Differences? 8800M GTX and 9800M GT

Very intersting. But GPU-Z reads a lot of the displayed values from a database. If the information about are not incoporate in the database you get crap values shown.
I had a 9600M GT notebook here. There I got wih GPU-Z also very strange values.

500 core / 1250 shader should be the clock of the 9800M GTS (240 GFlops). It would bet that GPU-Z read of the datebase the clocks of the 9800M GTS.

If nVidia really gets with a G9x with 4 active shader blocks (64 SPs/cores) and 1250 MHz around 360 GFlops, then they would release a bigger corresponding GPU immediately as a 9800 GTX++ since ATI is beating them pretty bad with the 4850 on the desktop markt and the GTX260 ist to expensive.

Perhaps nVidia make a error on the homepage and the 9800M GT ist based on a G92 and has 92 SPs/cores? Then the clocks would be correct. But I belive GPU-Z simply shows wrong values.

 

Err Gpu Z is constantly updated and it's usually pretty factual.
Dunno where you got the idea it isn't.
I'm glad you're 19 and you own a business. I was 14 doing that. I was so young doing this people thought I was smoking rocks.
Ifyou want to compare dongs I"m willin' to.
The bottom line is I'm giving hard numbers and Xeon thought I couldn't afford things that I, well, really easily can afford.

 

Dunno if people remember Paladin's post on page three:

So maybe the nVIDIA site isn't right and it actually is 96 SPs. I agree that something is up if nVIDIAs site has the correct specs, but Paladin and Justin haven't confirmed that they do or don't (to my knowledge -- please correct me if I'm wrong).

I really hope it's just as good as the 8800M GTX, but based on the nVIDIA specs and a couple screenshots Justin threw up, the GPU gurus here have ruled out the possibility that the differences between the 8800M GTX and the 9800M GT are "statistically insignificant".

Paladin and Justin have both been genuine with us since the start about this machine, so when they say that both cards are equivalent, I am going to take them at their word. In my experience they have only given us the facts or hints that wouldn't lead us astray in our extrapolations and suppositions (or if they did, accidentally, it was cleared up in no uncertain terms -- like the sticky on not supporting the 9800M GTX or EE processors).

 

Thats correct. The 9800M GTX is stripped down G92, same as the 8800M GTX. But because of the thermal and power efficiency of 55nm, they have been able to use 112 stream processors, rather than the 96 of the 8800M GTX.

The 9800M GT and GTS are G94.
http://i11.photobucket.com/albums/a170/vairish/20080715_ae1eae3cf442a38629587CLjzz.jpg

 

This could be an old table

 

No, it's been confirmed by the drivers.

NVIDIA_G94.DEV_0622.1 = "NVIDIA GeForce 9600 GT"
NVIDIA_G94.DEV_0623.1 = "NVIDIA G94 "
NVIDIA_G94.DEV_0624.1 = "NVIDIA G94 "
NVIDIA_G94.DEV_0625.1 = "NVIDIA G94 "
NVIDIA_G94.DEV_0626.1 = "NVIDIA G94 "
NVIDIA_G94.DEV_0627.1 = "NVIDIA G94 "
NVIDIA_G94.DEV_0628.1 = "NVIDIA GeForce 9800M GTS"
NVIDIA_G94.DEV_0629.1 = "NVIDIA G94 "
NVIDIA_G94.DEV_062A.1 = "NVIDIA GeForce 9700M GTS"
NVIDIA_G94.DEV_062B.1 = "NVIDIA GeForce 9800M GT"
NVIDIA_G94.DEV_062C.1 = "NVIDIA GeForce 9800M GTS"
NVIDIA_G94.DEV_062D.1 = "NVIDIA GeForce 9600 GT"

@V_C: You sure the 9800M GTX is a G92b?

 

Oh, okay. Thanks for clarifying.

 

But if the 9800M GT is clocked at 1875 (which seems high tbh), it would be pretty much equal to the 8800M GTX. No? 96x1250=1875x64

 

Yes, and a other thing which I forgot about the G94:

The G94 with all units active and the G92 configuration used for the 8800M GTX have the same amount of ROPs (Raster Operation Processors) units. So with a higher core clock the 9800M GT would have a higher pixel fillrate compared to the 8800M GTX. In situation where the pixel fillrate limits the 9800M GT would be faster then the 8800M GTX (depending on the core clock of the 9800M GT).

 

In what situations would a pixel fillrate be limiting?

