**official** 7950gtx Vs 8700gt

Stop insulting people's intelligence KernalPanic....it doesn't help your argument or in any other way. I'm starting to get the impression your defending your purchase decision when you shouldn't, the x205 is an amazing laptop. I also have to disagree about needing more benchmarks, I think enough have been posted to give us an approximate estimate of each GPUs performance. Finally, these benchmarks have been posted by reputable people (Anand, NBR, notebookcheck, etc) so I personally can't argue them.

7950 GTX:

Takt Chip: 575 Mhz
Takt Speicher: 700 Mhz
Speicher: 512 MB
DirectX: 9.0c
Stromverbrauch: 45 Watt
Busbreite: 256 Bit
8700M GT:

Takt Chip: 625 Mhz
Takt Speicher: 800 Mhz
Speicher: 512 MB
DirectX: 10
Stromverbrauch: 29 Watt
Busbreite: 128 Bit

Quake III Arena: 7950 GTX: 551,9/8700M GT: 561,5
Doom III min: 7950 GTX: 189,1/8700M GT: 164,6
Doom III med: 7950 GTX: 190,7/8700M GT: 158,1
Doom III high: 7950 GTX: 189,9/8700M GT: 150
Doom III ultra: 7950 GTX: 176,5/ 8700M GT: 123,6
Fear min: 7950 GTX: 405/8700M GT: 479
Fear med: 7950 GTX: 199/8700M GT: 169
Fear max: 7950 GTX: 85/8700M GT: 53

Its also interesting to note that at higher resolutions the 8700GT is limited by it's memory bus. For example when the resolution in Doom 3 was upped to 1600x1200 - Ultra High , the FPS for the 8700GT was cut in half (123.5 --> 64.4) while the 7950gtx fell by only one-third (176 --> 116.8).

 

at Kernal

although you claim that your verison is inferior and can peform better, there is no way that 2mb of L2 cache .2ghz and 256 more vram can make the difference your talking of. the 8700gt cant even utilize the extra 256mb of vram because of its limited memory bandwidth.

 

well narsnail, look at his name. Panic tells me something.

 

I think most people still have the mis-conception that "memory-bus-bit/bandwidth"
is still directly proportional to card performance in new-generation cards.
With the new nvidia 8x00 unified architecture, this not longer holds true,
so stop saying this is the #1 factor limiting the performance of 8700m.

As you can see, the 8800 Ultra desktop cards have 384bit mem bus,
but in this anantech test, they follow the same downward pattern in both Direct 9 and 10 games as 8700m which has only 128bit bus:
http://www.anandtech.com/video/showdoc.aspx?i=3029&p=5
In this case, the having of 384bit mem does not prevent the 8800 ultra from the drastic drop from 100+fps to 30+fps when moving to higher resolutions.

What I am saying is that mem-bus is no longer the #1 factor affecting performance in unified architecture.
It's something else that need to be optimized by Nvidia, and ATI when using unified architecture!
- which is what we like nVidia to fulfill their promise with better firmware/driver!
(It's not a simple matter of mem-bus!)

 

8800gtx/ultra desktop cards have a 384-bit memory bus planet.

 

TThanks. Corrected. So that says even more that mem-bus is not longer a accurate measure of performance in new-generation cards.

 

I am not sure if lost planet is really the best coded DX10 game out there, desktop 8800 ultra's fps does not have that huge fps drop in Bioshock:
http://www.firingsquad.com/hardware/bioshock_directx10_performance/page6.asp
from 76.5 (1600x1200) -> 45.9 (2560x1600!)...

No matter, memory bandwidth still matter, AA/AF do consume additional memory bandwidth apart from high res.

 

Great Tutorial, but with that being said, i agree the 7950GTX can overpower the 8700 but could current drivers be hindering it from flourishing to its maximum?
the 7950GTX has been out for awhile and when people saw it was a beast they went to it and wanted the best drivers for top performance. but the 8700 GT being semi-new and coming out later after everyone has already gotten their 7950, people still sticking to their 7950GTX there isnt a huge demand for good drivers for the 8700?

it that seems to make sence. its all about supply and demand, and it people fewer people upgrade why waste money/time on the new.

 

The current 8700M-GT drivers are largely matured. Of course they'll get better, with tweaks for a little more performance/compatibility as new game engines come out, but the drivers are no longer a major hindrance.

 

I agree with KernelPanic...

