Palit GeForce GTX 670 JetStream review

Posted by Hilbert Hagedoorn on: 05/09/2012 01:00 PM [ 0 comment(s) ]

The graphics architecture that is Kepler

The graphics architecture that is Kepler

As you can understand, the massive memory partitions, bus-width and combination of GDDR5 memory (quad data rate) allow the GPU to work with a very high framebuffer bandwidth (effective). Let's again put most of the data in a chart to get an idea and a better overview of changes:

Graphics card	GeForce GTX 580	GeForce GTX 670	Palit GeForce GTX 670 JetSTream	GeForce GTX 680	GeForce GTX 690
Fabrication node	40nm	28nm	28nm	28nm	28nm
Shader processors	512	1344	1344	1536	3072
Streaming Multiprocessors (SM)	16	7	7	8	16
Texture Units	64	112	112	128	128x2
ROP units	48	32	32	32	32x2
Graphics Clock (Core)	772 MHz	915 / 980MHz	1006 / 1084 MHz	1006/1058MHz	915/1019MHz
Shader Processor Clock	1544 MHz	915 / 980MHz	1006 / 1084 MHz	1006/1058MHz	915/1019MHz
Memory Clock / Data rate MHz	1000 / 4000	1502 / 6008 MHz	3054 / 6108 MHz	1502 / 6008 MHz	1502 / 6008 MHz
Graphics memory	1536 MB	2048 MB	2048 MB	2048 MB	4096 MB
Memory interface	384-bit	256-bit	256-bit	256-bit	256-bit
Memory bandwidth	192 GB/s	192 GB/s	192 GB/s	192 GB/s	192 GB/s
Power connectors	1x6-pin PEG, 1x8-pin PEG	2x6-pin PEG	2x6-pin PEG	2x6-pin PEG	2x8-pin PEG
Max board power (TDP)	244 Watts	170 Watts	180 Watts	170 Watts	300 Watts
Recommended Power supply	600 Watts	550 Watts	550 Watts	550 Watts	750 Watts
GPU Thermal Threshold	97 degrees C	98 degrees C	98 degrees C	98 degrees C	98 degrees C

So we talked about the core clocks, specifications and memory partitions. Obviously there's a lot more to talk through the GPU architecture for example. To understand a graphics processor you simply need to break it down into pieces to better understand it.

Let's first look at the raw data that most of you can understand and grasp. This bit will be about the Kepler architecture, if you're not interested in g33k talk by all means please browse to the next page.

GeForce GTX 680

So above we see the GK104 block diagram that entails the Kepler architecture. Let's break it down into bits and pieces. A fully operating GK104 will have:

1536 CUDA processors (Shader cores)
192 CUDA core clusters (SM).
8 geometry units
4 raster Units
128 Texture Units
32 ROP engines
256-bit GDDR5 memory bus
DirectX 11.1

Above is thus a fully operating GK104 as used on the GTX 680. The GTX 670 uses the same chip, but has one SM (CUDA / Shader core cluster) disabled. So the more important things to focus on are the SM (block of shader processors) clusters (or SMX as NVIDIA likes to call it for the GTX 680), which each has 192 Shader processors. That's radically different from Fermi, the GeForce GTX 580 for example had 32 shader processors per SM cluster. 1536 : 192 = 8 Shader clusters (SMs). Let's blow up one such cluster:

GeForce GTX 680

Above the block diagram for a single Shader processor cluster, aka SM or SMX as NVIDIA now calls it. The new SMX has quite a bit more bite in terms of shader, texture and geometry processing. 192 CUDA cores, that's six times the number of cores per SM opposed to Fermi. Now, at the end of the pipeline we run into the ROP (Raster Operation) engine and the GTX 680 again has 32 engines for features like pixel blending and AA.

There's a total of 128 texture filtering units available for the GeForce GTX 680. The math is simple here, each SM has 16 texture units tied to it.

GeForce GTX 580 has 16 SMs X 4 Texture units = 64
GeForce GTX 670 has 7 SMs X 16 Texture units = 112
GeForce GTX 680 has 8 SMs X 16 Texture units = 128

The GK104 host interface then - The Gigathread engine, four GPCs, four memory controllers, the ROP partitions, a 768 KB L2 cache. Each GPC has eight polymorph engines - ROP partitions are nearby to the L2 cache, each shader cluster is then tied to L1 and a shared L2 cache. Shading performance is going be increased quite bit, geometry performance will get a nice boost as well. NVIDIA is using 64KB Shared Memory/L1 per SMX please note that they have a 16/48 48/16 ratio here for graphics/compute, as before with Fermi. For L2, 128KB per 64-bit memory controller. So that adds up to 512KB L2

In regards to architectural changes, on top of the pipeline NVIDIA has now added new Polymorph 2.0 (world space processing) engines and raster (screen space processing) engines, they act like a mini CPU really.

24 pages « < 4 5 6 7 next »

Palit GeForce GTX 650 Ti Boost OC edition review
For this review we test and benchmark the Palit GeForce GTX 650 Ti Boost OC edition. The product comes customized with their own PCB design, a dual-fan cooler, 2GB of memory with both that memory and the core baseclock slightly overclocked.

Palit GeForce GTX 660 Ti Jetstream review
In this review we'll look at the GeForce GTX 660 Ti from Palit, it's their all beefed up version, the GeForce GTX 660 Ti JetStream version. The GTX 660 Ti again has been equipped with a JetStream series cooler yet which remains a 3-slot design. It runs at a core clock frequency of 1006 MHz, has a boost frequency of 1085 MHz and the effective memory data rate (192-bit) is 6108 MHz.

Palit GeForce GTX 670 JetStream review
We review the Palit GeForce GTX 670 JetStream graphics card. the JetStream version which comes pre-overclocked at 1006 MHz on the baseclock and an impressible 1084 MHz on the boost clock. More interestingly, the boost clock during our test sessions was actually closer to 1200 MHz most of the time (!). To give the card enough framebuffer to work with the cards are equipped with 2048 GDDR5 on a 256-bits wide bus. Palit clocks this memory at 6108 MHz.

Palit GeForce GTX 680 4GB Jetstream review
We review the Palit GeForce GTX 680 4GB Jetstream edition. Why 4 GB ? Well some of you like to game at extremely high resolutions or have 8xAA as a bare minimum. If a graphics card runs out of graphics memory it'll starts swapping frames back and forward in that framebuffer which decreases the overall framerate. So today we'll look at the 4GB model, we'll specifically place a focus at some tests at 2560x1600 with a good chunk of AA enabled to see what difference the extra 2GB graphics memory will bring us in terms of performance.