Hp 8510 P - upgrade from hd2600 to firegl v 5600

But I hope that, anyone will found any method (without signed in test mode) to disable the signed drivers on Windows Vista 64 SP1.

 

Shocko i ran spc viewperf here are results , not sure i ran it properly
3dsmax-04 29.36
catia-02 15.21
ensight-03 18.70
maya-02 62.86
proe-04 15.77 t
sw-01 34.37
tcvis-01 9.94
ugnx-01 13.21




Graphics Hardware Configuration
Graphics Accelerator fill in config.txt
Total Graphics Memory
Display Manufacturer/Model
Display Resolution 1280x1024

Viewperf Version 10.0
Viewperf Executable Standard
Price, availability, etc
Price
System Class
Test Date
General Availability
Submitted by
Comments

 

hmm, you have only about half the performance i use to get.
have you checked what clocks the card is running at ? are the settings for powerplay mixed up somehow ?

you should check the clocks of the card, cause that is not looking like 500mhz/600mhz the card should run at, but more like the limited power setting of 300/400mhz.

get everest from http://www.lavalys.com and see how the card is clocked.

these results show that the mod is working, but your card is not clocked properly.

 

I'm an idiot! was running power on demand ! I'll try again at FULL throttle !
i get same results ! and it's at 500-600 mhz ! what could be wrong ???

gpuz

 

this is my report-thanks schoko!!!!

 

here are my 3d report why am i seeing lesser numbers ?
Test # Weight () Frames/sec
1 5.50 26.20
2 5.50 25.70
3 5.50 20.10
4 5.50 19.80
5 17.00 42.80
6 5.50 28.10
7 5.50 27.80
8 5.50 17.10
9 5.50 17.00
10 17.00 32.30
11 5.50 36.40
12 5.50 36.40
13 5.50 29.50
14 5.50 29.40
Weighted Geometric Mean = 28.86

 

As promised. The thermal paste guide is here
http://forum.notebookreview.com/showthread.php?p=3339276

 

@ jerry66 and vietnam8510p

well, somehow i think your cards are still clocked down a little

my results with specviewperf are :

overall xp32 is a tiny bit faster than xp64 in viewportperformance, so your 3dsmax results should be around 35 fps.

did you check powerplay ? what does everest say about your card ?
i have the feeling your graphics card does not apply the 500/600mhz setting.
could you do a quick test with 3dmark06 ? your results should be between 3400 and 3500 with stock clocks. if it´s lower than 3100 there is definitly something wrong with your settings.

 

i can't get any 3-d mark to run . they all just freeze on my comp

 

huh ?
that is strange. i remember that in october last year a release fixed a problem with our c2d laptops. since then actually everybody can run 3dmark.

 

Check futuremark's website for the fix if you already haven't done it.

 

everest sees it as hd 2600 ??? everthing else reports file gl 5600 ?? wtf ??

 

thats ok. of course everest recognizes the device id of the card. 9581 is hd2600.

but the driver thinks the 9581 is a firegl. so it will report to the system that it´s a firegl.

don´t worry about it.

 

do these numbers make sense ?
3DMark Score
6718 3DMarks
CPU Score
7191 CPUMarks N/A N/A
Result name
Untitled
Not specified

Date May 10, 2008 12:23 CDT

 

hmmm, not really, unless it´s 3dmark05.
would be better to have tested 3dmark 06. but if it´s 3dmark05, the score of 6718points would not be that bad.

 

3-d 2005
3DMark Score
6718 3DMarks
CPU Score
7191 CPUMarks N/A N/A
Result name


here is from everest
Field Value
Graphics Processor Properties
Video Adapter ATI Mobility Radeon HD 2600 (HP)
BIOS Version 010.056.000.008.028028
BIOS Date 01/29/08,15:40:3
GPU Code Name M76
PCI Device 1002-9581 / 103C-30C5 (Rev 00)
Process Technology 65 nm
Bus Type PCI Express 1.0 x16 @ x16
Memory Size 256 MB
GPU Clock 500 MHz (original: 500 MHz)
RAMDAC Clock 400 MHz
Pixel Pipelines 4
TMU Per Pipeline 2
Unified Shaders 120 (v4.0)
DirectX Hardware Support DirectX v10
Pixel Fillrate 2000 MPixel/s
Texel Fillrate [ TRIAL VERSION ]