Besides, that anandtech review don´t used an nVidia specs 8700M GT(the one in the X205 is underclocked on the mem bus 700 vs 800). If they used an M570RU witch haves an 800mhz mem bus 8700M GT, that will be most accurate. Iven that the X205 haves an T7300 CPU, with is unfair to compare with other systems that have superior CPUs.

I have an 8700M GT in my M570RU and I compared with other M570RU with an 7950GTX. The differences don´t worth the money... besides the 8700M GT overclocks like well, and the 7950GTX is allready on its max. Another fact is the temperatures, mine runs so cool that I don´t even believe... barely toutchs 60ºC after hours on Bioshock@DX10 with an ambient temp of 25,5ºC. The 7950GTX on the other side, may get temps near 90ºC.

 

Traditional War...

7950 have a lot of potential when we talk about oc... when i puch the limits of my gpu the maximum (in DiRT) temp was about 90º but i got 40 fps @ 1600x1200. so take your conclusion...

What do you think about your machine audigy ?

regards

 

Wow ambient temp of 25ºC ! how did you do it?
My runs normal at 54ºC
and goes up to 65ºC when playing games....

 

How did he do what? "Ambient temp" refers to the ambient temperature, i.e. the air temperature of the room or environment where the laptop is. I think you may be confusing that with the idle temperature of the card.

 

OIC. my confused. thanks.

 

What are your maximum gpu clock limit and which drivers are you using? I oced mine from 575/700 to 605/835 using 163.71...

 

I've done mine to ~625/825, but only in an airconned room... (94.20) 88ºC on master and 82ºC on slave card... ambient temp of 20ºC

 

Well guys the articles are springing up and the products are showing up
SLI woohoo (prbably already posted by many)
Dell dual 8700 and Toshiba dual 8600

I am sooo tempted (buy now pay later LOL )

sorry if everyone has already whooped about these but its a nice performance boost and I have been waiting

 

Hi/Boas

Humm... I don´t think that 25/50Mhz of OC on core is a lot of potential, if you know what I mean... besides the temperature is a very limiting factor when overclocking, and the 7950GTX is not a very cooler card.

The cooling sistem used on the M570RU is the same on the 7950GTX and 8700M GT, so we can take the conclusion that the 8700M GT is almost 40ºC cooler than the 7950GTX on full, and thats a lot...

I haven´t started to OC my 8700M GT yet, but I think that I can reach with no problems 780/800Mhz on core and 950/1000Mhz on mem. As you can see it´s an OC of over 150Mhz on core and 150/200Mhz on mem. Now thats a lot of overclocking potential... the performance can reach something like 20/30% above the specs...

I am thinking to use NiBiTor to edit the core and mem frequencies on the BIOS. But now I don´t have much time to do it...

So far so good

Bioshock runs superb in DX10, it´s a lot more nice than in DX9... the shadows are just amaizing, so real :O ... the smog is very nice too, in DX9 you can´t see that quality smog. The water is more real in DX10 too...

I just recomend everyone to try on DX9 and DX10 and see the diferences...

The only thing that I don´t like so far, is the screen... it´s nice, the 1680x1050 ressolution is the ideal on 17" but the viewing angles are not good. The overall quality is nice, deep blacks, good contrast...

Cumps

 

omg look at this
http://forum.notebookreview.com/showthread.php?t=172564&highlight=8600m+gt+3dmark
OC'd 8600m GT just got me a 5100+ 3DMark score!
as good as a 7950!
I can't imagine sli dual 8600m GT in the new x205!

 

I agree with the stuff posted from Nvidia's tables... i agree with most of the technical info (that can be substantiated) with the exception that he left out the dual-rank 128-bit memory bus...

I argued that he has no basis for his conclusions because:
-He owns neither (or at least hasn't claimed to)
-He isn't comparing apples to apples in that he is taking the results of one laptop then comparing the price difference of the GPUs which are not sold separately. You must order a laptop with a graphics card in it. You can't just compare "$100 more" when the base price of the laptops in question and the performance differences in question are not being taken into account.
-He has no numerical data except Anand's very flawed comparison.


Let's say we are talking about the clevo unit where you CAN have either card in it.
the price is $100 different... now what about performance? The clevo unit ships with the 8700GT factory overclocked. The (underclocked) 8700 in the x205 Anand tested isn't a fair comparison.
We COULD compare the complete price of a laptop with a score like 3dmarks or fps and get a 3dmarks/dollar or fps/dollar
You might note that I posted that the x205 compared to the clevo unit (admittedly old prices now) the x205 with an 8700GT actually BEAT the clevo unit with a 7950GTX in 3dmark06/dollar.

