Meaker's Notebook graphics charts.

(Please note that the DX10 and DX9 sections performance levels are NOT comparable. ie just because a card is in the low end DX9 does not mean its similar to the DX10 lowend cards)

DX10:

(If you cant find your card, make sure you check the DX9 section).

High end - 512mb Useful
8800 GTX, 96 Stream processors, 16 ROPs, 256bit mem (500/1250/800)
8800 GTS, 64 Stream processors, 16 ROPs, 256bit mem (500/1250/800)

High mid range - 512mb Useful
8700 GT, 32 Stream processors, 8 ROPs, 128bit mem (625/1250/800)
HD2600XT, 120 Stream processors, 4 ROPs, 128bit mem (600-700/700-750)

Mid range - 256mb Useful
8600 GT = 32 Stream Processors, 8 ROPs, 128bit mem (475/950/700)
HD2600 = 120 Stream processors, 4 ROPs, 128bit mem (400-500/550-650)

Low mid range - 256mb Useful
8600 GS = 16 Stream Processors, 8 ROPs, 128bit mem (600/1200/700)
8400 GT = 16 Stream Processors, 8 ROPs, 128bit mem (450/900/600)

Low end - 128mb Useful
HD2400XT = 40 Stream Processors, 4 ROPs, 64bit mem (500-600/600-700)
8400 GS = 16 Stream Processors, 4 ROPs, 64bit mem (400/800/600)
HD2400 = 40 Stream Processors, 4 ROPs, 64bit mem (350-450/400-500)
8400 G = 8 Stream Processors, 4 ROPs, 64bit mem (400/800/600)

Business chips and their ordinary counterparts: (be aware clock speeds are likely to be different)

ATI:

FireGL V5600 = HD2600

Nvidia:

Qyadro NVS 320m = 8700M GT
FX 1600m = 8700M GT
FX 570m = 8600M GT
Quadro NVS 140M = 8400 GT
Quadro NVS 135M = 8400 GS
FX 360m = 8400M-GS
Quadro NVS 130M = 8400 G

Nvidia = (Core clock/Shader clock/Mem speed)
ATI = (Core clock/Mem speed)

Memory speeds quoted are actual frequencies, not DDR.
--------------------------------------------------------------

DX9:

Ultra High end - 512mb useful
--------
7900GTX SLI = 24 PSUs, 8 vertex units, 16 ROPs, 256 bit mem (500/600) *2 SM3 (90W)
7800GTX SLI = 24 PSUs, 8 vertex units, 16 ROPs, 256 bit mem (400/550) *2 SM3 (132W )

High end - 256mb/(512mb top cards) useful
--------
7950GTX = 24 PSUs, 8 vertex units, 16 ROPs, 256 bit mem (575/700) SM3 (45W)
7900GTX = 24 PSUs, 8 vertex units, 16 ROPs, 256 bit mem (500/600) SM3 [A] (45W)
X1900 = 36 PSUs, 8 vertex units, 12 ROPs, 256 bit mem (398/468) SM3 (clocks provided by Sunjhoon)
X1800 XT = 16 PSUs, 8 vertex units, 16 ROPs, 256 bit mem (550/650) SM3
7800GTX = 24 PSUs, 8 vertex units, 16 ROPs, 256 bit mem (400/550) SM3 [D] (66W)
7900GS = 20 PSUs, 7 vertex units, 12 ROPs, 256 bit mem (375/500) SM3 [A] (20W)
X1800 = 12 PSUs, 8 vertex units, 12 ROPs, 256 bit mem (450/500) SM3 (35W?)

Mid-High end - 256 mb useful
--------
x800xt = 16 PSUs, 6 vertex units, 16 ROPs, 256 bit mem (480/550) [E]
6800U = 12 PSUs, 5 vertex units, 16 ROPs, 256 bit mem (450/550) SM3
7600GT = 12 PSUs, 5 vertex units, 8 ROPs, 128 bit mem (600/500) SM3
7700 = 12 PSUs, 5 vertex units, 8 ROPs, 128 bit mem (~540/500) SM3 (450/400 reported by m447hew)
x800pro = 12 PSUs, 6 vertex units, 12 ROPs, 256 bit mem (400/400)
7800 = 16 PSUs, 6 vertex units, 16 ROPs, 256 bit mem (250/329) SM3 (35W)
6800 = 12 PSUs, 5 vertex units, 12 ROPs, 256 bit mem (300/300) SM3

