Page 1:What If A 768-Shader Pitcairn Existed?
Page 2:A Look At The Tech Specs
Page 3:Test Setup And Benchmarks
Page 4:Benchmark Results: 3DMark 11 And Unigine
Page 5:Benchmark Results: Metro 2033, AvP, And Crysis 2
Page 6:Benchmark Results: Mafia II, GTA IV, And Batman Arkham City
Page 7:Benchmark Results: DiRT3, StarCraft 2, And Battlefield 3
Page 8:Impressions: Overall Performance, Power, Thermals
Page 9:A Wistful Conclusion
A Look At The Tech Specs
We can’t help but wonder why AMD leaves such a large gap in its line-up instead of plugging it with a card that would practically be without a direct competitor. Even overclocking our Radeon HD 7770 to 1.15 GHz doesn't reliably push it to that half-way point between a stock 7770 and 7850. So, filling the void by simply increasing the clock speed of an existing product is not an option for AMD. How about tacking on a few more shader units to a Cape Verde GPU? That seems just as unlikely since it requires a redesign.
Now, we could speculate that the yields for AMD’s Pitcairn chips are apparently so good that there simply aren’t enough partially-defective chips to “harvest” and recycle into a card that slots in between the 7700 and 7800 families (a move we’ve seen from AMD and other companies in the past). AMD's Radeon HD 5830 and Nvidia's GeForce GTX 560 TI 448 Core are good examples.
On the other hand, it's just good business to address an audience looking for a better card at a certain price point, even if it means artificially paring down an otherwise healthy existing product. Again, that’s something we’ve seen in the past on a number of occasions.
The engineering sample itself.
Here’s how our engineering sample would fit into AMD’s existing portfolio of GCN-based cards:
|Radeon HD 7770||Radeon HD 7850|
|Radeon HD 7850||Radeon HD 7870|
|Full Color ROPs||16||32||32||32|
|GPU CLock||1000 MHz||860 MHz||860 MHz||1000 MHz|
|Texture Fillrate||40 Gtex/s||41.3 Gtex/s||55 Gtex/s||80 Gtex/s|
|Memory Clock||1125 MHz||1200 MHz||1200 MHz||1200 MHz|
|Memory Bandwidth||72 GB/s||153.6 GB/s||153.6 GB/s||153.6 GB/s|
|Video-RAM||1 GB GDDR5||2 GB GDDR5||2 GB GDDR5||2 GB GDDR5|
|Manufacturing Process||28 nm||28 nm||28 nm||28 nm|
|Aux Power||1 x 6-pin||1 x 6-pin||1 x 6-pin||2 x 6-pin|
|Power Consumption (max. measured)||80 W||120 Watt*||130 W||175 W|
(Newegg, May 08, 2012)
|~$149+||~$200+ (If we got our way)||~$250+||~$330+|
*Note: This is an engineering sample, not a final GPU.
This is what the card looks like without its cooler.
Actually, the tech specs represent a pretty plausible starting point for our analysis, since a 25 percent reduction in shader units should translate to between 20 and 25 percent less performance, depending on the workload. Since we’re dealing with an engineering sample on a prototype board intended to bring up full Radeon HD 7850 cards, attributes like power consumption and heat don't bear the weight they would if we were looking at a piece of production hardware. Let’s not forget, the express purpose of the card we have in-house is to give board partners a way to tweak and improve their designs, bringing down power consumption and optimizing cooling on final boards. With that said, we don’t think AMD has anything to worry about, based on the numbers we got from our engineering sample. For what it is, the board returned completely acceptable results.
A view from below.
Even in its prototype stage, our review sample does not disappoint. So, without further ado, let’s get cracking on those benchmarks!
- What If A 768-Shader Pitcairn Existed?
- A Look At The Tech Specs
- Test Setup And Benchmarks
- Benchmark Results: 3DMark 11 And Unigine
- Benchmark Results: Metro 2033, AvP, And Crysis 2
- Benchmark Results: Mafia II, GTA IV, And Batman Arkham City
- Benchmark Results: DiRT3, StarCraft 2, And Battlefield 3
- Impressions: Overall Performance, Power, Thermals
- A Wistful Conclusion