The point is they are comparable... with the 7950GTX having its ups and downs and the 8700 having its ups and downs.

If you are looking to game at 1920x1200, or absolutely MUST have that extra 5-10% at lower resolutions, then buy the 7950GTX.
If 1280x1024 is good enough, then the 8700GT is a comparable card offering comparable performance and DX10 at a lower price, less power, and less heat or noise.

My opinion is fueled with:
-personal experience with at least one of the cards in question
-scientific method (numbers/comparisons/ratios)
-fps/benchmark comparisons that I have done first-hand with the 8700 and 2nd-hand info from 7950GTX owners. (note, I want more results from the 7950GTX crowd... I am quite curious)
-access to a decent number of other laptop via my job

His opinion if fueled with:
-one 2nd-hand and flawed comparison
-theorycraft
-zero personal experience with the cards in question

Did I miss anything?
I am sure he is a fine guy and I realize that his stat comparisons do seem plausible... but I just brought 1st-hand experience to the table and people seem to be discarding it right away... while his opinion might have some merit still remaining, we discard real experience for theorycraft at the drop of a hat?

I argued that his opinions are most certainly not fact.
I said we should test his theories... you just argued that the 7950GTX is obviously superior in every way... I said the 8700 has more potential than Anand tapped...

I feel like the guy who came in to a room full of people who mutually agreed the world is flat.


Worst of all... the 8x00 series is designed for DX10 and Vista... neither of which is mature yet. You are still comparing looking at what the 8x00 series can do but isn't designed for...

 

There is no insult intended so any you detected is imagined...
I disagree with his opinions and I disagree that his opinions are fact.
I posted why.

I don't need to defend the purchase... I love my x205 and its 8700GT.
I wouldn't trade it for a 7950GTX even if the option was available.
it would be hotter and louder and would have cost more for a marginal (IMHO) difference in the resolutions I have access to on this machine.

I am arguing that his opinions are not fact and that his recommendations have no basis. (read the previous post for why)
I am posting that my real results contradict the second-hand results of Anand.

More benchmarks means more dots on the plot-line.
Anand made mistakes and I listed them.
His results do not do the 8700GT justice.

Anand usually does fine, but he isn't perfect.
I am only arguing that he did not represent the 8700GT well.

I am more than willing to read the other comparisons if you will post them.
You listed the numbers but not the methodologies... did they patch? Did they use up-to-date drivers?

 

And notebookcheck is not a reputable or reliable source.

 

Nars, try this...
take two identical computers with the exception being that one has a T7100 and the other a T7500... Now do a 3dmark06 on both.

The result will probably be 300-400 points different because 3dmark06 runs a processor-only test component.
(this is what happened when I did it... I admit the process wasn't a perfect test as I didn't have forever to run it but it gives us a baseline)

Anand's x205-S9359 is at least one processor level higher (in the same family)
Thus his scores should be 150-200 points higher (so he should be getting 4600-4700 or higher if his system is where mine is at and using the same drivers at stock clocks.
Anand's scores are mysteriously lower despite getting a processor-only bias from 3dmark06.

Have you ever run a 3dmark06 and looked at how it scores?
What basis are you using for the following statement?
"there is no way that 2mb of L2 cache .2ghz and 256 more vram can make the difference your talking of."
I want to know where you are getting your info from...

 

I've decided to stop being reasonable with you because your clearly delirious. "Why?", do you ask, "How could I be delirious when im clearly always right?". Here's why:

You talk about backing up results with fact (or you insult narsnail's intelligence by telling him to go look up "fact" in the dictionary....but maybe I just imagined that). Then you go on to completely contradict yourself by stating your opinions about the heat output/temperatures of the 7950gtx right after telling people to use facts. If that wasn't enough you go on to say EVERY SINGLE BENCHMARK WE POST IS INACCURATE AND SAY YOURS ARE BETTER WITHOUT ACTUALLY POSTING ANY OF YOUR OWN BENCHMARKS.

Now go on quote pieces of my paragraph and twist my words. Didn't we say to stop that and actually try forming coherent paragraphs already? I think ill give it a try though:

Maybe you could enlighten us about this dual-rank memory bus so he can add it to his post....

Yet you post your thoughts on a 7950gtx without actually owning one...

