Post Reply 
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Reporting statistics
05-26-2013, 12:45 AM
Post: #1
Reporting statistics
Some statistics in case anyone is interested. These are statistics of errors, so they more reflect what has problems than what is popular, but it's sorta in between both.

I couldn't be bothered to convert it to bbc, so find it here:
https://github.com/unknownbrackets/ppssp...Statistics

-[Unknown]
Find all posts by this user
Quote this message in a reply
05-26-2013, 07:41 AM
Post: #2
RE: Reporting statistics
That was an interesting read, thanks [Unknown].

The reports skew rather heavily towards the android devices, although PCs are getting a good representation up there too. Quite a healthy mix, some of the reported CPUs are cracking me up lol.

PPSSPP Modern Testbed:-
Intel Core i5 4690K @ 4.0 GHz
NVIDIA Geforce GTX 760 2GB GDDR5 VRAM @ 1138/6500 Mhz
16 GB DDR3 RAM @ 1600 MHz
Windows 7 x64 SP1

PPSSPP Ancient Testing Rig:-
Intel Core 2 Duo T9550 @ 2.8GHz
ATI Mobility Radeon 4670 1GB GDDR3 VRAM @ 843/882 MHz
8 GB DDR3 RAM @ 1066 MHz
Windows 7 x64 SP1
Find all posts by this user
Quote this message in a reply
05-27-2013, 08:23 PM
Post: #3
RE: Reporting statistics
Interesting Smile

♦ Intel Core i7-6700HQ | 16 GB RAM | NVIDIA GeForce GTX 960M | Debian Testing
♦ Intel Core i7-2630QM | 4 GB RAM | NVIDIA GeForce GT 540M | Debian Testing
♦ PSP-3004 | 6.60 PRO-C2
Find all posts by this user
Quote this message in a reply
05-28-2013, 03:53 AM
Post: #4
RE: Reporting statistics
It has quite some intresting stuff there!
* 99 reports from a 16-core setup with 3 issues! LOL

*Intresting to see that on the Intel reports the second most popular reports are from setups that support ALL instruction sets, which proves that atleast half of the people here use a pretty efficient setup that could get benefited from instruction set implementations!

*Something simmilar as above goes to ARM (Even though it's like 1/3rd of the users here), that's really cool!
Find all posts by this user
Quote this message in a reply
05-28-2013, 04:04 AM
Post: #5
RE: Reporting statistics
I can also do statistics on particular OpenGL extensions if anyone has a list of specifically interesting ones. Doing a full list by what's there is hard, unfortunately.

But yeah, cool data.

-[Unknown]
Find all posts by this user
Quote this message in a reply
05-28-2013, 05:00 AM
Post: #6
RE: Reporting statistics
Well, shouldn't OpenGL 4.x have different results to previous implementations? They said they pretty like re-invent OpenGL with those releases (4.x and after).
Find all posts by this user
Quote this message in a reply
05-28-2013, 05:32 AM
Post: #7
RE: Reporting statistics
Well, I don't know, I see things like:

3.1.11931 Compatibility Profile Context (ATI Technologies Inc. AMD Radeon HD 5500 Series), 4.20

I'm not sure what that means, but I assume it means not 4.x. I don't think we're asking for 4.x? There's one card reporting 4.x:

4.2.12217 Compatibility Profile Context 12.10.17 (ATI Technologies Inc. AMD Radeon HD 6700 Series ), 4.20

The latter 4.20 *should* be GLSL version.

-[Unknown]
Find all posts by this user
Quote this message in a reply
05-28-2013, 05:53 AM
Post: #8
RE: Reporting statistics
I don't think reports are working on Blackberry. Will try and track down why.
You said duplicates are hit for matching gpu/cpu/version but I change version and it's never recorded.
Find all posts by this user
Quote this message in a reply
05-28-2013, 06:07 AM
Post: #9
RE: Reporting statistics
It only checks duplicates based on message. The associated versions, gpus, and cpus for that message are all updated, but the message is not bubbled to the top.

It shows the highest version number for the log entry by the value of the version (example: 760879 for v0.7.6-879-gc9e931e.) If two have the same version, which is shown is indeterminate.

Per the stats, there have been 26 reports from Blackberry devices.

-[Unknown]
Find all posts by this user
Quote this message in a reply
05-28-2013, 06:12 AM (This post was last modified: 05-28-2013 06:14 AM by xsacha.)
Post: #10
RE: Reporting statistics
So that is reports altogether or reports that haven't been updated to newer cpu/gpu/version?

I think I've generated more than 26 reports myself which is why I think something is up :\
I even purposely hit some unknown syscalls and didn't see it appear in temp/recent.

I might have a look over the net code later. It should just be regular unix stuff though.

Could I see the most recent report from a Blackberry device?
Find all posts by this user
Quote this message in a reply
05-28-2013, 06:41 AM
Post: #11
RE: Reporting statistics
I'm counting platform hits, so that's all. There's a counter per message for each gpu, each cpu, each platform, and each version. The main intention is to be able to see quickly if the issue is only affecting certain gpu(s) or if it's affecting everyone.

Here are the unique reports from Blackberry devices (26 is just a number):
Code:
| id_game        | formatted_message                                                                    | hits |
+----------------+--------------------------------------------------------------------------------------+------+
| ULUS10457_1.00 | Wrong magic number 46535000                                                          |    4 |
| ULUS10045_1.01 | BREAK instruction hit                                                                |    1 |
| ULUS10457_1.00 | Unimplemented HLE function sceHprmRegisterCallback                                   |   18 |
| ULUS10236_1.00 | sceKernelCreateSema(SNDMUTEX) unsupported options parameter, size = 167769136        |    2 |
| ULUS10537_1.00 | Thread ended within an extended stack                                                |    1 |
| ULES00645_1.01 | sceIoIoctl(disc0:/sce_lbn0x65370_size0x10d73a, 01010005, 09f9fa90, 10, 00000000, 0)  |    1 |
| ULES00645_1.01 | sceKernelStartModule(): doing nothing for some reason?                               |    1 |
| ULES00645_1.01 | sceIoIoctl(disc0:/sce_lbn0x1c640_size0x11d5800, 01010005, 09f9fa90, 10, 00000000, 0) |    1 |
| ULES00645_1.01 | sceIoIoctl(disc0:/sce_lbn0xb6a0_size0x107a000, 01010005, 09f9fa90, 10, 00000000, 0)  |    2 |

-[Unknown]
Find all posts by this user
Quote this message in a reply
05-29-2013, 09:53 AM
Post: #12
RE: Reporting statistics
Thanks. Looks like it worked then. Smile It's just I am the only one putting in reports, lol.
Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump: