Help me compare MBX and the GoForce 4500
- Thread starter Shogmaster
- Start date
Shogmaster
Regular
DE~NIED! *pouts*
Luminescent
Veteran
Kristof, can you comment as to the shader abilities of the MBX. I see it has a color blender and a texture shader. Are any of these parts configurable, perhaps in a similar manner to NV1x/2x's register combiners?
Shogmaster said:
Calling something which cannot even do Dot3 or EMBM operations "programmable pixel shader" is missleading, IMO.
Funny how definitions are streched based on the actual marketing needs ... In some newer presentations about desktop products, Nvidia doesn't even count the NV2x with its über-register combiners as programmable pixel shader part ...
Shogmaster
Regular
ram said:
Not even capable of EMBM?!? Are you sure?
zeckensack said:
People can always correct me if I'm wrong, yet I think Elan could do 10M Tris/sec whether 1 or 6 lights were enabled. And that measurement must be obviously from a very CPU bound platform.
---------------------------------------------------------------------------
For the record I don't think there's EMBM support on MBX either.
I'm not 100% sure for the AR10, but I haven't seen anything close yet that could even indicate it.
Agreed. How bout showing off some of your curves? *snicker*
***edit: oh I almost forgot...any possibility of letting us know if the anisotropic filtering algorithm is by any chance performance optimized (ie adaptive)?
*cough*
Dear PowerVR,
please, don't be British. Please be aggressive. There's no reason why NVIDIA shouldn't be considered "dead" in this particular market space. You know you deserve the title of market leader. Don't let them get near any of your cake under any circumstance.
Thanks.
I still get some "grass roots" insights, and currently it looks like a lot of folks will give NVIDIA et al the benefit of the doubt, just because of the brand name. Go figure! Now get that "Mobile graphics competitive comparison" slide set out before I start feeling sorry for you.
Dear PowerVR,
please, don't be British. Please be aggressive. There's no reason why NVIDIA shouldn't be considered "dead" in this particular market space. You know you deserve the title of market leader. Don't let them get near any of your cake under any circumstance.
Thanks.
I still get some "grass roots" insights, and currently it looks like a lot of folks will give NVIDIA et al the benefit of the doubt, just because of the brand name. Go figure! Now get that "Mobile graphics competitive comparison" slide set out before I start feeling sorry for you.
I'll bring the booze and Simon the glasses PatrickL said:Whole europe can testify that nothing is worst than a drunk english supporter when it comes to be agressive
[sarcasm] Because, of course, no other country has football hooligans, do they?[/sarcasm]
Unfortunately, a handful of idiots causing trouble have led to the perception that many English football fans are hooligans.
I went to Portugal for two weeks in the summer with some friends and everywhere we went, people were asking us if we were hooligans!
The thing to realise is that the people causing trouble at football matches would also be causing trouble if they were just at home instead - the minority who are looking for a fight on a Saturday night just for the sake of it. Scumbags to the man, it must be said.
I do feel, however, that any misdemeanours by English hooligans (I don't class them as football fans) are hyped up by the media.
Sorry for the off topic post.
Kristof said:Ailuros said:
You interested in my curves ?
*cough*no*cough
PowerVR® Shows Off its Curves at GDC 2004
Teasy said:I think a very important part of any comparison between these chips has been left out. Power usage, or do Nvidia not specify this?
I wonder if Gizmondo lasts more then 2-3 hours without a recharge?
NV quote some power numbers here
Some pretty old ARM docs quoted MBX numbers nearer 200 mW.
Obviously there are big architectural differnces to consider, such as embedded RAM, quoted die sizes etc. which have cost implications. Also neither document is 100% clear whether the numbers relate to total power draw of CPU+GPU or just GPU - as the ARM / MBX combo is primarily aimed at the SoC sector so I think the ARM/MBX 200mW figure is for CPU+GPU, whereas the NV number may be for GPU alone?
That old pdf claims <1.3mW/MHz for MBX@130nm.
120 * 1.3 = 156mW
MBX Lite@130nm <0.5mW
50 * 0.5 = 25mW
(I took highest and lowest clockspeeds; without being sure I think Intel has clocked the 2700G @66MHz).
mboeller are you reading? If yes what does the "other" ARM pdf say about power consumption (you know the one you have and I lost)?
120 * 1.3 = 156mW
MBX Lite@130nm <0.5mW
50 * 0.5 = 25mW
(I took highest and lowest clockspeeds; without being sure I think Intel has clocked the 2700G @66MHz).
mboeller are you reading? If yes what does the "other" ARM pdf say about power consumption (you know the one you have and I lost)?
Code:
Product Technology Gate Embedded Power
(mm) count RAM (mW/MHz)
MBX HR-S 0.18/1.5v 660k 15.5k 2.3 max. 80MHz
(no VGP)
MBX HR-S 0.18/1.5v 870k 22k 3.0 max. 80MHz
(with VGP)
MBX HR-S 0.13/1.0v 660k 15.5k 1.1 max. 120MHz
(no VGP)
MBX HR-S 0.13/1.0v 870k 22k 1.5 max. 120MHz
(VGP)
MBX R-S 0.18/1.5v 365k 7.5k 1.5 max. 80MHz
MBX R-S 0.13/1.0v 365k 7.5k 0.7 max. 120MHzHope this helps