High-Performance Graphics with VMware Horizon View 5.2

High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper Dell Desktop Virtualization Solutions (DVS) Engineering September 13 A Dell Technical White Paper

Revisions Date September 13 Description Initial release v.5.5 THIS WHITE PAPER IS FOR INFORMATIONAL PURPOSES ONLY, AND MAY CONTAIN TYPOGRAPHICAL ERRORS AND TECHNICAL INACCURACIES. THE CONTENT IS PROVIDED AS IS, WITHOUT EXPRESS OR IMPLIED WARRANTIES OF ANY KIND. 13 Dell Inc. All rights reserved. Reproduction of this material in any manner whatsoever without the express written permission of Dell Inc. is strictly forbidden. For more information, contact Dell. PRODUCT WARRANTIES APPLICABLE TO THE DELL PRODUCTS DESCRIBED IN THIS DOCUMENT MAY BE FOUND AT: http://www.dell.com/learn/us/en/19/terms-of-sale-commercial-and-public-sector Performance of network reference architectures discussed in this document may vary with differing deployment conditions, network loads, and the like. Third party products may be included in reference architectures for the convenience of the reader. Inclusion of such third party products does not necessarily constitute Dell s recommendation of those products. Please consult your Dell representative for additional information. Trademarks used in this text: Dell, the Dell logo, Dell Boomi, Dell Precision,OptiPlex, Latitude, PowerEdge, PowerVault, PowerConnect, OpenManage, EqualLogic, Compellent, KACE, FlexAddress, Force1 and Vostro are trademarks of Dell Inc. Other Dell trademarks may be used in this document. Cisco Nexus, Cisco MDS, Cisco NX- S, and other Cisco Catalyst are registered trademarks of Cisco System Inc. EMC VNX, and EMC Unisphere are registered trademarks of EMC Corporation. Intel, Pentium, Xeon, Core and Celeron are registered trademarks of Intel Corporation in the U.S. and other countries. AMD is a registered trademark and AMD Opteron, AMD Phenom and AMD Sempron are trademarks of Advanced Micro Devices, Inc. Microsoft, Windows, Windows Server, Internet Explorer, MS-DOS, Windows Vista and Active Directory are either trademarks or registered trademarks of Microsoft Corporation in the United States and/or other countries. Red Hat and Red Hat Enterprise Linux are registered trademarks of Red Hat, Inc. in the United States and/or other countries. Novell and SUSE are registered trademarks of Novell Inc. in the United States and other countries. Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Citrix, Xen, XenServer and XenMotion are either registered trademarks or trademarks of Citrix Systems, Inc. in the United States and/or other countries. VMware, Virtual SMP, vmotion, vcenter and vsphere are registered trademarks or trademarks of VMware, Inc. in the United States or other countries. IBM is a registered trademark of International Business Machines Corporation. Broadcom and NetXtreme are registered trademarks of Broadcom Corporation. Qlogic is a registered trademark of QLogic Corporation. Other trademarks and trade names may be used in this document to refer to either the entities claiming the marks and/or names or their products and are the property of their respective owners. Dell disclaims proprietary interest in the marks and names of others. 2 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Table of contents Revisions... 2 Executive summary... 4 1 K1 graphics performance analysis results... 5 1.1 K1 tests: Fixed frame rate video component Single VM... 5 1.2 K1 test: Heaven Benchmark Single VM... 7 1.3 K1 test: Movie clip + companion workloads... 8 1.4 K1 test: Heaven Benchmark + companion workloads... 8 1.5 K1 NVIDIA GPU results... 9 1.6 K1 subjective tests... 1 1.7 K1 testing conclusions... 1 2 K2 graphics performance analysis results... 11 2.1 K2 tests: Fixed frame rate video component Single VM... 11 2.2 K2 test: Heaven Benchmark Single VM... 12 2.3 K1 / K2 comparison test Viewperf single VM... 13 2.4 K2 test: Viewperf + companion workloads... 14 2.5 K2 NVIDIA GPU results with K1 comparison... 15 2.6 K2 subjective tests... 16 2.7 K2 test: Dell Wyse P25 zero client comparison Single VM... 16 2.8 K2 testing conclusions... 17 3 Conclusions... 18 A Configuration details... 19 B Additional resources... 