UPDATED - The Mobile Graphics Card Info Page - Most GPU Qs answered

Diablo 2 should run fine on max. I mean, a netbook can run Diablo and Diablo 2 >.>

Everything else should also run, but as ahl said, may have to lower the settings a bit.

 

hey guys can u plz tell me whether i should go for Nvidia gt120 1 gb or ATI 3200(896 mb shared)+3450(256 mb DDR3) in Crossfire?

 

gt120 is a lot (2.5x) better than the 3450.

 

Which one is better?

Thanks!

 

Well if you believe notebookcheck the 9800M GS is better - http://www.notebookcheck.net/NVIDIA-GeForce-8800M-GTS.6934.0.html


If your lucky like a bunch of us ASUS users have you can edit the 9800M GS bios to GTS Clocks or higher.

 

They are the same except 9800M is a tiny bit higher clocked, like 30mhz core clocks more.

 

Its also die shrunk (8 Series 65nm,9 Series 55nm), produces less heat and can run at higher clocks stably.

 

Sorry but your wrong, there is only one die shrunk card in the 9m series and thats the 9500m gs and its only considered a die shrink because they used 65nm for this model instead of 80nm. The 9800m cards have been proven already to be nothing but rebranded 8800m cards. The 55nm nvidia cards start with the 100 series the 160m replaces the 9800m gs.

 

same dif, either way it wont matter

 

Actually the Geforce 9800M GS/GTS is G94M architecture with a 65nm die running at 600 Core and 800 Memory (GTS) while the Geforce 8800M GTS is G92M with a 65nm die running at 500 Core and 800 Memory. The difference between the two isn't much (something around 15% higher performance). You are correct about the die being the same, so other than clocks they aren't much different.

 

hi i currently have 8500 gt 1gb on desktop comp but i want to but laptop.
found one with 9600 gt 1gb... is there a big difference in speed and fps

thx

 

Chaz! 9800m GS is better than 8800m GTS.Both have 64 Unified Shaders, but 9800m GS have higher core and shader frequency.

 

Hi.. I have an HP Pavilion dv6-1125ee notebook with ATI Mobility Radeon HD 4650 graphics card.. can you help me to get official windows XP driver for it.
I try hard but all I get it not work with XP.. Thanks in advance

 

The 9600m GT is MUCH better. Twice the shaders, faster core speed, double the memory speed. If it is a DDR3 card, all the better.

 

hi

i have a question about ATI Mobility Radeon HD 4530

is it good with games ? can it run GTA IV , crysis ,assassins creed, Need for Speed Undercover ... etc


thanks

 

The 4530 is a low to medium card, you should be able to run all the games mentioned on low resolution and detail settings.

http://www.notebookcheck.net/AMD-ATI-Mobility-Radeon-HD-4530.13972.0.html

 

crysis ....mehhh low settings (low res to)
GTA 4 depends on CPU and GPU so i say low or none
AC--med low(unsure)
NFS (med)

 

thanks ,,, and what about

Nvidia GeForce G103M
ATI Mobility Radeon™ HD 3650 supporting HyperMemory™ technology

 

the ATI card is low end as well and
the ge force is yet to be tested that i know of but its gonna be low to med settings


if your looking for a decent laptop your gonna wanna find a 9600m gt or higher

 

The 3650 is the best of the cards you've mentioned so far. Not as good as an Nvidia 9600 GT, but still a midrange card.

 

i'm looking got laptop , and been looking a long time now between many differnet models..its very hard thing to decide which laptop i should buy.
i want it for normal using , and about games i only care of FIFA GAMES LIKE FIFA 2008, FIFA 2009 .
and according to my money , i have found 2 Hp laps to choose between them , one has vedio card , Nvidia Gforce 9200M , and the other is ATI mobile 4530
i don't know which model i should choose.
some say the Nvidia 9200 is not bad , some say ATI 4530 is much better

so what you think?
hope someone can help me

 

How does the ATI 4570 compare to an 8600M GT DDR2?? with the 4570 being a 64 bit card???

 

It doesnt, the 4570 even a 4670 would under perform it but not a great deal. I dunno about the 4570, but I think a 4670 would have been okay for me.

 

An HD4670 wouldn't underperform a 8600M GT >_>

The HD4670 outperforms the 9700M GT on certain games so I wouldn't really say it performs under a 8600M GT GDDR2 >.>

But yeah the HD4570 would be limited by its bus size and perform less than a 8600M GT. However I've heard of GDDR3 versions and even of a 128bit version of the HD4570.

