Jump to content

GFX-EFI string for Nvidia Geforce 8800GT 512MB (G92)


agrafuese
 Share

112 posts in this topic

Recommended Posts

When will y ou guys understand that xbench is {censored} ? 7300GT perf in it is the same cause xbench is limited and old.

 

Go try some real games, and you'll see how 7300GT is a piece of {censored}. 8800GT IS optimized so don't expect to have any better drivers in 10.5.3.

Link to comment
Share on other sites

@verendus ,

I have Gigabyte 8800GT, turboforce edition (the one with Zalman cooler).It is factory OC-ed that's probably why I have a bit better results than others. But you ARE RIGHT.That card is great in some Windows environment, but under OS X is worst than my old card. Believe me, I had Sapphire X1800GTO which was better performer than 8800GT.I believe too it's due to poor support. Maybe in future we will be able to take something more out of this card, but for now it's good as it is.

 

 

@cuclonefr,

You made your point with xbench man, we heard it all.If 8800GT is so well optimized for OS X, could you explain to me why Call of Duty 2 is soooooo crappy now.I had it with all maxed out, at 1680x1050, it was perfectly playing with my old X1800GTO.All I did was to change graphic card, did nothing to the rest of system.Reinstalled Leopard, just to be sure there was nothing to brake of old system.Now CoD 2 has choppy graphics (like it has problem with acceleration), I couldn't fix with anything. Also for example Collin mcRae Rally is not as fast as it was with old card.I have no other explanation except that Leo is still not optimized for 8800 series....

Link to comment
Share on other sites

I haven't noticed anything like what you are describing with my 8800 GTS 512, it nearly the same card. Something might be amiss with your system. I know X bench is a poor test program because of its age, but my scores are quite a bit higher than yours in the graphics sections.

Link to comment
Share on other sites

I haven't noticed anything like what you are describing with my 8800 GTS 512, it nearly the same card. Something might be amiss with your system. I know X bench is a poor test program because of its age, but my scores are quite a bit higher than yours in the graphics sections.

Post up the scores brother... :P

Link to comment
Share on other sites

Hmmmm, I didn't test WITHOUT QuartzGL, but I didn't notice any slowdown.Before I had ATI X1800GTO which I must tell you was better performing in my system, but I was into experimenting so I took 8800GT.With nvinject, card had choppy graphics, not exactly good.Now everything is good.I will try it without QuartzGL and let you know...[edit]Actually, you are right.I tried now with and without QuartzGL, and system is better performing WITHOUT.Here are results[/edit]@cyclonefr , I tried your string but no luck on my card.I changed device path, but again nothing.If I will have time I will test it at my friend's system, which differs only by motherboard (945 chipset) and let know about results.
Great thread, was able to get my strings up n running in 20 minutes, first shot...thanks allIn regards to the Quartz GL, leave it off... which is not exactly off. you don't want it running system wide, it will slow down your performance. It is there, and as your programs need it...it is switched on on a "as needed" basis. :P just a little info for ya!with my strings ... btwMy scores are nearly the same as before... was using a beta NVinject 2.2But... my performance is much better visually... i'm really happyworking on my sound and ethernet nowwill post files, info, etc...very soon
Link to comment
Share on other sites

I haven't noticed anything like what you are describing with my 8800 GTS 512, it nearly the same card. Something might be amiss with your system. I know X bench is a poor test program because of its age, but my scores are quite a bit higher than yours in the graphics sections.

 

I would like you to write your specs of the machine.Only thing which could have ANY problem are BIOS settings, but old card was working with exactly the same settings.As I said, I didnt add anything else.Took out one card, popped in the other.Generally performance is good, but as I said, before CoD worked well, now it's all....well it is not good.I found somewhere that I shout put vsync on in game settings, but it was just a bit better.However, I repeat, old card was way better working in same system.

 

So, make one screenshot of your bencmark, that we could compare...

Link to comment
Share on other sites

When will y ou guys understand that xbench is {censored} ? 7300GT perf in it is the same cause xbench is limited and old.

 

Go try some real games, and you'll see how 7300GT is a piece of {censored}. 8800GT IS optimized so don't expect to have any better drivers in 10.5.3.

 

cyclonefr is 100% right, xBench should be named as the first official bit of malware for OS X; mostly for wasting everyone's time. Honestly, a random number generator would be more accurate.

Here's an OLD!!! thread that shows how to test graphic cards in OSX.

Please read through it (only 4 pages) and then we can post screenies till cows come home... Just remember to put correct settings into OGLEV when benching.

