Hi all,
Over in the X1 forum, they have a utility & thread to speed up SD card access, and they are benchmarking it with SKTools. It also briefly discussed in our graphics performance thread, but NeoS2007 justly noted it was a bit off-topic, as that thread is actually about OpenGL and D3D (maybe he can change the title)
Anyway, what really bugs me is not that the X1 utility does not work for HD, but that the HD has soo much less RAM (and flash) performance then the X1. Why? Is it not supposed to be nearly identical base machine? Have they used a cheap RAM chip, or is the timing of it wrong. Would that be a bootloader setting?
In fact, I suspect comparing flash speed between X1 and HD is irrelevant if the RAM is so different, you are in fact still measuring RAM.
For example, this is my HD:
(posting from others show it is a fairly normal HD, only a bit high in storage write, maybe because I have a Ultra-card):
Code:
RAM access: 544
Bitmaps: 785
Main write 1776
Main read 6413
Storage write 1303
Storage read 8657
The typical X1:
Code:
RAM access:1519
Draw Bitmaps:797
Main (write):11070
Main (read):17066
Storage (write): 483
Storage (read): 16832
Thats 3 times the RAM speed
And as everything depends on RAM, it does not even matter to mention the 6 times flash write speed
In this thread, I am hoping that people who have significant other findings (you can test this with the trail version of SKTOOL) post so, just to see what really is normal.
Or better yet, if someone smart is able to somehow change the RAM timing, by modding a bootloader or poking some bytes into a chip register somewhere. (I suspect the RAM timing is not a windows setting, but deeper than that)
If you do post your setting, please include if you applied any tweaks, and the brand/type/class of microSD.
Thanks
if you look at the index in sktools benchmark the HD gets given a score of 560 v 472 for the diamond on ram access.
i get on my x1
Integer 329.2257
FP 7.233
Ram 1502
Bitmaps 799
MS Write 8727.27
MS Read 15170.39
SD Write 487
SD read 16168.42
it's curious as I also thought that the two devices were internally very similar, what tweaks are you using?
Do you get different results in integer and floating point, as some people believe their X1s run at 400mhz
Wow thats huge difference in ram speeds between blackstone and hd. But interestingly enough blackstone seems to perform better with in coreplayers benchmark. That maybe suggests that sktools is not accurate, or its buggy etc.
Integer 316.9227
FP 7.294
Ram 461
Bitmaps 769
MS Write 1860.13
MS Read 6849.50
SD Write 461.71
SD read 4740.74
Click to expand...
Click to collapse
Tested using SKTools
Am running SPB Benchmark at the moment to compare
my Touch HD:
Integer 330.3489
FP 7.605
Ram 576
Bitmaps 814
MS Write 1897.47
MS Read 7177.57
SD Write 192.25
SD read 6729.46
what the....?
why am i getting so low result for SD write? i have installes SD tuneUP and still....
noris08 said:
what the....?
why am i getting so low result for SD write? i have installes SD tuneUP and still....
Click to expand...
Click to collapse
Annoying, but I suspect the top 4 benchmarks are much more critical in whether the device feels snappy than the latter 4, and of those, I think the read has more influence than the write. So maybe it makes you feel less bad to know the SD write is probably the least important one?
Also, it may simply be the SD card you have. I just bought an 'ultra', most people have a class II, maybe you have a class I?
yeah, i know that this is the less important but it still bugs me.
that and the huge difference compared to x1
i have the 8gb Transcend microsd card that came within the package. i have no ideea what class it is. how can i find out?
it has a circle with a 6 inside of the circle printed on. that means it's a class 6? then the results i am getting worries me a lot
Since I kind of started this discussion in the other tread mentioned by cybermaus in his initial post, I think, I can put this to rest also. Run the Memory card benchmark in pocket mechanic (Trialversion available here: http://www.wizcode.com/downloads ) and you will see that the read/write performance of the 8GB card that came with the HD is quite remarkable, my results were even better than the ones posted over in the X1 thread here: http://forum.xda-developers.com/showpost.php?p=3154958&postcount=257.
It's a more sophisticated bench for the card performance and we are talking about read performance of 22.5 MB/s unbuffered, 42 MB/s buffered and write performance of 1.80 MB/s unbuffered, 5.33 MB/s buffered, so all is good. The readings in the SK Tools bench seem to be false, I would suggest that this also goes for the ram performance readings. This is also consistent with the observation that the HD is the slightly faster device overall.
razorblader said:
Since I kind of started this discussion in the other tread mentioned by cybermaus in his initial post, I think, I can put this to rest also. Run the Memory card benchmark in pocket mechanic (Trialversion available here: http://www.wizcode.com/downloads ) and you will see that the read/write performance of the 8GB card that came with the HD is quite remarkable, my results were even better than the ones posted over in the X1 thread here: http://forum.xda-developers.com/showpost.php?p=3154958&postcount=257.
It's a more sophisticated bench for the card performance and we are talking about read performance of 22.5 MB/s unbuffered, 42 MB/s buffered and write performance of 1.80 MB/s unbuffered, 5.33 MB/s buffered, so all is good. The readings in the SK Tools bench seem to be false, I would suggest that this also goes for the ram performance readings. This is also consistent with the observation that the HD is the slightly faster device overall.
