Monday, November 30th 2009
AMD Preparing Radeon HD 5950 for Q1 2010?
Close to two weeks after launching the industry's fastest graphics card with the Radeon HD 5970 2 GB, it looks like AMD will back the release with another high-end graphics card in Q1 2010, ideally to stack up a lineup against NVIDIA's performance DirectX 11 offerings that are slated for around the same time. The new release comes in the form of Radeon HD 5950, aimed to occupy the gap between the Radeon HD 5870 and Radeon HD 5970.
The Radeon HD 5950 will retain the design methodology of the HD 5970. It will use two AMD Cypress GPUs with the same configuration Radeon HD 5850 uses. It has 1440 stream processors enabled per GPU, 72 TMUs and 32 ROPs enabled, and 256-bit wide GDDR5 memory interfaces per GPU probably to hold 2 GB of total memory. Just as the dual-GPU HD 5970 uses lower clock speeds compared to the single-GPU HD 5870 that uses the same GPU, HD 5950 keeps up with the trend. It is expected to have its core clocked between 650~675 MHz, and memory at 900~1000 MHz (3.60 GHz to 4.00 GHz effective).
Source:
NordicHardware
The Radeon HD 5950 will retain the design methodology of the HD 5970. It will use two AMD Cypress GPUs with the same configuration Radeon HD 5850 uses. It has 1440 stream processors enabled per GPU, 72 TMUs and 32 ROPs enabled, and 256-bit wide GDDR5 memory interfaces per GPU probably to hold 2 GB of total memory. Just as the dual-GPU HD 5970 uses lower clock speeds compared to the single-GPU HD 5870 that uses the same GPU, HD 5950 keeps up with the trend. It is expected to have its core clocked between 650~675 MHz, and memory at 900~1000 MHz (3.60 GHz to 4.00 GHz effective).
41 Comments on AMD Preparing Radeon HD 5950 for Q1 2010?
this also looks cool
IMO they did it to taunt massive overclockability, and to keep the noise down a little, even though I hear its damn loud anyway, oh and maybe to keep the power just under the 'rated' 300w of the power connectors and pci-e slot, even though as weve seen they can clock much higher than a stock 5870 with some volts and are fine power wise.
In the end it does give it massive over clock ability, but I don't really count that as it should be since it's 2x 5870 GPU's that were clocked down. But hopefully that helps oyu understand how it does come into it.
On a separate note, even though I bought one, I don't think a 5870 is $400 worth of card, even more of a rip at $~600 AUD at the time, considering we trade at over 90 US cents, they want ~$1000 AUD for a 5970 :ohwell:
My feeling is they have price drops ready and waiting for a Fermi counter-attack.
All I'm really saying is on the entire lineup I don't like their pricing (can see why tho, sell em for as much as they can get away with while they're still the best), and for the dual GPU card/s, I personally don't like their choice of clock speeds.
And who says I liked their choice of clock speeds? (did not btw) Not to mention you couldn't SLi it with a GTX280/285 anyway, and have roughly the same dilemma as 5970+5870.
Keep in mind Nvidia never really taunted overclockability and GT200 chips clock very well, especially GTX260/275 variants.
At the end of the day they do what they do to maximize sales/profits, doesn't mean I like some of their choices, that's really all I'm saying here.
and hey not liking aspects doesn't mean I won't buy it...
bought a GTX295, and a 5870, probably will look at a 5970 or Fermi soon enough :p
We have reached the performance plateau required and now we are working on dedicated hardware to perform the pretties/physics calculations offloading it from the CPU to GPU core shaders. Much like UVD and using Stream or Cuda to perform tasks, we have a tesselator in place, and multithreading.
Now just for the next round of real DX11 games to utilize the tech. Unfortunately the more I look at games the more I realize we will need a Xbox refresh to push the games into DX11 territory. Perhaps Valve will save us.
One thing is certain though, the difference between 60FPS and 120 is crap, but the difference between angle independant AA, larger bitmaps, tessellation, and other new hardware tech is going to be the driving technology soon.