Parhelia Competitive Analysis

[PVR_Extremist]

http://www.hardocp.com/articles/parhelia/analysis.html

Check it out

 

[Kristof]

I think it contains some mistakes and a whole bunch of "trademarked" names that are obviously not supported by anyone else (they also have the same thing in there multiple times, like including "surround gaming" and "tripplehead" which is basically the same thing).

Potential mistakes : Radeon8500 AFAIK has 2 vertex units, not 1 as the table indicates. Also AFAIK GF4 can do 4 dual textured trilinear filtered pixels per clock, at least in theory: BW might have some impact on this.

 

[Dave Baumann]

Radeon also does 6X AA, not just 4.

But, yes, that slide is pretty pointless because becuase of their own implementations of things. Also, if you add P10 into that it doesn't seem quite as impressive right now.

 

[Tagrineth]

Heh, also comparing Pixel Shaders versions, they supposedly win with 1.3 while the NV25 has the same version and the R200 has version 1.4...

Gotta love marketing

 

[Panajev2001a]

Thinking about NV2A... I think I recollect the card being able to ( at 250 MHz ) do:
2 GTexles/s with bi-linear

1GTexles/s with tri-linear

1 GTexels/s / 4 pipelines / 250 MHz = 1 pixel/pipe... GF4 might be different...

I heard that was due to cache BW constraints for the XGPU

 

[muted]

http://www.tomshardware.com/graphic/02q2/020514/parhelia-01.html

a pretty average preview ..

i don't know why i keep checking this site out but

hmm
just look who wrote this preview, and you'll realise when he says something good about the parhelia , he says something good for the gf4ti line ..

Parhelia has a whopping memory bandwidth of 20 GB per second. By comparison, NVIDIA's top product, the GeForce 4 Ti4600, just reaches 10.4 GB/s. However, with its Lightspeed Memory Architecture II, NVIDIA has provided its graphics cards with a very effective Crossbar Memory Controller, lossless Z-compression and further optimizations

bah

I want some more numbers , guess i'll have to wait