 

why no one didnt answer me?!! Nvidia 9200 or ATI 4530 ?
i have read everything at the first page and it show that the ATI 4530 IS BETTER ... but is it much better than the Nvidia 9200 ???

 

ATI 4530 is far better than the 9200. Refer here for more information

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

 

OOPS you know what i did Melody, lol I thought it said 9800 I keep doing that.... anyways yeah im wrong on account of the wrong card.

 

How do you do that? Do you invert '86' so it becomes '98'? lol

Oh I have a GPU question, I've heard that there is a 128bit version of the HD4570 and that apparently if its GDDR3 that it actually performs quite decently. Can anyone confirm this? This might make it an interesting alternative over the older HD2600/HD3650 GPUs.

 

thanks for the respond, i have looked at the link and information in the table
but it doesnt mention any information about the 3Dmark 01 , 3Dmark03 for the card ATI 4530 ... and its mentioned for the Nvida 9200

why is that? i just would like to understand this point

does it mean it wasnt tested for 3d or what?

 

No, that they have to update their charts

http://www.notebookcheck.net/Review-HP-Pavilion-dv6-Notebook.17209.0.html

 

There's no 128-bit version. There were two things causing confusion:

1. Notebookcheck originally listed 64-bit/128-bit. They've since revised it to list only 64-bit.

2. GPU-Z incorrectly read/s the Studio's 4570 as a 128-bit card. This is or was a driver issue.

 

ATI4530 is relatively new card when compared to the 9200, certain benchmarks has to be updated in that chart.

 

Yeah honestly I wasn't sure about that because ATI's site doesn't list bus size in their specs for the HD4570(or I can't find it at least). It'd make more sense if it was 64bit since it's the successor to the HD3470 which was 64bit.

I'm surprised at its benchmarks though for a 64bit GPU hence why I considered the possibility of it being 128bit.

 

It only affects memory bandwidth last i checked and while some games can be memory intensive many arent. The amount of memory used doesnt always relate to how fast its needed. A racing game would benifit from a very high bandwidth as its constantly needing to read from memory. Not so sure about FPS games which dont use vehicles.

Older games like fear used very little ram and ram bandwidth.

 

So I take it that the jump from 64-bit -> 128-bit isn't as drastic as the jump from 128-bit -> 256-bit.

 

Well in terms of numbers, the jump from 128-->256(a 128 difference in bus size) is twice that of a jump from 64-->128(a 64 difference) so yeah, I guess you could say it'd be smaller.

The bus size incapacitates the bandwidth and how much information can be written on VRAM in a single cycle so you the jumps always induce a 2x increase in terms of that.

However, as Clutch points out, not all games will require large amounts of information to be stored on the VRAM in which case the bus width might not be so crippling so long as the other components are up to par(shaders, clocks and such).

Mind you, usually components go hand in hand with bus width sooo...

 

I dunno, there is a huge benchmark difference between the 4570 and the 4650, which is probably due in large part to the bus width (though 1/4 of the shaders is also huge). All of the 4xxx series blow away their 3xxx series counterparts, but the 4570 is still far below even a 3650, so it's still very much on the 64-bit tier.

For newer cards:

Memory Bus............. 3dmark 06
64-bit.........................2-3k
128-bit.......................5-7k
256-bit.......................8-10k

 

I haven't measured this sort of stuff in a long time. If i remember fear in multiplayer only used 75 mb of vram.

GTA4 gets around 400mb on my system i think and while driving and drawing the background it would require moving data pretty fast from the vram. Though that game in particular is a CPU killer.

 

forever you will get repped for that. but i would also like to point out the bit bus was ruled out with gddr5 rightm the 4860 mobolity claims to out perform the a 4850 and the desktop 4770 cf out performa a 4890 and is almost equal to a 280 desktop.

can you explain that. i know the mobility isn't released but the desktop version is. and i know that crossfire helps out alot but still 128 bit gddr5 vs a 4890 or a 280 gtx desktop.

 

also i forgot to mention one 4770 desktop is eqully to my 3870 mobily crossfired.

 

@Neil: Well the HD4650 is the successor to the HD3650 form the previous generation. The HD4570 is supposed to replace the HD3470 and arguably from what I've seen in benchmarks, it does a good job at that.

