Old Nvidia Flowfx (official?) video
- Thread starter Simon82
- Start date
rendezvous
Regular
http://www.youtube.com/watch?v=PFZ39nQ_k90
Well, the Geoff Ballew looked like the one one you see if you do an image search at google.
Dan Vivoli could be the same but he looks younger in the video.
Those were the ones i cared to look up.
Well, the Geoff Ballew looked like the one one you see if you do an image search at google.
Dan Vivoli could be the same but he looks younger in the video.
Those were the ones i cared to look up.
Certainly if they really was internal people, I think that it was a big thing that people don't make a visit to the offices to search them and tell what they thoke about this...
The guys are real. They were just poking back at the community who made all the photoshops at the time. Its pretty obvious thats not how they have product meetings. Actually most conversations like that would occur over blackberries these days
Chris
Chris
But in fact, that's not the really fun part of that video. They apparently didn't make it just to poke back at the community and make fun of themselves a bit. They mostly wanted to make fun of the man who was most responsible for FlowFX: Jen-Hsun Huang himself. Don't believe me? There you go: http://216.239.59.104/search?q=cach...:www.computerpoweruser.com&hl=en&ct=clnk&cd=1
hoom
Veteran
That was a surprisingly good natured response by NV to criticism
Shame about the hardware & the poor shmucks who actually bought one...
BTW, mention of VLIW there reminds me: Has anyone actually worked out how we would classify NV30 in modern parlance (ALU:TEX:ROP, SIMD/VLIW etc) rather than what was used at the time 4/8 pipes (8 according to NV marketing, 4 'actual')?
Presumably it was something like: 8 FP16 ALUs that were able to pair up for FP32, 4 TEX & 4 ROPs?
DX7/INT was a separate fixed function section of hardware? (rumor?)
Shame about the hardware & the poor shmucks who actually bought one...
BTW, mention of VLIW there reminds me: Has anyone actually worked out how we would classify NV30 in modern parlance (ALU:TEX:ROP, SIMD/VLIW etc) rather than what was used at the time 4/8 pipes (8 according to NV marketing, 4 'actual')?
Presumably it was something like: 8 FP16 ALUs that were able to pair up for FP32, 4 TEX & 4 ROPs?
DX7/INT was a separate fixed function section of hardware? (rumor?)
NV30 had 4 dual-tex units (one per pipe, 4 pipes, so 8 tex overall). 8 FP16 ALUs sounds way too high considering it was outperformed by the 4 ALU RV350--surely register pressure wasn't so brutal as to make 8 perform worse than 4 lower-clocked ALUs? Maybe the NV35, which was supposed to double NV30's FP performance, had 8 FP16 ALUs. But I could be wrong.
Similar threads