Memory Bus Properties
Bus Type GDDR3
Bus Width 128-bit
Real Clock 594 MHz (DDR) (original: 600 MHz)
Effective Clock 1188 MHz
Bandwidth [ TRIAL VERSION ]

Miscellaneous
Fan Speed 30%
Utilization 0%

ATI PowerPlay (BIOS)
State #1 GPU: 500 MHz, Memory: 600 MHz (Boot)
State #2 GPU: 500 MHz, Memory: 600 MHz
State #3 GPU: 125 MHz, Memory: 252 MHz
State #4 GPU: 600 MHz, Memory: 600 MHz (UVD)
State #5 GPU: 500 MHz, Memory: 600 MHz
State #6 GPU: 300 MHz, Memory: 400 MHz (Limited Power)

 

hmm, everything looks good now. am wondering waht the problem might be.

would you mind running the maya02 test from spec viewperf ?

 

I know what might be the problem! Check in Control panel if you have 3D options set to Maximal Performance. By default they are set to Optimal!
Check if mapping is set to Maximal performance as well.

 

I'll get 3-d 06 and try that . sorry bout that

not running control panel but ccc , should i get rid of ccc and run control panel ?

Shocko i'll run maya

 

CCC is Catalyst Control Panel That is exactly what I meant

 

Destiny , i only have maximal perf in drop box for options with plugged in , and i'm aleways plugged in
results maya-02
Test # Weight () Frames/sec
1 11.00 67.00
2 11.00 93.80
3 11.00 64.10
4 11.00 94.90
5 12.00 170.00
6 11.00 137.00
7 11.00 82.30
8 11.00 82.30
9 11.00 188.00
Weighted Geometric Mean = 101.93

 

well, that looks better now.

you have it.

 

great thanks !

thanks Destiny , have changed settings

 

I meant that:

 

hi schoko! i think something wrong,
Viewperf 3dsmax-04 Report

Viewperf 10.0
fill in config.txt
fill in config.txt
Test # Weight () Frames/sec
1 5.50 19.30
2 5.50 16.10
3 5.50 10.40
4 5.50 10.30
5 17.00 22.60
6 5.50 13.80
7 5.50 21.30
8 5.50 10.50
9 5.50 9.46
10 17.00 21.20
11 5.50 20.30
12 5.50 15.20
13 5.50 17.50
14 5.50 17.40
Weighted Geometric Mean = 16.73
Viewperf maya-02 Report

Viewperf 10.0
fill in config.txt
fill in config.txt
Test # Weight () Frames/sec
1 11.00 62.60
2 11.00 53.80
3 11.00 40.80
4 11.00 91.40
5 12.00 160.00
6 11.00 122.00
7 11.00 83.70
8 11.00 84.00
9 11.00 165.00
Weighted Geometric Mean = 87.49

 

and the everest

 

same "problem" as jerry.

check in CCC what your settings for powerplay are. and look at the 3d settings.
your card does not effectivly use the 500/600mhz settings.

 

@ Destiny, and anyone who's interested.

I'm not sure if you have noticed, but there are only a few 3d-software suites that support Vista 64. Driver support is even scarser. Here's why:

http://arstechnica.com/news.ars/post/20060201-6098.html

Vista 64 (as the only MS os) REQUIRES signed drivers. This means that the provider needs to get a Publisher Identity Certificate (PIC) from MS. This PIC is free, but the issuer needs a VeriSign licence which is $500,- pr year.