@donda:Well I remember someone else posted the analogy with the highway so I'll use that.

Think of it like this: rendering information is like parcels that need to be delivered by trucks. Bus width would be equivalent to the size of a highway on which those trucks can move.

For ease's sake, assume a 64bit GPU has 64 lanes and a 128bit has 128 lanes. So obviously in a nutshell higher bus width would advantage more transfer for larger amounts of data. However if your "lanes" weren't already filled, then the difference would lie in the other components(clocks, shaders etc.).

Now memory interface I suppose we can think of as the type of vehicle running on the road. Each type of memory induces more speed of data transfer. I'm not an expert at this, but to put it simply, GDDR5 has improved data streaming in terms of input/output(through some form of parallel traveling) as well as other things. So in a nutshell, you could say that GDDR5 is a "faster truck" than GDDR3. Therefore, you can transfer data in a shorter laps of time, making the bus size less of a hindering factor in terms of how much data is transferred. I suppose that at higher resolution were maximum amounts of data must be transferred that even with GDDR5 the bus width might become a bit of a bottleneck. At least, that's to the best of my understanding on the topic. >.>

Your example comparing the 128bit GDDR5 HD4770 in comparison to the 256bit GDDR3 HD4850 is indeed correct . Higher memory bandwidth allows the HD4770 to yield similar if not slightly better performance than the HD4850.

 

mmmmmmm faster trucks or more lanes. i think i would would prefer both but more lanes and forceing the trucks to go faster(overclocking) seems the best.


i just remembered there was a desktop 4850
vs 4870 at the same clocks(4850 heavily overclocked.) and the 4870 won by 10 percent.


so i think that having more lanes(bit bus) is better then faster trucks(memory speeds) cause you can make the trucks faster(memory speeds overclocking) but you can make more lanes(bit bus)

 

Well ideally you'd have both.

That's why the HD4770 by itself can't best an HD4870 because the HD4870 has both a 256bit bus and GDDR5 memory.

It all depends what type of gaming you're doing though. Bus interface wins at higher resolutions(when paired with higher VRAM). What made the HD4770 such a big deal was the pricing it had. Manufacturers were able to shave off lots of money by keeping a 128bit GPU. SO we're left with an HD4770 that costs ~100-120$. Buy 2 of these and Crossfire and you can even outperform high end cards that cost more than the dual setup.

 

ok you seem to know alot and this has been bugging me. ati and nvidia

ati has alot more shaders even though i know it's boss 1 and 4 slave and you divide by 5 to equal nvidia's. but they have lower clocks and it's locked up with the core clock but they can't be overclocked high.

as nvidia's clocks are highly overclocked (there shaders are extremely high) they don't have as much shaders. and they can really push.

what are the pros and cons about each manufacture?

 

I should point out you dont use the word overclocked if refering to the manufacturers clock rate as overclocking is raising it above the factory clock rate.

On the 9 series the Shader clocks are 2.5 times the Core and as nvidia guru pointed out to us 9800M GS overclockers when overclocking should be raised according to that. Different series and chips have a different ratio though.

I will give you a con on ATi, Poorly written drivers, though Im not a fan of either company ive had both.

 

yeah i had both also. i pointed out overclocking because i having the 3870 crossfired and the 9800m gtx single(looking for another card). i have seend that i can push my 9800m gtx alot farther in terms of overclocking thatls why. like my 3870 was 660/850 and i couldn't pass 700/900 as to the 9800m gtx is (i don't remember the core right)700?/1250/800 and i brought it up to 800?/1600/950 and i javen't push it higher yet.

 

My card stock was 530-1325-800 @ 1.11v (some cards are 1.07v). I have pushed it to 640-1600-800 @ 1.15v but currently run it at the Undervolted GTS clock 600-1500-800 @ 1.07v.

 

my card was at 1v but i have tried 1.05 but that's all i can add. i don't know what's the temps the card is recommeded for so that why i didn't really push them any higher.

 

I had to push to 1.15 to make my highest stable. But temps soared a bit there. Seems like the 9600 and 9800's and possibly all Dedicated 9 series throttle at 110C. Despite having personally observed this with my own card when i forgot to plug the fan in, my bios is actually set to 105C.

GPU-Z can rip ur bios (though i dont think u have yours in a notebook anymore?) and Nibitor can view and edit it.

 

yeah that's what i did to make the voltage higher. but what you mean mines is in a notebook???