This week Matrox, the eclectic pioneering Canadian graphics company who developed a 3D graphics controller in 1998 they called the G200, is celebrating its 20th anniversary. At the time it was a killer new part, coming on the heels of the programmable TI TMS34010 (which was not called a GPU at the time) and just ahead of the introduction of … Read more
Nvidia just won’t quit and has built a double-decker server rack-mount console of Volta AIBs that can be used for AI training among things. The hardcore specifications are: AIBs: 16x Tesla V100s, providing 10,240 Tensor cores, plus 81,920 CUDA cores with 500GB of GPU memory Performance: 2 petaFLOPS AI | 250 teraFLOPS, FP32 | 125 teraFLOPS FP64 NVSwitch Communication Channel powered … Read more
The following is a collage of stories about AIBs that occurred in Q1’18. Not meant to be an exhaustive listing of every mention of an AIB, or product announcement, it will give you some view of the general activities and areas of interest. MSPR finally means something, part 2 No longer a cheap discount disguise We’ve heard and used … Read more
Q1’18 GPU and PC shipments broke a lot of records. Q1 is traditionally a down quarter for PCs (see Figure 1). Q1’18 was down 14.1%, not a record, it went down 16.2% in Q1’16, but a big drop nonetheless. Figure 1: PC shipments and quarter-to-quarter change As the chart indicates, the PC market has been exhibiting a steady seasonal … Read more
They’re monitors, not models To the best of my recollection, Dell was first with their beautiful U3415W curved 34-inch 21:9 IPS WQHD 3440 × 1440 monitor. They then topped that with the U3818DW 37.5-inch, 21:9 IPS WQHD+ 3840 × 1600. Then HP introduced the Envy 34, an AIO, and a workstation version. And now Fujitsu has joined the ranks and … Read more
We’ve been racing to keep pace with Moore’s Law for decades. GPUs have become our daily workhorse for visualization and aspects of compute. We’re close to adding FPGAs and quantum computing to that mix. This is the era of processors. Hardware is important again. The story of the last twenty years has been that everything is software; as long as … Read more
Fastest boot-to-shoot Enmotus, as you will recall, has designed some clever storage-blending coupled with machine learning to make frequently used applications and files first to load, and run. Is that important? And if it is, can you measure it, can you prove it? Yes, yes, and yes. We took an AMD Threadripper with an EVGA 1080Ti, and ran some tests … Read more
We have been evaluating Dell’s S718QL projector, and what an amazing piece of technology it is. We used the machine with our 120-inch high-gain screen, and placed it on the floor, 9-inches from the screen. The 4k HDR laser projector is extraordinarily bright—5,000 ANSI lumens (~1,460 Nits). For video projectors, 1000 ANSI Lumens is the minimum that a projector should … Read more
The triumphant now wear blue badges Raja Koduri was the vanguard who left AMD in December and popped up at Intel in January, and he’s lured a few of his old friends to join him. This month Sr. Director, Global Product Marketing, Chris Hook, left AMD after 20 years of service, and is now at Intel, working with Koduri. And … Read more
What Nvidia is doing to Intel, Wave is trying to do to Nvidia, and it might just work We got a look at Nvidia’s DGX2 datacenter accelerator at GTC a few weeks ago and about the same time Wave Computing started dropping hints that their DPU (Dataflow Processing Unit) is about ready to ship in datacenter-ready systems and that we … Read more
Act two of the Zen One year after the release of the AMD’s Ryzen processors, the company announced the 2nd generation AMD Ryzen desktop processors. This second-generation Ryzen desktop processor lineup includes two 8-core, 16-thread models and two 6-core, 12-thread models, all offering major upgrades, AMD SenseMI technology, and other features. AMD’s 2nd generation Ryzen processor line up The entire … Read more
The evolution of future processors Is it a revolution — or just evolution? The GPU with its hyper-dense compute capacity and relatively low cost, is an amazingly powerful workload accelerator for certain classes of problems — those that lend themselves to massive parallel processing and multi-threaded workloads. When programmable vertex shaders were first introduced in 2002 (by 3Dlabs), and the … Read more
