3d mark 06 is out

[Guest EVIL-SCOTSMAN]

yup, you have read it properly, 3d mark 06 is out, get it from a host of places, guru3d.com have it also.

 

be prepared for shiat scores, and i mean shiat :mad:

 

example:

 

fx55 with sli 6800ultras = baddabing 4500

[Guest Cali]

WOW I would get 2000 :(

[Guest EVIL-SCOTSMAN]

i aint even gonna think about using it on this pc. i will wait til i get my good one back, but the scores will fall dramatically from 12000k in 3d05 to probably 5-6000 max = wtf 3dmark peeps fuxed up, as everyone likes teh high numbers when doing benchies.

[Mustang]

i had a go and got 2155 marks :o man, the CPU test made me cringe, my poor poor CPU is gonna need a rest after that slideshow

[Guest EVIL-SCOTSMAN]

[Guest EVIL-SCOTSMAN]

i dunno whats up with this forums, but it wont allow me to copy n paste links to anything via firefox, when i copy a link, there is no option to paste and it will not paste, does the same on all 3pc's.

 

get it fixed

 

ok, the above pic is my new crap that i got today and my first run of 3d mark 06, everything at stock.

 

the cpu tests = 0fps

[zenTera]

Hi Scotsman,

nice to see you got your 'good one' back-hope it runs without problems now.

Just ran the bench again at stock settings (last time I selected all tests). My results:

score 5437

SM 2 2253

SM 3 2266

CPU 1795

Same result as last time with all the tests selected. So, what are the other tests for if they seem to do nothing?:confused:

Well, I really don't know if this 3DMark2006 is trustworthy. Of course, with two 7800 GTX 256 in SLI you should get better graphics scores and hence overall scores than with one 7800 GTX 512. But what makes me wonder is the CPU score. If you get 1797 with a 4800+ @ 2400, shouldn't an FX-60 @ 2700 score more than 1795? Can't really understand that.

And the CPU test itself is pain in the a**, I also get a 0 FPS slideshow. What monster CPU out of hell did they invent that test for? Would like to run the test on the notebook I've got from work - but I'm afraid that rendering 1 frame would either take half an hour or even the whole thing would go off in flames;)

[emenance]

Really though why would 3d mark 2006 have anything much to do with LOCKON? Seems to me this new bench mark doesnt match the game software. Maybe a better clue for fancy stuff like FEAR or the like?? Like I have always said for years and even more important today. Benchs are becoming worthless unless its compared apple to apple with NEW respect to what software a person is running on the hardware. Sorry I just cant see a match here.

 

I mean do the tests of 2006 really mirror Lock on?

[Guest EVIL-SCOTSMAN]

dude, 3d06 is meant to be used with that new ageia ppu cards, as its got it plastered all over it.

 

as for 06 having anything todo with lockon, it has nothing in common, with lockon or any other game out there atm,.

 

Basically you are 100% right, benchies mean shit and should be taken with a pinch of salt.

 

me, i am not a lover of benching, I do it once every now and then after some big changes, then i wont touch it for another 6 months to a year.

 

This was first time i tried 06 since it came out and I had to get it running to see what it would do to teh pc.

But overall, I dont trust em, cuz if I did, then there is this lil hoe on another forum with a 3800cpu and 2 78s and she absolutly pwns my score. 9k 3d06

 

now that i do not believe thus 3d mark is way off in some respects and cannot be trusted.

 

Zen, i looking at your cpu score, damm, you got pwnd. that is strange why mine at stock beats urs 300mhz faster ?

 

i am gonna have to look into that, maybe something your end, or something my end, as i only ran it once just to see what the pc would get. but yea, something strange happening ????

 

TBC

[Guest EVIL-SCOTSMAN]

its definetly something ur end zen, as my lowest score at 1280x1024 is 7800ish, so i gots no idea why your would be as low as what they are ???????

 

my lowest score is still 2000 above yours ?

 

you have got cool n quiet turned off i presume ?, also when you overclock amd64 cpu's by only 1/2 of a multiplyer then your memory isnt running at what it says it is. thats why you should never ever overclock amd64 cpu's with just 1/2 a multiplyer, its either a whole or nothing, never just bump the multi up by half as it seriously fux with your memory and your memory will not run at what its says its running at, its due to some weird thing that cannot adjust to a half thus it affects the memory settings big.

[zenTera]

its definetly something ur end zen, as my lowest score at 1280x1024 is 7800ish, so i gots no idea why your would be as low as what they are ???????

