screen is 56 fps and thus you can only go up to 56 fps as the v-sync is on. This proves that the sgs is indeed a much more powerful device that is actually being held back. If you can disable the v-sync then you can get a higher fps score
appelflap said:
But the topic is about "what's in a score". Maybe one can generally say that is pointless to compare devices this way. I think that such benchmark scores are only (a bit) relevant at the two poles of the benchmark score spectrum. Everything in between can be neglected due to the uninformed way sub-scores are evaluated.
Click to expand...
Click to collapse
I just read in a post that the Galaxy S gets a 0 on the 2D score:
"JIT isn't fully enabled in the current froyo versions, and quadrant, frankly, is bull**** (for exmple, 2d acceleration gets the same weight in the final result as 3D. Due to the fact that the SGS doesn't have a dedicated 2D accelerator, quadrant doesn't try to use the cpu- it just gives a round zero in that part)"
I can't confirm this, but that definitely seems like a terrible set-up, seeing as how I'm pretty sure I have games run in 2D, so to say that it can't do it just seems wrong regardless of if the SGS has a dedicated 2D accelerator or not (so if you aren't testing the way it performs in real-world, why are you testing?)
http://forum.xda-developers.com/showthread.php?t=737787&page=3
Qazz~ said:
You got 55.7 FPS on Neocore as the sgs has vertical sync enabled, the refresh rate on the sgs'es screen is 56 fps and thus you can only go up to 56 fps as the v-sync is on. This proves that the sgs is indeed a much more powerful device that is actually being held back. If you can disable the v-sync then you can get a higher fps score
Click to expand...
Click to collapse
It isn't really being held back - the screen can't display more than 56 fps as you say, and it wouldn't really be visible even if it could. Disabling v-sync isn't really that important, we need a benchmark that can actually use the advanced features in the SGS GPU (Neocore just pushes a fairly small amount of polygons with no real extras.) Using current 3D benchmarks to benchmark the SGS is like using quake 1 to benchmark the brand new ATI/nVidia cards.
The benchmark is what is at fault here, not the device
RyanZA said:
It isn't really being held back - the screen can't display more than 56 fps as you say, and it wouldn't really be visible even if it could. Disabling v-sync isn't really that important, we need a benchmark that can actually use the advanced features in the SGS GPU (Neocore just pushes a fairly small amount of polygons with no real extras.) Using current 3D benchmarks to benchmark the SGS is like using quake 1 to benchmark the brand new ATI/nVidia cards.
The benchmark is what is at fault here, not the device
Click to expand...
Click to collapse
I don't want to speak for the other poster, and I agree with your premise, however, it isn't actually solving the issue at hand. Better FPS wouldn't be noticed, however, it would give a better score and, more importantly, indicate it's potential. So, getting 56FPS isn't doing the phone any justice within the score, which is what reviews are using, giving it an artificially low score, and putting it more in line with units that can't compete on higher end games. So, when a site like anand pushes 150FPS on a game, I know that means that their rig is entirely too powerful for the game in question, but it still means something when you compare it to the lower end graphics card that only gets 90...then when they run Crisis you see these results play out more with differences that we can notice with the eye.
I think the HD2 gets that score because, as I can see in the video, the CPU tests run faster compared to my SGS, probably because of Froyo, and I know, from the time I had the Diamond and the HD2, that the internal memory and RAM are very fast. Sadly SGS has a slow internal memory, atleast when used by the phone`s software, when copying from PC is faster than my class 6 microSD. Luckily, we have mimocan`s fix. Hope this will be fixed in future FW`s.
NexusHD2 with-FRG83D V1.7 with hastarin r8.5.1 On my HD2 got 1920 in quadrant,31.5 on neocore, and 37 on linmark.
The lag might be because you are using launcher pro, I use launcher pro and sometimes it makes the the lock lag on my phone but it doesn't happen when I use the default lock also if you have alot of Widgets on your screen it will cause lag also
appelflap said:
Look at this (from 1:44 on):
It's a quadrant benchmark run on a android port on the HD2. Graphics are really bad, but in the end it has approximately the same score as the benchmarking score of the Galaxy with the original firmware. I mean what is in a score? If I look at the beginning of the movie, the UI is very slow and not as responsive as the Galaxy
(BTW i got 55.7 FPS with the neocore benchmark on JM2)
This is not to say that I don't have deep respect for what the HD2-android development team is doing. Really amazing job. I just can't wait to get my HD2 back from repair.
Click to expand...
