NVIDIA GTX 1080 FP64 PERFORMANCE PROFESSIONAL
Coming in 10% slower and 20% behind in perf/W is still completely fine for consumers and many professional workloads if it's a cheaper option.
NVIDIA GTX 1080 FP64 PERFORMANCE FULL
Well, as long as it's behind Pascal in either perf/W or perf/area you can consider it a failure without being a complete nutjob, and realistically it's probably not going to beat out full GP102. Even a 3584 shader part at 1.5 GHz should beat the 1070 pretty nicely. I wouldn't expect anything below 1080 for full Vega, even with only a clock speed bump from Fury X. TL DR: if you see reviews with Firestrike Extreme & Timespy having over GTX 1080 Ti out-of-the-box OC scores it's a good prospect for the card.
For the cut down RX VEGA to compete it'd have to be a roughly 250W rated card. Games may paint a different picture due to optimizations of course.Īnother concern should be the power / cooling required: the GTX 1080 is a mere 180W rating. In this benchmark it's about +30-40% from GTX 1080 to GTX 1080 Ti.Īll of the RX Vega cards have to be above the GTX 1070 OC to make any sense whatsoever, thus the top part is not going to be slower than the GTX 1080. RX VEGA 3rd SKU? possibly as replacement to R9 Fury X, presumably 225W TDP RX VEGA cut down SKU goes here, likely cut 10-20%, presumably 250-275W TDP rating
RX VEGA 4096 shader SKU goes here likely or above here if it's good, presumably 300W TDP rating RX VEGA 4096 shader / 64 compute unit SKU goes here or above here at ~ 1500MHz, presumably 300W TDP rating Right now AMD's lineup in terms of 3DMark Firestrike Extreme is like so: The same goes for Timespy, a 8K score (2x RX 570) would barely be faster than a GTX 1080. If it turns out that the 4096 shader RX VEGA comes out with > 13K Firestrike Extreme instead of ~11K Firestrike Extreme it would be a good sign (1250MHz with 4096 shaders ala 2xRX 570 should get you 10.5K which is barely better than a GTX 1080 OC). In addition to that a Galax HOF card comes with over 12.5 TFLOPs out of the box. (So right between a GTX 1080 and GTX 1080 Ti which have roughly a 30 to 40% performance delta.) The stock GTX 1080 Ti may be a ~11.5 TFLOP part boost but the issue is that TFLOPs are just one part of the picture : the AMD Radeons have been held back by pixel fill due to lower ROPs. Click to expand.I just hope that it competes with the stock clocked GTX 1080 Ti in graphics within 10-15% without excessive voltage while also providing half decent FP64 performance.
0 kommentar(er)
