Possible? If so, how? FX5900->Quadro / 6800->Quadro

[terrablader]

Once again the Master's words ring true!

I turned off v-sinc and disabled monitor 2 (on pci 5200 and 3 on the seccond head of the 6800U).

new results!

Run All Summary

---------- SUM_RESULTS\3DSMAX\SUMMARY.TXT
3dsmax-02 Weighted Geometric Mean = 34.50

---------- SUM_RESULTS\DRV\SUMMARY.TXT
drv-09 Weighted Geometric Mean = 130.7

---------- SUM_RESULTS\DX\SUMMARY.TXT
dx-08 Weighted Geometric Mean = 124.8

---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
light-06 Weighted Geometric Mean = 28.57

---------- SUM_RESULTS\PROE\SUMMARY.TXT
proe-02 Weighted Geometric Mean = 48.55

---------- SUM_RESULTS\UGS\SUMMARY.TXT
ugs-03 Weighted Geometric Mean = 79.16


Also I just got the motherboard in that lets me put the e6600 Conroe with an AGP GPU so going to post that up soon!

[pedrolito]

I have been doing some overcloking tests on my mobo, and the cpu frequency doesn't affect a lot the opengl performance. You should get the same results with your Conroe.

By the way I bought myself a E6300 and my 6800GT PCIe will soon be arriving, I will also post some results.

[terrablader]

that makes sense, how about things like rendering time and stuff wouldnt the conroe help with that?

[AlecRyben]

Conroe kicks ash in rendering...

[pedrolito]

This is different. During rendering times, the "future" is known, so many processors can work at the same time and at full speed. Typically, you will see that a core renders the top half of an image and the other core renders the other half.

While doing "real time" stuff, the future is unknown and therefore is it harder to program multithreaded softwares.

By the way in games I think only Call of Duty 2 and maybe Quake 4 use a little bit the second core and the same applies for SLI.

Same applies for CAD: Autodesk is absolute **** in using multicore systems!!!
UGS NX is the best programmed regarding multicore/SLI systems.

[pedrolito]

Holly crap!!! Using Clokkevi's tips for unchecking Vsync, MemCAD went through the roof!!! 300fps!!!

Actually the 60fps limit comes from the refreshing speed of my monitor which is as you guess 60Hz, sorry i didn't spot this one.

[AlecRyben]

Originally Posted by chalamadad

even while editing in the viewport set to wireframe/shaded mode?

Yes, tested and confirmed. No difference between a Quadro and a GeForce in Cinema4D.

[pedrolito]

Specviewperf9 results on my 6800GT AGP@QuadroFX4000

ugnx-01 Viewset
-------------------------------------------------------------------------------
Test Weight Frames DLB Visual Double Frame Buffer Depth Stencil
# Per Sec Sec ID Buffer Red Green Blue Alpha Buffer Buffer
-------------------------------------------------------------------------------
1 9.00 47.0 N/A 7 True 8 8 8 0 24 0
2 9.00 24.2 N/A 7 True 8 8 8 0 24 0
3 9.00 4.36 N/A 7 True 8 8 8 0 24 0
4 9.00 4.35 N/A 7 True 8 8 8 0 24 0
5 9.00 9.69 N/A 7 True 8 8 8 0 24 0
6 9.00 5.60 N/A 7 True 8 8 8 0 24 0
7 9.00 3.30 N/A 7 True 8 8 8 0 24 0
8 9.00 21.4 N/A 7 True 8 8 8 0 24 0
9 9.00 15.4 N/A 7 True 8 8 8 0 24 0
10 9.00 2.14 N/A 7 True 8 8 8 0 24 0
11 10.00 6.22 N/A 7 True 8 8 8 0 24 0
-------------------------------------------------------------------------------
ugnx-01 Weighted Geometric Mean = 8.483

[chalamadad]

Originally Posted by AlecRyben

Yes, tested and confirmed. No difference between a Quadro and a GeForce in Cinema4D.

That is strange because I was talking with the tech support at Maxon today and he told me he would experience better performance using a Quadro. So he told me it's a good idea to get one. Maxon is listed on the Nvidia Quadro homepage too.

Did you also copy the ...ws...inf files over like clokkevi suggested?

[pedrolito]

Could someone with

[pedrolito]

Could someone with a 6800GT PCIe @Quadro post some results with Specviewperf9 or 8.1?

I think the FX4000 is being held back by the AGP bus, I'd like to check to be sure.