3 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Executive summary VMware virtual dedicated graphics acceleration (vdga), also known as pass through graphics support, refers to the technology of mapping a VMware Horizon View desktop directly to a graphics processing unit (GPU) on a high-end display adapter such as an NVIDIA GRID K1 or K2 card. These adapters, when used with vdga mode, enable the user of a virtual desktop to run high-end, graphics-intensive applications such as CAD or other graphics editing and authoring applications. This solution offers a greater density of high-end graphics users per server when compared to the traditional one-to-one model of a dedicated graphics workstation. As such, cost benefits can be realized by using VMware vdga, VMware Horizon View, and Dell hardware technologies. The capability for 3D graphics and video in VMware Horizon View further expands the use cases and user groups to which IT can deliver virtual desktops. vdga allows the IT department to deliver virtual desktops to users who traditionally would have required high-performance workstations in order to get the graphics performance that they required for running applications like AutoCAD and edrawings. The critical difference between vdga and virtual shared graphics acceleration (vsga) is that the virtual machine (VM) has full usage of the assigned GPU. The dedicated GPU is passed through the hypervisor to the VM and the driver is installed locally on the VM, whereas in vsga mode a single GPU is shared between a pool of multiple virtual machines. Throughout the extensive testing and validation of this Reference Architecture, 2 graphics adapters were used at any given time (either dual K1 adapters or dual K2 adapters). Additionally, the typical graphicsintensive users were broadly classified as belonging to one of the following categories: Premium Plus users are users who regularly consume high-end graphics through high frame rate applications such as Google Earth, graphics-rich HTML5 pages, and may also review electrical or mechanical computer aided design (CAD) drawings. The term Premium Plus is used to distinguish this user category from the Premium user category referenced in other DVS Engineering documentation. Workstation users are users who would have traditionally required a dedicated, high-end system such as a Dell Precision Workstation to accomplish daily tasks. Users in this category typically perform activities such as 3D modeling and rotation using applications commonly found in the gas and oil industries which utilize large amounts of system resources. For the purposes of this Technical Whitepaper, the validation efforts described in this document pertain to Workstation users. It is also understood that all VMs used in these tests are configured to run in vdga mode, no other VMs are provisioned on the compute servers except those that are involved in testing and that each server is configured to support the same number of users as GPUs installed in vdga mode. It should also be noted that as of this writing vdga support with NVIDIA GRID adapters is a tech preview option expected to be released for general availability in the coming months. vdga is supported at the ESXi hypervisor level but not currently with the NVIDIA GRID K1 or K2 adapters and Horizon View 5.2. 4 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 39 58 77 96 115 134 153 172 191 21 229 248 267 286 35 324 343 362 381 419 438 457 Frames Per Second 1 K1 graphics performance analysis results 1.1 K1 tests: Fixed frame rate video component Single VM When considering the end-user experience in terms of perceived video or graphics smoothness, a useful medium for assessment is broadcast video. NTSC (US analog television system) broadcasts are transmitted at 3 frames per second (FPS), while PAL (used widely in Europe) broadcasts are transmitted in 25 FPS. For each of the following tests, a 3 FPS movie trailer of The Hobbit was used. The graph below shows that while the OptiPlex 71 displayed many more FPS than the Dell Wyse Z9D7 during playback of the 1 minute video clip, both were above the 25 FPS mark. NVIDIA K1 Single VM - Movie Clip Test - GPU enabled 8 7 6 5 3 1 OptiPlex 71 Dell Wyse Z9D7 Duration in Seconds Figure 1 K1 single VM movie clip test with GPU enabled 5 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 24 47 7 93 116 139 162 185 8 231 254 277 3 323 346 369 392 415 438 461 Frames Per Second 1 39 77 115 153 191 229 267 35 343 381 419 457 495 533 571 Frames Per Second In the following graphs, we see the difference in FPS from both endpoints during the same video clip playback while the VM does and does not utilize the GPU. NVIDIA K1 Single VM - Movie Clip - OptiPlex 71 8 7 6 5 OptiPlex 71 With GPU 3 OptiPlex 71 W/Out GPU 1 Duration in Seconds Figure 2 K1 single VM movie clip on the OptiPlex 71 NVIDIA K1 Single VM - Movie Clip - Dell Wyse Z9D7 7 6 5 3 Dell Wyse Z9D7 With GPU Dell Wyse Z9D7 W/Out GPU 1 Duration in Seconds Figure 3 K1 single VM movie clip on the Dell Wyse Z9D7 From these graphs it is clear (more so with the OptiPlex 71 than the Dell Wyse Z9D7) that the VM is able to achieve higher FPS when utilizing a GPU. 6 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 13 25 37 49 61 73 85 97 19 121 133 145 157 169 181 193 5 217 229 241 253 265 Frames Per Second 1.2 K1 test: Heaven Benchmark Single VM Heaven Benchmark is a GPU-intensive benchmark that significantly stresses graphics cards. This benchmark tool can be used effectively to determine the stability of a GPU under extremely stressful conditions, as well as validating the characteristics of the card s thermal subsystems. 6 5 NVIDIA K1 Single VM - Heaven Benchmark 3 1 OptiPlex 71 Dell Wyse Z9D7 Duration in Seconds Figure 4 K1 Heaven Benchmark - Single VM As shown above in the benchmark graph, the Dell Wyse endpoint is not able to maintain sufficiently high FPS during the Heaven Benchmark so it was not used in any further heaven tests for K1. The Dell Wyse P25 and P45 zero clients showed significant performance gains over the Z9D7, and while they were only used for K2 testing, they show to be a perfect match for K1 workloads as well. Notes: Heaven Benchmark produces a score after each benchmark test; however, the scores can be unreliable so they are not included in this document. Heaven benchmark was executed at the low quality setting with a resolution of 6 x 48 (lowest possible). If the tests had been conducted at a higher quality and resolution, both endpoints would have displayed consistently less than 25 FPS. 7 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 26 51 76 11 126 151 176 1 226 251 276 31 326 351 376 1 426 451 476 51 526 551 576 Frames Per Second 1.3 K1 test: Movie clip + companion workloads As a companion workload for K1 testing, the edrawings Viewer for SolidWorks was used, set to Advanced Animation, and configured to rotate indefinitely. The edrawings activity was not very high in terms of graphics resource utilization, thus it is appropriate for use as a companion workload for the K1 card which is targeted at lower end graphics scenarios. Companion workloads were executed on 7 of the 8 VMs running on the host to establish whether or not it would have any impact on the performance of the Heaven Benchmark VM. As expected, the additional workload placed on the host had very little impact as each VM had access to a dedicated GPU. Figure 5 shows almost identical performance to the single VM configuration representing in Figure 1 confirming the limited impact of the companion workloads on the tested VM. NVIDIA K1 Movie Clip Test - With Companion Workloads 8 7 6 5 3 1 OptiPlex 71 Dell Wyse Z9D7 Duration in Seconds Figure 5 K1 movie clip test with companion workloads 1.4 K1 test: Heaven Benchmark + companion workloads Following the same results pattern above, Figure 6 is comparable with the earlier graph of the Heaven Benchmark test (Figure 4) which further shows that the companion workloads had very little impact on the test results of the Heaven Benchmark VM. It should also be noted that there were no host resource issues throughout these tests. CPU usage spiked briefly when the Heaven Benchmark ran but was well within the acceptable range. Data store latency and network usage were also well within the capabilities of the host system. 8 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Usage % 1 14 27 53 66 79 92 15 118 131 144 157 17 183 196 9 222 235 248 261 274 Frames Per Second NVIDIA K1 Heaven Benchmark - With Companion Workloads 5 45 35 3 25 15 1 5 OptiPlex 71 Duration in Seconds Figure 6 K1 Heaven Benchmark test with companion workloads 1.5 K1 NVIDIA GPU results GPU usage during the Heaven Benchmark test is shown in Figure 7. Note that the GPU utilization is at 1% indicating that the K1 card is not capable of running highly intensive graphics programs on a regular or sustained basis. In contrast, Figure 8 shows that the K1 GPU is very capable of handling the test video clip with less than half of the available GPU resources. 1 1 NVIDIA K1 Heaven Benchmark - GPU Usage 8 6 GPU1 Memory Usage GPU1 Utilization 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 Duration in seconds Figure 7 K1 Heaven Benchmark GPU usage 9 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Usage % NVIDIA K1 Movie Clip - GPU Usage 6 5 3 1 GPU1 Memory Usage GPU1 Utilization 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 21 22 Duration in Seconds Figure 8 K1 movie clip GPU usage 1.6 K1 subjective tests While the companion workloads were running on the hosted VMs, the testing engineers were able to log in to the virtual desktops and perform several subjective tests. Performance was generally good on both, but the OptiPlex was superior to the Dell Wyse endpoint which occasionally encountered some jitter and lag. The OptiPlex was more capable of handling the workload and therefore did not demonstrate the same lag and jitter issues resulting in a better end user experience. 1.7 K1 testing conclusions The K1 card worked well in a vdga environment. Each VM had dedicated access to its assigned GPU and workloads on other VMs had little or no impact on each other. However, the K1 card was not well suited to running highly-intensive graphics program workloads such as those produced by Heaven Benchmark. When running these types of workloads, neither the OptiPlex 71 nor the Dell Wyse Z9D7 was able to maintain an FPS value of greater than 25. Endpoint selection is very important when pairing them with these GPUs as performance can vary greatly. The results from this testing indicate that the OptiPlex 71 performed much better than the Dell Wyse Z9D7 when paired with the K1 GPU. It should also be noted that the Dell Wyse P25 was not tested with the NVIDIA GRID K1 card but was tested with the K2 card where it demonstrated excellent performance on par with the OptiPlex 71 client. 1 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Average Frames Per Second 1 26 51 76 11 126 151 176 1 226 251 276 31 326 351 376 1 426 451 476 51 526 551 576 Frames Per Second 2 K2 graphics performance analysis results 2.1 K2 tests: Fixed frame rate video component Single VM Figure 9 shows that the OptiPlex client is able to maintain a higher level of FPS when using the K2 card when compared to the K1 (see Figure 1). In contrast, the Dell Wyse Z9D7 displays roughly the same performance regardless of GPU (K1 or K2). NVIDIA K2 Single VM - Movie Clip Test - GPU Enabled 1 9 8 7 6 5 3 1 OptiPlex 71 Dell Wyse Z9D7 Duration in Seconds Figure 9 K2 single VM movie clip test with GPU enabled NVIDIA K1 / K2 Single VM - Movie Clip Test - Endpoint Comparison 8 7 6 5 3 K2 - OptiPlex 71 K2 - Dell Wyse Z9D7 K1 - OptiPlex 71 K1 - Dell Wyse Z9D7 1 Figure 1 K1 / K2 movie clip test comparison 11 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 11 21 31 41 51 61 71 81 91 11 111 121 131 141 151 161 171 181 191 1 211 221 231 241 251 Frames Per Second The chart in Figure 1 compares the average FPS during the single VM movie clip tests run on both clients and both GPUs. The graph clearly shows that the OptiPlex 71 + K2 configuration offers the best performance in this application. 2.2 K2 test: Heaven Benchmark Single VM When paired with the K2 GPU, the OptiPlex 71 is capable of maintaining higher FPS than the K1 GPU. Figure 11 compares the performance of each configuration. Note: the Dell Wyse Z9D7 endpoint used in these tests was incapable of maintaining greater than FPS and therefore was excluded from the Heaven Benchmark tests. 