8400m GS vs 9800 Pro

I'm guessing the desktop 9800 Pro is still better than the mobile 8400 GS?

 

i believe the 9800 pro bests the 8400m gs at older games (2005 and before) but falls short at the newest games (crysis, cod4, etc)

 

I would say yes because mobile 9800 is a little faster than 8400m GS so yea I would guess 9800 pro even more on games (3DMark 05 and b/4)

 

"yeah, since 9800>8400 by 1400" lmao

 

The Mobility 9800 is actually based on the X800 series and not the 9800 series.

It's hard to compare the two because their architectures are so radically different. Think about the difference between a Radeon X1950 Pro and a GeForce 8600 GT on the desktop, where the X1950 Pro can best it in raw power but the instant aggressive shader use comes into play, the 8600 GT cruises past it. For all intents and purposes, though, the X1950 Pro is the faster of the two.

Now take that odd comparison and make it a couple hundred times more complex.

Final answer: "Your mileage may vary."

 

Yeah... it honestly depend on what you're trying to play.

From a raw power perspective, the 9800 Pro is definitely more powerful... But the newer 8400M GS might have more shader power when push comes to shove.

However, just for the sake of a comparison, the 8600M GT performs about on-par with the 9800 Pro (desktop) in FEAR with all the highest settings... so who knows.

 

My ddr3 8600m gt is a good bit better in FEAR than my x800 desktop card, for what its worth.

 

http://www.gpureview.com/show_cards.php?card1=19&card2=531


The ponly place the Mobility 9800 wins out is memory bandwidth, otherwise the 8400M GS has it licked. Practically gaming performance would be better on the 9800 for games up to around 2005, after 2005 the 8400M GS would be better. I'd expect the 8400M GS to prove an all-round better card for today's gaming world.

 

8400GS had only 64-bit bus vs 9800 which has 256 bit.

9800 to 8400 GS is NOT an upgrade
8400GS would win in games such as GOW and R6V because 9800pro lacks SM 3 support and would ultimately have trouble starting the game, but if the gameplay is enjoyable on 800x600 with everything turned off and settings set to minimum well you tell me...

 

9800>x700>8400M GS in older games. New games that are more shader intense will be in the 8400M GS's favour. OC the 8400M GS and it will beat the x700 in the older games as well.

 

The 8400m GS's only weakness is memory bandwith, which greatly limits it at resolutions above 1024x768. But if you don't mind playing at 800x600, where bandwidth doesn't matter, it will be the best of the too IMHO.

That's a matter of personal taste. I, for instance, play crysis at all low settings at 848x480 and i think it looks great. i even played ut3 at 640x480 and it didn't bother me much.

 

I can tell someone either has eye sight issues, or will soon!

 

Congratulation you have found the 8400m GS bottleneck, Achilles heal, why it will not run better than the mobile 9800 and even more on desktop 9800Pro except on newer games that it can't run well anyway. Not enough bandwidth!

 

I don't see why every card can't have at LEAST a 256bit bus width.

A 3-4 year old video card has more bandwidth than a new one, but the new one has faster memory that's limited by an extremely low bandwidth.

Hopefully they fix this in the 9 series of Nvidia cards.. It's rediculous.

 

I agree. One would think that with the technology out there it would be cheapest to stick with similar architecture (256-bit) but use slower GPUs/memory for the lesser model graphics.

 

Actually, there's a reason for less bus bandwidth... It relates directly to physical manufacturing process and the fact that a 256-bit bus is quite huge, physically.

It does, literally, take 4 times the amount of physical connections of a 64-bit bus, and not all mobile cards have the liberty of freeing up that much physical space to plunk down just a memory bus. Quite different from CPU/GPU manufacturing process where the physical chip are still the same size no matter how many transistors are present.

The more bus connection you have, the less space there's left on the PCB to house other components, and the more crowded the components will be, thus raising the temperature, and requiring a more powerful cooling solution.

And it's quite obvious the manufacturer needs to cripple something from a business standpoint.

If all the differences were simply reduced clock speeds and less memory, then a weaker GPU OC'd to the same speeds would be running at comparable speeds to their more expensive breathrens... except for a lot cheaper.

 

Posts like these kind of bug me. Don't be an armchair engineer, there's a reason 12.1" ultraportables aren't running 8800Ms. nVidia and ATI aren't out here just to screw you out of performance, these people are professional hardware engineers working with microscopic transistors.

Besides, a 256-bit memory bus on an 8400 core would be overkill, just like the 512-bit bus on the Radeon HD 2900XT turned out to be.

 

I was thinking something else on this topic.

The Nvidia Geforce 8400M G/8400M GS/8400M GT/8600 GS all seem to use the same G86M core technology but have different clock/memory rates.

If the Core is the same and only the bus is the only major difference then this might explain why people have been able to get such high overlocks for the 8400M GS reaching 8600 GS Core speed (600mhz) without problems. Of course it doesn`t equal the same performance due to the bus difference.

Is it possible in the physical manufacturing process all G86M are produced the same and then QA graded based on tests and then given their model allocation? Much like CPU`s are produced all the same speed and then just firmware locked to a slower speed for the `value` products but easily overclocked to their real speed if the chip used is of higher grade?

So in a sense is a G86M core produced for a 8600 GS the same as a G86M core in 8400M GS but on a reduced physical bus and firmware set reduced clock speed and memory speed/type?

 

And less stream processors in the 8400M GS