Rage3D Discussion Area

Rage3D Discussion Area (http://www.rage3d.com/board/index.php)
-   Radeon Tweaking, Modding, and Overclocking (http://www.rage3d.com/board/forumdisplay.php?f=22)
-   -   official 3D MARK 11 Score Thread (http://www.rage3d.com/board/showthread.php?t=33971454)

shane_p Dec 7, 2010 03:34 PM

official 3D MARK 11 Score Thread
 
8,500+ from the red side here


moshpit Dec 7, 2010 06:31 PM

Broke 6K with i7 860 + single GTX 580, Green Team shows strong for a single GPU and default clocked i7 860:

[IMG][/IMG]

In fact, memory is significantly underclocked right now, running DDR3 1066 instead of the rated 1600.

john dolan Dec 8, 2010 10:50 AM

P7605 3DMarks

http://3dmark.com/3dm11/57324

shane_p Dec 8, 2010 12:40 PM

Quote:

Originally Posted by moshpit (Post 1336440484)
In fact, memory is significantly underclocked right now, running DDR3 1066 instead of the rated 1600.

why oh why?


why not underclock the cpu too?
LOL

moshpit Dec 8, 2010 01:32 PM

Quote:

Originally Posted by shane_p (Post 1336441139)
why oh why?


why not underclock the cpu too?
LOL

Because something is screwy with the XMP profiles when filling all 4 slots of a motherboard. If I only use 2 of the 4 slots, I can run 1600mhz memory, but if I use all 4, I cannot set any XMP profiles or it won't boot. 8Gb of ram is more useful for the 64bit future then 600mhz, IMO. Single app performance isn't as important as being able to run more apps at once, more memory is better then faster in that scenario.

I've been eyeballing some of the new Corsair Vengeance 4Gb modules, perhaps a pair of them. That way I can keep 8Gb and only fill 2 slots.

http://www.newegg.com/Product/Produc...82E16820233144

caveman-jim Dec 8, 2010 02:00 PM

just set the timings and speeds manually, nudge up qpi and memory controller voltage

also Cat 10.11 CAP #4 is now available, with performance improvements in 3DMark11 for Crossfire users.
http://www.rage3d.com/index.php?cat=75#newsid33971493

moshpit Dec 8, 2010 02:04 PM

Quote:

Originally Posted by caveman-jim (Post 1336441207)
just set the timings and speeds manually, nudge up qpi and memory controller voltage

I guess I could, but that would kill my excuse for some new ram. I have to admit the thought of 1600mhz 4Gb modules are causing me great temptation.

john dolan Dec 8, 2010 02:49 PM

P7763 3DMarks

http://3dmark.com/3dm11/66037

shane_p Dec 8, 2010 07:46 PM

Quote:

Originally Posted by caveman-jim (Post 1336441207)
also Cat 10.11 CAP #4 is now available, with performance improvements in 3DMark11 for Crossfire users.
http://www.rage3d.com/index.php?cat=75#newsid33971493

Thanks, it does help (not a whole lot though)

but broke 8,600 now: (seems a lot smoother now)

Uploaded with ImageShack.us

caveman-jim Dec 10, 2010 08:34 AM

If you're feeling adventurous, try Cat 10.5 hotfix with 10.11 CAP 4.

demo Dec 10, 2010 10:17 AM

3DMark Score7168
--------------------------------------------------------------------------------


Not bad for a single GTX580 I guess.

shane_p Dec 10, 2010 12:34 PM

12,000+ with 10.5



Uploaded with ImageShack.us

moshpit Dec 12, 2010 11:09 AM

Quote:

Originally Posted by shane_p (Post 1336442849)
12,000+ with 10.5



Uploaded with ImageShack.us

WHAT??? Are you saying that 3Dmark11 loses 4000 points for using the current driver over using Cat 10.5???? What the hell? Something HAS to be not rendering right then. I call bunk on a 4K performance change for using an older driver (by the way, not calling bunk on YOU, just the driver thing and AMD for allowing such a freaky scenario, your discovery of it is pretty cool actually).

Edit: Just for clarification, kudos on you for finding such a cool trick. My complaint isn't lodged at you. I'm calling bunk on the score's relevance OR AMD has severely nerf'd something hardcore! I'd prefer it to be the former, but if it's the later, WOW! Hella nerfing!