[pedrolito]

My computer is an A64 3000+ @2.52Ghz s754 with 1Gb of PC3200 DDR. The driver version is 91.36

SpecviewPerf 8.1

POV 6800GT AGP @ Quadro FX4000
3dsmax-03 = 35.55
catia-01 = 26.06
ensight-01 = 25.42
light-07 = 23.90
maya-01 = 52.15
proe-03 = 43.01
sw-01 = 25.77
ugs-04 = 34.66

Genuine Quadro FX4500, taken from here:
Tested on a Dual Opteron 275 with 2GB of dual channel PC3200, 77.56 drivers
3dsmax-03 = 41.51
Catia-01 = 30.71
Ensight-01 = 23.64
Light-07 = 23.69
Maya-01 = 57.47
Proe-03 = 64.11
Sw-01 = 32.12
Ugs-04 = 50.29



Specviewperf 9

POV 6800GT AGP @ Quadro FX4000
3dsmax-04 = 18.44
catia-02 = 22.28
ensight-03 = 13.93
light-08 = 20.41
maya-02 = 67.99
proe-04 = 16.48
sw-01 = 25.84
ugnx-01 = 8.494
tcvis-01 = 6.809

Genuine Quadro FX4500, taken from here:*
Tested on a Core2Extreme X6800 with 2Gb of ECC PC4200 DDR2, 91.36 drivers
3dsmax-04 = 35.63
catia-02 = 44.66
ensight-03 = 24.41
light-08 = 39.5
maya-02 = 106.16
proe-04 = 36.92
sw-01 = 54.95
ugnx-01 = 11.03
tcvis-01 = 13.63

*: this values are not very accurate, sometimes one FX4500 does better than a SLI of 4500s in applications optimised for SLI...

[pedrolito]

Looking at this link: http://www.nvidia.com/object/IO_24673.html

Show that FX4000 AGP and PCIe are slightly different.

AGP8x, 3D Primitive Perf:
Triangles per Second 133 Million
Texels per Second/Fill Rate 4.5 Billion

PCIe, 3D Primitive Perf:
Triangles per Second 150 Million
Texels per Second/Fill Rate 5.1 Billion

[AlecRyben]

Of course they are different, the FX4000 has 4 locked pipelines and much lower clocks than the FX4400.
IMHO, the Maxon guy was suggesting you buy a Quadro because of the stability promised by using certified drivers, and not because of the increased speed. You can easily run their own benchmarks and see for yourself, there will be no visible difference between a GeForce and a Quadro.
My modded 6800 Ultra AGP easily beats the PCI-E Quadro FX 4400, so the AGP interface is definitely not a bottleneck.

[pedrolito]

But let's say, if I buy myself a 6800GT PCIe and mod it to Quadro, will it be an FX4000 or an FX4400?

The problem is that windows still sees a 6800GT. Some people say that the 6800GT PCIe becomes an FX4000, but I have seen someone doing SLI of softmodded 6800GT (by flashing the bios), and that is only possible with an FX4400.

So could someone post Specviewperf 9 results of a modded 6800GT PCIe? So we can be sure of which Quadro it becomes. As results only differ in SPec9 and not in 8.1 or lower.

EDIT: you say your 6800Ultra AGP beats an FX4400, but in which benchmark? Did you try the Spec9? This is the only benchmark where the true power of FX4400/FX4500 kicks in!

[AlecRyben]

Name is not important, as long as the chip is NV40, the cards are practically the same thing.
You cannot do SLI on AGP only because the downstream bandwith is much lower than the PCI-E and because there are no dual AGP slots on any motherboard.

[pedrolito]

Why is there almost no difference between an FX4500 and my quadro under Spec8.1, and so much difference (twice as much) under Spec9? Are the bench results "tweaked" to suit the FX4500?

Concerning SLI I was talking about PCIe, I know that SLI doesn't exist for AGP cards.

EDIT: ok I got it, FX4500 is based on g70 gpu.

[pedrolito]

I have just found an excellent website showing test results under Specviewperf9 and different quadros.

Moreover, the CPU actually is VERY IMPORTANT!!! There is a comparison between a PentiumD930 and a C2DE6600, and there is almost +50% increase in the test results!!!!

These Conroes are tha' bomb'! I am glad I got myself one

The link to the tests: http://www.servodata.com.pl/testy/cad.htm

[chalamadad]

Originally Posted by AlecRyben

Of course they are different, the FX4000 has 4 locked pipelines and much lower clocks than the FX4400.
IMHO, the Maxon guy was suggesting you buy a Quadro because of the stability promised by using certified drivers, and not because of the increased speed. You can easily run their own benchmarks and see for yourself, there will be no visible difference between a GeForce and a Quadro.
My modded 6800 Ultra AGP easily beats the PCI-E Quadro FX 4400, so the AGP interface is definitely not a bottleneck.