 

my lowest score is still 2000 above yours ?

 

you have got cool n quiet turned off i presume ?, also when you overclock amd64 cpu's by only 1/2 of a multiplyer then your memory isnt running at what it says it is. thats why you should never ever overclock amd64 cpu's with just 1/2 a multiplyer, its either a whole or nothing, never just bump the multi up by half as it seriously fux with your memory and your memory will not run at what its says its running at, its due to some weird thing that cannot adjust to a half thus it affects the memory settings big.

Whooo...will have a look into it tonight. Might it really be the 13.5 multiplier? But what confuses me: it would make sense if the memory score would be low then, but that one is ok, it's only the CPU-score that's strange. Anyway, will try it. And Cool'n'Quiet: it's activated in the BIOS, but there are no drivers installed on my gaming partition, so it shouldn't run. Taskmanager as well as benchmark programs tell me it's running at 2700, so I guess it does. On my application partition I have the drivers installed and get of course different CPU speeds in taskmanager etc because of C'n'Q running.

[Ice]

My Score 4692

[Guest EVIL-SCOTSMAN]

comparin ice's score to yours zen, it makes me think something is up big, cuz calis rig is good, but its no fx60 with 7800/512 and he isnt that far behind you, i would think you would get atleast 7000

[zenTera]

comparin cali's score to yours zen, it makes me think something is up big, cuz calis rig is good, but its no fx60 with 7800/512 and he isnt that far behind you, i would think you would get atleast 7000

 

WTF:mad:

Just did a new test again. This time I disabled Cool'n'Quiet in the BIOS and also put the clock rate back to auto, which means the CPU running at 2600. No way! Tested 3DMark2006 and PCMark2005 again. Results are what I expected: In 3DMark2006 the scores for SM2.0 and 3.0 stay the same, while CPU score goes down a bit and hence the overall performance. With PCMark2005 I chose the standard settings with additional CPU and Memory tests. Memory dropped from about 4750 to 4550, while CPU stayed around 5337 (5338 on 2700:confused: )

 

So, I guess something is really odd, but I don't think it's on my side. The tests showed, that it doesn't seem to cost any performance to set the multiplier to xx.5 values or to have C'n'Q enabled. Furthermore, have a look at some article on guru3d: http://www.guru3d.com/article/article/313/7/

They are testing with a FX-57 and different graphics cards. Look at the last benches, FX-57 with 7800 GTX 512; comparable to mine, but I beat it in any category:

score 5437 (4787)

SM 2.0 2253 (2176)

SM 3.0 2266 (2220)

CPU 1795 (1107)

And look at the comparison between the 6800 GT and the same in SLI. You can see that the CPU score doesn't change there, but the overall score does dramatically because of the shader models.

Which brings me to two conclusions:

1) Guess there's nothing wrong with my rig or setup

2) Don't waste no more time on those fu**ing benchmark programs, 'cause they don't say nothing. Scotsman, do you get 150% of my performance in games and applications? Don't really think so. And me as well sure don't have double the FPS in LockOn e.g. than people with half the score would gain. So, what are all these benchmarks good for? Nothing!! A guy with a single core 4000+ and two 7800 GTX in SLI might me much happier than me with a game that makes extensive use of SM2.0 or 3.0, whereas I would score if GPU would be important, but CPU power also (especially with games/apps that use dualcore).

Ok, I stop wasting time on those useless benchmarking, better use it for some useful things like flying or :icon_jook .

Over and out;)

[Guest EVIL-SCOTSMAN]

lol, its frustrating as fux :)

 

I see where your coming from, they are a waste of time, thats why i only do them when I do a massive change of hardware or a new 3d program comes out, I normally test 1 or 2 times then thats it for roughly 6 months to a year.

 

to be honest with ya tho, we have a long list of 3d06 benchmarks for the notebookforums, and a rig like your should have atleast 7000, well thats what there scoring, its only the intel cpus that get around what your getting.

 

the list is huge with all different kinds of pc's etc, but the dualcore and 78s are always at the top, most of the dualcores with 78s score between 7300-8300, thats why I said yours should be higher, but you have to take into account a good few things, like what else is running at the time of the benchmark etc etc.

 

dont beat yourself up about it, cuz it will drive you nutz, aslo you say an fx57 with same vid card is comparable to yours ?

