Project "Xtreme Analysis" Information.

[Lestat HWL]

XtremeSystems will be embarking on a project to create its own in-house system benchmark. This benchmark will consist of 2-3 OpenGL tests, a Pi Calculation test, along with a memory bandwidth test.

Not many details have been emerged, as this is a brand new concept. This benchmark is being written to fill in the missing spots in 3DMark / Aquamark, etc...

There are several benchmarks out there, but very few that can give truely comparable results between different CPUs, memory, or GPUs. If you think about it, when you want to benchmark your system, chances are, you run 3DMark, Super Pi, and Sandra. Project "Xtreme Analysis" will be the benchmark to put an end to a need for multiple benchmarks to accuratly measure full system performance.

This benchmark will utilize the OpenGL API to allow this benchmark to be ran on Linux/Unix based systems.

Currently we have few developers, so if you are wanting to help out, contact me on AIM: Lestat107Z, or Email: Lestat@Xtremesystems.org

Our current developers include:

Lestat - Project Manager
Eshbach - Development Supervisor
Kunaak - Security Supervisor
coldequation256 - C++/Open GL Development

Currently not many details are in stone, but we are still in the concepting process.

If you have any questions, feel free to post in this forum. As always, your ideas are welcomed and greatly appreciated!

[echow87]

sounds like a good idea :)
no more multiple benchmark :D

however i will still continue benching with 3d01 and 3d03 ;)

[MaxxxRacer]

very coool. But some DX9 stuff woudl be nice as well.

[kryptobs2000]

yeah, why no dx9?

[Lestat HWL]

DX9 is a possibility, but it would be for a Win32/64 only version.

[Lestat HWL]

When testing video cards, I agree, DX9 tests are needed definitely.

[chunkylover77]

Sounds like a great idea. Wish I had some skills to help you out.

[skate2snow]

That is an Idea I had a long while ago.... I even think I posted about it:) I am happy that it will become something active. And thanks to who are working on it:).

I dont know much in coding and all so I wont offer my help, but be sure I support your project at 100%. If you need a beta tester or something just ask me... It is important that there is no bugs that can influance cheating, although, Kunaak is very good at that;).

[craig588]

Use the OGL2 spec, don't even bother with the old 1.x versions, we already have more than enough apps that do stuff with those.


I know how to break stuff, but not fix it so I don't think I will be able to help with development in any way. (Unless you just want to see random exploits without any solutions)

[eshbach]

It occurred to me that we will need someone to do db management and write a frontend for DB access over web. probably MySQL, so if anyone has experience with that let one of us know.

[madman2233]

this is a great idea, but i agree that there should be some DX9 tests.

thank you every one thats working on it.

[craig588]

DX9 is not needed, just use OGL2

[Sentential]

Bout time... futurmark is so bloated with :banana::banana::banana::banana: that a change is needed :woot:

[reject]

i can beta bench :D
what we need is a way to check the core of the cpu, so newcastles eg. dont have to compete driectly with venice or clawhammer

[perkam]

This is quite an undertaking guys and I'm sure none of us would like for it to get outdated very fast. So why dont u wait for WGF 1.0 and then make the benchmark ??

Perkam

[recons]

Is this going to be building on the LX Project? If not will it still be open source?

[D_o_S]

Alright, I'm up for betatesting as well.

[shadowing]

I'm up for betatesting too!

[Bloody_Sorcerer]

count me in. I can do 3D art, (mediocre) 2D art, and I'm learning graphics programming.

[matt9669]

I can be of assistance with most anything in C/C++ :up:

[_Eduard_]

if you need betatesting i'll be willing to help. sounds like a very nice idea, unfortunately my programming skills aren't good enough to help you in your benchmark. if you want some artwork or design or so done then i could make a little contribution

[jjcom]

Willing to beta test. I don't know anything about coding, but I do like the benchmark. :up:

jjcom

[magg]

As everybody else has said and probably will say, I'm willing to beta test as well... sounds like a great project, what a way to really put XS on the map! :D

[jrw]

That sounds sooooo cool. If it turns out nice, Im willing to pay for it. Make sure to implement the possiblity og spitting out a screen-jpg in the end, balanced for webpublishing :D

[DGeNeRaT3]

Fantastic idea. It will be so nice to have a community based benchmark that the results actually mean something.

Dgen

[Nubius]

Freakin awesome. Can't wait to see what you guys can pull off

[Revv23]

im in as a beta tester...

hows it going to be scored? overall or will it be seperate scores for each area... that would probably be better....

also, why not have a DX7/DX8/DX9 benchmark as well?