Click to collapse
same galaxy s scores 6000+ in quadrant with custem roms
The HD 2 is a better fit for quadrent then the sgs as quadrent was made for the snapdragon processor which the hd2 has and the sgs does not. Comparing apples to orenges in an apple juice contest doesn't really prove much. Use real life feel. If you care about the scores a rom can be made to get you over 3000 quad score but is laggy as hell. Don't believe me? Look at my sig
interesting... I was using quadrant to see how a stock xxjvo and gingerreal compared. Surely that would indicate a real speed difference and not just be some kind of "hack" ?
zelendel said:
The HD 2 is a better fit for quadrent then the sgs as quadrent was made for the snapdragon processor which the hd2 has and the sgs does not.
Click to expand...
Click to collapse
That's right.
HD2 uses two android OS :
- Cyanogenmod, that is faster than our samsung os..
- Nexus one's port to HD2, greatly optimized by google...
It's really fast. I upgraded my father's HD2 last month, replacing windows in the NAND with CM7. It really makes a big change, the phone is like brand new ^^
http://forum.xda-developers.com/showthread.php?t=1012556
Quadrant is pretty flawed. And I say that being someone who had a phone (before modifications) that was mid-range in Quadrant (Galaxy S), and having a phone that's right top of the heap (Galaxy S II)
One of my coworkers has a tmobile vibrant with some lag fix according to him.. he did a quadrant benchmark right in front of me and it was showing 2500 plus everytime.. Im very curious as to what is making his phone so fast. And can it be dont to ours. Hes not running a custom rom or overclocking. Im only getting 1030 with mine clocked at 1.2ghz. Any Ideas? I couldnt get into too much details with him yesterday and I dont know whens the next time ill see him..
If you were to look at a test break down you would see generally all the scores are identical or the epic a little ahead except in the read/write area. The scores from their read/write are just inflating their overall score. It's a issue with quadrant and how it handles its overall score. Basically it just makes the system easy to abuse/cheat. So I wouldn't worry much about the difference in your score and his.
Sent from my Samsung Epic
The reason other Galaxy S phones score high in quadrant is because of the lag fix they use. The lag fix mounts a different file system on the phone with DRAMATICALLY increases read-write times. That portion of the quadrant benchmark gets inflated beyond reason. Using this game technique, Cyanogen was able to score more than 3000 on a snapdragon phone.
All of the Galaxy S phones have the same processor. Also, quadrant is a terrible benchmark. It's the most over-quoted and abused benchmark for android phones
Ahh ok.. thats good to know.. so what would be a better benchmark to use? Linpack?
jok3sta said:
Ahh ok.. thats good to know.. so what would be a better benchmark to use? Linpack?
Click to expand...
Click to collapse
Linpack is good for measuring raw CPU processing power... but only on devices running the same version of android. Phones with 2.2 will score insanely high due to the JIT compiler. For example, a snapdragon phone with Froyo can score ~40 Mflops. A snapdragon phone with eclair scores around 7 Mflops. Does Froyo make the phone run 5-6X faster? Hell no. In some cases, the difference is almost unnoticeable to the human eye.
Here is a rundown of what I believe to be the pros and cons of various benchmarks:
Linpack
Pros:
- Good for measuring CPU processing power on the same version of Android
- Great tool for measuring the performance gain from overclocking
Cons
- Scores are boosted unreasonably by Froyo's JIT compiler on snapdragon phones
Quadrant
Pros:
- Great tool for measuring the performance gain from overclocking
- Decent tool for measuring 3D graphics performance (just pay attention to FPS, not the end result)
- Decent tool for measuring 2D graphics performance (again, look at FPS)
- The paid version ("Quadrant Pro" I believe) shows which parts of the benchmark contributed to the score. Easier to spot the inflated CPU or I/O inflation
Cons:
- I/O portion isn't valued as much as others, but can boost scores beyond reason via exploits, hacks, fixes, etc.
- CPU portion is inflated on phones running 2.2. A Nexus One is not faster than any Galaxy S, Droid X, Droid 2, etc.
Neocore
Pros:
- Good tool for measuring graphics processing power
Cons:
- Graphics are not intense enough to push the power of very fast GPU's. Some phones will hit their FPS limit
- Only measures graphics processing power.
Nenamark1
Pros:
- Great tool for measuring graphics processing power
- Effects are advanced enough to show the performance of faster GPUs in relation to phones with lesser GPUs.
Cons:
- Only measures graphics processing power.
Sweet thanks for all the info man..