7 6 NVIDIA K1 / K2 - Heaven Benchmark - OptiPlex 71 5 3 K2 K1 1 Duration in Seconds Figure 11 K1 / K2 Heaven Benchmark comparison OptiPlex 71 Single VM The K2 GPU also ran at much lower rates of utilization during the Heaven Benchmark when compared to the same tests using the K1 GPU. When comparing the data in Figures 7 and 12, it is quite clear that the K2 GPU has a greater overall capacity than the K1 GPU, Note: The K2 card was able to run the Heaven Benchmark in high-quality mode at a resolution of 1366 x 768 while maintaining over 25 FPS. 12 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Viewperf Result Usage % 8 7 6 5 3 1 NVIDIA K2 - Heaven Benchmark - GPU Usage 1 3 5 7 9 11131517192123252729313335373941434547495153 GPU1 Utilization GPU1 Memory Usage Duration in Seconds Figure 12 K2 Heaven Benchmark GPU usage 2.3 K1 / K2 comparison test Viewperf single VM The graphs in Figures 13 and 14 compare the results between the K1 and K2 cards running various Viewperf benchmarks. Both endpoints show better results when using the K2 card, supporting the claim that the K2 is a higher performing GPU as compared to the K1. NVIDIA K1 / K2 - Viewperf - Single VM - OptiPlex 71 5 45 35 3 25 Catia Ensight Lightwave Maya Proe 15 1 5 K2 K1 Sw Tcvis Snx Figure 13 K1 / K2 Comparison Viewperf single VM OptiPlex 71 13 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

Viewperf Score Viewperf Result NVIDIA K1 / K2 - Viewperf - Single VM - Dell Wyse Z9D7 6 5 Catia Ensight Lightwave 3 Maya Proe Sw 1 Tcvis K2 K1 Snx Figure 14 K1 / K2 Comparison Viewperf single VM Dell Wyse Z9D7 2.4 K2 test: Viewperf + companion workloads As a companion workload for K2 tests, AutoCAD was configured to execute a continuous orbit activity on a sample 3D drawing. The increased workload on the K2 GPU generated by AutoCAD was appropriate in this case as the K2 card is intended for use in higher end graphics environments. In this configuration, there were 3 VMs running the AutoCAD companion workload on the host. 6 NVIDIA K2 - Viewperf - With Companion Workloads - Dell Wyse Z9D7 5 3 1 Catia Ensight Lightwave Maya Proe Sw Tcvis Snx Figure 15 K2 Viewperf score (FPS) with companion workload Dell Wyse Z9D7 14 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 17 33 49 65 81 97 113 129 145 161 177 193 9 225 241 257 273 289 35 321 Usage % Viewperf Score NVIDIA K2 - Viewperf - With Companion Workloads - OptiPlex 71 6 5 3 1 Catia Ensight Lightwave Maya Proe Snx1 Sw2 Tcvis Figure 16 K2 Viewperf score (FPS) with companion workload OptiPlex 71 2.5 K2 NVIDIA GPU results with K1 comparison The graphs in Figures 17 and 18 show that the GPU utilization and memory usage is much higher for the K1 as compared to the K2 when running the Viewperf benchmarks thereby reinforcing the greater performance and capacity of the K2 GPU. 1 1 NVIDIA K2 - Viewperf - Single VM - GPU Usage - OptiPlex 71 8 6 GPU1 Memory Usage GPU1 Utilization Duration Figure 17 K2 Viewperf benchmark test GPU usage 15 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 9 17 25 33 41 49 57 65 73 81 89 97 15 113 121 129 137 145 153 Usage % NVIDIA K1 - Viewperf - Single VM - GPU Usage - OptiPlex 71 1 1 8 6 GPU1 Memory Usage GPU1 Utilization Duration Figure 18 K1 Viewperf benchmark test GPU usage 2.6 K2 subjective tests Similarly run in the K1 testing, subjective tests were conducted successfully on the OptiPlex 71 and Dell Wyse Z9D7 while companion workloads were running on other VMs on the same host. While Viewperf ran, repeated zoom in/out activities were executed in AutoCAD. These activities were designed to generate graphics loads visible to an end-user directly running subjective tests. Performance using the OptiPlex 71 was better than the Dell Wyse Z9D7 in that there was less jitter and greater responsiveness. Both endpoints, however, displayed some lag when executing multiple activities simultaneously. 2.7 K2 test: Dell Wyse P25 zero client comparison Single VM The Dell Wyse P25 PCoIP zero client for VMware View is a secure, easily managed zero client that provides outstanding graphics performance for advanced applications such as CAD, 3D solids modeling, video editing and advanced office productivity applications. Smaller than a typical notebook, this dedicated zero client is designed specifically for VMware View. It features the latest Teradici processor technology to process the PCoIP protocol on-chip rather than in software. The P25 also utilizes client-side content caching to deliver the highest level of performance to 2 HD displays in an extremely compact, energyefficient form factor. The Dell Wyse P25 delivers a rich user experience while resolving the challenges of provisioning, managing, maintaining and securing enterprise desktops. Note: At the time of writing this document, multiple monitor configurations are not supported with vdga and K1 / K2 cards. However, a new firmware version is planned which will support this feature. The graph in Figure 19 shows that the Dell Wyse P25 is capable of displaying more FPS than the Dell Wyse Z9D7 and is comparable with the OptiPlex 71 client. 16 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

