The Radeon HD 4850 & 4870: AMD Wins at $199 and $299
by Anand Lal Shimpi & Derek Wilson on June 25, 2008 12:00 AM EST- Posted in
- GPUs
One, er, Hub to Rule them All?
With R500 AMD introduced its first ring bus, a high speed, high bandwidth bus designed to move tons of data between consumers of memory bandwidth and the memory controllers themselves. The R600 GPU saw an updated version of the ring bus, capable of moving 100GB/s of data internally:
On R600 the ring bus consisted of two 512-bit links for true bi-directional operation (data could be sent either way along the bus) and delivered a total of 100GB/s of internal bandwidth. The ring bus was a monster and it was something that AMD was incredibly proud of, however in the quest for better performance per watt, AMD had to rid itself of the ring and replace it with a more conventional switched hub architecture:
With the ring bus data needed to be forwarded from one ring stop to the next and all clients got access to the full bandwidth, regardless of whether or not they needed it. For relatively low bandwidth data (e.g. UVD2 and display controller data), the ring bus was a horrible waste of power.
With the RV770 all that exists is a simple switched hub, which means that sending data to the display controller, PCIe and UVD2 (AMD's video decode engine) traffic are now far less costly from a power standpoint. Another side effect of ditching the ring bus is a reduction in latency since data is sent point to point rather than around a ring. With the move to a hub, AMD increased their internal bus width to 2kbits wide (which is huge). Maximum bandwidth has increased to 192GB/s (in 4870) but this depends on clock speeds.
With nearly double the internal bandwidth and a point to point communication system, latency between memory clients should be decreased, and huge amounts of data can move between parts of the chip. Certainly getting enough data on to the GPU to feed 800 execution units is a major undertaking and AMD needed to make a lot of things wider to accommodate this.
The CrossFire Sideport
Although AMD isn't talking about it now, the CrossFire Sideport is a new feature of the RV770 architecture that isn't in use on the RV770 at all. In future, single-card, multi-GPU solutions (*cough* R700) this interface will be used to communicate between adjacent GPUs - in theory allowing for better scaling with CrossFire. We'll be able to test this shortly as AMD is quickly readying its dual-GPU RV770 card under the R700 codename.
One thing is for sure, anything AMD can do to assist in providing more reliable consistent scaling with CrossFire will go a long way to help them move past some of the road blocks they currently have with respect to competing in the high end space. We're excited to see if this really makes a difference, as currently CrossFire is performed the same way it always has been: by combining the output of the rendered framebuffer of two cards. Adding some sort of real GPU-to-GPU communication might help sort out some of their issues.
Facebook
Twitter
RSS
215 Comments
View All Comments
BusterGoode - Sunday, June 29, 2008 - link
Thanks, great article by the way Anandtech is my first stop for reviews.jay401 - Wednesday, June 25, 2008 - link
Good but I just wish AMD would give it a full 512-bit memory bus bandwidth. Tired of 256-bit. It's so dated and it shows in the overall bandwidth compared to NVidia's cards with 512-bit bus widths. All that fancy GDDR4/5 and it doesn't actually shoot them way ahead of NVidia's cards in memory bandwidth because they halve the bus width by going with 256-bit instead of 512-bit. When they offer 512-bit the cards will REALLY shine.Spoelie - Thursday, June 26, 2008 - link
Except that when R600 had a 512bit bus, it didn't show any advantage over RV670 with a 256bit bus. And that was with GDDR3 vs GDDR3, not GDDR5 like in RV770 case.JarredWalton - Thursday, June 26, 2008 - link
R600 was 512-bit ring bus with 256-bit memory interface (four 64-bit interfaces). http://www.anandtech.com/showdoc.aspx?i=2552&p...">Read about it here for a refresh. Besides being more costly to implement, it used a lot of power and didn't actually end up providing provably better performance. I think it was an interesting approach that turned out to be less than perfect... just like NetBurst was an interesting design that turned out to have serious power limitations.Spoelie - Thursday, June 26, 2008 - link
Except that it was not, that was R520 ;) and R580 is the X19x0 series. That second one proved to be the superior solution over time.R600 is the x2900xt, and it had a 1024bit ring bus with 512bit memory interface.
DerekWilson - Sunday, June 29, 2008 - link
yeah, r600 was 512-bithttp://www.anandtech.com/showdoc.aspx?i=2988&p...">http://www.anandtech.com/showdoc.aspx?i=2988&p...
looking at external bus width is an interesting challenge ... and gddr5 makes things a little more crazy in that clock speed and bus width can be so low with such high data rates ...
but the 4870 does have 16 memory modules on it ... so that's a bit of a barrier to higher bit width busses ...
JarredWalton - Wednesday, June 25, 2008 - link
I'd argue that the 512-bit memory interface on NVIDIA's cards is at least partly to blame for their high pricing. All things being equal, a 512-bit interface costs a lot more to implement than a 256-bit interface. GDDR5 at 900MHz is effectively the same as GDDR3 at 1800MHz... except no one is able to make 1800MHz GDDR3. Latencies might favor one or the other solution, but latencies are usually covered by caching and other design decisions in the GPU world.geok1ng - Wednesday, June 25, 2008 - link
The tests showed what i feared: my 8800GT is getting old to pump my Apple at 2560x1600 even without AA! But the tests also showed that the 512MB of DDR5 on the 4870 justifies the higher price tag over the 4850, something that the 3870/3850 pair failed to demonstrate. It remains the question: will 1GB of DDR5 detrone NVIDIA and rule the 30 inches realm of single GPU solutions?IKeelU - Wednesday, June 25, 2008 - link
"It is as if AMD and NVIDIA just started pulling out hardware and throwing it at eachother"This makes me crack up...I just imagine two bruised and sweaty middle-aged CEO's flinging PCBs at each other, like children in a snowball fight.
Thorsson - Wednesday, June 25, 2008 - link
The heat is worrying. I'd like to see how aftermarket coolers work with a 4870.