Database Users

What's up with my bench scores?

I'm running Cyanogen Mod 7 v 2.5 Gingerbread on my HD2 which is clocked at 1.536Ghz.
Best quadrant score I could pull was 1418, with results being inconsistent every time I run the test.
Best score in linpack was 41 MFLOPS.
Best CPU bench score was 672ms.
Now, don't get me wrong. I know these scores are a def improvement over stock and nothing to sneer at, but I've seen HD2's that only run at 1.1 - 1.2Ghz with Froyo in the 1600+ Quadrant/60 MFLOPS+ CPU bench range.
I'm kinda stumped here. What's up?

Low Quadrant scores on over clocked kernel

Im only scoring mid 900's, is that normal?
Swyped from a fascinating over clocked droid using XDA App

Well it'd help if we knew how many mhz overclocked the kernel you're using is. But in most cases, if you're not on Voodoo and on an overclocked kernel, a 900 score sounds about right.

I'm going to guess it hangs up at the i/o test, correct?

SirGatez's blazing eclair voodoo kernels seem to score low on quadrant. Is that what your using?
Rockin' DJ05, Liberty Gingerbread, Clocked at 1.2ghz Voodoo5
-----------------------------------
DD username: TheSonicEmerald

Skrazz said:
I'm going to guess it hangs up at the i/o test, correct?
Click to expand...
Click to collapse
You know what, it never did until just today, why's that?
Kernel is in Signature, Geeknik Stupid fast Test Kernel 12.19.10, 1150mhz

Quadrant scores are harmed by rfs, the default filesystem on SGS devices.
Voodoo5 converts everything to ext4, which dramatically improves that section of the test.
Sent from my SCH-I500 using XDA App

Quadrant scores in the 800-900 range are common without the Voodoo lag fix installed. As Alex pointed out, the RFS filesystem causes lots of write lag, which significantly drags down the Quadrant score (if you had the full version of Quadrant, you'd see that). With Voodoo converting the filesystem to ext4, you'll see your Quadrant score practically double. Unfortunately, Quadrant doesn't use tests that are equal to all types of CPU architectures, filesystems, etc., so the results aren't really a fair comparison of real world performance.
Smartbench 2010 is much more accurate, according to everything I've read, although they're having a little trouble with it getting accurate readings on the Galaxy S series for some reason. They're working on it though, and will likely have an updated version soon that will fairly test the Galaxy S series alongside all other Android phones.

Post your Linpack and Quadrant scores

along with setup please.
Figured we might as well record benchmark progress as kernels and ROMs get faster. I'll start off.
Linpack -- 56.264 high, 52-55 consistent
Quadrant -- 2090 high, 2000-2050 consistent
Setup:
bcnice's 1.5 OC kernel
NFX's stock deodexed ROM
SetCPU 245/1516 interactive
Dread's GB theme modified
Thanks to all the devs working on this phone. Can't wait to see what else this thing is capable of.

my scores
Running:
Bcnices 1.8 Gingerbread AOSP kernel
Evervolv Alpha Update 3
Deodexed
Quadrant:
Highest: 2968 , 26-2800 consistently
Linpack:
Highest: 60.XXX, 55-58 consistently

Its also important to note which quadrant version you are using as well. Scores vary widely between the paid and free versions.

Jeezimus said:
Its also important to note which quadrant version you are using as well. Scores vary widely between the paid and free versions.
Click to expand...
Click to collapse
This is very true. The test that scored 2968 was run using Quadrant Advanced 1.1.1 Rev 1352
New screenshot is the same system as above but run on the newest version of Quadrant Standard (1.1.7) available.

I get about 2000 on Quadrant at 1.8GHz. How is there such a big difference?
Sent via carrier pigeon.

He's on an aosp rom with no sense framework
Sent from my PG06100 using XDA App

this was on regular quadrant OC'd to 1.8 on stock sense ROM and rooted

Is this any good?