He was referring to performance - said it was actually more comfortable. I ordered a 6800 Ultra today and will try to find out what the truth is. After all, even unmodded, it's still a good card given the low price. Worth a try I guess.

And if you're interested in this, I also asked the guy if SLI would enhance Cinema4D performance. This is negative and is not supported in the near future.

Did you use the latest version of Cinebench? Do you think it is possible Cinebench is behaving differently compared to real Cinema4D? I wonder if those quadro driver profiles are valid for both apps. If not, it is fairly clear that there is no difference being shown.

[AlecRyben]

Yes, it's possible Cinebench is behaving differently but since I don't use Cinema4D, the only thing i can run is Cinebench.

[terrablader]

ok well i am reporting back after trying the

6800 U OC @ FX4000 with new strange motherboard Asrok Dual-VSTA this lets you put the conroe with AGP card. My idea was/is to get top performance for rendering, and still have Quatro like graphic OpenGL power.

using also 2gb of 667mhz ram. 500 watt power supply (wondering if this is the problem or the strange $50 mobo or ????

The frame rate in Specview Perf. 7.1 did not change much. However in Solidworks rendering time is about half, opening closing files half the time. Very nice.

I am getting some stability issues. Specview is crashing half way through some of the tests! (pc reboots)

Open GL games crash to desk top.

I had to reinstall drivers, and solidworks once solidworks started crashing during render. (freeze up)

was not getting this kind of problem with my old intel 875pbz motherboard and P4 3.4ghz. I am wondering what is up. Where might this instalbilitly suddenly be coming from this frankenstien motherboard?, i dont think the CPU is too hot its about 35cel.-50cel.

Looked around in RivaTuner again at what options could be tinkered. V-Sinc is off in nvidia panel. using 77.56 drivers.

kind of an open ended question but anyone got an idea why getting some crashes like this suddenly.


side note.
Also other info. my FX5950 on the other machine appears to be rendering soldiworks in realview without having modified the 77.56 drivers. just rivatuner was enough. however not getting true 3500fx frame rates so its not really true Quadro mod. This machine had the 6800 in it before so something inadvertantly started working there, as result of previouse hardware mod? strange.

R

[AlecRyben]

NV30 and NV35 cards can also be modded with the same NVStrap procedure as the NV40 cards. That's why the mod can be "inherited" if you put an NV30 into a slot previously occupied by NV40.

GeForce FX 5950 cannot be compared to Quadro FX 3500 because FX 3500 is a 20 pipeline G71 card with 7 vertex shaders while the GeForce is a 4x2 pipeline, 3 vertex shader NV35 card.
You can compare GeForce 5900 and 5950 to Quadro FX 3000.

[terrablader]

** edit ** got it stable now was a setting in bios enable flexibl RAM..

quick question:

in RivaTuner when toggling back from the Quadro to the Geforce should the card have professional unchecked? and uncheck straps? or just select Geforce and apply? Also is a reboot manditory?

also should Reinstall be clicked!! or is apply enough. Seems I am getting real stability issues now when gaming with it on Geforce select. Thanks again..

[pedrolito]

** edit ** got it stable now was a setting in bios enable flexibl RAM..

This means you were overclocking your Ram to timings/fsb it couldn't handle. "Flexible" option means that if the Ram can't keep up, it can underclock itself automatically thus avoiding crashes.

Can you post some Spec 8.1 or 9 results using your E6600?

[terrablader]

Bechmarking the E6600 / 667 Ram 2gb / 6800U@FX400 / Asrok dual vsta mobo
upgrade $465 (pc3200 ugrade optional since mobo takes both dd2 and dd1)

did a couple animation renderings in sw07 last night works fine quite stable, after tweeking the bios (off onboard stuff etc) game test was not so stable but that might be the rivatuner hack etc. game might not like 77.1 forware Quadro base drivers.

1280x1024 32bit color no programs runing clean install xp pro sp2 vert sinc off all other options default..

Here is results for 7 8 9 spec views: what do you think? there is huge x2 speed increase opening closing, saving, rendering solidworks files.

my older system P4 3.4, 875 chipset, pc3200 2gb. i posted below in quote my 7.1 tests with my older system dont remember my fps in 8 and 9 test with older system.

