The first proper benchmarks of the AMD Radeon Fury X “Fiji” GPU based graphics card have been revealed by Videocardz. The benchmarks show the performance results which we all have been anticipating for the upcoming Radeon Fury X graphics card that delivers increased graphic performance but the more important thing is that it does so while using the latest tech that include next generation HBM memory which pumps out significant amount of bandwidth.
AMD Radeon Fury X 3DMark FireStrike Benchmarks Unveiled
The AMD Radeon Fury X will be the latest and greatest graphics card from AMD featuring their new Fiji GPU core which is the first to utilize high-bandwidth memory. We covered the specifications of the card yesterday and found out what really lies beneath the GPU core. The Fiji GPU which measures around 560mm2 is the biggest chip AMD has ever produced but it is also the fastest. Packed inside the Fiji GPU core are 64 Compute Units, each featuring 64 stream processors. After running the numbers, we get a total of 4096 stream cores, 128 raster operation units and 256 texture mapping units. These are insane specifications for a card that is meant to push 4K performance.
The Radeon Fury X also comes packed with HBM along side the GPU die. The GPU and HBM are actually allotted space on an interposer die which measures around 1000mm2. While the complete chip is larger than any previous GPU ever produced, HBM saves a lot of space around the PCB which would otherwise have been taken over by several GDDR5 chips. This leads to significantly smaller PCB and design schemes like the one we have seen in renders, reducing the cost and heat output of flagship graphics cards. The card comes packed with 4 GB of HBM memory that operates along a 4096-bit wide bus interface, clocked at 500 MHz and pumps out 512 GB/s bandwidth. The core clock is maintained at 1050 MHz for reference designs and leads to a total of 8.6 TFlops of FP32 compute performance. AMD has managed to squeeze more performance per watt out of their Fiji GPU compared to their last gen Hawaii GPU that is rated around 28.7 GFlops/W versus 19.4 GFlops/W.
But the real test of the Radeon Fury X would be how it manages to hold up against its rival, the GM200 based GeForce graphics cards. NVIDIA’s GM200 are also the biggest GPUs ever produced with a die size of 601mm2. Their parts are focused solely on single precision with over 7 TFlops of compute performance but also a pure gaming oriented series since Pascal is reserved for the double precision oriented Tesla lineup. The launch of Fiji GPU takes place three months after GM200 arrived in the market and it is expected to feature new technology set such as full support of DirectX 12 and HBM memory but let’s see how well these features translate in bench marks.
Image Credits: Videocardz
AMD Fury X Edging Past The Titan X at Firestrike Ultra 4K Benchmark
As usual, we have several synthetic benchmarks covered by Videocardz which were taken from 3DMark’s data base which has validated several new cards. Previously, the GeForce GTX 980 Ti synthetic benchmarks covered by Videocardz turned out very true in the sense that they were close to the real benchmarks and we can again expect these results to be close to the cards which hit the market later this month. The benchmark shows the Fiji XT based Radeon Fury X performing faster or 100 3DMarks better than the GeForce GTX 980 Ti in Firestrike Ultra and Extreme (both cards are clocked at reference speeds). The GeForce GTX 980 Ti OC shows additional performance due to higher clock speeds but we can expect the same from the Fury X which is designed for OC with its unique water cooling design which is a beauty. The GeForce GTX Titan X is faster in 3DMark Firestrike Extreme but is slightly slower in 3DMark Firestrike Ultra compared to the Radeon Fury X due to its higher ROP count which benefit in gaming at 4K and higher resolutions.
The Radeon R9 390X and Radeon R9 390 can also be seen in the benchmarks with the Hawaii Pro with faster clock speeds performing slightly ahead of the GeForce GTX 970 while the Radeon R9 390X is shown faster than the Radeon R9 290X due to larger VRAM and higher clock speeds but lags behind the GeForce GTX 980 in both 3DMark benchmarks.
AMD Radeon Fury X Is a Beast In CrossFire Benchmarks
AMD is also fine tuning CrossFireX performance for their upcoming beast and we can see the card performing a slight notch ahead of the GTX 980 Ti SLI and just a few points below the GTX Titan X in SLI. The Radeon Fury X has been rumored to cost around the $999 US range but we are expecting a much lower cost in the case AMD really wants this card to compete against the GM200 based rivals.
AMD Radeon 300 Series Product Lineup:
| WCCFTech | Fury X (Water Cooled) | Fury X (Air Cooled) | Fury (Air Cooled) | R9 290X |
|---|---|---|---|---|
| GPU | Fiji XT | Fiji XT | Fiji Pro | Hawaii XT |
| Stream Processors | 4096 | 4096 | 3584 | 2816 |
| GCN Compute Units | 64 | 64 | 56 | 44 |
| Render Output Units | 128 | 128 | 128 | 64 |
| Texture Mapping Units | 256 | 256 | 224 | 176 |
| GPU Frequency | ≥ 1050Mhz | 1050Mhz | 1000Mhz | 1000Mhz |
| Memory | 4GB HBM | 4GB HBM | 4GB HBM | 4GB GDDR5 |
| Memory Interface | 4096bit | 4096bit | 4096bit | 512bit |
| Memory Frequency | 500Mhz | 500Mhz | 500Mhz | 1250Mhz |
| Effective Memory Speed | 1Gbps | 1Gbps | 1Gbps | 5Gbps |
| Memory Bandwidth | 512GB/s | 512GB/s | 512GB/s | 320GB/s |
| Cooling | Liquid, 120mm Radiator | Air, 3 Axial Fans | Air, 3 Axial Fans | Air, Single Blower Fan |
| Performance (SPFP) | ≥ 8.6 TFLOPS | 8.6 TFLOPS | 7.2 TFLOPS | 5.6 TFLOPS |
| TDP | 300W | 300W | 275W | 290W |
| GFLOPS/Watt | ≥ 28.7 | 28.7 | 26.2 | 19.4 |
| Launch Price | TBA | TBA | TBA | $549 |
No comments:
Post a Comment