[eshbach]

im in as a beta tester...

hows it going to be scored? overall or will it be seperate scores for each area... that would probably be better....

also, why not have a DX7/DX8/DX9 benchmark as well?


there will be seperate scores as well as a total score that is a function of the componenet scores.

i cannot say yet whether it will be a simple summation or if different parts of the test suite will have different values in contributing to the total score.

one thing, though, the DB will certainly be sortable by partial scores, meaning, if you have a very fast CPU and a slow video card, you can still potentially get the #1 cpu score and sort the DB as such.

there will still be only one "#1" overall though, this way we can have a clearly defined champion.

[R.Rabbit]

w00t!! this sounds like an awsome idea, XS is kicking so much ass at benching we need to make our own bench to continue to kick more ass!!
also like previously stated, it should auto save a screan-shot(possibly security built in?) for those suicide runs where paint wont work!
good luck guys!! oh yeah, i'm willing to beta bench!

[blinky]

i dont think it should be in linux, because people with ati cards wont have good drivers, and i dont want to have to optimize my linux install just to get a good bench, hell i dont even wanna mess with linux for anything more than surfing and basic stuff.

just something to ponder

[kryptobs2000]

i dont think it should be in linux, because people with ati cards wont have good drivers, and i dont want to have to optimize my linux install just to get a good bench, hell i dont even wanna mess with linux for anything more than surfing and basic stuff.

just something to ponder

then don't bench in linux :p:

Just cause you can dosn't mean you have to. And I'm glad it'll suck on ati cards, maybe it'll give ati a reason to get their :banana: together and make some actual decent drivers for linux.

[Magnj]

Ill test anything :D

[Saliman]

Thats a nice project guys, but DirectX 9.0b & c will have to take part in the tests in my opinion.

Continue the good work!

[craig588]

Thats a nice project guys, but DirectX 9.0b & c will have to take part in the tests in my opinion.
Why? There is no benefit of DX9 over using OGL2. Have you even seen something made with OGL2?

[kryptobs2000]

Why? There is no benefit of DX9 over using OGL2. Have you even seen something made with OGL2?

cause 99% of retail big budget games use dx for the graphics api :stick:

[Revv23]

cause 99% of retail big budget games use dx for the graphics api :stick:


not to mention that certain graphics cards favor opengl and certian favor DX.

[matt9669]

On OpenGL vs DirectX:

It would certianly be best and ideal to have benchmarks with both APIs, running in both Windows and Linux. Creating a good graphics benchmark is by no means an easy task however. To truly utilize the features of DX 9.0 / OGL 2.0 requires complex pixel and vertex shaders, which must be coded in assembly language since no widely used HLSL exists.

So, patience folks, I'm sure the priorities right now lie with incorporating currently available benchmarks into the test (PiFast, memory b/w etc.), a good front end and a stats database/server. Creating good, impartial, thorough hardware accelerated graphics scenes will likely be a massive and ongoing project.

[Revv23]

On OpenGL vs DirectX:

It would certianly be best and ideal to have benchmarks with both APIs, running in both Windows and Linux. Creating a good graphics benchmark is by no means an easy task however. To truly utilize the features of DX 9.0 / OGL 2.0 requires complex pixel and vertex shaders, which must be coded in assembly language since no widely used HLSL exists.

So, patience folks, I'm sure the priorities right now lie with incorporating currently available benchmarks into the test (PiFast, memory b/w etc.), a good front end and a stats database/server. Creating good, impartial, thorough hardware accelerated graphics scenes will likely be a massive and ongoing project.


very good point, seems like a benchmark to showcase overall system power would be much easier, with the graphics benchmakr part coming in later... like you said.

[STEvil]

Actually you just hit on a very good idea matt!!!


Creating good, impartial, thorough hardware accelerated graphics scenes will likely be a massive and ongoing project.

Making the benchmark somewhat "modular" so updates and new tests can be added without recoding the entire benchmark (making for smaller downloads and updates too maybe?) would be neat.

[matt9669]

Making the benchmark somewhat "modular" so updates and new tests can be added without recoding the entire benchmark (making for smaller downloads and updates too maybe?) would be neat.Good programming style would make the benchmark portions modules already, shouldn't be difficult to do. :cool:

[eshbach]

Good programming style would make the benchmark portions modules already, shouldn't be difficult to do. :cool:

to an extent, yes.

there is a point where modular coding can become seriously form over fuction, and that's really not what we're going for here.