Agreed, this is great info thanks. I think the quadrant score is the most quoted becuase it provides a very easy to read graph built in with it for instant comparing/gratification. I guess I am gonna start going by linpack and nenamark1.
hydralisk said:
Linpack is good for measuring raw CPU processing power... but only on devices running the same version of android. Phones with 2.2 will score insanely high due to the JIT compiler. For example, a snapdragon phone with Froyo can score ~40 Mflops. A snapdragon phone with eclair scores around 7 Mflops. Does Froyo make the phone run 5-6X faster? Hell no. In some cases, the difference is almost unnoticeable to the human eye.
Click to expand...
Click to collapse
Linpack is ok for when your using same CPU comparison, different CPU's can cause issues...The reason why snapdragon gets scores of 5-6x is for some reason the snapdragon utilizes the VFP rather then using raw processing power..aka snapdragon cheats on the Linpack.
In reality our I/O scores should be a lot higher then it is as even in the Epic some of samsung's crappy file system still exists. But not as high as the lagfixed Vibrant of course.
Quadrant Pro is probably best indicator out of them all(The non-pro version is pretty much useless unless your comparing the same phone)...the con of having 2.2 show is higher is expected as it is a measure of efficiency of JIT in comparison to the current. The OS always played a role in Benchmarks so it is expected.
it can be faked by using a different partition to test on. IIRC the data partition making the speeds much faster than they should be so be careful when accepting those high scores
rjmjr69 said:
it can be faked by using a different partition to test on. IIRC the data partition making the speeds much faster than they should be so be careful when accepting those high scores
Click to expand...
Click to collapse
It is not exactly faking it..as you are increasing performance..thing is you cannot see at what it performs well at unless you see the individual scores from the Pro version....
I know about the differences in benchmarks and how they arent set up for dual core, but I just ran smartbench 2011 and my gaming score is off by 1000 points on a stock xoom, I am rooted and running stock kernel. I am not sure why, maybe something is wrong with it.
My quadrant scores are lower than my dx but my linpack score is 64mflops! Don't know why our quadrant scores are so low but I'm having the same problem.
Sent from my Xoom using XDA App
Off from what? A phone? Synthetic benchmarks say almost nothing about real world performance, and they will always be different with devices at different resolutions.
A 1280x800 tablet will always score unusually low on a graphics benchmark that scales to resolution compared to a phone.
Usmc7356 said:
My quadrant scores are lower than my dx but my linpack score is 64mflops! Don't know why our quadrant scores are so low but I'm having the same problem.
Sent from my Xoom using XDA App
Click to expand...
Click to collapse
Quadrant also places a lot of weight on the filesystem, and my Xoom always hangs quite a while during database writes.
Also I think that getting ~6 FPS on their first 2D animation test can't help.
The Xoom is really zippy, take the benchmarks with many, many grains of salt.
I was just talking about smart bench, everything else is working fine, but the smartbench 2011 shows a galaxy s as being more powerful than my xoom, and the half the speed of a stock xoom. I am just wondering if other people were showing that they are below what a stock xoom should be too.
you have to make sure that these benchmarks are compatible with dual core processors. otherwise the results are moot.
I know that, but it is shows below the average xoom, thats the problem I am seeing, average xoom gets like 2k I get 1k
I was having the same issue, I believe it is because of spare parts for gaming full screen. I factory restored my xoom and scored higher than average. The benchmark ran on a much smaller screen when I ran it on a fresh xoom.
joepfalzgraf said:
I was having the same issue, I believe it is because of spare parts for gaming full screen. I factory restored my xoom and scored higher than average. The benchmark ran on a much smaller screen when I ran it on a fresh xoom.
Click to expand...
Click to collapse
Confirmed that is it, thanks for this I guess I was overlooking it and thinking my xoom wasnt up to par.
I had always read that CPU performance increases were logarithmic with respect to frequency, so I decided to do some testing myself and share the results, which do confirm this (I wasn't doubting it, but wanted to see some actual hands-on data, what can I say, I'm a science student). I thought the forum might be interested in seeing it.
System Info:
Ziggy471 BFS Kernel, 4.4.11 build
Das BAMF 1.3.2 ROM
Procedure:
At each frequency I tested, I did 3 SetCPU Long Benchmark tests and 3 Linpack tests. I know these are not necessarily representative of performance, but they do provide a nice numerical comparison. I averaged the 3 results for each frequency and plotted them. The SetCPU Benchmark shows an especially distinctive exponential decay curve, (Lower is better for SetCPU, higher is better on Linpack) but the Linpack one also displays a vaguely logarithmic shape, although not as distinctive as the SetCPU one.