And that is why this mod will be extremely dificult to do in Vista 64. We need:
a1) Newer drivers from HP that are certified by SW, Autodesk..(8.44 are certified, HP's are 8.383)
or
a2) Patched FireGL drivers
b) Someone to mod the drivers for 8510p(not the biggest obstacle)
c) Someone to sign the drivers (needs PIC and VeriSign licence)
d) Some way to distribute the drivers (easy enough)

The alternative is to disable the Integrity check:

http://www.geekzone.co.nz/freitasm/3135

Which is a little bit scary imho as it conflicts with KB932596 for Vista x64.
source:
http://www.mstalkonline.com/vista-install-unsigned-hardware-driver-in-x64-bit/

and it doesn't seem to work for everyone (maybe because of KB932596)
source:
http://chris123nt.com/2007/04/28/driver-signing-in-x64/
As I understand it Destiny has experience on modded catalyst drivers for Vista, right?
Have you tried any of these loopholes/fixes for Vista 64? Are you running modded rivers on your Vista64-partition?

 

Everybody knows, I think.
But you can't disable integrity check with SP1. Yet.

 

Im not entirely experience with modded catalysts, as it was all schoko that modified them and provided them. I am going to try to do same thing, either on my own or with schoko. I dont know how well is it going to go

Anyway, what we really need, is to modify the latest FireGL driver from Ati to add our id so it would be detected as FGL5600. They will already be signed, so no problems with the rest of things.
Now... the modification can take a while

 

I'm not quite sure I folllow you. I tried the same ting with the old (8.383-based) drivers from HP, but as soon as you change as much as a single bit in the original file you wil destroy the "signing". All I did was to swap the deviceID of FireGL Mobility 5600 and 2600HD in the .inf file to get the card to be recognized as FireGL. (I did not patch the ogl-driver) CRC should be the same but for some reason the signing got ruined and Vista 64 refused.

@ ed_co
KB932596 is part of SP1 for Vista, so yes, you are right in that there is no fix for SP1. I use the norwegian version, so I don't have to worry... yet

Anyway. I had no luck modding the drivers myself. As I understand it I would need to re-sign the drivers after the changes to get them to work. And for that one needs:

a) PIC
b) VeriSign license

I have neither.

There exists a "crack" to sign unsigned drivers with and it is found here:
http://rapidshare.com/files/101082713/Test.Certificate.rar
(read the .nfo-file)

Also see here:
http://drivercert.oxyhostsfree.com/source/drivercert.zip
http://rapidshare.com/files/101869157/drivercert.zip.html

I do NOT personally vouch for any of these methods, as I've only just DL'ed the first one. Came up clean in KAV 7.0, so I trust there to be no viruses. However there is a big BUT to these drivers (no pun intended):

Apparently driver signing has to be switched off to install the root certificates needed. That means no SP1, guys..


It is good to see that everyone has gotten their FireGL mod to work and that people are posting specviewperf scores. If there are any newer drivers for XP32 (than 8.391.2.111) out for testing, then I'm up for benching (Specwievperf 10 and 3D Mark 06) and testing (SW2007 SP2.2, 3DS Max 8).

As for my experiences with the firegl-mod in XP, it's still all positive. Modelling in SW2007 and 3DS Max 8 is a breeze. The only thing that gives me a bit of lagging is rebuilding complex surfaces after editing a sketch, and I guess that comes down to the number-crunching abilities of the CPU more than anything.

As I've said, I'd like to do a Vista vs. XP comparison but I guess it'll be a while until the Vista64 drivers get modded in a useable fashion. Maybe someone who has Vista32 installed can do a bench of the Vista32-drivers and post?? I'm in the middle of my master-thesis now and do not have the time to reformat my HDD.

The reason is that the OpenGL performance under Vista is said to be 70% worse than that of XP... BUT that is with gaming-drivers. I am curious if the FireGL-drivers fixes this problem and to what extent. I mean, there realy is no point in getting modded driver to work on Vista64 if the performance is going to be less than a third of the performance in XP... right?

Anyways, great to see this thread blossoming. I'm sorry for lack of participation lately, but it's been a hell of a couple of weeks. Anyone up for the Win32 or Win64 drivers (8.391.2.111 schoko), PM me and I'll send a link.

 

Ya, signage of a driver means the whole driver is hashed, likely with CRC32, and the result is encrypted using M$'s secret key.