Mid range - 256mb useful
--------
x1700 = 12 PSUs, 5 vertex units, 4 ROPs, 128 bit mem (?/?) SM3 [x1600 with Avivo (video) support] (17W)
x1600 = 12 PSUs, 5 vertex units, 4 ROPs, 128 bit mem (470/470) SM3 [C] (17W)
7600 = 8 PSUs, 5 vertex units, 8 ROPs, 128 bit mem (~450/500) SM3 (16W)
x700 = 8 PSUs, 6 vertex units, 8 ROPs, 128 bit mem (350/300)
6600 = 8 PSUs, 3 vertex units, 8 ROPs, 128 bit mem (300/300) SM3

Low end - 128mb useful
--------
x2300 = 4 PSUs, 2 vertex units, 4 ROPs, 128bit mem (?/?) SM3
x1400 = 4 PSUs, 2 vertex units, 4 ROPs, 128 bit mem (432/400) SM3
x1300 = 4 PSUs, 2 vertex units, 4 ROPs, 128 bit mem (hyper mem) (400/325) SM3
7400 = 4 PSUs, 3 vertex units, 4 ROPs, 64 bit mem (450/450) SM3 (10W)
x600 = 4 PSUs, 2 vertex units, 4 ROPs, 128 bit mem (400/250)
x600se = 4 PSUs, 2 vertex units, 4 ROPs,128bit mem (358/277)
9700 = 4 PSUs, 2 vertex units, 4 ROPs, 128 bit mem (450/213)
9600 = 4 PSUs, 2 vertex units, 4 ROPs, 128 bit mem (333/? (240 turbo))
6400 = 4 PSUs, 2 vertex units, 4 ROPs, 64 bit mem (400/350) SM3
7300 = 4 PSUs, 3 vertex units, 4 ROPs, 64 bit mem (350/350) SM3 (7W)
7200 = 4 PSUs, 3 vertex units, 4 ROPS, 64 bit mem (?/?) SM3 (reported 450/350 by mobius1aic)
x300 = 4 PSUs, 2 vertex units, 4 ROPs, 64 bit mem (300/230)
6200 = 4 PSUs, 2 vertex units, 4 ROPs, 64 bit mem (300/300) SM3

Ultra low end - N/A
--------
xpress1250 = 4 PSUs, 0 (software) vertex units, 2 ROPs, system memory only (400/ N/A) SM2b [F]
7050 = 2 PSUs, ? vertex units, ? ROPs, system memory only (425/N/A) SM3
6150 = 2 PSUs, 1 vertex unit, 1 ROP, system memory only (475/N/A)** SM3 (5.6W)
6100 = 2 PSUs, 1 vertex unit, 1 ROP, system memory only (425/N/A) ** SM3 [F]
xpress1150 = 2 PSUs, 2? vertex units, 2 ROPs, 128^ bit & 32* bit mem (400/ (SYS MEM & ?*)) (<10W)
xpress1100 = 2 PSUs, 2? vertex units, 2 ROPs, 128^ bit & 32* bit mem (300/ (SYS MEM & ?*))
xpress200 = 2 PSUs, 2? vertex units, 2 ROPs, 128^ bit & 32* bit mem (300/ (SYS MEM & 350*))
5x00 = 4 PSUs, (Array) vertex units, 4 ROPs, 128 bit/64 bit mem (%/%)
All other integrated solutions (GMA 900, 950 etc.)

[A] = http://www.anandtech.com/mobile/showdoc.aspx?i=2766
= http://www.anandtech.com/mobile/showdoc.aspx?i=1957
[C] = http://www.anandtech.com/video/showdoc.aspx?i=2632
[D] = http://www.anandtech.com/video/showdoc.aspx?i=2557
[E] = http://www.anandtech.com/video/showdoc.aspx?i=2437
[F] = http://techreport.com/reviews/2007q1/amd-690g/index.x?pg=9

*Dependant on memory attached to side port
** Performance of these two is identical
^system mem bus width
% Covers entire series, pretty awful due to high power consumption and generally poor arc, 5700go would rate a little higher but those above it perform much better in DX9 titles such as HL2 therefore they get stuck at the bottom.

Business chips and their ordinary counterparts: (be aware clock speeds are likely to be different)

ATI:

FireGL v5000 --> x700
FireGL v3200 --> x600
FireGL v3100 --> x300
FireGL T2 --> 9600
FireGL 9000 --> 9000 (old) (DX 8)
FireGL 7800 --> 7500 (V. old) (DX 7)

Nvidia

Quadro FX 2500M --> 7900GTX
Quadro FX 1500M --> 7900GS
Quadro NVS 510M --> 7900(not sure which suffix)
Quadro Go1400 --> 6800
Quadro NVS 300M --> 7600
Quadro FX 350M --> 7300
Quadro NVS 120M --> 7300
Quadro NVS 110M --> 7200
NVIDIA Quadro NVS 210S --> 6150LE

Credit where credit is due, recently Neeyik (http://www.neeyik.info/3dspecs/) has really helped get core details.

