NVIDIA GeForce 8800 Ultra review
Posted by Hilbert Hagedoorn on: 04/27/2007 01:00 PM [ 0 comment(s) ]
Wazzuuup! Welcome to page two of 15 (yeah, really). Six months people, that is how long it's been since NVIDIA released its GeForce 8800 GTS and GTX. It really took a lot of us by surprise as the performance was and is breathtaking. In-between the 320 MB model of the 8800 GTS was released and a week or two ago the 8500 / 8600 series of DirectX 10 products.
Question - why are there no DX10 titles available on the market yet? What the heck was Microsoft thinking releasing Vista propagating the new era in DX10 gaming? Microsoft put out some really good (and I still say photoshopped) MS Flight-Simulator screenshots. Microsoft has its own game-studio... so again Microsoft what the hell are you guys doing? Give us at least one DX10 game, dudes I'm begging you. A quarter of an entire year has passed and after 10+ updates and three reinstalls of Vista finally is crashing just once a week. Give us a game, please? We bought four business licenses at top-dollar, come on just one game... just one... ? *sighs*
Alright back on topic. So the new big pappa of graphics cards is called the GeForce 8800 Ultra that comes with no less than and precisely the same as the GTX, 768 MB memory. So how does the new and current GeForce product line shape up? Have a look:
- GeForce 8800 Ultra
$999- $829- $699
- GeForce 8800 GTX - $599
- GeForce 8800 GTS 640 MB- $449
- GeForce 8800 GTS 320 MB - $300
- GeForce 8600 GTS - $219
- GeForce 8600 GT - $149
- GeForce 8500 GT - $99
- GeForce 7600 GS - $89
- GeForce 7300 GT - $89
- GeForce 7300 LE - $79
- GeForce 7100 GS - $59 but they should give it away for free ;)
Now obviously the minute ATI's R600 & 8800 Ultra will become available I expect another shift in manufacturer suggested retail prices. Small hint, expect the 320MB to drop in price soon. The MSRP for the Ultra is 829 USD - but since nobody will buy it at that price expect it to be 699 by the end of this month.
The GeForce 8800 Ultra
Ultra. Try to image how that would sound out of the mouth of Arnold Schwarzenegger : "I just bought this GeForce 8800 Oeltra". No clue why I just wrote this? Well guns, action, gamers ammo, muscle power, see the parallel here ? Wouldn't it be fun to have a Terminator edition of cards called Oeltra? Just like the Dodge or Pontiac its a muscle car for PC gaming.
Ahem let's be geek again and do transistors; yay! So did you know that G70/G71 (GeForce 7800/7900) each had nearly 300 Million transistors? Well, G80 is a 681 million transistor and counting product. Which means performance. And the faster you clock these transistors the faster it'll perform... or do something like spark, boom... smoke. Now the 8800 Ultra has the 128 streaming cores (Unified Shader processors) and it comes with 768 MB of gDDR3 memory.
The main differences between the GTX and Ultra: memory was clocked at 1800 MHz (2x900) on the GTX resulting into 86 Gigabyte per second memory bandwidth. This has changed. Memory is now defaulting at 2160 MHz, which equals a theoretical bandwidth of 101.3 GB/s. So do the math, that's roughly 15% extra bandwidth, which is one of the most limiting factors for a high-end GPU. The memory is still on that weird 384-bit (12 pieces of 16Mx32 memory) bus.
Now here's where we'll go on a quick side-track. All reviews are rambling on Unified Shaders, on last gen hardware it was Shader's model 2 and 3, Pixel Shaders, Vertex Shaders and now we have geometry shaders.
But do you guys even know what a shader is? Allow me to show give you a quick brief on what a shader operation actually is as very few consumers know what they really are.
Demystifying the shader.
If you program or play computer games or even recently attempted to purchase a video card, then you will have no doubt heard the terms "Vertex Shader" and "Pixel Shader" and with the new DirectX 10 "Geometry shaders". In these today's reviews the reviewers actually tend to think that the audience knows everything. I realized that some of you do not even have a clue what we're talking about. Sorry, that happens when you are deep into the matter consistently. Let's do a quick course on what is happening inside your graphics card for to be able to poop out colored pixels.
What do we need to render a three dimensional object as 2D on your monitor? We start off by building some sort of structure that has a surface, that surface is built from triangles. Why triangles? They are quick to calculate. How's each triangle being processed? Each triangle has to be transformed according to its relative position and orientation to the viewer. Each of the three vertices that the triangle is made up of is transformed to its proper view space position. The next step is to light the triangle by taking the transformed vertices and applying a lighting calculation for every light defined in the scene. And lastly the triangle needs to be projected to the screen in order to rasterize it. During rasterization the triangle will be shaded and textured.
