The final benchmark will be Luxmark v2.0, which focuses on GPU compute performance via OpenCL. The benchmark is one of five used by AMD when they launched the Radeon HD 7970. The benchmark itself is a great test of a GPU’s compute performance and we will be running all three possible scenes to get a good grasp on where the FirePro W5000 stands in terms of GPU compute performance and whether or not the FirePro cards offer a performance advantage in this particular situation.
The first test is the simple benchmark: LuxBall HDR, which is a simple 262,000+ triangles scene.
It’s obvious in this test that more cores is better, with the Radeon HD 7970’s 2048 GCN stream processors simply dominating the FirePro W5000’s 768 stream processors. The performance difference here is massive, with the HD 7970 being 133% faster than the FirePro W5000.
The second test is the medium benchmark: Sala, a scene comprised of 488,000 triangles that was designed by Daniel “ZanQdo” Salazar and was adapted for SLG2 by Michael “neo2068″ Klemm.
Again we see the Radeon HD 7970 dominate in this Open CL benchmark, outpacing the FirePro W5000 by 144%.
The final test is the complex benchmark: Room, which is a 2,000,000 triangle scene that was designed by Mourelas “Moure” Konstantinos. This scene really stresses the GPU’s compute performance. bringing weaker cards to their knees.
Once again, the Radeon HD 7970 comes out on top in the most complex scene Luxmark v2.0 offers. The Radeon card outpaces the FirePro W5000 by 145% here.
It seems driver optimizations that gave the FirePro W5000 the advantage in previous tests don’t come into play in this OpenCL benchmark. Simply put the W5000 just doesn’t have the specs to overcome the Radeon HD 7970 when it comes to OpenCL, this is a job for the W9000 instead.