Ive noticed some people using this list elsewhere, please link to this thread, mostly to ensure people can stay up to date.

PSU = Pixel shader units
ROP = Raster Operations Units, takes the info from the Pixel shader about each pixel and assembles it into the final picture, this is where AA is added.

I will ask when this is stickied for the thread to be left unlocked, HOWEVER, please only post corrections/suggestions. If you have any other questions feel free to make a thread or PM me

TDP - Not provided by me, do NOT message me about it, mujtaba is your man there. Remember ATI and Nvidia can use different measurement practises.

 

once again very nice of Meaker to put this together, thanks M!

 

great and very concise. Thanks for sharing.It will help ones who confuse about notebook's graphic cards.

 

It's about time this was made a sticky. Nice work Meaker.

 

*smiles* Well thats made my somewhat boring day at work a bit better.

Gah so many lines of C code....

 

Added in a general amount of ram that each group can use.

 

Nice work! Sticky!!!

 

http://forum.notebookreview.com/showthread.php?t=71264
just an update for your list and im sure it beats a 6800,
i have a 256meg 7600gt shown in the link above, im not sure how to find out how many pipes it has but all i know is that it has 256 meg 128bit ram and speeds at stock are 500/600
cheers

 

Geforce go 7600 gt information:
http://www.gpureview.com/GeForce-Go-7600-GT-card-436.html

7600gt = 12 PSUs, 5 vertex units, 8 ROPs, 128 bit mem (500/600) SM3

Seems like it should beat atleast the x1600..

 

Added 7600GT.

 

what about adding the 7200 to the low end ?

 

Any details on it would help all I can find for now are of old rumours to an 8 pipe card and I assume they got confused thinking the 12 pipe card would be a 7600 when infact nvidia held it off. All I could get from nVidia is that is uses shared memory ONLY, it does NOT support dedicated memory.

Therefore I have assigned it to the ultra low end and placed it at the top as if it shares the same core spec as the 7300 it would be the fastest.

 

Is there any notebook out there that has the 7600GT. I don't think there is any.

 

The Sony AR series is the only one that has a Go7600GT at the moment.

 

for that price i might as well buy an XPS notebook

 

Meaker: I got the 7200 and i ran the 3dmark05 straight out the box (meaning i havent optimized windows at all straight from HP with all bloatware/etc i got a 1505 it has 32mb dedicated and 96 shared for 128 total. I think thats should allow it for the low end - useful

after i format, optimize windows, i bet i could get like mid 1500s ?

 

Any chance of getting clocks? Also can you post that benchmark?

 

What do you guys think of the review linking? Worth taking forward? Want it deleted as it messes things up?

These are just a first few, I dont have time to go searching atm so they are all from anandtech.

 

Reviews are nice of course..

 

isn't the nvidia 7400 better then the x1300 performance wise?

 

Is there a comprehensive list of which notebooks have which chipsets? I'm looking specifically for GF7600 based notebooks.

 

Should be in the specs of the notebook...

 

Added 7700, note it is slower than the 7600GT, Nvidia at their best again.

 

What about 512MB NVIDIA® Quadro® NVS 120M ?

 

it isn't a gaming card...it is a mobile workstation card. so it isn't in the charts.

that being said, it should offer performance nearing that of a GeForce 7400 256MB. some people have reported that the Quadro doesn't play some games either, so be careful with that

 

what about the

256MB NVIDIA Quadro FX 1500M, (dedicated), OpenGL
and
Nvidia Quadro FX 2500M

what are their specs and where do they rank?

*edit, i just found out that the 1500m has the 7900 GS core and the 2500m has the 7900 gtx core

 

Please note this chart is for normal gpus only, I would assume those looking for professional solutions would be people able to find out details themselves before buying.
If I were to make a chart for these chips they would be on a different list and I dont want to waste my time on something that only a few people would use.

 

Edited with business to desktop chip comparisons. Be aware of clock differences and possible performance issues due to the drivers being optimised for CAD rather than gaming.

 

7900 = 500/600

7900 = 38.4 GB/sec
7950 = 44.8 GB/sec

44.8/38.4 (these are the mem bandwidths for the 7950 and 7900 respectivly) * 600 = 700mhz mem clock
1150 (7950 vertex/sec) = 1000 (7900 vertex/sec) + 15%
500 * 1.15 = 575mhz

Just to let you know how to calculate clocks from raw values Oh and 7950GTX added.