See, this way the public key in Winduhs can decrypt it, ensuring that the driver maker has paid M$'s astronomical fee.

 

norwegian version doesn't have signed drivers requirement with SP1?

 

well we are already one step ahead now. i guess the drivers for xp32 i send you were 8.44 ( release date december 2007), the ones on hp´s driverpage are a modified 8.391 ( original release august 2007 / and the 8.391.2.1.1.1 was released in october 2007)

now we have a new driver available ( 8.453.1) since 1.may.2008. for this one i managed to make a patchscript for xp 64 and xp 32.

send you a pm with the new xp 32 driver.

 

Stupid M$. Why they are so f***** greedy?
Anyway... seems like we need a cracker now to break the secret key Why do they make it so difficult...

 

They are not greedy, read the article. The PIC from MS is FREE! as in gratis. The only point is to keep the hardware/driver manufacturers accountable. If a signed driver causes a crash it is very easy to trace the driver to the source (Company, programer) and thereby make them accountable (and hopefully fix it). I guess MS got tired of getting bad rep over WinXp and Vista crashes which were mainly caused by bad drivers (31% Nvidia, 12% MS own fault, can't remember the rest..). What costs money is to get the certificate from VeriSign, but there are also certs in the files i provided links for.

The PIC is free, as in free to distribute. It kan be DL'ed from MS for free (as in gratis). I also think all one needs to sign is contained in the links I added in my earlier post.

Norwegian version does not get SP1 for a long while I hear. MS has (as I understand it) prioritized a few versions, among them english. Norwegian is a small language..

@schoko
Thank you very much. I've put it up on my webspace.
Anyone keen on the new xp32-driver (or the aforementioned original 32 and 64 bit XP-drivers from schoko) can PM me for a link.

@all
I will get to testing as soon as possible and post results here.

'till then, guys.

 

Here are my Specviewperf 10 scores with the new-schoko drivers:

3dsmax-04 34.83
catia-02 34.09
ensight-03 23.36
maya-02 109.00
proe-04 33.91
sw-01 70.54
tcvis-01 21.17
ugnx-01 18.14


looks good compared to my results with the original schoko-driver:

3dsmax-04 32.31
catia-02 30.73
ensight-03 23.22
maya-02 108.03
proe-04 28.95
sw-01 64.77
tcvis-01 20.11
ugnx-01 18.18

improvements with new drivers:
3dsmax-04 + 7.8%
catia-02 +10.9%
ensight-03 + 0.6%
maya-02 + 0.9%
proe-04 +17.1%
sw-01 + 8.9%
tcvis-01 + 5.3%
ugnx-01 - 0.2%

I'm very pleased to see an increase of 17.1%, 8.9% and 7.8% in ProE, SW and 3DS Max respectively. The 10.9% increase in catia is also far better than my expectations, but as it's not my fave CAD-suite it's not as great (to me) as the other results.
Supprisingly Ug-NX sees a drop in 0.2% which I believe can be attributed to CCC taking up some system rescources (I didn't run CCC with the old drivers)

Now, on to 3D Mark 06..

 

well, your results for xp32 are looking good.
i made the specviewperftest in 32 bit emulation under win xp64,and the results are very much the same.

the 3dmark results for me were around 3400 points, which is nearly the same performance i had with cat 7.11. since 3dmark06 under xp 64 seems to be slower compared to xp 32 bit, i guess your result will be somewhere around 3450-3500 points with stock clocks.

 

And the scores are in...

Test system:
HP Compaq 8510p
C2D 7700 @ 2.4GHz
4GB OCZ DDR2-6400 @ 5-5-5-15
Ati Mobility HD2600 (256MB GDDR3, no TurboCache ) w/ FireGL-mod

3D Mark 06 on the new schoko drivers: 3420

In comparison to the older drivers:
XP32 stock (HP): 3295
XP32 Omega: 2939
XP32 Schoko (8.391): 3302
XP32 Schoko (8.4): 3420

This means an increase of 3.6% over the original schoko drivers and 3.8% over the stock drivers from HP.