Plots are attached.
Hopefully, this information will at least be found interesting by some of you who were not already aware of the logarithmic curve of performance and maybe even guide decisions on how high of a frequency is worth it. But obviously my results are not the gospel or anything, just the results I obtained for these tests.
I did some similar testing with roughly the same results. After 1.4ghz you typically get diminishing returns. Tested with several roms and several kernels all with same results. currently running perfect storm 1.2 with ziggy's kernel pre-loaded and is very fast...but let's be honest...the phone is plenty fast stock...we really need more battery!!!! lol
thanks for sharing.
Yeah, it seems across all benchmarks the gains really drop out past ~1.4-1.5.
..Coming from the Eris where I was extremely excited to see 5.1 mflops, well, nuff said I guess....
Luv this fone!!
Perfect Storm stock kernel (ziggy's) is set and 1408mhz...my phone benchmarks faster at 1382mhz so that's what i have settled on.
patdroid1 said:
I did some similar testing with roughly the same results. After 1.4ghz you typically get diminishing returns. Tested with several roms and several kernels all with same results. currently running perfect storm 1.2 with ziggy's kernel pre-loaded and is very fast...but let's be honest...the phone is plenty fast stock...we really need more battery!!!! lol
thanks for sharing.
Click to expand...
Click to collapse
Yup. I'm done with the quad wars. I really hope we get better battery life fast.... Also, I noticed the same diminishing returns as well.
Sent from my thunderbolt
Its to the point where speed is a non issue. I'm not running matlab on this phone, I'm playing pokemon. I'm happy with how fast it is, now if i could get my battery up from 12 hours to ~ 18 I would **** bricks.
athorax said:
Its to the point where speed is a non issue. I'm not running matlab on this phone, I'm playing pokemon. I'm happy with how fast it is, now if i could get my battery up from 12 hours to ~ 18 I would **** bricks.
Click to expand...
Click to collapse
Hahaha. You echoed my sentiments precisely.
Sent from my thunderbolt
We might not be seeing diminishing returns.. what if Linpack isn't properly generating a end result because the phone begins to be too fast for it?
We need much longer tests to get more accurate results.
Diversion said:
We might not be seeing diminishing returns.. what if Linpack isn't properly generating a end result because the phone begins to be too fast for it?
We need much longer tests to get more accurate results.
Click to expand...
Click to collapse
do smartbench
It's nice to know that I"m not the only one nerdy enough to do this. I saw similar results.
Nbench will be good for that, it is very accurate and it takes about 10min to complete, its a stability test rolled into one. Its the only way I can accurately gauge my xoom dual core. OOOO, thunderbolt vs xoom in nbench, give me 10 min...
XOOM
Memory 2.86
Integer 3.89
Float 1.17
Thunderbolt 2.0 kernel, no OC
Memory 2.73
Integer 4.07
Float 0.94
surprising results, thought the xoom would be better, must not be fully implementing dual core in 3.0
Based on the graph, looks like all real gains are minimal for the risk above the chips 1.4ghz rating. Good to know that its almost useless to really OC, and just unlock it to the rated frequency.
This is interesting, I'll have to do a little testing myself. Thanks.
I only used results on which Linpack claimed to have a highly accurate result, but it's probably possible that it's not accurate at the higher clock speeds. I kind of doubt it though. The reason I used quicker tests (Linpack & SetCPU) is because I figured it would be a good enough statistic to compare performance of different clock speeds but on the same ROM and kernel, while allowing me to do multiple trials of each clock rate without the "experiment" being overly time-consuming.
The sticking point for me on chalking it up to inaccuracy is the fact that if you fit a logarithmic function to the SetCPU plot, the R² value is .9947, meaning it's almost a perfect fit.
Additionally, I think you start getting more errors and near kernel panics, as I believe they're referred to in *nix systems, at the higher clock rates (1.7, 1.8, etc) where the voltage, in this particular case, cannot be safely increased but isn't sufficient for that rate. Most of what I know about this is just from my experience in undervolting Macs, though. I'm far from an expert.
dacre said:
I only used results on which Linpack claimed to have a highly accurate result, but it's probably possible that it's not accurate at the higher clock speeds. I kind of doubt it though. The reason I used quicker tests (Linpack & SetCPU) is because I figured it would be a good enough statistic to compare performance of different clock speeds but on the same ROM and kernel, while allowing me to do multiple trials of each clock rate without the "experiment" being overly time-consuming.
The sticking point for me on chalking it up to inaccuracy is the fact that if you fit a logarithmic function to the SetCPU plot, the R² value is .9947, meaning it's almost a perfect fit.
