HardOCP refuses to review GTS 250

[Richard]

I'm sure you know this because you've seen the architecture, right? No? Well at least you wrote batteries of directed tests to measure its behavior and compared it to a GF2 and a GF4? No? Hmm, ok, so you basically looked at a single feature in a tiny part of the pipeline, and used that to decide that your point was true?

That may very well be but for the end-consumer (what JR was pointing out) it was a GF2 on steroids. You got "SM 0.5" instead of SM 1.1, you got the same amount of video memory (if you had the GF2Ti), etc. Yes there were differences and the clock/mem was higher but in the real world:

http://www.tomshardware.com/reviews/pc-graphics-xbox,423-13.html

http://www.tomshardware.com/reviews/pc-graphics-xbox,423-11.html

You got slightly better perf than a GF2 but lower than a GF3, not to mention that once games started requiring DX8 you were out of luck. Which was JR's whole point: the name "GF4" led people to believe it was better than GF3 when it was actually worse and little better than a GF2.

Carmack himself blasted nvidia for naming the NV17 a GF4 card and actively told people not to buy it and get a GF3 instead.

 

[Davros]

But I would expect them to understand that a mid range card of todays generation is not necessarily going to be faster than the top end card of the previous generation.

Actually i lot quite a few noobs (for lack of a better name) that believe the higher the number the better

 

[pjbliverpool]

Actually i lot quite a few noobs (for lack of a better name) that believe the higher the number the better

Yeah I know quite a few myself. My point though was that I wouldn't hold NV to blame if someone did make that assumption. I would blame the individual for not doing at least a basic amount of research into what they were buying.

The only other option would be to have graphics cards being named in a never ending upwards scale of numbers based on their performance.

There is always going to be someone out there stupid enough to think a GeForce 4MX 440 must be more powerful than a Geforce GTX 285

 

[Blazkowicz]

I'm not aware of msaa on geforce 4MX, lol, that would make it a better card than those atrocious radeon 9200SE/9250 . (well, perhaps a MX440 already is..)

by the way it's actually geforce 2MX on steroids. it's not too unlike the voodoo3, that was called the voodoo3 instead of banshee 2.

 

[Arun]

It's a mix of GF2 & GF4 IP, with the most important parts from an end-user POV being GF2-based but still plenty of GF4 technology. That does make it problematic to call it a GF4 since it is inferior to GF3 in several key ways feature-wise, but architecturally it's unfair to call it a GF2.

There, happy?

 

[Bob]

There is always going to be someone out there stupid enough to think a GeForce 4MX 440 must be more powerful than a Geforce GTX 285

Imagine the awesome power of the GeForce Ti 4200!!

 

[homerdog]

So am I alone in thinking RV740 will be the final nail for G92? Don't see how GTS250/240 can be competitive on a price/perf level at all against that little devil. Perhaps I'm underestimating the power of the Evil Green Marketing Machine...

 

[pjbliverpool]

Whats the actual price difference between the 3? I have no idea what that level of GPU is going for these days!

 

[homerdog]

HD4770 (?) is expected at ≤100USD; GTS250/240 I dunno, all we've got is Charlie claiming they'll be jacking it up compared to the 9800s. FWIW the cheapest 9800GTX+ is $145 on Newegg and $110 for the 9800GT and it's doubtful they'll be lowering it for the rerehash.

Edit: Ah but it appears the GTSs are due a couple months before RV740.

 

[Jawed]

GTS250 won't have Tri-SLI

http://xbitlabs.com/news/video/disp..._approaching_geforce_gts_250_due_in_days.html

In fact, Nvidia bets heavily on the low price of the GeForce GTS 250: the 512MB version should cost $129 and 1GB version is projected to be available at manufacturer suggested retail price of $149, considerably lower compared to price-points of the 9800 GTX+ models.

Jawed

 

[homerdog]

$129's not bad (and lower than I expected ).

 

[Silent_Buddha]

For people that are looking for video cards down in that range however, 129 is a whole lot more expensive than 99. For marketting to consumers 99 is a lot more attractive than 100.

Regards,
SB

 

[neliz]

new rumor, I heard they're going to sell the "new" GeForce GTS 256 for about $125 (and that without any rebranding)

 

[pjbliverpool]

new rumor, I heard they're going to sell the "new" GeForce GTS 256 for about $125 (and that without any rebranding)

I hadn't realised how amazingly close those names actually are!

 

[Davros]

and a GeForce GTS 256.003 for $125.32

 

[ChrisRay]

Well considering the SLI perk. It's clear Nvidia isn't trying to hide anything.

 

[neliz]

but there is some smoke and mirrors since they'll now be running on "CUDA processors" instead of stream processors/shader cores/intahwibbles and fail to mention that there is a functional difference.

 

[Jawed]

Bizarrely, G92b will also have only 4 clusters enabled for GTX160M

http://www.fudzilla.com/index.php?option=com_content&task=view&id=12295&Itemid=1

what's the matter with using G94b

Jawed

 

[AnarchX]

Latest Mobile FW 179.44 says:

NVIDIA_G94.DEV_0631.1 = "NVIDIA GeForce GTS 160M"

Also it looks like there will be a G96-based GeForce 200M:

NVIDIA_G96.DEV_0653.1 = "N10P-GV1"

 

[CarstenS]

Might there be a difference between "Enthusiast" GTX 160M and "Performance" GTS 160M if they even coexist?



BTW, IMO it'd make sense wrt to cost-effectiveness: For mobile you only have to bin (=lower costs) on only one of the two GPUs, you have the ability to sell I dare say everything off of the G92-wafers (= higher revenue per fixed-price wafer) and finally, you have more of the same package (= lower cost per piece of this package).

What doesn't make sense though, is to keep selling another Chip (G94) with basically the same specs in mobile too.