Graphic processors like the GeForce series are able to perform a very large amount of these tasks. The first generation was able to draw shaded and textured triangles in hardware. The CPU still had the burden to feed the graphics processor with transformed and lit vertices, triangle gradients for shading and texturing, etc. Integrating the triangle setup into the chip logic was the next step and finally even transformation and lighting (TnL) was possible in hardware, reducing the CPU load considerably (everybody remember GeForce 256 ?). The big disadvantage was that a game programmer had no direct (i.e. program driven) control over transformation, lighting and pixel rendering because all the calculation models were fixed on the chip.
And now we finally get to the stage where we can explain shaders as that's when they got introduced.
A shader is basically nothing more than a relatively small program executed on the graphics processor to control either vertex, pixel or geometry processing and it has become intensely important in today's visual gaming experience.
Vertex and Pixel shaders allow developers to code customized transformation and lighting calculations as well as pixel coloring or all new geometry functionality on the fly, (post)processed in the GPU. With last-gen DirectX 9 cards there was separated dedicated core-logic in the CPU for pixel and vertex code execution, thus dedicated Pixel shader processors and dedicated Vertex processors. With DirectX 10 something significant changed though. Not only were Geometry shaders introduced, but the entire core logic changed to a unified shader architecture that is a more efficient approach to allow any kind of shader in any of the stream processors.
GeForce 8800 GTX and Ultra have 128 stream processors. These are the shader processors I just mentioned. And it's very unlikely that you understand what I'm about to show you, but allow me to show you an example of a Vertex and a Pixel shader. A small piece of code that is executed on the Stream (Shader) processors inside your GPU:
Example of a Pixel Shader:
float2 tc0 : TEXCOORD0; // base
half4 c : COLOR0; // diffuse
half4 main ( v2p I ) : COLOR
return I.c*tex2D (s_base,I.tc0);
Example of a Vertex Shader:
float4 P : POSITION;
float2 tc : TEXCOORD0;
float4 c : COLOR0;
float4 hpos : POSITION;
float2 tc : TEXCOORD0;
float4 c : COLOR0;
vf main (vv v)
o.hpos = mul (m_WVP, v.P); // xform, input in world coords
o.tc = v.tc; // copy tc
o.c = v.c; // copy color
Now this code itself is not at all interesting for you and I understand it means absolutely nothing to you (Ed: Hey! Some of us are software engineers!) but I just wanted to show you in some sort of generic easy to understand manner what a shader is and involves.
Okay now back to the review.
Last week we arrived at Sin City not only to cover CES but there was something else going on as well. In Las Vegas, NVIDIA had organized a briefing for a select group of the press. From Europe perhaps ten to fifteen people where invited for this somewhat privileged preview -- the topic, a technical overview of project Fermi. Fermi is of course the family name of the latest generation of GPUs from NVIDIA. The first chipset deriving from Fermi will be called the GF100 GPU which will likely be used on what we think will be called products like GeForce 360 and GeForce 380. Join us in a nice technology preview.
NVIDIA GeForce 3D Vision review
In this article we will test and review the NVIDIA GeForce 3D Vision stereo kit. NVIDIA teamed up with Samsung to optionally bundle 120 Hz LCD monitors with their all new 3D stereo shutter glasses technology. NVIDIA on their end got driver support up and going to a state where it's really good. Next to that, they redesigned the approach to the overall gaming experience. A set of shutter glasses that is wireless and rechargeable, games that are supported in the new drivers will automatically kick in 3D mode and next to that, NVIDIA really wanted a cool looking kit.
NVIDIA GF9300 (ECS GF9300TA) mainboard review
A test on the ECS GF9300T-A motherboard. Today NVIDIA is introducing their more budget conscious mainboard chipsets. The GF9300 and GF9400 based integrated graphics chipset motherboard products.
NVIDIA GeForce 8800 Ultra review
Today is the day that NVIDIA is launching it's GeForce 8800 Ultra. Now, NVIDIA tried to keep this product as secret as can be ... why ? Two reasons, to prevent technical specifications leaking onto the web. Secondly; obviously to change specs at the last minute. See ATI is releasing their R600 graphics card soon and the Ultra is the product that NVIDIA prepared to counteract in the market, an allergic reaction tothe R600 so to speak.