Windows 7 DX11 Cards

Was the wait worth it? Will you be buying one?

  • Price is too high so no to Fermi..

    Votes: 2 16.7%
  • I bought a 5000 series and am happy..

    Votes: 5 41.7%
  • Both Fermi and 5000 series way too expensive

    Votes: 0 0.0%
  • At last! Can't wait to get my Fermi.

    Votes: 0 0.0%
  • I'm waiting until the price drops.

    Votes: 4 33.3%
  • I'm going to wait for the refresh and 512 cores

    Votes: 1 8.3%

  • Total voters
    12
kemical said:

Yeah, I have heard about that. :) I just forgot about optimization.

Trilinear most likely refers to trilinear optimization in this case. There is such an option in NVIDIA control panel, I reckon. No such setting in ATI Catalyst though.

Athlonite said:
trilinear or bilinear filtering is done on mipmaps where as anisotropic and anti-aliasing is done on the final textures

Must be what Athlonite says. :)

TweakGuides.com - Nvidia Forceware Tweak Guide
 
Last edited:
Bilinear And Trilinear Filtering : ATI's Optimized Texture Filtering Called Into Question

This has some pictures to show the differences...as you can clearly see bilinear and trilinear are very poor compared to anisotropic filtering, also worth stating Ati 5800's do have the same features it just doesn't have all the manual settings in catalyst to tweak...athough rivatuner will likely sort that.



From tech specs for cypress :-


80 texture address units and just as many bilinear filtering units that can filter FP16 textures at full speed, trilinear and anisotropic filtering for all texture formats

32 ROPs supporting antialiasing with programmable sample patterns (over 16 samples per pixel), including FP16 or FP32 formats of the frame buffer. Peak performance is up to 32 samples per cycle (including MSAA 2x/4x and FP16 buffers), 128 samples per cycle in Z only mode
 
Last edited:


cybercore said:
Trilinear most likely refers to trilinear optimization in this case.

Highwayman said:
This has some pictures to show the differences...as you can clearly see bilinear and trilinear are very poor compared to anisotropic filtering, also worth stating Ati 5800's do have the same features it just doesn't have all the manual settings in catalyst to tweak...athough rivatuner will likely sort that.



So, I understand the trilinear filter in Unique Heaven benchmark means trilinear optimization done along with anisotropic filtering, correct?



 
GTX 490 throws down to 5970



61Eatc.jpg

May 23rd, 2010 at 9:01 am - Author faith
Behind closed doors at Computex, nVidia is likely to be showing its most important partners a very serious piece of new kit. The GTX490. Remember when the 7950 GX2 was considered power-hungry for pulling 110w in 3D testing? Looking back at that era, it seems that ATI and nVidia just weren’t trying hard enough.
As we continue to look under the rocks in Taiwan, more interesting information about nVidia’s new high-end card has come to light. To be honest, water-damage and exposure to UV rays means the ink has run in places and we’re looking at this document over a weak video link, but we’re pretty sure that the key info is intact. Question remains: Is it accurate?
Firstly, it shows that the GTX490 is a dual-GPU card. We gathered that.
It’s likely to use the Fermi GTX470 processors, but we’re unable to confirm the exact clocks etc.
GTX490’s power draw looks to be around 375w.
Given that the GTX470 is a 215w part when shipped on its own PCB, there would appear to be some serious down-clocking in action.
We know from the big notebook manufacturers that nVidia sells GTX chips that can run on as little as 100 watts (if you have plenty of spare money and can wait for them to be speed binned).
The key question won’t be “Can the GTX490 beat the original 2GB Radeon HD 5970?”.
The actual question will be “Can GTX490 beat the new 4GB Radeon HD 5970?”.
KitGuru says: Like watching a 20 year old Mike Tyson looking to re-unify the heavyweight belts, we just can’t wait for these boyz to get into the ring and knock seven shades of crap out of each other!

Ref: GTX490 throws down to 5970 | KitGuru
 
So, I understand the trilinear filter in Unique Heaven benchmark means trilinear optimization done along with anisotropic filtering, correct?