[matt9669]

there is a point where modular coding can become seriously form over fuction, and that's really not what we're going for here.True, but the actual benchmark portions could easily be external modules, the input/output requirements are minimal - pass the settings to the program, recieve the results.

[trakslacker]

count me in for beta benching!

[don_vercetti]

looks like a nice idea...i for one will use it, once it's up and running, and i'll be happy to beta test as well (endless benching over and over...deja vu).

Have you had any ideas for what sort of scene the games will be showing?

[shadowing]

well, this is what i think we should do.

Have linux support.
Have both OpenGL and DirectX implemented for hardware.
To push the other parts aka the cpu and ram, is there any way we can bench both?

I hope there's a loop option too...

[perkam]

In the words of freecableguy, "Gotta pay to play" :D . If any donations are required for 3d programming or use of a professional company to finalize and make the benchmark flashy, I'm in for that too... as well as the beta test ofcourse.

Maybe u can post progress in here on what ur preliminary plans are etc. You dont have to start 3d animation right away, some 2d illustrations of possible scenes would be a good start along with, first and foremost, a listing of the type of tests that are going to be run.

Perkam

[alexio]

I can work on the Pi calculator, and I'd love to be a beta tester.

[[XC]thewildblue]

Sounds a fantastic idea, just what we all need. Lets set the new standard.

I would love to alpha and beta test.

Are the xtremesystems regulars going to get first looks ?

[Flib]

I'm for including SSE/SSE2/SSE3/HT etc. Support.
That's not because I'm an AMD odr Intel fanboy, no!
I think it is better to show the real performance you can achieve.
There are only few programms without SSE2 Support.

[recons]

HT etc. Support.

Hmmm.

Seeing as there aren't many games that are multithreaded, its not a huge issue. But it would be nice two see single and multithreaded versions of tests (yeah I know lots more work), but it would be interesting to see how a real world test, stacks up to a test of what the real world might be heading for.

Also people saying "I will betatest" maybe should be dealt with later, as it detracts from the topic at hand (making the software).

[saaya]

uh, why opengl?

can we add at least one d3d test? :/

[Nerm]

Great idea! Any clue on an ETA for finishing?

[recons]

uh, why opengl?
Because one of the major game engines uses it.

[dippyskoodlez]

Because one of the major game engines uses it.

In order to have a useful benchmark, we need OGL **AND** D3d.

Some games use d3d, and some use OGL. only good way to compare is with a little of both. :)

Using just one would defeat some of the purpose of a benchmark- to compare performance.

Some cards are better with opengl, some are better with D3D.

[kryptobs2000]

And you can't have it on linux or anything with d3d, since that won't run on linux at least not natively. And if something supports hyperthreading that will lead into dual core processors, which will soon be hitting the market so it'll be taken advantage of.

[dippyskoodlez]

And you can't have it on linux or anything with d3d, since that won't run on linux at least not natively. And if something supports hyperthreading that will lead into dual core processors, which will soon be hitting the market so it'll be taken advantage of.

it isnt really "ht" support, as it is SMP support. Multithreading should be tested, though.

Linux has poor OGL speeds, and lacks D3D altogether, so linux benching wont gain anything useful.

Benching linux would end up more of a how well can you tweak WINE and if your able to properly install drivers than it will a hardware competition.

Some Ideas for the "utility" menu would be:
-Integrating CPU-Z like things, possibly in the main UI like 3dmark.
-Integrate clockgen!
-ever use overclockulator? useful little thing..
-Gas boiling points/conversions?
-CPU wattage calculation
-Heatsink C/W
-HTT/FSB and ram divider calculator
-Atitool/Rivatuner
-Stress testing

As for 3d tests, I was thinking like 2-3 OGL and D3D tests, a cpu test, and maybe a HDD test? perhaps more 3d tests, too, though, incorporating more of a focus on physics like UT2k3, and maybe one or two having heavy memory usage, etc...


Heres a UI I made up in VB real fast, of a basic Idea for stuff...

:toast:

[STEvil]

If there is a HDD test we need to be able to select the HDD we want to test instead of just defaulting to the main OS drive or the one with the swap file on it.

[Lestat HWL]

Actually you just hit on a very good idea matt!!!



Making the benchmark somewhat "modular" so updates and new tests can be added without recoding the entire benchmark (making for smaller downloads and updates too maybe?) would be neat.