If you have Advanced you can go into the Custom run and disable volumetric shadowing for the hallway. Scores will jump by 300-400. How the hell did you get over 4000?

Benchmark Bragging

Most people on this forum have tinkerer with their xoom, and at some point ran quadrant and linpack. I'm curious as to who has the fastest benchmark xoom.
We all know benchmarks don't mean squat in the real world but its still cool to OC it and see what it will do.
So post up your quadrant and linpack scores and at what speed
Mine @ 1.6
Quadrant 4513, it ran 4517 however was unable to get a screen shot.
68.772 linpack. Ranks 124 of all devices

I got 4157 as my highest on quad (1.6 ghz) but I got a 69.4 on linpack (screenshot lost when I had to flash stock images to recover, also at 1.6 ghz) ( its a GIF, so the app doesn't like it)
http://db.tt/GQ8XznI
Sent from my Xoom using XDA Premium App

This is kinda weird for a brag I guess, but I get 3758 on quadrant and 60 on linpack with the 1.4ghz 1.4.2 kernel.

Performance of Quarx 5/12 CM10 compared to other ROMS

I have a question regarding performance. I have a MB525 (red lense), with the JORLA_U3_3.4.2_108-2_R02_SIGNED_USAJORDMR1B25RETLALA01F.0R_JORLAFRORETLALA_P008_A005_M001_HWp3_WIG147146_Service1FF as my base ROM. I'm currently on Quarx's 5/12 CM10 ROM, and performance is really good, the user experience is the smoothest I've had in all the ROMs I've flashed. I'm using Interactive Governor, and my OC settings are 300/20, 600/35, 800/45, 1100/59. Now the deal is that for some reason Quadrant benchmarks give me a really low score, lower 2D and I/O scores, even though I feel that the actual user performance is a lot better with the 2nd Boot ROM.
I must admit that with my current Vsel settings, this is the lowest I've ever undervolted, and in my previous benchmarks I might have been OC'd to 1.2 ghz, but I don't think a 100 mhz difference would change the scores that much, especially regarding the I/O score.
Should I be worried about this drop in score? Or is Quadrant simply acting up?

Antiga Prime said:
I have a question regarding performance. I have a MB525 (red lense), with the JORLA_U3_3.4.2_108-2_R02_SIGNED_USAJORDMR1B25RETLALA01F.0R_JORLAFRORETLALA_P008_A005_M001_HWp3_WIG147146_Service1FF as my base ROM. I'm currently on Quarx's 5/12 CM10 ROM, and performance is really good, the user experience is the smoothest I've had in all the ROMs I've flashed. I'm using Interactive Governor, and my OC settings are 300/20, 600/35, 800/45, 1100/59. Now the deal is that for some reason Quadrant benchmarks give me a really low score, lower 2D and I/O scores, even though I feel that the actual user performance is a lot better with the 2nd Boot ROM.
I must admit that with my current Vsel settings, this is the lowest I've ever undervolted, and in my previous benchmarks I might have been OC'd to 1.2 ghz, but I don't think a 100 mhz difference would change the scores that much, especially regarding the I/O score.
Should I be worried about this drop in score? Or is Quadrant simply acting up?
Click to expand...
Click to collapse
I'm in the same situation.
Same phone, same cm10 quarx rom...
1200 in quadrant, i have
400 32vsel
700 40vsel
1000 54vsel
1300 67vsel
Lowest score: 2d graphics and i/o.

Very interesting Would be nice if someone could explain it.
I tested some roms with Antutu and Quadrant.
WiuiGB:
Antutu 2912-3131
Quadrant 1883
MiuiICS (Braun's):
Antutu 2272
Quadrant 1755
CM10 (Quarx's Early April Nightly 2ndboot):
Antutu 3064
Quadrant 1167
Quadrant score is much lower on CM10 with 2ndboot, but Antutu is pretty high.