1 12 23 34 45 56 67 78 89 1 111 122 133 144 155 166 177 188 199 21 221 232 243 254 265 Usage % NVIDIA K2 - Heaven Benchmark - Single VM - Endpoint Comparison 8 7 6 5 3 1 OptiPlex 71 Dell Wyse Z9D7 Dell Wyse P25 Duration Figure 19 K2 Heaven Benchmark test GPU usage 2.8 K2 testing conclusions The K2 cards offer improved performance over the K1 cards. An increase in FPS was seen during the playback of the movie clip and was particularly noticeable during Heaven Benchmark tests. The K2 cards are capable of running more graphics-intensive workloads and are better suited to high end workstation type users than the K1 cards. As with the K1, choice of endpoints is important as better performance was seen with the OptiPlex 71 than the Dell Wyse Z9D7. 17 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

3 Conclusions vdga works well and offers very good graphics performance in virtualized environments. Setup and installation are straight forward and easily accomplished by the average VDI administrator. While the NVIDIA GRID K1 GPU is best suited to entry-level high-end graphics usage, the K2 card is capable of much greater performance when working with intensive, high-end graphics applications such as Heaven Benchmark. K2 GPU usage was much lower than the K1 GPU usage when running the more demanding applications. The K1 card was only capable of running Heaven Benchmark at the lowest resolution, and the lowest quality setting available in the application; whereas the K2 was able to display many more FPS when running Heaven at a higher resolution and quality. These test results clearly demonstrate that the K2 GPU is much better suited to handling more graphically-intensive workloads. Endpoint selection is also very important when designing a vdga environment. Large differences measured in FPS between the Dell Wyse Z9D7 and the Dell Wyse P25 demonstrate that one will be much more suitable that the other when using certain high-end applications. The P25 offers frame rates comparable to the OptiPlex 71 and is much more efficient at rendering the displayed graphics than the Z9D7. As expected, performance of the VMs running in vdga mode were unaffected by added workloads that were placed on co-hosted VMs. The data collected in each of these tests confirm that when GPUs are dedicated to VMs using the vdga technology present in VMware Horizon View 5.2, performance was maintained at a high level (as measured by frame rates and host metrics). 18 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

A Configuration details Table 1 Configuration details Solution Configuration Hardware Components Description Virtual desktops VMware compute host VMware management host Windows 7 64-bit 4 x CPU 32 GB RAM 1 x Dell PowerEdge R7 Server VMware ESXi 5.1 U1 Intel Xeon E5-267 @ 2.6GHz 3 GB RAM @ 16MHz 1 x 146GB 15K SAS HDDs Broadcom BCM57 1GbE NIC PERC H71P RAID controller 2 x NVIDIA GRID K1 2 x NVIDIA GRID K2 Apex 28 PCoIP card 1 x Dell PowerEdge R7 Server ESXi 5.1 U1 Intel Xeon E5-269 @ 2.7 GHZ 128 GB RAM @ 16 MHZ 1 x 146GB 15K SAS HDDs Broadcom BCM57 1GbE NIC PERC H71P RAID Controller Configuration of VMs for vdga testing For VMware ESXi environment 1 x 146GB drives were configured in a RAID-1 array For VMware ESXi environment 1 x 146 GB drives were configured in a RAID-1 array. Each VM hosted the following workloads on Windows Server 8 R2: VMware vcenter VMware Horizon View Connection Server Microsoft SQL Server (View Connection Server and vcenter databases) File server 19 High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5

B Additional resources Referenced or recommended sites and publications: DVS Enterprise for VMware Horizon View 5.2 Reference Architecture Dell Wyse Dell OptiPlex VMware Horizon Suite NVIDIA GRID cards High-Performance Graphics with VMware Horizon View 5.2 Technical White Paper v.5.5