Additionally, I think you start getting more errors and near kernel panics, as I believe they're referred to in *nix systems, at the higher clock rates (1.7, 1.8, etc) where the voltage, in this particular case, cannot be safely increased but isn't sufficient for that rate. Most of what I know about this is just from my experience in undervolting Macs, though. I'm far from an expert.
Click to expand...
Click to collapse
Holy crap! Im sure glad there are guys/gals like the ones in this thread to figure this stuff out because I feel like I walked into Trig class all over again. BTW I hated that stuff but in my own dumba$$ speak, this is kinda what I though. What I mean to say is this is a mobile, comparably lowres, linux based system, so how much speed is enough? One would think that at 1.5-1.8 it would be getting "maxed". If not "maxed" then at least enough that 90% of all users would be so thrilled and never see a need for more, that it would be fine. I (like so many others w TB's) would love to see a way to make that battery last just one full day with average usage. Thanks for all the effort you guys are putting in this! Look fwd to great ROMs as the dev work progresses.
Just using the phone, the higher frequencies actually can feel slower. Am I the only one that noticed? I run 1.4 gHz on the LeanKernel 1.9. Did anyone else realize you can bring it to 1.9? It gets really hot, really fast though.
Nope, mine is the same way. My phone is much snappier at ~1.4-1.5GHz than at 1.6GHz+.
I believe this is because the CPU starts derping up when it's at a clock speed higher than it can handle well (which varies from chip to chip); I think eventually when a big enough error happens it results in a kernel panic (where the phone freezes and eventually reboots), but I'm not exactly sure on how it works.
A quick note on the theoretics from someone who studied embedded systems...
Theoretically performance gain should be proportional to CPU frequency (double your frequency, halve your processing time). In practice you often see what you are seeing here, which is the effect of another system on the phone being the performance bottleneck. Without knowing anything about the cpu architecture and exactly how linpack is designed, it's hard to judge where that bottleneck is on the TB, but the point is that performance should be linear with cpu frequency.
Instead of averaging over 3 trials it would probably be better to take the *best* out of, say, 10 trials.
I'm assuming it's somewhat the same as testing runtime of particular algorithms on a computer--you want the best result you get because it actually is the closest to the actual answer. Processors won't "glitch" to being faster than they really are, so a slower result just means that some of the processor was used for something else (another background program perhaps). You will never get all of the processor's "attention" when doing tests like these on your own phone, so go with the result you get when you get *most* of the processor's attention!
Just a thought
Looking good, battery ok, feels smooth.
2.3.4 Quadrant
At last 2.3.4 seems to be very fast ! Quadrant score with out ext4 is 1600-1800 amazing!
PDA:XXJVP
MODEM:XXJVO
CSC:JV4XEN
Why are you all using Quadrant?
Run it 3 times and you won't even get the same score :| Pretty not reliant as a benchmark!
you won`t get same score cuz first time i like a cold start , it loades files etc.. second time you will get a better score
When you run 3dmark on a PC, you get the same (or around the same) score, same goes to a lot of benchmark software out there. Caching a benchmark goes against benchmarking itself...
t1mman said:
When you run 3dmark on a PC, you get the same (or around the same) score, same goes to a lot of benchmark software out there. Caching a benchmark goes against benchmarking itself...
Click to expand...
Click to collapse
Even that probably isn't truly indicative of performance. Nvidia and ATI both "optimise" *cough cheat* on benchmarks. Nvidia may blitz ATI in a certain game, but another game might be optimised for ATI. 3D mark also doesn't go into enough detail about what its testing (similar to Quadrant), which makes them both toys, because you don't get a breakdown. It doesn't tell you even basic stuff such as average random I/O speeds and very few benchmarks ensure that the test is run in a consistent fashion. You running the same test a second time leads to significantly better results, then caching is taking place, and the effect of caching should be added to the results too.
Benchmarks have unfortunately grown from an optimisation tool for developers, to a marketing tool for salespeople, and endusers who wish to boast. My suggestion, is that you should ignore them, and if you think the ROM feels nice, just use it. But, it seems most people are simply deciding on whether a ROM is nice AFTER they've seen the quadrant score, and that is a step backwards.
Its entirely possible to have dreadful real-world performance, but have the worlds highest Quadrant result (just like in a car, you might be able to drive 90km with 1litre of petrol. But if you only obtain that rate at 1km/h, its useless to almost everyone, especially if at 50km/h (normal speed), it starts gulping fuel like a ferrari).