Cast your vote Do you agree or disagree with GPUBoss?

Thanks for adding your opinion. Follow us on Facebook to stay up to date with the latest news!

Differences What are the advantages of each

Front view of Radeon HD 9970

Reasons to consider the
AMD Radeon HD 9970

Report a correction
Much better floating-point performance 8,192 GFLOPS vs 4,494 GFLOPS More than 80% better floating-point performance
Higher clock speed 1,000 MHz vs 836 MHz Around 20% higher clock speed
Significantly more shading units 4,096 vs 2,688 1408 more shading units
Higher pixel rate 64 GPixel/s vs 46.8 GPixel/s More than 35% higher pixel rate
Higher memory bandwidth 307 GB/s vs 288.4 GB/s More than 5% higher memory bandwidth
More render output processors 64 vs 48 16 more render output processors
More texture mapping units 256 vs 224 32 more texture mapping units
Wider memory bus 512 bit vs 384 bit Around 35% wider memory bus
Front view of GeForce GTX TITAN

Reasons to consider the
Nvidia GeForce GTX TITAN

Report a correction
More memory 6,144 MB vs 4,096 MB 50% more memory
Higher effective memory clock speed 6,008 MHz vs 5,000 MHz More than 20% higher effective memory clock speed
Higher texture rate 187.3 GTexel/s vs 168 GTexel/s More than 10% higher texture rate
Higher memory clock speed 1,502 MHz vs 1,250 MHz More than 20% higher memory clock speed

Features Key features of the Radeon HD 9970  vs GeForce GTX TITAN 

memory bandwidth Rate at which data can be read from or stored in onboard memory

Radeon HD 9970
307 GB/s
288.4 GB/s

pixel rate Number of pixels a graphics card can render to the screen every second

Radeon HD 9970
64 GPixel/s
46.8 GPixel/s

texture rate Speed at which a graphics card can perform texture mapping

Radeon HD 9970
168 GTexel/s
187.3 GTexel/s

floating point performance How fast the gpu can crunch numbers

Radeon HD 9970
8,192 GFLOPS
4,494 GFLOPS

shading units Subcomponents of the gpu, these run in parallel to enable fast pixel shading

texture mapping units Built into each gpu, these resize and rotate bitmaps for texturing scenes

Specifications Full list of technical specs


Radeon HD 9970  vs
GPU brand AMD Nvidia
GPU name Hawaii XT GK110
Market Desktop Desktop
Clock speed 1,000 MHz 836 MHz
Is dual GPU No No
Reference card None None

raw performance

Shading units 4,096 2,688
Texture mapping units 256 224
Render output processors 64 48
Pixel rate 64 GPixel/s 46.8 GPixel/s
Texture rate 168 GTexel/s 187.3 GTexel/s
Floating-point performance 8,192 GFLOPS 4,494 GFLOPS


Radeon HD 9970  vs
Memory clock speed 1,250 MHz 1,502 MHz
Effective memory clock speed 5,000 MHz 6,008 MHz
Memory bus 512 bit 384 bit
Memory 4,096 MB 6,144 MB
Memory type GDDR5 GDDR5
Memory bandwidth 307 GB/s 288.4 GB/s

noise and power

TDP 255W 250W

Read more


Showing 12 comments.
Your English was spot on :)
ATI Radeon HD 9970 is coming.
Well the hd 9970 hasnt been officially released. And all of the info on the hd 9970 might be incorrect. But the of course there are "pro" gpus like Nvidia quadro k6000 that are stronger. Sorry for my bad english.
Based on these stats and comparing other card differences one can guess the performance gain in FPS from these 2 cards. i mean thats all most people want to know anyway. What gets me more frames? Well if you crunch the numbers the answer is the 9970 should get 22-25% more FPS at 1080p gaming which is amazing considering nvidia only really gained 0-5% better FPS from 600-700 series. Remember to take into account that the 780 is not a 680 replacement, its a whole new type of card. 680/770, 670 equal was eliminated then you have 660/760 etc.... Nvidia played a dirty trick on us this gen of cards. I'm sure with the release of the AMD 9000 series Nvidia will actually step it up for real.
Yes, for a two or three more years. GCN will be long gone ina few years and with it every optimization you're writing about. Just like VLIW or Fermi or Kepler.
8,192 Gflops 4,494 Gflops, 1,000 mhz vs 836 mhz, 64 Gpixels vs 47 Gpixels, AND HOW might i ask can they claim to be "the fastest" ?
"Based on NVIDIA's flagship GK110 silicon, it delivers outstanding single-GPU performance, easily claiming the title of "world's fastest GPU." Are they mocking them?
Agreed. The HD7950 (non-Boost) runs 1250MHz on 384-bit, with a bandwidth of ~240GB/s. Also with the memory of the HD7970GHz at 1400Mhz already, why would AMD release their next-gen flagship video card with anything below the current gen flagship, heat and cost issues not withstanding?
Also consider the next gen games are all optimized for Radeon. (what PS4 and Xbox1 are using).
How can a 512bit wide bus, working at 5ghz (1250 TIMES 4, GDDR5 IS QUAD PUMPED), yield 144GB/S and a 384bit bus working at 6ghz yield the double? For god's sake, a 512bit bus @ 5ghz yields a 320gb/s bandwidth. If Texture fill rate is calculated by the number of TMU's times the core clock, how can a HIGHER CLOCKED CARD with a HIGHER NUMBER OF TMU's yield a LOWER TEXTURE FILL RATE than the other one? The right math raises the HD9970 Fill rate to 256GT/S This is an embarrassment.
Hmm I think that's what we have up there right now. Please let me know if i'm missing something though.
Typo: Titan has 6gb not 4gb
comments powered by Disqus