 

For example in very high resolutions with high AA modes active or older games without much shaders. Modern games are more limited trough the shader output. Normally a 8800M GTX has enough pixel fillrate i realtion to the shader power. But in some situations and games the 9800M GT (if it really clocks on 740 MHz) could be faster.

 

i will ask again....

is there in any way the possibility of the 64 vista allowing for better usage/scores since most other comparisons were run on 32 vista?

or am i missing the point? is there in anyway the OS can affect any of this?

 

@ryujin:

The OS can affect things, but not significantly. A lot of it depends on whether the software is optimized for 64-bit vs. 32-bit. If 3DMark's latest versions aren't optimized for 64-bit, having a 64-bit OS could actually lead to a DECREASE in performance as Vista uses 32-bit emulation. If they ARE optimized, there could be a slight increase in performance, but I think Futuremark tries to rule out advantages/disadvantages caused by OS versions (I could be wrong).

 

thx Ele.
from what i saw of the screenies though, the test was done under 64-bit OS.

this is why i was curious.

 

Here are benchmarks from Sager

"Attached are benchmark scores of Crysis and DevilMayCry4 of the 8800M GTX and 9800M GT.

These were run on exactly same spec'd laptop with the same software including the driver.

NP5796, X9100, 4GB (2GBx2) 200GB 7200RPM, Vista ultimate x64, using 176.02 driver."

First, the 9800M GT Crysis benchmarks. These results will be similar to the differences in the DevilMayCry4 screen shots below

==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 120.58s, Average FPS: 16.59
Min FPS: 11.82 at frame 1938, Max FPS: 19.54 at frame 990
Average Tri/Sec: -15036093, Tri/Frame: -906557
Recorded/Played Tris ratio: -1.01
!TimeDemo Run 1 Finished.
Play Time: 116.83s, Average FPS: 17.12
Min FPS: 10.85 at frame 1984, Max FPS: 19.77 at frame 78
Average Tri/Sec: -15319768, Tri/Frame: -894928
Recorded/Played Tris ratio: -1.02
!TimeDemo Run 2 Finished.
Play Time: 116.81s, Average FPS: 17.12
Min FPS: 10.85 at frame 1984, Max FPS: 19.89 at frame 64
Average Tri/Sec: -15317772, Tri/Frame: -894659
Recorded/Played Tris ratio: -1.02
!TimeDemo Run 3 Finished.
Play Time: 116.85s, Average FPS: 17.12
Min FPS: 10.85 at frame 1984, Max FPS: 19.89 at frame 64
Average Tri/Sec: -15336628, Tri/Frame: -896005
Recorded/Played Tris ratio: -1.02
TimeDemo Play Ended, (4 Runs Performed)
==============================================================

Next, the 8800M GTX Crysis benchmarks...

==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 124.59s, Average FPS: 16.05
Min FPS: 10.01 at frame 1909, Max FPS: 19.27 at frame 974
Average Tri/Sec: -14556376, Tri/Frame: -906816
Recorded/Played Tris ratio: -1.01
!TimeDemo Run 1 Finished.
Play Time: 125.75s, Average FPS: 15.90
Min FPS: 10.01 at frame 1909, Max FPS: 19.27 at frame 974
Average Tri/Sec: -14233392, Tri/Frame: -894957
Recorded/Played Tris ratio: -1.02
!TimeDemo Run 2 Finished.
Play Time: 125.61s, Average FPS: 15.92
Min FPS: 10.01 at frame 1909, Max FPS: 19.27 at frame 974
Average Tri/Sec: -14266192, Tri/Frame: -895997
Recorded/Played Tris ratio: -1.02
!TimeDemo Run 3 Finished.
Play Time: 125.80s, Average FPS: 15.90
Min FPS: 10.01 at frame 1909, Max FPS: 19.27 at frame 974
Average Tri/Sec: -14255367, Tri/Frame: -896665
Recorded/Played Tris ratio: -1.02
TimeDemo Play Ended, (4 Runs Performed)
==============================================================

Some may have a different opinion, but this appears to be statistically insignificant.

 

9800m GT is a bit faster?

 

>.> still don't wanna give up the 9800m gt specs eh?

 

Thanks Paladin, I think that should clear up any questions about the performance of the 9800m GT against the 8800m GTX. At least it does for me.

 

We'll wait for nVIDIA to do that.

Besides, we have already seen what people can do to twist and spin specs, I have always held the position that what is important is the performance.