 

disappointing with the x1400

 

Why is that?

 

x1700 added.

 

Sorry to bug you, but according to Dreamer the business cards are off. From what I can tell, it should be:

110M ==> 7300
120M ==> 7400
350M ==> 7600

Can anyone verify these are right? Dreamer get in here!!!

 

Calm down. I am here.

According to Dreamer the business cards are not off, but they should be in different chart because they have different purpose. I just had no time to make one more chart. Anyway, I am gonna do that later, so the chart of the professional cards is temporary unavailable.

Sorry, I did not mean to provide misleading information.

Otherwise, that's the information about the professional GPUs that I have.

Meaker sorry that I use your thread for that.

 

Updated......

 

While thats true, this is a gaming list.

 

Pipe count is too low, the graphic card does handle the newest techniques but it cant handle much gaming.

 

A small suggestion... you could add the Quadro Go1400 featured in a few Alienwares and Dells of yesteryear, such as the Precision M70. It was basically a 6800, but it was in 15.4" notebooks, and was generally a really strong card.

 

slotted in though i doubt it will be of much use :s

 

Isnt X600 64 bit? You listed as 128bit. correct me if im wrong

 

Consider yourself corrected then

Of all the ones on the list you picked the one I owned, it was 128bit.

 

GeForce Go 7200 has a core clock speed of 445 MHz and memory clock of 350 MHz.

Now if only the Forceware driver and RivaTuner could oc' the core past 600 MHz and let me oc' the memory past it's default 350 MHz setting.......

 

1st off...great review

the 7700 in the asus g1 has 512mb and runs at 450core/400mem...according to http://forum.notebookreview.com/showthread.php?t=91603 i dont know if asus did that, or if thats the general, but that's the only laptop currently using the 7700.

2 bad it's slower than the 76gt...tis sad...but least in the g1 it's got 512 instead of 256

 

What on EARTH are they thinking!? Clocking the 7700 LOWER than the 7600 non GT!? Thats got to be illegal.

 

This is bizare, I cant see what makes the 7200 worse than the 7300, it has higher clocks, it scores lower, but I cant see why, it seems to have the same core specs....

 

That was my conclusion as I could find no hard data on the 7200. I suspect it's just a 7300 core, with a particular overall configuration that calls for being called "7200".

The 7200 and 7300 both have a 64 bit interface. When I ran RivaTuner, it told me the speed (before I did any tweaking) was 445 MHz, which I concluded really was 450 MHz as 445 seemed unusual. Much of my experience as has been compared to the GPU in my previous machine with an ATi Mob. Rad. X600, of which my 7200 has noticably better performance, especially in polygonal performance. The extra vertice pipeline makes a big difference.

My final hypothesis is the 7300 and 7200 are the same core speed. The one thing though that makes the difference in performance I'm guessing is that the GF Go 7300 uses GDDR2 as opposed to DDR2 on the 7200. I suspect that makes a noticeable difference, especially when dealing with a 64 bit memory interface.

 

gfx cards have more to do with architecture than clock...like procs, a 7200 architecture is probly more efficient than a 7300

all the same, the 7700 core clock could be slower, but with 512 and mabe a better architecture it could be faster than a 7600...mabe not a 7600 gt though

o and i dont know if you alerady seen this but :

http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html great place

 

The 7200 and 7300 GPUs both use the G72 core, in the 4 x 3 x 2 configuration of pipelines. Pretty much the same core architecture, just different settings. It's like how GFX card manufacturers sell an "OC'd" version of a card, charging you an extra $30 bucks for it. The only reason to perhaps charge you more is maybe for a different heatsink, but a smart person could overclock a card, increase voltage, and even add their own heat sink/cooling system.

 

The cores are identical, and I think the 7300 may even use GDDR3, while the 7200 uses DDR2 but thats not the point since they perform equally at the same clock, so given that they share:

The same core
The same memory interface

Then how come the 7200 is clocked higher than a 7300? A 7200 will outperform a 6200 they are a different core and the 7200 is more efficiant but here they are too similar.

Bah its always muddy at this end of the charts :/

 

The Dell UK website has an Inspiron with an option to add a Geforce 7900, but the 'GS' bit isn't mentioned. Where does this card fit into your chart?

Or it is really a 7900GS?

Here is a link:

http://configure.euro.dell.com/dell...s1&kc=305&l=en&oc=N12945&s=dhs&sbc=inspn_9400