What makes his review flawed? Because you got 200 points higher in 3dmark? Did you fail to notice the 7950gtx got 300 points under what it normally does?

1) 3dmarks/dollar ratio is a ludacris performance test, real-world gaming performance is much better.
2) FPS/dollar ratio would be won by the 7950gtx easily at only $100 more.

I may be imagining things again but i'd almost say you forgot to back up any of those statements with facts....or you based them off the benchmarks you just called inaccurate.

I feel like I'm trying to verbally teach a deaf person....

gee that wouldn't bring a biased opinion into it at all...

your ENTIRE post listings doesn't show any benchmarks....why lie?

 

Well, the M570RU can have either graphics card, with everything else exactly the same, and the price difference there is at most $100 (depending on reseller).

Well, the X205 is a very popular laptop, so while it may not be a fair comparison, it certainly is a valid comparison for people considering the X205 or a 7950GTX based laptop. And if you want to compare the Clevo's two graphics cards in performance per dollar, you'll have to decide on a performance guide. If CPU is the same, then 3dmark06 is an okay test of the performance difference, so using that we get:
7950GTX: 3.14 marks/dollar
8700MGT: 2.84 marks/dollar

Using high end configs we get:
7950GTX: 2.45 marks/dollar
8700MGT: 2.42 marks/dollar

These are found assuming maximum 3dmarks at stock clocks (taken from NBR members) and current prices at XoticPC. So the difference may not be much, but the 7950GTX is still always the better buy if performance is the only concern.

You're right, I agree with this. But when looking only at performance, their is no reason to settle for the 8700M-GT.

The "theories" he has about the influence of particular aspects of a GPU on overall performance are indeed fact. His conclusions aren't fact, but they are supported by all available info right now. Yes, the Anand review was detrimental to the 8700M-GT, but even ignoring that review there are no results I've ever seen (and I've looked) showing the 8700M-GT even close (within 5% or so) to the 7950GTX at high settings and resolutions.

No one here can predict the future. Both of these cards are available now so comparing them in modern games is all we can do. Also keep in mind that the 8700M-GT drivers are really quite mature for modern games, nVidia has really done a great job getting the 8m drivers out. No one can predict what better DX10 codepaths will bring, so we really can't base any review off of the assumption that the gap between the 8700M-GT and the Go7950GTX will close with time.

Those results were from a German review IIRC. I'll try to find it again so that anyone who understands German can read about their methodology.

 

kernal you are the ONLY one who seems to completely disagree with what I am saying. although the review maybe flawed, I did not even consider it AT ALL in my guide and threw it in at the end, after I had written the conclusions. Your rebuttles, while you claim does not insult anyones intellgences, are implying that I am stupid and I do not know what im talking about because you think my conclusions are false. I didnt make you think anything, I dont care if you read the conclusions, but the compiled info is very true and the results show themselves. So make up your own friggin conclusion, make your own **** review, because your constantly scrutinizing my every word and I have had it. I did not base my conclusions on any results, my purpose was to show the internal differences between the two cards and what makes each one better. Youve come the the race without a car. Make your own review and support your own "theories" and "conclusion" and see how you like it when I come over ans tell you youre wrong, your dumb, you dont know what your talking about, im the only one that is right and no one else knows what there talking about. I really crave to see your review so get to it.

 

As you can see, odin here knows what he's talking about.

If some super power drivers come out for the 8700 GT that can beat the 7950 GTX, then this thread wouldn't exist. the only thing the 8700 GT has over the 7950 is low res playing and DX10 compatibility

kernels review would be:

Introduction
8700 GT is better because of DX10 compatibility and low resolution gaming
But the 7950 can totally beat me in high res and is better at everything... but none of this counts , obviously, since newer is always better.
conclusion:
Refer to above statement

 

I think Kernal has a very valid point in that he OWNs a 8700m GT, and writes his personal experience "First-Hand".

Whatever reviews we have read, we all get our findings and conclusions from them "Second-Hand".
So is always good to read those reports with these in mind...

Anyway here's is another report that show another powerful aspects of 8700m GT.
=> It scales almost 200% DOUBLE with a SLI setup scoring 3dmarks06 of 8000+!
http://www.notebookjournal.de/tests/327/9

I have not seen such scaling magnitude with 7950 SLI...

 

well, check this thread out. 2nd message there left by kernal...
http://forum.notebookreview.com/showthread.php?t=173294

 

Not sure he's referring in this case to a desktop SLI 7950GTX or Go 7950 SLI....
If it is, it's quite high indeed...