Edit 2: Just for more clarification, I AM reading that right, right? All you did was change drivers back to 10.5, and your P score in 3Dmark 11 jumped nearly 4K points? If I'm reading that right, did you notice any kind of rendering issues during the benchmark run? If everything looked perfect, I wonder if something is accidentally causing a similar issue to what happened with Nvidia back on 3Dm03 where they were caught not rendering the whole scene, just what was seen? Getting off the benchmark's scripted path allowed one to see rendering shortcuts that shouldn't have been happening, but this time it's more of an accident or bug with early driver not doing what the benchmark wants of it?

shane_p Dec 12, 2010 02:06 PM

Quote:

Originally Posted by moshpit (Post 1336444299)
WHAT??? Are you saying that 3Dmark11 loses 4000 points for using the current driver over using Cat 10.5???? What the hell? Something HAS to be not rendering right then. I call bunk on a 4K performance change for using an older driver (by the way, not calling bunk on YOU, just the driver thing and AMD for allowing such a freaky scenario, your discovery of it is pretty cool actually).

Edit: Just for clarification, kudos on you for finding such a cool trick. My complaint isn't lodged at you. I'm calling bunk on the score's relevance OR AMD has severely nerf'd something hardcore! I'd prefer it to be the former, but if it's the later, WOW! Hella nerfing!

Edit 2: Just for more clarification, I AM reading that right, right? All you did was change drivers back to 10.5, and your P score in 3Dmark 11 jumped nearly 4K points? If I'm reading that right, did you notice any kind of rendering issues during the benchmark run? If everything looked perfect, I wonder if something is accidentally causing a similar issue to what happened with Nvidia back on 3Dm03 where they were caught not rendering the whole scene, just what was seen? Getting off the benchmark's scripted path allowed one to see rendering shortcuts that shouldn't have been happening, but this time it's more of an accident or bug with early driver not doing what the benchmark wants of it?

yah well the same thing happens with 3d Vantage, 10.6 I get a couple thousand more points than the current driver

and yah Mark11 gains 4,000 points, just by reverting back to an older driver is insane, I see no visual difference between 10.5a and 10.11 in Mark 11 (ATI needs to apply that to 10.12)

and my 4,000 isn't all that much, at Guru3D Hans Dampf gained over 5,000 points reverting to 10.5a!!!! (thats with Quadfire though)
http://forums.guru3d.com/showpost.ph...&postcount=258
LOL



another update:
10.5a worked so awesome for Mark 11, I tried to play Need for Speed Hot Pursuit (2010) and I actually got a few more FPS than with 10.11, but some missing textures

eva2000 Dec 13, 2010 11:58 AM

sorted my psu issues by adding 2nd psu http://i4memory.com/f54/my-first-ret...70/#post138699 :D

14,619 http://3dmark.com/3dm11/164326


demo Dec 13, 2010 12:10 PM

Very impressive score, system and overclocks! :up:

eva2000 Dec 15, 2010 01:37 PM

Quote:

Originally Posted by demowhc (Post 1336445229)
Very impressive score, system and overclocks! :up:

thanks

finally, broke 15k 3dmark11 with HD5970 + 2x HD5870 http://i4memory.com/f54/my-first-ret...70/#post138789 :D

3dmark11 = 15,049 http://3dmark.com/3dm11/189647

i7 980X @4.7ghz
[email protected]/1297 + 2xHD5870 @1030/1282
Asus P6T6 WS Revolution
HX1000 + OP1200

Powermate measured 785 watts from HX1000 psu, so ~340-350w from OP1200 = 1,135 watts for system :)


moshpit Dec 16, 2010 01:54 PM

I honestly feel that scores shown with Cat 10.5 are invalid. Showing them for the sake of pointing out the flaw in 3DM11 is one thing, to try to count it as a valid score is something else all together.

Do not cut out the part saying "Your result has the following problem(s) and will not be shown on leaderboards: Graphics driver is not approved"

Shane_p left this part in and I think that should be required. One can obviously look to the GPUz to see the driver version, but we shouldn't have to be doing that if everybody is posting in good faith.

demo Dec 16, 2010 05:59 PM

Quote:

Originally Posted by moshpit (Post 1336449105)
I honestly feel that scores shown with Cat 10.5 are invalid. Showing them for the sake of pointing out the flaw in 3DM11 is one thing, to try to count it as a valid score is something else all together.