 

you should wipe a fx57 comparison, as 3d06 is designed to take use of dualcore, thus roughly any highend dualcore cpu will wipe an fx57 in 3d06 for the simplefact its multithreaded and fx57 isnt.

 

on my other forums, the one with the huge list of 3d06 scores, some notebooks beat you, not by much but they should not be doing with your leet rig.

 

also, have you did the tests with AA and AF on ?

 

i dont know how you did the tests, but make sure, the resolution is 1280x1024 in 3d06, you cant change res if its the free version, you can if you payed for 3d06, make sure vsync is off as if thats on your score will be shit, aa and af both to off/application controlled, basically, make sure everything in nvidia control panel is off and its set to high performance.

 

but dont beat yourself up over it, as i said in an earlier post, take them with a pinch of salt, as long as you get roughly what others get with same cpu same type of vid card and roughly same amount of ram, then your good, if its say 1000 lower then start wondering as 1000 in 3d06 is HUGE margin , not like it was in 3d05.

 

tell me, are you the dude with sli 78s 512 ? cuz i am going by the assumption that you have sli ?

 

also, i would read up about 1/2 multiplyer settings on amd64 rigs and see what it says about it. the dfi boards have a real good piece on it, its in there overclocking help section.

 

i can link you if needed.

[zenTera]

to be honest with ya tho, we have a long list of 3d06 benchmarks for the notebookforums, and a rig like your should have atleast 7000, well thats what there scoring, its only the intel cpus that get around what your getting.

Yeah, reading that again is really frustrating:( :confused:

 

 

also, have you did the tests with AA and AF on ?

 

i dont know how you did the tests, but make sure, the resolution is 1280x1024 in 3d06, you cant change res if its the free version, you can if you payed for 3d06, make sure vsync is off as if thats on your score will be shit, aa and af both to off/application controlled, basically, make sure everything in nvidia control panel is off and its set to high performance.

Will have a look at nVidia panel, but I didn't change anything there except a special profile for LockOn with max AA and AF; but that should only apply when LO is loaded. VSync should be off usually and AA and AF set to application, but maybe that's a clue...will see.

 

 

tell me, are you the dude with sli 78s 512 ? cuz i am going by the assumption that you have sli ?

SLI board: yes. SLI cards: no:( Maybe someday when prices for the 7800 GTX512 drop to 200 EUR (well, probably in 6 to 8 months;) ) and I can save some money till then I might upgrade. But in the moment I dont't see the slightest need for a second card...except 3DMark benches around 8000-9000.

 

also, i would read up about 1/2 multiplyer settings on amd64 rigs and see what it says about it. the dfi boards have a real good piece on it, its in there overclocking help section.

 

i can link you if needed.

yeah, link would be appreciated. Saves me some time to search myself.

Shit, didn't I say over and out? Man, you just got me into it again;) And I've got no time for that. Just received that 'Ruggbutt toy' Ergodex DX1 and am hot like teacher to try that thing:horseback

[Guest EVIL-SCOTSMAN]

There ya go, I eventually get the link to paste, retarded way of doing things....

 

http://www.dfi-street.com/forum/showthread.php?t=20823

 

 

quote from the site:

 

CPU/FSB Frequency Ratio

This is your CPU multiplier.. This times your HTT\FSB is how you get your CPU frequency..

 

**Important note**

On all a64 systems it is not recommended to use half multipliers ( i.e. 8.5,9.5,10.5) because this will put your ram on a ratio which means it will be running slower then you think and not the same as you have your HTT\FSB set to no matter what..

 

A quote here from Wild shows what happens to your ram speed when using half multis…

Quote:

Originally Posted by Wildstyle

(DRAM frequency) X (RAM divider) = CPU frequency

 

Dividers are all whole numbers, so if you use a half multiplier your memory speed is calculated by using the next highest whole number.

 

An example...

 

200X10 = 200/10 (divider) = 200MHz

 

200x10.5 = 2100/11 (next highest divider) = 190MHz (effective memory frequency.)

 

 

 

 

 

 

 

 

[Prophet_169th]

PC specs in sig.

 

I get 4300.

 

And regarding OC'in. My MoBo is the bomb. I dont have to worry about ratios/dividers and all that.

 

There is a spot for FSB speed, and then a seperate one for RAM. The BIOS handles the rest!

[TucksonSonny]

 

 

This was not done with the free version I guess?

Can we get a key for the CPU? For free? Anyone:confused:

[Guest EVIL-SCOTSMAN]

^ was done with the free version, u really think i be buying 3dmark 06 ?