I guess I misunderstood they way things work: before textures get processed to final, either trilinear or bilinear method is done on mipmaps along with anisotropy and anti-aliasing. Trilinear optimization can optionally be done, but not necessarily in Unique Heaven Benchmark 2.0. So, I got it. : )


Mipmap - Wikipedia, the free encyclopedia
Link Removed due to 404 Error

...thats the way it reads to me, that trilinear has to be on for anistropic mode

trilinear or bilinear filtering is done on mipmaps where as anisotropic and anti-aliasing is done on the final textures

So, you were all right, and what Athlonite said is precisely correct. ; )
 
Nvidia has officially unveiled its latest Fermi GF100 derivative, the GTX 465. Featuring 352 CUDA Cores and support for DirectX 11, PhysX, SLI, CUDA and 3D Vision Surround, the new GTX 465 is also the most affordable GF100 based graphics card with it US $279 price tag.

As noted, the GTX 465 features 352 CUDA Cores and works at 607MHz for the core and 1215MHz for the shaders. It has 1GB of GDDR5 memory clocked at 3206MHz and paired up with a 256-bit memory interface. The card still needs two 6-pin PCI-Express power connectors and has a 200W TDP. The new GTX 465 looks a lot like the GTX 470 and it still uses the dual-slot cooler and feature two dual-DVI and an HDMI output.

The performance of the GTX 465 is far from being impressive as in most tests it is placed somewhere in between the Radeon HD 5830 and Radeon HD 5850. The card still deals nicely with DirectX 11 tessellation, as it is something that Nvidia's GF100 GPUs are really good at.

Ref:Link Removed due to 404 Error
 

The real problem with NVIDIAs offerings, as we all know, and not to be the master of the obvious here, is that they are critically expensive cards for the small improvement increase they provide. What ATI did with the 5xxx series is well defined and great for the price. There are no games you are going to buy anymore that require the nearly one thousand dollar investment that a good NVIDIA card, which beats out ATI somehow, requires.

It seems that NVIDIA is hell-bent on defeating ATI and once again reclaiming the throne of video card king. But if the software industry is any indication, they will need a major, revolutionary card, that beats ATI in pricing. The problem also arises when you start to realize that most studios are now developing games and graphic applications and testing said software primarily on ATI hardware, and not NVIDIA brand stuff. Of course some testing will be done to ensure compatibility, but not maximum performance. As long as ATI maintains an edge, you won't see this change. And they definitely have an edge.

Why would kemical pay so much quid to buy a NVIDIA card when he could go the ATI route and be able to play pretty much any graphic intensive game with a low framerate? Well, the reality is, there really is no reason, unless he was producing a movie and editing CG models for a multi-million dollar production!

We are in a recession, so no one really wants to pay over $500 for a video card. Even if NVIDIA comes out with a solid and cheap competitor that works great, they are going to have to move mountains, with a whole line of cards that do the same thing, before they can "reclaim the throne" so to speak.
 
I tried every link on that page, and it looks that NVIDIA Fermi is an extremely rare video card, so rare that Microsoft didn't bother to put it on their list.




The real problem with NVIDIAs offerings, as we all know, and not to be the master of the obvious here, is that they are critically expensive cards for the small improvement increase they provide. What ATI did with the 5xxx series is well defined and great for the price. There are no games you are going to buy anymore that require the nearly one thousand dollar investment that a good NVIDIA card, which beats out ATI somehow, requires.
Yes, I absolutely agree.


Why would kemical pay so much quid to buy a NVIDIA card when he could go the ATI route and be able to play pretty much any graphic intensive game with a low framerate?
That's why I also bought ATI 5870, I didn't want to wait another year or two before GTX would become available at a reasonable price, and I didn't want to buy a new power supply either. I believe NVIDIA has advantages, but damn it - it's too expensive now.
 
Just to add that if one looks at the benchmarks between a 480 and a 5870 then in 'most' games the difference in framerates is roughly about 5 to 10 FPS. I simply cannot imagine anyone in their right mind paying an extra 150 dollars/pounds for that kind of performance increase.
 
Just to add that if one looks at the benchmarks between a 480 and a 5870 then in 'most' games the difference in framerates is roughly about 5 to 10 FPS. I simply cannot imagine anyone in their right mind paying an extra 150 dollars/pounds for that kind of performance increase.