This has already been set as the plan. Adding a test will work like a "snap on" module.

[craig588]

-CPU wattage calculation
-Heatsink C/W

The first would require guessing and the second would reguire temperature measurement from an inaccurate mobo probe. Keep them out.

[SDX]

i am more than willing to test =)

[dippyskoodlez]

The first would require guessing and the second would reguire temperature measurement from an inaccurate mobo probe. Keep them out.

Im speaking more as a calculator than live results..

CPU wattage can easily be calculated using datasheets. I did it with a bunch and have it lying around here somewhere.,.. lots o' numbers.

it is NOT thermal dissipation-- its watts consumed, running.

(its not 100% accurate, as everythiung varies, but according to specs, thats what it should be.. AFAIK.... known wattage of a 2500+M, and voltage, you should be able to derive the wattage at 2.5ghz, no?)

[makatee]

I don't know what all you guys want, but i have a pretty extensive background with programming. mysql is a breeze too, if you need any help give me a pm.

[SuperDude]

How about an optimization for users of 2k and server/64 bit?

[skate2snow]

Why an optimization for certain OS???

[recons]

How about an optimization for users of 2k and server/64 bit?
You mean support right?

[dippyskoodlez]

You mean support right?


I don't think there should be any specific OS optimizations-- as it'll cause everyone to eat up win2000 CD's, and screw some of the people that only own windows XP. Not like there really is anything special to make, is there?

Other than 64bit, ofcourse... 64bit version could be nifty, though.

[eshbach]

I don't think there should be any specific OS optimizations-- as it'll cause everyone to eat up win2000 CD's, and screw some of the people that only own windows XP. Not like there really is anything special to make, is there?

Other than 64bit, ofcourse... 64bit version could be nifty, though.

yes i am learning to code for 64bit windows as we speak...

[kryptobs2000]

You could have options as to run with what support: sse, sse2, sse3, 64-bit etc...

[dippyskoodlez]

You could have options as to run with what support: sse, sse2, sse3, 64-bit etc...

I think that option would be cool, but I dont think that should be selectable on the scored run at all..

Just gotta remember to get the scored run as reproducible, yet not... too much, as to cause driver cheating... :rolleyes:

Having a nice little list of tests to run each seperate run with SSE2, etc would be very useful though. Especially doing reviews ;)

[kryptobs2000]

You could have the 'standard' run but then also have alot of seperate options, this would be good for reviews or comparison of different processors etc..

[Quazi]

If you would like to have an ordinary, run of the mill, end user to beta test for you all, please let me know. I'd be happy to be able to help in that way. Mainly because it is the only way I can help.

[masterofpuppets]

it isnt really "ht" support, as it is SMP support. Multithreading should be tested, though.

Linux has poor OGL speeds, and lacks D3D altogether, so linux benching wont gain anything useful.

Benching linux would end up more of a how well can you tweak WINE and if your able to properly install drivers than it will a hardware competition.

Linux can't compete with Windows for 3D since both nVidia and ATi write completely pathetic drivers, the latter especially. Now if ATi and nVidia got off their asses and made a driver that actually works "properly", then Linux would make a great gaming platform, although, I play Doom3 with most settings up flawlessly I have to say. I really can't stand the manafacturer's negative (read: lazy) attitude to Linux. Many people want to use Linux as a gaming platform, and those idiots can't even see that. I pray for the day ATi ever make a decent Linux driver, if I am even alive by that time.

[reject]

why dont u make a thread geting everybody to forward an email to ati/nvidia?

[masterofpuppets]

It would never make an impact. They get complaints all the time, I imagine.

[skate2snow]

Thats probably what everyones say? ...

And those "idiots" are still those who powers your 3D;)

[reject]

dude, i dont use linux so im not gonna make the thread, but if u do ill forward the email.
like s2s says wat if everyone had ur attitude? theres a good chance that they dont have enough feedback to care

[masterofpuppets]

skate2snow, dude, they work for the same company, but that doesn't mean they are the same people.

[erorr404]

Very cool project!

I will be happy to betatest this benchmark on a few older machines (66 MHz 486, 233 MHz PII, 600 MHz PIII, 2.26B P4, 2.4 GHz P4 laptop, 2.8E P4).

It is probably a good idea to test the program on various systems to see if it works correctly on older platforms as well as new ones.