Link to comment
Share on other sites

post-170453-1204003724_thumb.jpg

 

Edit:Sorry it might just be you are CPU limited?

 

 

My 7900GS KO pretty much gets close to beating your card I get 304 Quartz, 268 OpenGL and 537 User interface. When I put in my new 8800gt today I got similar to what you have which is not the performamce increase you would expect ie almost none at all in my situation. This is just not in Xbench either as the followup poster notes in the OpenGL Extension Viewer Test it is actually even worse 7900GS KO gets ~1800s for lowest score of the tests 8800GT gets ~1300s on all of them so that is roughly a third drop in performance not impressive at all.

Link to comment
Share on other sites

@Long Time Listener: congrats, glad we could help you out.

 

@eddie11c: can you please post your specs? (preferably, make a sig for yourself too ;))

 

@consolation: thanks for that info on OpenMark. here's a link to the file since it looks like the original author's site is down: OpenMark 1.60

 

If anyone else is interested, we should rehash this benchmarker and start posting results. It's not very intuitive, but it's kind of fun if you let it run while you take a break, read the paper and have a cup of coffee :D Basically, what you do is run the full bench mark test and it will progressively add triangles up to 18,000,000 (EDIT: or until it hits 9 FPS?). If you start up the app and let it run on its own without doing the full benchmark test, it will just run a constant benchmark and give you the current frame rate. Just for kicks, here is my (probably crappy) score:

 

OpenMark_result.jpg

 

And for my constant score, I get around ~2000.

 

Yeah, yeah. Let's not take it too seriously :P

Edited by agrafuese
Link to comment
Share on other sites

OpenMark is fine, the OGL 2.1 spec has been around since mid 2006. The 3.0 spec has only been finalised at the end of last year is going to be finalised sometime this year (after yet another delay); I'm pretty sure it's irrelevant for OS X. The problem with xBench isn't that it's old, it's that it was utter {censored}e to begin with.

) Basically, what you do is run the full bench mark test and it will progressively add triangles up to 18,000,000.
I think it adds triangles till your FPS drop below a certain point (9FPS?), iirc, my old G4's mx4 would choke at about 2mln. ;) THAT, was a spectacularly {censored} card...
Link to comment
Share on other sites

OMG, noone is LISTENING to what I am saying.I said about three times that I get WORSE performanse from 8800GT than from X1800GTO!!!

 

I don't care about Xbench, I say what I see generally.I sticked to one game, but the rest too is not playing very well.Sims 2 should work flawlessly now, but they are choppy worst than before! CoD 2 too! Colin mcRae Rally is so-so, but it was BLASTING with old card...

 

I could say that plenty of games and apps are not any faster, even worst than before.THAT is my problem, not the Xbench results....

 

After all, anyone could make better results.I just OC-ed my E6600 to 3GHz, raised PCI Express fsb to 110, and got result of Xbench over 250! But again those games I just said, and I am taking for reference, are playing the same.This will say that bottleneck is not the processor, it's something else that it is braking the system.But I don't know what.... ;)

Link to comment
Share on other sites

OMG, noone is LISTENING to what I am saying.I said about three times that I get WORSE performanse from 8800GT than from X1800GTO!!!

 

ZOMG!!!! We like SO need some objective measurements, OK!!!!! ;)

 

Then, we can see where the problem is. In OS X the graphics performance is only as good as the openGL implementation. SOOOOO... if you use the OGLV, and play with aniso/clip planes etc, you might find the answer. Unless you are using cider versions of games - in which case that's probably the culprit.

Link to comment
Share on other sites

@PPZ: I hear ya man. I just don't have any suggestions, as I am in the same boat as you - minus having ever experienced the X1800GT, so I don't have that to compare to, but regardless, I can see that my 8800GT isn't where it should be.

 

I came from a perfect Tiger installation with an old GeForce 6200. Now I am working with a buggy Leopard installation and this 8800GT (which is the only reason that I switched to Leopard in the first place). So, as you can imagine, it's a bummer for me. It still beats the hell out of having no OS X on my system at all, but sometimes I have to admit that the Tiger installation was running a lot better (in regard to stability and smoothness). That's the only thing that gets me down. Still it's not TERRIBLE...just not where it should be at all. Leopard should be flying with this card, when in actuality it barely gets off the ground :thumbsup_anim:

 

@consolation: are you still running the 10.5.2 beta kexts as indicated in your sig? just curious...Also, I am laughing at myself right now, because my previous post about OpenMark was made before I actually clicked on your link :) In fact, I randomly came across OpenMark on some other thread in my quest for a better bench marker. Anyway, I just edited that post so it makes more sense :weight_lift: Thanks again.