March 25, 2010

Introduction - NVIDIA Fermi - GTX 470 - GTX 480 - GTX 480 SLI Review | [H]ard|OCP

126962492671BZgJ5ZxI_6_4.gif


126962492671BZgJ5ZxI_4_2.gif


126962492671BZgJ5ZxI_3_4.gif


126962492671BZgJ5ZxI_5_4.gif


126962492671BZgJ5ZxI_5_6.gif





Power - Temp - Sound - NVIDIA Fermi - GTX 470 - GTX 480 - GTX 480 SLI Review | [H]ard|OCP

126962492671BZgJ5ZxI_7_1.png


126962492671BZgJ5ZxI_7_2.png
 
Translated from Hardware Infos: See link below.


Beginning of may we reported about a new NVIDIA high end card with the full 512 Shader-and 64 text ureinheiten. Bill Dally, Chief developer for NVIDIA, has now confirmed the card to the page kitguru and said even has some details.​

Therefore the GTX to 485 named graphics along with 32 other Shadereinheiten and 4 additional text ureinheiten to GTX 480 a 20 MHz higher chip clock, and subsequently also 40 MHz of higher Shadertakt. This suggests with the following specifications:

Link Removed due to 404 Error


.


The specifications true, GTX 485 in approximately 10 percent would have more Shader-and texture performance and nearly 3 percent more grid power. The memory it has lost nothing so far, however, GTX 480 suffers anything but under bandwidth problems, so it certainly is possible that it leaves them at 924 MHz.​


In General, selling 512 Shader-64 text ureinheiten cards is now possible because the TSMC has improved 40 nm manufacturing and it was now time to select. We met with an improved cooling system, which is hopefully not only leistungsstärker the GTX 480, but also quiet increased chip and Shadertakt.​



GTX 485 to expected to be left in the autumn in the market. The selling price should then be $ 20 the GTX 480 introductory price was at $ 499.​



 
Last edited:
geforce.jpg


Possible GTX 475 at a later date

In addition to the specs of the GTX 460 card that were leaked a while back, we managed to score additional bits and pieces of the information and it looks like that the GF104 will be crippled from the start. This leaves room for a new card that will be based on the non-crippled GF104 with full specs.

The GF104 should have 384 SPs organized as 16 clusters of 24 SPs. The leaked screenshot that Expreview.com managed to grab showed the GTX 460 with 336 SPs/CUDA Cores which means that it has 2 disabled cluster which leaves it at 336 SP. The full GF104 has a 256-bit memory interface, but with one MC block disabled you get the 192-bit memory interface seen on the leaked specs list for the GTX 460.

The leaked specs from Expreview show the GTX 460 working at 675MHz for the GPU, 1350MHz for shaders and 1800MHz (3600MHz effective) for the 768MB of GDDR5 memory paired up with a 192-bit memory interface. According to their info, the GTX 465 should have a lower TDP and it should be priced at around US $230.

This sounds a lot like the G92 which had full 128 SPs but as you remember the first card, the 8800GT had only 112 SPs, and according to our sources, there has been some talk about the possible GTX 475 based on fully enabled GF104.

We will keep our ears on Nvidia's door as more info has to appear sooner or later. You can check out the leaked GTX 460 specs Link Removed due to 404 Error.

Ref: Link Removed due to 404 Error
 
geforce.jpg


Replaced by GF104 architecture


Nvidia's mainstream Fermi generation got cancelled. It's not tragic, as the promising GF104 architecture takes over and Nvidia's future mainstream as well as entry level chips will be based on this improved architecture.

There are two chips to come after the GF104, the GF106 and finally the cheapest and slowest GF108.

Both of these chips are expected for back to school so Nvidia can have the final DirectX 11 portfolio ready for August, just before the start of the shopping spree in western hemisphere.

The chips based on Fermi that were supposed to have two times or four times fewer shaders have been cancelled due to high TDPs and the fact that the GF104, Nvidia's new chip architecture, was brought in earlier than expected. You can view a performance chip as ½ of the original high end design and entry level as ¼ of the original big chip.

The GF104 is coming to the rescue, but despite it Nvidia will be some six month behind ATI on the mainstream and performance front.

Still some customers will wait for Nvidia and we are hearing that notebook manufactures are very excited about GF104 and its derivatives.

Ref: Link Removed due to 404 Error
 
Dude can't count either... "Looks & Price" thats 2 points in my book.... lol, and Physx is hardly Nvidia exclusive as can be hacked to run on Ati numerous ways. I guess it's hard to be objective when you're a FanBoy...lol speaking as a former Nvidia fanboy myself.
 
Back
Top