 

look at the taskbar, "basic edition"

[Wombat1940]

Yuor SLI system and other

 

Scotty .......

 

What's you're SLI system ...... dual 512's or 256's?

 

My 3DMARK06 was 5840 with both cards in SLI a markup of 45% from having just one card on (see my signature for system details).

 

I tried many test runs with different backgrounds turn on

eg: AA 4x AF 8x, ......AA & AF Application Controlled ...... etc and got the same 3BMARK06 values each time. Also did the same for 3DMARK03 and 05. Results all in the same ballpark irrespective of how I set the background.

l

3DMARK03 27462

 

3DMARK05 11067

 

As the readme file says ...... 3DMARK sets everything to the same in its own background (Res 1024 x 768, AF off, AA off, etc). It took me heaps of testing to convince myself of this. I even did cold boots before running each test! Even sat on the same chair each time the test was run ;).

 

I do have an issue though with the background setting in the nVidia SLI dual setting areas.

 

There are two ....... which don't make the same changes to each when you change just one.

Let me explain ...... if you go to the Window's Setting window (right click mouse on desktop .... click on your NVIDIA Display - Monitor ....click on Performance & Quality Settings ... change AA 4x ... AF 8x .... check box to store both .... click OK. Now go the NVIDIA Settings Icon in the Taskbar of the Desktop ... left click on it .... click on ... Performance & Quality Settings .... if AA is on 4x and AF is on 8x that's good, that's what it should be. IF NOT WHY NOT :(:(. If they are on 4x and 8x change them to both to Application Contolled and exit. Go to the Windows Setting window and check P & Q Settings ........ IT WONT HAVE CHANGED, IT WILL STILL BE ON 4x and 8x :(:(:(:(. This doesn't affect the 3DMARK but it does affect Fsims like 3FAF where AA is set in Windows and not in the game like FC. The NVIDEO Settings Icon is what sets the default mode and overrides what is set in the Windows Display Settings. Well at least that's how I have found it.;)

 

One final point, I can overclock my Mobo by up to 10% (I have no idea what you are talking about .... tweaking the CPU is it? :rolleyes:) But when I 3dMARK it when OC, it freezes thru the test. Freezes is LOMAC1.02 as well. Although looking at the fps as its testing they seen to increase by 1 fps for each 1% of OC.

I think "something is short of voltage" in the bios setup. But that all gobbldegoop to me :rolleyes:.

 

Sorry ....one other thing do you use FRAPS? (it's a free program that measures fps ) in LOMAC1.02 I get an average of 70.466fps for that full demo of "F15 Low".

 

Gee's I talk too much ....:redface::redface::redface::redface: sorry .... (Please don't forget my ? at the start of the post)

 

Wombat

[Guest EVIL-SCOTSMAN]

256s, it was done with a fresh install of windows and just firefox and 3d06 installed, even with all the crap i have installed now, i get min of 7500.

 

with older cpu, my 3d05 score was 11600 and that was with the 4400 instead of the 4800 i have now.

 

when doing 3d marks, go into the nvidia control panel, and turn aa and af off make sure vsync is off basically just turn everything off as with vsync and aa and af on you will affect the score dramtically, also i found that sli mode AFR was better than SFR for 3do6.

 

my score aint that great, i get beat by lower cpu's and also by dual 512s.

 

Also 3d06 is multithread and is built for dualcore cpu's, so there will be a significant margin between a single core cpu and dualcore cpu.

 

your score is in the ballpark of where your cpu and cards should be. nice score might i add ;)

 

if you use sli, never use nvidia conytrol panel, as it will do what it wants to do, use nhancer. www.nhancer.com, it allows you to mess with sli profiles and stuff that cant be done with the normal nvidia control panel. basically it overrides what nvidia wants todo and helps you get better fps etc.

[Wombat1940]

 

512's ...... even with all the crap i have installed now, i get min of 7500.