In comparison, on Vista 64 I get about 3200 with the really old 8.383 drivers from HP, which is supprisingly good, IMHO.

On another note: SW 2007 SP2.2, 3DS Max 8 are working perfectly with the new driver. RealView is still available. No complaints despite this driver not being officially supported by Ati/AMD.

Autodesk Inventor 2008 no longer complains about the drivers not being certified.

that's it for now..

 

hmm, i was expecting a little bit more under xp32. with the old catalyst 7.11 i already had 3439 points in 3dmark06 when i still was having xp 32 installed.

since then i tested cat 7.12 and 8.1 and i even got a little bit more performance with the 8.1. and this new firegl driver 8.453.1 should contain the cat 8.4 files for directx.

i guess you could squeeze a little bit more performance out of your machine. might just be a few percent, but if you switch off some of the not necessary services, disable file indexing and disable system restore, you could gain a little bit more performance.

 

I think you have a point. I'm not sure what services I have runing and what I have installed since the last bench, but the CPU-tests were seemingly more choppy than what I remembered from the previous runs. I guess I'll shut down some services and have another go.

Not that I'm not happy with 3400.. Any improvement is welcome, and I plan to work rather than play.

Edit:
Retested now. I shut down some unneeded services from Autodesk and Stopped Avira's Active Guard. I still get 3420. I'll do a run of GRAW2 to see if I experience some gameplay enhancement later.

GPU-Z info:
ATI FireGL V5600

Driver version: ati2mtag 6.14.10.6764 /XP

Clocks stock: 500/600
Clocks actual: 499/594

 

I'm only getting 3201 on 3-d mark 06 . but that is on my work comp with t-7300 , would that lower the score compared to t-7500 ?

 

As two of the tests are CPU-specific I would say it is safe to assume that you could loose a few points, yes.

I would believe the scores to be overall lower as well. By how much, I can't say.

3200 on a T7300 isn't too shabby considering my 3420 on a T7700..
2.0GHz vs 2.4GHz equals a 20% increase...

 

OK, I am out of date, thinking of the WHQL driver certification process. Forgive me, I haven't used Winduhs for ten years. (Debian)

But it does look like Winduhs64 users are SOL unless that crack works.

 

Schoko, it might be hdd, as i think you have 7K200, don't you?

 

yes, you are right, i´m having the 7.2k seagate momentus installed.

but i always thought 3dmark would load into memory and then run the actual benchmark. thats how it should be, but you never know...

 

i notice the hd activity light flashing during test , i have to install the seagate 7200 drive , lazy i am . at least it has 3gb of mem .

 

I guess, with the penryn processor our notebook would be one of the best, if not the best?, CAD workstations

 

hmm, well most likely it´s amongst the very best mobile workstations.

just found this specviewperf result of a quadro fx 3600m in a hp 8710w.

the results that are listed are official, but you never know who tested this, but anyway here are the results


3DSMAX 34.65
Catia 43.58
Ensight 40.04
Maya 151.18
PROE 43.76
Solidworks 84.25
TCVis 23.66
UGNX 24.06

of course these results are a little higher, since the fx3600m is based on the Nvidia 8800mobile chip, and of course its in a 17" brick. so i guess the modded fglv5600 is somewhere between the fx570m and the fx3600m.

not bad. and indeed it´s a cheap and "powersaving" graphics chip. i guess it´s far ahead the other competitors in the performance/watt comparison.

 

Here's a head to head between FireGL V3600 and Quadro FX570. The Performance of our "Mobility FireGL V5600" is damn near equal that of the workstation V3600.

The Quadro FX570 lags a little behind in the SW, 3DS and Maya. A lot in tcvis and ugux. Nvidia takes the win in ProE, Catia and Ensight.

http://www.3dprofessor.org/Reviews Folder Pages/EntryLevelHead/EntryLevelP5.htm

for those of you who don't like to read the whole thing:

"What is conclusive; is that the base line tests from SPECviewperf® 10.0 have returned substantial amounts of indisputable results for the AMD's FireGL V3600 which is exceptional for an entry level professional card. "


naka