Click to expand...
Click to collapse
did that, got
read unbuffered 11,86 mb/s
read buffered 32,75 mb/s
write unbuffered 1,47 mb/s
write buffered 3,38 mb/s
Couldn't do the RAM readings as it kept complaining it needed 21,5MB free even tho 70% was free. But indeed quite a difference to SKTools (phew).
rhtizzy said:
did that, got
read unbuffered 11,86 mb/s
read buffered 32,75 mb/s
write unbuffered 1,47 mb/s
write buffered 3,38 mb/s
Couldn't do the RAM readings as it kept complaining it needed 21,5MB free even tho 70% was free. But indeed quite a difference to SKTools (phew).
Click to expand...
Click to collapse
Does the same thing for me on the RAM bench, the low level tests for the the card bench also crash the program on my HD, so we are left with the file read/write tests, but good enough for me at the moment.
would be nice to do that benchmark before and after the install off sd_tuneup
Vinski- said:
Wow thats huge difference in ram speeds between blackstone and hd.
Click to expand...
Click to collapse
Blackstone and HD are names for the same device .
noris08 said:
why am i getting so low result for SD write? i have installes SD tuneUP and still....
Click to expand...
Click to collapse
The "tuneUP" doesn't improve the writing speed, it just increases the reading speed.
johnpatcher said:
The "tuneUP" doesn't improve the writing speed, it just increases the reading speed.
Click to expand...
Click to collapse
i've noticed that 'cause i've made the test with and without tuneUP.
that's is too bad because one may need fast writing speed when taking movies or burst photos
i still can't understand the huge differences compared to X1
btw, i have got better results testing with pocket mechanic. i did it last night and i don't remember the numbers, but, for writing test, there were something like 2.35 mb unbuffered and 4.95 mb buffered.
menwhile i found out that he sd card is class 6.
but i am still bugged by the high numbers X1 scored
johnpatcher said:
Blackstone and HD are names for the same device .
The "tuneUP" doesn't improve the writing speed, it just increases the reading speed.
Click to expand...
Click to collapse
Oh, looks like I mistyped it yesterday, I meant X1 versus blackstone
But still wondering is the benchmark buggy, or why so huge difference. X1 doesnt feel that snappier, and looses Coreplayer benchmark, which uses software decoding in both, meaning pure processing power.
johnpatcher said:
The "tuneUP" doesn't improve the writing speed, it just increases the reading speed.
Click to expand...
Click to collapse
Hi guys,
I found two interesting things when I tested my HD with SKtools.
1) I compared the results of SKTools SD card reading test before I install SD tuneUP and after. It was very suprised for me. The results are following:
Without SD tuneUp: ~5500 KB/sec
With SD tuneUp: ~ 2500 KB/sec
2) I made RAM SKTools read/write tests on HTC Touch Diamond and I was very-very suprised.
HD results:
Main storage (read): 5779 KB/sec
Main storage (write): 1692 KB/sec
Diamond results:
Main storage (read): > 8000 KB/sec
Main storage (write): > 8000 KB/sec
Who can explain that?
razorblader said:
Since I kind of started this discussion in the other tread mentioned by cybermaus in his initial post, I think, I can put this to rest also. Run the Memory card benchmark in pocket mechanic (Trialversion available here: http://www.wizcode.com/downloads ) and you will see that the read/write performance of the 8GB card that came with the HD is quite remarkable, my results were even better than the ones posted over in the X1 thread here: http://forum.xda-developers.com/showpost.php?p=3154958&postcount=257.
It's a more sophisticated bench for the card performance and we are talking about read performance of 22.5 MB/s unbuffered, 42 MB/s buffered and write performance of 1.80 MB/s unbuffered, 5.33 MB/s buffered, so all is good. The readings in the SK Tools bench seem to be false, I would suggest that this also goes for the ram performance readings. This is also consistent with the observation that the HD is the slightly faster device overall.
Click to expand...
Click to collapse
But this, and may posts above, are focussing back to the SD card.
I really think the SD card is far less important than the RAM for general performance of the device. Checking the SD is like wanting a speedy car by checking how fast you can open the booth.
I'd like to indeed think that it may be a artefact in SKTool that causes the difference between HD and X1 RAM benchmarks (500 vs 1500). Especially since people tell me practical benchmarks like Coreplayer tell a different story.
But the difference is too significant to ignore. Does anyone know a different RAM/CPU (not SD) testing program?
so you don' think writing speed of the sdcard is important!
i'll say it again - when you record a movie or burst some photos writing speed will count or not? maybe a drop more then opening the booth?
Actually im thinking RAM is the issue here, the reason being yes the read speeds of the SD card would impact on the performace of loading the data which is why we have buffers, the problem i see is that im not getting buffer underruns so we dont have a shortage of data to play. So we are loading the data in to RAM without much issue.