[(sin)morpheus]

Yeah, I would have to say this looks just plain awesome. I will also be happy to beta test for you on all 16 of the systems in my house. (everything from an overclocked 486sx to a high end athlon64 gaming machine)

[Qkjhfhaiguihfma]

I'm sure they appreciate the offers for help, but as of right now beta testing is prolly the last thing on their minds. When they are ready to test something I'm sure they will post it here, swamping the thread with offers really doesn't help your chances, it just adds clutter.

[jrw]

If you want some nice linux 3d drivers from ATI, here's what you ought to do:
http://sh.webhire.com/servlet/av/jd?ai=405&ji=1502385&sn=I

[kempez815]

I'm in for any beta's that are going :)

[Entity_Razer]

Lestat, if you need some 3D models, I model in Milkshape3D and can provide you with .max, .3ds, .obj, .x (direct X), and a whole lot of other types including valve models.

So if you need some models, lemme know. MS3d is a low poly modeler though so i can't provide you with a model that has 90 000 polygons but I'm sure someone with max can make high poly models. Well I can to if I can just use meshsmooth to make it high poly but .... :)

Just PM me if you need a 3D model of something.

[[XC] leviathan18]

something like pcmark test would be nice compressing email sending spell check all the test from pc mark 04 would be nice nice project guys sad im not too advanced in programing :(

[craig588]

I made a police car model (Basicly a copy of a police car from GTA) in 3DSMax and it can be scaled from like 300 polys to well over a million. The only problem is that I have it backed up on a cd somewhere and I have over 400 cds backed up with stuff I have made. (I think alot of them are just windows installs, but I have so many I'm scared of taking on the massive orginization project to make sense of all of them, especially considering at least 75% of them are unlabeled)

[-Acid-]

Amazing Idea and i love to beta test

[don_vercetti]

any progress on this? looks like a decent idea.

[Entity_Razer]

yhea any progress update? Maybe we should have a small website with (for now) the update logs?
Maybe with a small wiki where people can "donate" Code, models, art and I don't know what else :)

[dippyskoodlez]

checking up on this, if you need some help cooridnating this thing, I'd be glad to help with overseeing it, and possibly building a page to cooridnate this project, as I'd love to see it get off the ground and do well.

[Roger_D25]

I just wanted to say I think this is a great idea and would also love to see it happen. Also like everyone else here I'd love to help in anyway I can. Just drop me a P/M!

I also like the idea's some have had about building a page to post progress, news, beta downloads, etc.... Let's make it happen!

[Freebo]

sweet, i just spotted this, looks like it will be good!

[unclean]

It was mentioned that a db person and web programming would be useful?

I'd like to offer up my services if the above is still the case. Looks good guys :)

[Revv23]

holy thread resuurection!


i dunno in my opinion the best benchmarks are the ones that stick with one thing and are good quick and easy at it.

like pi as a 2d performance app, or 3d mark as a 3d app,

ok 3d mark can be mch improved so maybe stick to that? pi isnt really neccesary to incorporate as we already have a great pi tool. its even owned by XS. Im not sure how much improving you can do on that.

[Ray_GTI-R]

HT is a must - all the best processors do HT :D

Multitasking is the Real World - think simultaneous "automatic updates & virus scanning" ... and that's without ever touching the keyboard/mouse.

Look at your Task Manager and smell the coffee :stick:

Beta-ing would be an honour, I've got a test rig all set up and waiting.

:toast: Ray

[[cTx]Philosophy]

So why did this thread die?
Is this program still in the makings, or what?
Im down for some testing

[Fred_Pohl]

Version 1B needs a LOT of work!

[dig412]

Is this project still running?
I have some good php/mysql experience, so if you'd like any help with the web front end just PM me.

[Nosfer@tu]

I just wanted to say I think this is a great idea and would also love to see it happen. Also like everyone else here I'd love to help in anyway I can. Just drop me a P/M!

I also like the idea's some have had about building a page to post progress, news, beta downloads, etc.... Let's make it happen!

HAHAH Spamming for a 100 posts so he can acces the sales section :slapass:

[NickS]

HAHAH Spamming for a 100 posts so he can acces the sales section :slapass:

Your post was more spammish than his. Cut it out. :slap:

[RaZz!]

any news on this project or is it really dead? *cough*

[Kunaak]

Unsticking this old thread...

this thread should have been gone a long time ago, the project died... when the original poster had to leave XS for various reasons.

[AAbenson]

:( :dammit:this is very sad.it was a great idea:scope:to have a XS benchmark,guess as alwys in life too good to be true...

[TouGe]

Does anyone know if the project will be started again?