Do not cut out the part saying "Your result has the following problem(s) and will not be shown on leaderboards: Graphics driver is not approved"

Shane_p left this part in and I think that should be required. One can obviously look to the GPUz to see the driver version, but we shouldn't have to be doing that if everybody is posting in good faith.


I disagree, 10.5 is an official driver and I used it on my Tri-CFX system for best performance even up until after 10.11 was released. Nothing wrong with using the best available official driver for CFX imo.

moshpit Dec 16, 2010 06:49 PM

Quote:

Originally Posted by demowhc (Post 1336449372)
I disagree, 10.5 is an official driver and I used it on my Tri-CFX system for best performance even up until after 10.11 was released. Nothing wrong with using the best available official driver for CFX imo.

It may not even be anything wrong. It may be just the opposite. But, as it stands, Cat 10.5 shows unapproved for leaderboard posting, so even if it's doing everything right, it's still "iffy" to trust at best.

There is still the issue of Crossfire though. It may be that Crossfire has been borked by a recent driver in some way.

demo Dec 16, 2010 07:45 PM

Quote:

Originally Posted by moshpit (Post 1336449421)
It may not even be anything wrong. It may be just the opposite. But, as it stands, Cat 10.5 shows unapproved for leaderboard posting, so even if it's doing everything right, it's still "iffy" to trust at best.

There is still the issue of Crossfire though. It may be that Crossfire has been borked by a recent driver in some way.

Fair enough it says unapproved on the leaderboards, but apart from being released long before 3Dmark 11 what is the issue with using older official drivers? the still are official after all.

moshpit Dec 16, 2010 09:10 PM

Quote:

Originally Posted by demowhc (Post 1336449473)
Fair enough it says unapproved on the leaderboards, but apart from being released long before 3Dmark 11 what is the issue with using older official drivers? the still are official after all.

In fact, nothing. Looking more into the issue, it seems something in current drivers may be disabling the third GPU's scaling for some people, while still showing it enabled. If this is the case, then it's only prudent to use the older driver.

Sasquach Dec 16, 2010 10:39 PM

http://3dmark.com/3dm11/202909
P4065
all stock (using 10.10e) and whatever default P setting is in 3dmark basic.

noko Dec 30, 2010 12:39 PM

P4748
Default 10.12 driver settings
Using official drivers 10.12 I still get the warning! :nuts:
Anyways for Q6600 systems single 5870 I am second fastest overall
Had my 5870 at 1095mhz clock speed :drool:, cpu 3.6ghz.
My graphics performance exceeded top runner but physics test dropped me to second place.

I should be able to get in the mid 5000's with a very good cpu with my 5870.

http://3dmark.com/3dm11/313514

moshpit Dec 31, 2010 02:36 AM

Increased overclocking nets new results:

P6682 - http://3dmark.com/3dm11/321167

[email protected], [email protected], [email protected], [email protected]

bigtabs Dec 31, 2010 07:25 PM

A sane entry from me:

P4108 - http://3dmark.com/3dm11/326849

Maverick123w Jan 4, 2011 11:22 AM

http://3dmark.com/3dm11/338099

XFire'd 6950's Flashed to 6970's

3DMark Score P8331
Graphics score 9760
Physics Score 6098
Combined Score 5379
GraphicsTest1 42.75 FPS
GraphicsTest2 51.54 FPS
GraphicsTest3 63.29 FPS
GraphicsTest4 28.04 FPS
PhysicsTest 19.36 FPS
CombinedTest 25.02 FPS

Phenon II [email protected] 4Ghz is really holding my score back. Bet I could hit 10k with Sandy bridge.

greasyhands Jan 5, 2011 05:41 PM

P3498, lower is better, right?;)

http://3dmark.com/3dm11/360605

Tyler582 Jan 6, 2011 12:51 AM

I got P6232 w/the settings in my sig. It's late, maybe I'll try cranking the 580 over 900Mhz later.

KAC Jan 6, 2011 01:57 PM

P4854
Q6600 @ 3.0 Ghz
4 GB ram
Windows 7 64
10.12a hotfix
6950 @ 6970 w/ 950 core / 1425 memory


All times are GMT -5. The time now is 03:13 PM.

Powered by vBulletin® Version 3.6.5
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
All trademarks used are properties of their respective owners. Copyright 1998-2011 Rage3D.com