 

Thanks, I am glad I could help

 

Looks like 1-2 FPS faster in Crysis, slightly over 6% I guess. Still, it might be only Crysis, just pointing out the possibility.

 

Thank you thats the last confirmation that I needed.

 

Thanks for those benchmarks Donald. This is certainly a useful real-world look at the comparative power of the cards. However without wanting to whinge endlessly, I still want to see more (more... MOAR...!!!) extensive benchmarks from whoever can provide them.

As far as the Crysis bench goes, in truth, an average of 16FPS or so is pretty sluggish. I'm not saying that reflects badly on either card since Crysis tends to murder any kind of hardware - just that it suggests these benchmarks were taken with all details set to High (or Higher) and the resolution also probably set pretty high. It would be edifying to know what both cards scored in relation to eachother at settings that more people were likely to use for Crysis (since IMO once frame rates start to get that low, people will start to lower the settings to compensate).

The DMC4 comparison is better in this regard since 30+FPS can be pretty playable. Nevertheless, the more games we see the 9800m GT tested in, the better.

I'm certainly grateful for this info, but still eager to see more as and when it becomes available.

 

Notice the Devil May Cry 4 speed-tests, too (the screenshots are the results pages from each time-demo run). Except for the first scene, where the 9800M GTX averaged about 10 FPS more (which could have been caused by any number of external variables), there is no more than a 1.3 FPS variation in average framerate either way, which would fall SQUARELY under the definition of "Statistically insignificant" (e.g., if you stuck the time-demos side-by-side, you would never know which was which without the hardware ID).

That's why Sager ran time-demos in two different games...that makes it a lot harder to say, "Well it's just game x" .

The point of the time-demos isn't to say that either card is better than the other, although that may actually be the case (if you consider 1-2 FPS better)...it's to prove that what we've been saying all along was truth: the performance difference between the GeForce 8800M GTX and the GeForce 9800M GT is statistically insignificant.

These aren't 3DMark scores...these are real-world benchmarks from two very demanding games dialed up to 11 on the same notebook with the ONLY difference between the tests being the GPU (no different processor, no different RAM, no different hard drive, no different driver, and not even a different serial number on the motherboard).

 

You're right. Sager took these benchmarks at the highest settings possible...which is where you would be most likely to see the "huge" differences between the 8800M GTX and 9800M GT that other posters in this thread were claiming. Having virtually identical performance at those maximum settings proves more than having, say, 30-35 FPS on each card at 1680x1050, because the maximum resolution is where you would really notice the supposed performance difference between the shaders on each model, and that performance difference is exactly what the timedemo results DON'T show.

 

thanks for setting noel straight paladin

edit: language =)

 

Thank you

edit language back

 

I was debating on whether or not to go over a hundred bucks over my preorder payment and sacrifice the 4 gb memory. With these benchmarks, I guess I'll probably take the 9800gt and treat myself to the new 320gb 7200rpm and still be under the preorder price.

 

Thanks a ton Donald! I wasn't really worried about the whole thing since I really don't need a 8800M GTX or comparable card, but it sure put my mind to rest about my $2200 investment.

Plust, Noel was being a jerk (stop the name calling).

 

on topic: nice to see some stats for the sake of the post.

off topic: why is that people slam someone when they were trying to get across what they believe. someone provides a bit of research (both noel and paldin) for everyone to digest however you see fit.
then paladin provides some new information and all of a sudden noel is a jerk and *&%#[email protected] gives? just cause someone has a different understanding and point of view does not make them wrong or a jerk per se.

bah! (ok off my soap box)....

edit: disclaimer - i do not know noel nor paladin and am not taking sides either way when it comes to personalities or expressed feelings in thier posts....did that make sense? lol

 

I didn't see Donald slam anyone, ever, did he?

 

Noel did more than just try to get his point across, read further back, mods had to delete his post.

 

i know....i have read the whole post...
i am not taking sides (thus the attempt at my disclaimer), i am simply stating that noel felt he was right and offered his information (of which everyone seemed attached to until paladin posted the fraps)
if paladin and donald are one and the same this is news to me. nothing was directed personally at either one.
i was simply pointing out that people have views and just because he felt strongly about them does not warrant the attitude expressed after the fact.
noel had an attitude from either thier ponit of view and/or yours as well, but that should not make him a target afterwords....
he flamed as well and i did not like it then either, but i am not moderator nor was i offended as i filtered the flame from the information i was seeking.

way off topic....lol

my bad....

as i said - i am glad paladin posted fraps of the tests, it helps me to be a more informed consumer now when the topic comes up again....