RAM would also explain any lag issues people might be gettings with the over all performance of the device.
noris08 said:
so you don' think writing speed of the sdcard is important!
i'll say it again - when you record a movie or burst some photos writing speed will count or not? maybe a drop more then opening the booth?
Click to expand...
Click to collapse
I am not saying it is not important. I am saying it is less important than RAM speed:
If you record a video at full VGA resolution, it would mean a 640x480x30x3 or 27 MB/s stream into RAM/CPU/GPU (depending which does the compression). Next, it is compressed (so cycles around between RAM and CPU/GPU a bit) and finally needs to be written to a file in the compressed stream of maybe 1 or 2 Mbps. Which is 0.125 or 0.250 MB/s
If you play a XViD video, the same thing in reverse, except now the RAM bandwidth is even 35 MB/s because we want the full WVGA
So, SD requirements: between 0.1 and 0.3 MB/s. RAM requirements, way above 35 MB/s (the CPU also needs to calculate, not just move data)
Like you, I'd also like a faster SD. I would hope it would speed up loading a new program after selecting it from the launcher. Also, interrupt calls from the SD could cause jerkiness in video. And that one guy with 0.192 MB/s SD defenitely would like a faster SD.
But the fact SD is important, does not take away the fact RAM is more important, which brings us back to the original question: Is the SKTool test flacky, or is X1 significantly faster in RAM?
cybermaus said:
And that one guy with 0.192 MB/s SD defenitely would like a faster SD.
Click to expand...
Click to collapse
that guy happens to be me
i am not arguing about bigger is better (i am speaking about RAM)
but considering the fact that X1 is also made by HTC, and it would be logic that they keep the best part of the cake for themselves (HD) i can only imagine that SE found a better management software of the given hardware.
can this be an explnation for the almost 1 year delay of the X1?
Hello!
I really dont know whats normal for a X8 in Quadrant but my X8 are sometimes lying around 682 but I've also been on 720-750 a couple of times. How can those numbers be so different?
Phone info:
Used since: 2010-12-24
ROM: Rachts CM7 Rc4 2.3.3 (Latest update)
CPU: Overclocked to 710 MHz
I can't post my question in the Development thread about overclocking because I don't have 10 posts but I've now overclocked it just to 710 (Normal with that "program") but how can i try to go up to 730 Mhz now?
Thanks
RoseN
Added some more info to first post...
The number show you the speed of your phone the time you test is. This means that if the next time you make the test your phone search for network or you play with your screen or you download something or you have many programmes running then the result of your speed will be lower.
What quadrant does is called benchmarking. Basically, it measures your phones performance in several aspects and at the end you could compare it to some of the big guns in android.
It analyses the CPU, 2D and 3D performance, and memory among others. The higher you score, the better our phone's performance has been rated.
Some of the things that affect your rating are: the number of apps that are running, overclock or clocking of CPU, and available memory.
You phone could misbehave from time to time so the results may vary...
I want to raise question is it possible to overclock the CPU, or GPU or memory speed. We all know that ze550kl / z00l performance with msm8916. Is almost in the limit to run apps and games, wanted to know if possible or not, and if there is any development has this respect. I like the device and it is great but it is clear that it is getting weak in the matter of hardware, if someone can work in this respect to give a breath / power to our device
Ermes.mt/brasil said:
I want to raise question is it possible to overclock the CPU, or GPU or memory speed. We all know that ze550kl / z00l performance with msm8916. Is almost in the limit to run apps and games, wanted to know if possible or not, and if there is any development has this respect. I like the device and it is great but it is clear that it is getting weak in the matter of hardware, if someone can work in this respect to give a breath / power to our device
Click to expand...
Click to collapse
i don't think so , if its possible to overclock the processor to its frequency,
you can costumize it too max frequency of it or to lower , as u need
But if is then i would love like to know too
I am already extracting the maximum, from him that I can of him, already put using ex kernel, with governor ordenad, frequency of 200mhz @ 12000mhz, and simple GPU ordered frequency of 200mhz to 400mhz, active zram of 512mb, and entropy of 128/256, then Of course, here with standard frequency of the paddle extracting nothing more than that, the problem and that we have a large bottleneck in memory lpddr3 of 533mhz, that opera and 32bit, there is a single channel giving bandwidth of 4.2gb / s that It has to be split between CPU and GPU. What is clearly one of the pro soc power limiter, I do not know much about the arm architecture, but I can not do over clock in lpddr3 memory, since I did not see any software or tutoring Citing this. Although this would be a solution, if you do over clock in the CPU or ns GPU will force the clock controller up what will increase the memory performance consequently, something else, not good will lock the clock at maximum in 1200mhz and GPU at 400mhz, as there will be a large power consumption and the performance will be limited by the cords of the tempering control and the low frequency of the RAM, # excuse me to have been so far more enthusiastic about hardware,
even with my old htc explorer i could increase cpu clocsk from 600mhz to 900mhz why we cant do it on zenfone 2 just we need good developer and kernel
it's up to the hardware
If you look at the specs from qualcom it has a max speed and not every device is possible to reach the max, but if you have the proper kernel you can reach that speed and "overclock"