Link to comment
Share on other sites

Yeah we all know CoD 2 is buggy : too much frameskip, it's not a 8800 GT bug, it's a bug with all 8xxx cards. Won't be fixed till Aspyr provides a fix.

 

I dunno about Colin Mc Rae, I tried the demo and it ran at 60fps constant everything maxed. And so is Tomb Raider Anniversary Demo (except when I look at the waterfall at the beginning, it drops at 40fps)

 

I know I tried WoW on lots of GFX cards, and running at 180 fps sometimes was pretty impressive to me, it surely bit any other cards I tried it with (HD2900 XT, etc).

Link to comment
Share on other sites

I'm assuming you didn't use my hex file, right? If you open up my plist file, you will see a line for the VRAM size. Take that value and apply it to your plist, then re-make your hex file and re-add it to your com.apple.Boot.plist. That will do it.

 

Thank you, that was indeed the problem! :(

Link to comment
Share on other sites

Yes, CoD 2 IS BUGGY, but it was working fine with ATI card.Colin mcRae Rally is working ok, but again I can see that it has no framerate as it had with old card.WoW I am not playing, so I can't say anything.Do we know for some good benchmark for Mac ? I still have old card, I will just make an experiment.I will install on external HDD Leopard with ATI drivers, and pop-in my old card to see the difference.Then we can speak. Don't get me wrong.I had X1800GTO, then I gave 200 euro for new card, which is suppose to be at least 200% faster than the old one, and finally I dont see any improvement.It's killing me....

Thank you, that was indeed the problem! ;)
Interesting! My card was recognized automatically.I did no corrections to VRAM size...
Link to comment
Share on other sites

On paper my 8800 should fly...and in some tests...the specs look wonderful, but in real world performance (this is a few days since installing the strings, I'm still not satisfied. My Mac Book Pro with its ATY,RadeonX1600 is twice as snappy on screen.

 

When I really want to make myself cry, i open up my stickies. I have about 40 of them loaded with nothing important. On my macbook, it opens so fast that sometimes it hurts my neck... on my hack, its like watching paint dry.

 

I installed WOW today, tried to start it... got to the log in screen and then my whole system froze :(

 

So while my hack runs pretty nice and at times is so fast its wonderful, there are still things like slow video performance and well... no ability to play wow that make me bum and sleepless in Thailand looking for a solution

 

so yes, i'm not so happy with this 8800 - I have an old 7300GT and am tempted to see how it performs

 

Other than that... its been a fun project and it seems like I'm getting closer and closer to "perfection" Getting this 8800 to do what it should do would be nice

 

:unsure:

Link to comment
Share on other sites

cyclonefr, PPZ, & LTL:

 

Can you guys give OpenMark a try and post your scores? It's by no means a complete or contemporary test of your system, but it will at least give all of us some idea of what you guys are experiencing. If cyclonefr's 8800GT performance is really all that much better than PPZ's & LTL's, then we should see some kind of difference between the two. It's really important to do things like this instead of just simply saying, "my graphics are great!" or "my graphics are not great!"...those are subjective statements that hold no means of comparison. What seems like good performance to one person could be totally {censored}e performance for another.

 

HINT: PLEASE run OpenMark after a fresh reboot and no other programs running!

Link to comment
Share on other sites

Just for comparison:

-> orderd a 8800gt and will post Results with new one as soon as possible.

 

- E6600

- 4GB Ram

- MSI 7900 GTO

 

Openmark 1.6

NVIDIA GeForce 7900 GT/GTO OpenGL Engine

1920x1200x32

Score 23557

 

1152 triangels = 1002 FPS

 

 

Xbench 1.3

Results 211.43

System Info

Xbench Version 1.3

System Version 10.5.2 (9C31)

Physical RAM 4096 MB

Model Mac Pro

Drive Type WDC WD5000AAKS-00TMA0 WDC WD5000AAKS-00TMA0

CPU Test 202.69

GCD Loop 377.96 19.92 Mops/sec

Floating Point Basic 182.81 4.34 Gflop/sec

vecLib FFT 150.30 4.96 Gflop/sec

Floating Point Library 201.40 35.07 Mops/sec

Thread Test 302.36

Computation 291.96 5.91 Mops/sec, 4 threads

Lock Contention 313.54 13.49 Mlocks/sec, 4 threads

Memory Test 243.84

System 289.30

Allocate 382.07 1.40 Malloc/sec

Fill 241.01 11718.31 MB/sec

Copy 277.54 5732.43 MB/sec

Stream 210.73

Copy 195.42 4036.38 MB/sec

Scale 204.34 4221.61 MB/sec

Add 224.30 4778.02 MB/sec

Triad 221.61 4740.78 MB/sec

Quartz Graphics Test 306.77

Line 235.81 15.70 Klines/sec [50% alpha]