7.1

Run All Summary

---------- SUM_RESULTS\3DSMAX\SUMMARY.TXT
3dsmax-02 Weighted Geometric Mean = 42.80

---------- SUM_RESULTS\DRV\SUMMARY.TXT
drv-09 Weighted Geometric Mean = 186.5

---------- SUM_RESULTS\DX\SUMMARY.TXT
dx-08 Weighted Geometric Mean = 157.6

---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
light-06 Weighted Geometric Mean = 36.56

---------- SUM_RESULTS\PROE\SUMMARY.TXT
proe-02 Weighted Geometric Mean = 56.15

---------- SUM_RESULTS\UGS\SUMMARY.TXT
ugs-03 Weighted Geometric Mean = 79.27


Here is previose results with P4 3.4 system.

Run All Summary

---------- SUM_RESULTS\3DSMAX\SUMMARY.TXT
3dsmax-02 Weighted Geometric Mean = 34.50

---------- SUM_RESULTS\DRV\SUMMARY.TXT
drv-09 Weighted Geometric Mean = 130.7

---------- SUM_RESULTS\DX\SUMMARY.TXT
dx-08 Weighted Geometric Mean = 124.8

---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
light-06 Weighted Geometric Mean = 28.57

---------- SUM_RESULTS\PROE\SUMMARY.TXT
proe-02 Weighted Geometric Mean = 48.55

---------- SUM_RESULTS\UGS\SUMMARY.TXT
ugs-03 Weighted Geometric Mean = 79.16

8.1

Run All Summary

---------- SUM_RESULTS\3DSMAX\SUMMARY.TXT
3dsmax-03 Weighted Geometric Mean = 46.41

---------- SUM_RESULTS\CATIA\SUMMARY.TXT
catia-01 Weighted Geometric Mean = 34.51

---------- SUM_RESULTS\ENSIGHT\SUMMARY.TXT
ensight-01 Weighted Geometric Mean = 23.82

---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
light-07 Weighted Geometric Mean = 32.59

---------- SUM_RESULTS\MAYA\SUMMARY.TXT
maya-01 Weighted Geometric Mean = 75.59

---------- SUM_RESULTS\PROE\SUMMARY.TXT
proe-03 Weighted Geometric Mean = 0.00000

---------- SUM_RESULTS\SW\SUMMARY.TXT
sw-01 Weighted Geometric Mean = 30.24

---------- SUM_RESULTS\UGS\SUMMARY.TXT
ugs-04 Weighted Geometric Mean = 41.12

8.1 on older p4 system

Run All Summary

---------- SUM_RESULTS\3DSMAX\SUMMARY.TXT
3dsmax-03 Weighted Geometric Mean = 16.73

---------- SUM_RESULTS\CATIA\SUMMARY.TXT
catia-01 Weighted Geometric Mean = 11.85

---------- SUM_RESULTS\ENSIGHT\SUMMARY.TXT
ensight-01 Weighted Geometric Mean = 15.86

---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
light-07 Weighted Geometric Mean = 9.199

---------- SUM_RESULTS\MAYA\SUMMARY.TXT
maya-01 Weighted Geometric Mean = 21.48

---------- SUM_RESULTS\PROE\SUMMARY.TXT
proe-03 Weighted Geometric Mean = 16.63

---------- SUM_RESULTS\SW\SUMMARY.TXT
sw-01 Weighted Geometric Mean = 14.44

---------- SUM_RESULTS\UGS\SUMMARY.TXT
ugs-04 Weighted Geometric Mean = 19.13

9.0.3

Run All Summary

---------- SUM_RESULTS\3DSMAX\SUMMARY.TXT
3dsmax-04 Weighted Geometric Mean = 23.60

---------- SUM_RESULTS\CATIA\SUMMARY.TXT
catia-02 Weighted Geometric Mean = 27.34

---------- SUM_RESULTS\ENSIGHT\SUMMARY.TXT
ensight-03 Weighted Geometric Mean = 13.55

---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
light-08 Weighted Geometric Mean = 26.77

---------- SUM_RESULTS\MAYA\SUMMARY.TXT
maya-02 Weighted Geometric Mean = 22.93

---------- SUM_RESULTS\PROE\SUMMARY.TXT
proe-04 Weighted Geometric Mean = 23.99

---------- SUM_RESULTS\SW\SUMMARY.TXT
sw-01 Weighted Geometric Mean = 29.88

---------- SUM_RESULTS\TCVIS\SUMMARY.TXT
tcvis-01 Weighted Geometric Mean = 5.421

---------- SUM_RESULTS\UGNX\SUMMARY.TXT
ugnx-01 Weighted Geometric Mean = 9.262

can anyone can advise on posted quote below? thanks much!

in RivaTuner when toggling back from the Quadro to the Geforce should the card have professional unchecked? and uncheck straps? or just select Geforce and apply? Also is a reboot manditory? also should Reinstall be clicked!! or is apply enough. Seems I am getting real stability issues now when gaming with it on Geforce select.