I always have the impressions that SLI setup averagely scale on average 30% higher than single setup in 3dmarks06,
as per notebookcheck table
http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html

So double is very amazingly.

 

Why hasn't anyone compared these cards to the desktop 8600GT and 7950GT... In my personal experience with desktop 6600s and mobile 7950GTX's SLi gives 50%-80% in fps. As for >200%...think about it... I won't state the obvious...

PS thats unless the 8700 SLi drivers are better than the single card drivers, which is... unlikely...

 

im not sure if this was posted

 

Hello all
YES XPS 1730 is OUT Now on dell site

But is it better then the Alienware m9750

from both these which has a better graphic card ?

is true tht the Dell we will be able to upgrade its 8700 to the 8800 cards when they are out ?
If so then isnt the Dell a better buy ?
OR IS THERE ANY OTHER BETTER THAN THIS LAPTOP
Money is not aproblem need a Extream & Fast machine
Please Help

 

i wouldnt buy the dell until the newer cards are released personally, but two 8700gt's will still be very good.

 

When will this happen?

 

im gunna say winter...maybe november at the absolute earliest, im just guessing of course.

 

The 8700M-GT is a very new card and I doubt Nvidia will replace it that soon - it wouldn't be economical for them to do that. If I had to make a guess I'd say Nvidia will release new cards with the Montevina platform in Q2 2008.

 

So you think it won't be untill then that they'll release a 256-bit DX10 style card? I'm betting they get the first 65nm 256-bit card out before then, whatever they decide to call it.

 

It's hard to say but if they're going to release an 8-series mobile card with a 256-bit bus they'll probably do it soon. That's only a guess. I've been hearing the end of the year for the supposed 8800M from some Clevo resellers but that is not concrete at all.

 

Its a shame really... the 8700 GT has the shading power to destroy the 7950GTX but the 128bit bus really cripples it to the point where it can only compete at resolutions that 17" laptops [normally] don't have...

 

People forget how great of a price they are getting an 8700 for. The 128 bit bus drastically lowers the price though I concede that it is not a solution for those looking for the ultimate package. Thats what the new gpu thats coming out sometime this year is for. Stll you can't look at the negatives, its not like the 8700 is more expensive than the 7950 and performs worse.

 

But Dell promises a upgrade when the 8800 cards are out
Also Dell has a better processor x7900 than any in Sager or Cleo 7800 i have seen
Worst part in them is all have same body diff name & layout SUCKS cos all usb ports on same side
While Dell layout is better & has a small LCD on top for gaming soooo KOOL
all im saying is looks do matter YES if the diff is ALOT in speed I wont go for look But here its not much & is upgradble
so wht do others think ?

 

i think this isnt the right thread for discussing this...

and really how much more could it cost to add a 256bit bus to the 8700gt? i dont know if it is an expensive procedure.

 

What about the mobile 9 series? The last 7 series card was released in like.... September? So if the 8 series mobile card gets released in Q2, then that would most likely push back the 9 series.

Remember, it's all about competition. They have NO competition at the moment with ATI so they can slack off. But if ATI's R series (is it?), which they've been working on very hard, comes out with the 9 series and whomps the 9 series, then Nvidia will have to work harder, all the better for us

 

Uh... The Clevo D900C has desktop processors, which are definately better than whatever the XPS can take... And even with Centrino Pro the XPS can't keep up with the m9750 despite the inferior processors...
"Adding" a 256bit bus is expensive as you need to double the size/amount of memory chips on the card, which are the most expensive components...

 

k why do you keep spamming this in multiple forums, you dont like the clevo models you say, but you want to know the difference. if you dont like them then why are you considering them. your just wasting your and other peoples time.

 

Hey men thanks for this I'm confuss what to buy I'm going to buy a notebook for gaming and I'm doing my research. And good job for this, me I'm 30 and i don't know as much as you of computers. I know about racing cars Hondas tuners stuff like that but this no. Feel proud.

 

Narsnail, whatever that means, when the 8800's come out, add them in to the list

 

haha sure thing man, i think i was going to anyways.

 

Eactly, it's 100$ less! and the 8700GT is DEFINATELY fast enough to wait until the 9700GT comes out,, or the 8700ULTRA then you can just SWAP THEM! Use the $100 you saved towards the (most likely) $400 upgrade GPU.