I WOULD HAVE THOUGHT 7500 WAS VERY GOOD. I SUPPOSE YOU ARE EXPECTING BETTER WITH THE DUAL CPU'S. HAVE YOU TRIED TO GET A FIGURE WITH ONE CARD TURNED OFF? MAYBE DON'T, ITS DEPRESSING. AS I SAID IN MY PREVIOUS POST I ONLY GET 45% IMPROVEMENT WITH TWO CARDS. I ASKED MY SUPPLIER OF THE CARDS FOR A 55% REFUND ON THE SECOND CARD ;). ITS CRAP WHAT NVIDIA ARE PUSHING: 'UP TO 100% GAIN WITH SLI .....' ITS TRUE ..... JUST MISLEADING!

when doing 3d marks, go into the nvidia control panel, and turn aa and af off make sure vsync is off basically just turn everything off as with vsync and aa and af on you will affect the score dramtically,

I DIDN'T FIND THIS. MY SCORES WHERE (WITH VSYNC APPILATION CONTROLLER):

 

03: 27430 AND 27451 WHEN AA 4X AND AF 8X

03: 27463 AND 27420 WHEN AA AND AF APPLICATION CONTROLLED

 

05: 11059 AND 11067 WHEN AA 4X AND AF 8X

05: 11050 AND 11067 WHEN AA AND AF APP. CON.

 

06: 5820 AND 5840 WHEN AA 4X AND AF 8X

06; 5839 AND 5840 WHEN AA APP. CON.

 

V

 

also i found that sli mode AFR was better than SFR for 3do6.

WHAT IS SFR?:confused:

 

if you use sli, never use nvidia conytrol panel, as it will do what it wants to do, use nhancer. www.nhancer.com, it allows you to mess with sli profiles and stuff that cant be done with the normal nvidia control panel. basically it overrides what nvidia wants todo and helps you get better fps etc.

THANKS I'LL CHECK THAT SITE OUT.

 

GOOD FLYING

 

WOMBAT

[Guest EVIL-SCOTSMAN]

in order to get more sli options, goto www.guru3d.com and goto the downloads section and download coolbits 2.0

 

its a nvidia registry tweak that shows you more options in the nvidia control panel, in order for you to be able to pick the sli mode, which there is

 

AFR = alternate frame rendering

SFR= split frame rendering

AFR2 = same as afr but with some tweaks for games that dont like afr

single gpu = 1 vid card running

sli AA = antialiasing of 16x instead of the normal max of 8x

 

never leave the aa and af to application controlled when benching, as the program will use that setting to turn aa and af on, make sure aa and af are off when you do the tests, do not set it at application controlled, also make sure vsync and all other fancy options are turned off.

 

so, get coolbits 2.0, click on it and it will auto insert it into the nvidia registry and then open the nvidia control panel, you will now see more options, and in the performance option place where you set aa and af, you will now find that you can have more sli options instead of the normal single gpu rendering and multi gpu rendering, coolbits 2.0 allows you to get the different types of sli mode that i mention above, and it also opens up the nvidia overclocking tool which allows you to overclock the video cards, you dont have to use it if you dont want and there is no harm in not using, if you do use the overclocking tool, little tiny steps at first til you know what your cards can handle.

 

also nhancer is a must for sli, it allows you to override nvidia settings and get better performance out of games, as nvidia sli profiles are shit and most of the time dont work properly, well thats what i found.

[Wombat1940]

in order to get more sli options, goto www.guru3d.com and goto the downloads section and download coolbits 2.0

 

its a nvidia registry tweak that shows you more options in the nvidia control panel, in order for you to be able to pick the sli mode, which there is

 

AFR = alternate frame rendering

SFR= split frame rendering

AFR2 = same as afr but with some tweaks for games that dont like afr

single gpu = 1 vid card running

sli AA = antialiasing of 16x instead of the normal max of 8x

 

never leave the aa and af to application controlled when benching, as the program will use that setting to turn aa and af on, make sure aa and af are off when you do the tests, do not set it at application controlled, also make sure vsync and all other fancy options are turned off.

 

so, get coolbits 2.0, click on it and it will auto insert it into the nvidia registry and then open the nvidia control panel, you will now see more options, and in the performance option place where you set aa and af, you will now find that you can have more sli options instead of the normal single gpu rendering and multi gpu rendering, coolbits 2.0 allows you to get the different types of sli mode that i mention above, and it also opens up the nvidia overclocking tool which allows you to overclock the video cards, you dont have to use it if you dont want and there is no harm in not using, if you do use the overclocking tool, little tiny steps at first til you know what your cards can handle.

 

also nhancer is a must for sli, it allows you to override nvidia settings and get better performance out of games, as nvidia sli profiles are shit and most of the time dont work properly, well thats what i found.

Thanks Scotty ...... looks mind boggling to me :rolleyes: ...but just what I need :D ....have to go and babysit my grandson now, so will explore later and I,m sure I'll be back.

 

All the best

 

Wombat