Stream Processing Units and memory interface of ATI Radeon HD 2000 Series @OCWORKBENCH

	<![if lt IE 4]> <SCRIPT SRC="http://imgserv.adbutler.com/adserve/;ID=131075;size=468x60;setID=61435;type=js" LANGUAGE="JAVASCRIPT"> </SCRIPT> <NOSCRIPT> <a href="http://imgserv.adbutler.com/go2/;ID=131075;size=468x60;setID=61435"> <img src="http://imgserv.adbutler.com/adserve.ibs/;ID=131075;size=468x60;setID=61435;type=img" border="0" height="60" width="468"></a> </NOSCRIPT><![endif]>
中文版 \| News \| Archives \| Reviews \| Forum \| $ Deals \| Links \| History \| Contact \| Privacy

Stream Processing Units and memory interface of ATI Radeon HD 2000 Series
Bluetooth 14 May 2007

Pricing of
Radeon HD 2900XT

>>>

Stream Processing Units and memory interface

The Stream Processing Units are arranged as a 5 way superscalr shader processors. It can co-issue up to 5 scalar MAD (multiply-add) instructions per clk. One of the 5 stream processing units handles transcendental instructions as well (SIN, COS, LOG, EXP,etc). It has 32 bits floating point precision. There is integer and bitwse operation support.

The brand excution units handle flow control and conditional operations. This frees stream processor units from handling this task and practically eliminate flow control perforamnce overhead.

Processors	Stream processors	FLOPS per unit	GPU CLK	Processing Rate
ATI Radeon HD 2900 XT	320	2	742 MHz	475 GigaFLOPS
ATI Radeon HD 2600 XT	120	2	800 MHz	192 GigaFLOPS
ATI Radeon HD 2400 XT	40	2	700 MHz	56 GigaFLOPS
High-end dual core CPU	8 (4 per core)	2	3000 MHz	48 Giga FLOPS

The 512bits memory interface ensures more bandwdith with existing memory technology. it also lower the memory clk required to achieve target bandwidth. The ring bus design enhances the routing to improve scalability and reduce lantecy.

As we can see from the chart below, bandwidth drives performance. HDR performance is very bandwidth intensive. The chart below compares the Radeon 1950xtx and HD 2900XT.

Next >>>

Discuss in our Graphics forum.