 

Hmmm, look at my Avatar, then my User Name, then my signature...

 

omg!!!!

and all this time i always looked at - i even linked to PWNB thru yur sig too...

*crawls into an ice cave*

 

why avatar?

 

ok - hope this is on topic...
can you provide the op temps for both cards under the same type of stress?
is this possible?

ie - fraps/jpgs....lol not sure what pgm one would use either....

 

^--^
For the moment you will just have to trust me. The performance, heat, and power consumption are essentially equal with no statistically significant difference.

 

Nirvana...My Avatar is a paladin. In medieval literature, the paladins, or Twelve Peers, were known in the Matter of France as the retainers of King Charlemagne.

What is a paladin? I offer the following answer, borrowed from Paladin Ryo. Perhaps best to first explain what separates a paladin from a warrior:

"To some we are Holy Warriors, destroyers of evil, but to me it is deeper than that. Paladins are symbols of all that is good and right in the world. We are the defenders of the defenseless. We are the voice of law, but we are the bearers of Justice. We fight for law and order, but laws can be corrupted, and twisted to fit the needs of the enemy. We are the voice of pure justice when the laws no longer work." "Paladins are the best of two worlds. We hone and partake of the scholarly pursuits, so as to better control and use our warrior abilities."

"Any barbarian can lead a mob, but a paladin will turn a mob into an army."

"From this balance of mind and body a Paladin derives the divine power of Leadership, which radiates out to those we lead, causing them to fight more fiercely, skillfully, and defend better."

"The Calling of the Paladin is of the most high and ancient origin. Long ago, when Men of all races strove against Chaos and Darkness, the Gods sent down grace and leadership unto those among us who were pure of heart and purpose. Those so chosen became shining beacons of leadership and courage to all around them. Each Paladin throughout the ages has been a focus for the forces of Light against Darkness. Thus, each Paladin must ever seek within for the dedication and purity of spirit required to lead and to inspire. We lead, not by shouting orders or commanding from afar, but rather by our example. A Paladin shall ever be the first into battle and the last to retreat, never quavering no matter the odds lest others fail in their own courage as well. A Paladin must ever be a shining example of courtesy and honor -- 'tis better never to have been born than to forswear your oath once you give it as a Paladin. Yet all is not grim, for a Paladin is not some hermit locked in a tower and forbidden the taste of life. No, far from it! For each Paladin may choose to wed in good time and raise a family, such that honor and courage may be taught even to the babe on his mother's knee. The one abiding purpose of a Paladin is to live as an example to others of courage, virtue, and leadership. A wise Paladin never seeks judge over others, but shall never refuse to aid in settling a dispute if aid may be given. If ever the Darkness should conquer the Light, the last gleam shall come from the uplifted blade of a righteous Paladin.

"It is important that you understand that to be a Paladin is not just a profession, or a calling, it is a code and path for living."

 

Do you *actually* fancy yourself as a Paladin?

Petrov.

 

have you been to powernotebooks.com? The tag line is "we live to serve Him, and you!"

 

him = donald = paladin ??

 

I was given that nickname for work I did in a previous life, I did not dream it up myself.

Some may find some of my posts a bit "prickly" as someone once described it, but if you look behind the curtain maybe now you can understand why.

 

Yes, that is our Mission Statement.

 

sorry to label it so inappropriately. lol. I couldn't think of anything better to call it at 2am.

 

I don't understand how a G94 (b??) with 64 sp clocked @ 500 1250 800 can beat a G92 with 96 sp @ same clock and same ROPs (G94 have less ROPs in desktop..)
GPU-Z read the frequency directly from bios not trough a database...SO??

 

We'll im pretty sold. Just going to wait now till the first set of owners get their 5796's and review them before I pull the trigger.

I guess the noise on the 5796, since the GPU's are pretty identical, should be the same at the 5793.

 

Surprising results, at least for me.
Then the 9800M GTX will be a tad better than these 2 card for sure...

 

Oh men... this conversation is getting quite interesting now!

Apart from all of the "fantasy" "metaphysical" kind of talk...
What is going to be the most powerful? The AW M15x (with an 8800GTX) or the M860TU (with the 9800GT)?

 

dx.9/ 10? settings? resolution?