|
|
#1 | |
|
Registered User
Join Date: Feb 2004
Posts: 585
|
Hi,
For me its time to buy a new graphic card and 2D performance is all that counts for me. I read the nvidia driver is able to rasterize trapezoids using the GPU, which is a very interesting feature for me. Which GPUs actually support that? Is it really only supported on G200, or does the newer/cheaper GT218 also accelerate it? Are there any benchmarks available indicating how well it works? Thank you in advance, Clemens |
|
|
|
|
|
|
#2 | |
|
NVIDIA Corporation
Join Date: Mar 2005
Posts: 2,487
|
GT200 and higher support it, which includes GT218 and the MCP79 chipset that's used in Ion. It's also supported on the new Fermi architecture (GF100 and the newly-released GF104).
|
|
|
|
|
|
|
#3 |
|
NVIDIA Corporation
Join Date: Mar 2005
Posts: 2,487
|
Oh yeah, and I don't know of any published benchmarks anywhere, but when I was testing it, it reduced some cairo-perf-trace results from thousands of seconds to tens of seconds.
|
|
|
|
|
|
#4 | |
|
NVIDIA Corporation
Join Date: Sep 2007
Posts: 254
|
For the record, the 256 driver release series supports trapezoid acceleration by default, without having to enable the AccelerateTrapezoids NV-Control attribute.
Since I don't know if it's been mentioned anywhere before, it also supports accelerating all types of RENDER gradients. Gradient acceleration works on all Geforce 8 series products and newer while trapezoid acceleration is only supported by anything higher than GT200 as Aaron pointed out. If you have any numbers of how that impacts your Java2D back-end or other things, I'd be interested to see a before/after comparison or against other drivers. |
|
|
|
|
|
|
#5 | ||
|
Registered User
Join Date: Feb 2004
Posts: 585
|
Quote:
Quote:
However I won't compare directly against other drivers. I think AMDs binary driver is _crap_, but they are officially supporting open-source driver development which I after all prefer, so I don't want to do marketing in favour of nvidia. Thanks a lot, Clemens |
||
|
|
|
|
|
#6 | |
|
Registered User
Join Date: Sep 2002
Posts: 623
|
If you can you should compare directly against other drivers. People say AMD binary is _crap_ because this is true. But until there is no independent, repeatable proof like any benchmark results no one in AMD will take such jabbering seriously. Benchmark results comparing Nvidia binary vs AMD opensource and binary will show how it looks like for real and may be motivating for AMD to do something with slow 2D. If AMD will speed up Nvidia will be motivated to speed up more to keep 2D speed crown. Finally the customer wins because Linux has better 2D speed thanks to improved drivers on both sides.
|
|
|
|
|
|
|
#7 |
|
Registered User
Join Date: Feb 2009
Posts: 138
|
I'd also like to see comparisons to other drivers - open source or not.
|
|
|
|
|
|
#8 | |
|
Registered User
Join Date: Feb 2004
Posts: 585
|
Quote:
![]() After all phoronix does already a lot of benchmarking, and they produce a lot bold headlines and judgements - I don't have the energy to do the same. - Clemens |
|
|
|
|
|
|
#9 |
|
Registered User
Join Date: Feb 2009
Posts: 138
|
Saying that Phoronix does benchmarks is a pretty bold claim already.
![]() |
|
|
|
![]() |
| Thread Tools | |
|
|