Rectangle 329.50 98.37 Krects/sec [50% alpha]

Circle 266.67 21.74 Kcircles/sec [50% alpha]

Bezier 254.88 6.43 Kbeziers/sec [50% alpha]

Text 740.78 46.34 Kchars/sec

OpenGL Graphics Test 267.11

Spinning Squares 267.11 338.84 frames/sec

User Interface Test 522.11

Elements 522.11 2.40 Krefresh/sec

Disk Test 84.41

Sequential 135.10

Uncached Write 135.83 83.40 MB/sec [4K blocks]

Uncached Write 129.54 73.29 MB/sec [256K blocks]

Uncached Read 121.67 35.61 MB/sec [4K blocks]

Uncached Read 158.53 79.68 MB/sec [256K blocks]

Random 61.38

Uncached Write 23.38 2.47 MB/sec [4K blocks]

Uncached Write 203.63 65.19 MB/sec [256K blocks]

Uncached Read 90.61 0.64 MB/sec [4K blocks]

Uncached Read 155.24 28.81 MB/sec [256K blocks]

Link to comment
Share on other sites

Ok, here it is.Full screen, with default number of triangles (1152).But this benchmark is not very reliable.It's changing speed as you open some window or do anything (eg move mouse)....

 

 

But however, here it is...

post-88863-1204066358_thumb.png

Link to comment
Share on other sites

@consolation: are you still running the 10.5.2 beta kexts as indicated in your sig? just curious...Also, I am laughing at myself right now, because my previous post about OpenMark was made before I actually clicked on your link
OOPS... /goes to edit signature/ Reading through that old thread made me realise I went through seven video cards in two years, I should buy some stock in nVidia & ATI -get some of my money back :P.
Ok, here it is.Full screen, with default number of triangles (1152).But this benchmark is not very reliable.It's changing speed as you open some window or do anything (eg move mouse).... But however, here it is...
Uhm... You are supposed to select benchmark from the menu, it will then run a "benchmark." What you are posting is the current FPS....

 

 

BTW, new OGLEV is here http://www.realtech-vr.com/glview/download.html

 

Select test, use these settings so everyone tests the same

 

1280X1024X32 (60Hz) Display Mode

Tick multisample set to 4

Untick LOD, set slider to 0

Anistropy set to 16

Tick Use Fog

 

As you reduce options your results will get higher - but it's just pushing raw polygons. Makes crappy cards look good...

Then new 3.0 version just came out last week, so it has some options I'm unfamiliar with...

 

Attached are some results, check the difference between fastNV and normal render options. Also, attached is a one with the "benchmark" option selected - maybe that one is more useful as it would load up the card more and standardise the settings?

Not much of an improvement on my old 7900gt; seems like nVidia has optimised the card heavily for D3D rather than oGL...

post-57474-1204070901_thumb.png

post-57474-1204070928_thumb.png

post-57474-1204073457_thumb.png

Link to comment
Share on other sites

Select test, use these settings so everyone tests the same

 

1280X1024X32 (60Hz) Display Mode

Tick multisample set to 4

Untick LOD, set slider to 0

Anistropy set to 16

Tick Use Fog

post-24999-1204073681_thumb.jpg

 

What does the benchmark checkbox do? (Since you still need to click the test button)

 

consolation, on your second 2 pics you used multisample 2, not 4 like you suggested.

 

Maybe a better OpenGL benchmarking tool would be using SantaDuck which works with both UT2004 Retail and Demo and gives an average FPS which is easier to translate into end user experience.

 

Here's an example from my system:

 

post-24999-1204073478_thumb.jpg

Link to comment
Share on other sites

What does the benchmark checkbox do? (Since you still need to click the test button)

 

consolation, on your second 2 pics you used multisample 2, not 4 like you suggested.

 

It seems to generate more items at set settings - so the results are more meaningful. IRC it was added in one of the revisions and that's why it wasn't used as it would make it difficult to compare benchies. But, it does seem like a more meaningful test...

 

Fixed the pic, nVFast makes marginal difference now, bugger...

 

BTW, when running 2.1 spec, do your cube textures look like noise? I think it used to show a mandlebrot (sp?) fractal in previous versions.

Link to comment
Share on other sites

 Share

×
×
  • Create New...