Aug 17

My NVIDIA GRID 5.0 Testing

In my previous post I covered some of what is new with NVIDIA GRID 5.0 and the NVIDIA Pascal cards. In this post I’m going to cover some of the testing I’ve done with GRID 5.0 and the NVIDIA Tesla P4 GPU.

NVIDIA was nice enough to provide the NVIDIA GRID Community Advisors (NGCA) access to P4 GPUs and beta candidates of GRID 5.0 for testing in our home labs. This is something I couldn’t pass up. Which means I had to stand up a home lab.

Tesla P4 GPUIt’s interesting, this cutting edge frontier is one of the few spaces that still requires you to have physical equipment to do testing. I can’t go out and run this in a cloud somewhere and test it. So for the last month, my wonderful wife has tolerated the sound of servers running in our basement while I put GRID 5.0 through its paces.

That said let’s talk about my lab setup. Data for the post actually comes from two different labs, one was a work lab that has since been purposed. Currently I just have a home lab. (Skip ahead 5 paragraphs if you aren’t interested in my home lab hardware.) [add anchor link to skip ahead]

What am I running to do all this GRID 5.0 testing… maybe something like some Dell R730s loaded with NVIDIA P40s or P100s? Not really, my home lab, like most folks home labs, are ebay specials. That said it’s worth noting this entire setup CAN NOT be found on any HCL list for VMware or NVIDIA.

My Home lab consists of an R610 running a management environment (VSA, Jump Box, AD, connection server, Security Server, NVIDIA License Servers, etc.). It has dual E5620 Intel quad core processors. It also has 24GB of RAM and 5-146GB SAS drives in a RAID 5 configuration. All of the VMs are running on local storage. It works as a system just to support management.

This is connected to the world using a 1GB ZyXEL 16 port switch and an old D-Link router I had sitting around. Standard networking setup for 1GB ports, nothing really special to tell you about.

The other system in my arsenal is a Cisco C240-M3. It’s running the Intel E5-2640 procs at 2.50GHz with 6 cores each. The system has 64GB of memory in it and I have it loaded with 2-74GB SAS drives in a RAID1 for my boot volume and 3-146GB SAS drives in a RAID 5 for my storage volume. The nice thing about this sever is that it supports the NVIDIA K1 and K2 cards should I need them for testing. I picked this up on ebay for about $500  + drives (http://www.ebay.com/itm/Cisco-UCS-C240-M3S-v02-Server-2x-E5-2640-2-50-GHz-6-Core-64GB-Dual-PS-No-HDD/262840746028).

If you hadn’t guessed by now my testing was done using VMware Horizon 7.1 (build-5170113). My vCenter is version 6.5 (build 5705665) and the ESXi hosts are at version 6.5 (C240 is build 5310538 and the R610 is build 4887370).

This is where the standard stuff ends and the fun begins…

The P4 GPUs that NVIDIA was nice enough to provide the NGCA members with went in my C240-M3 host. I’m going to save the install and setup of the GPU for another post. The C240-M3 host only runs my test VMs so that I can avoid artifacts caused by other VMs that aren’t part of what I’m testing. With the P4 installed and configured in the C240-M3 I built some CentOS 7 VMs. NVIDIA X Server Settings Screen

To test my VMs I used GFXBench and Unigine-Heaven 4.0 for both I used the Linux versions of the software. I chose GFXBench because of its testing methodology, how it has test for several different GPU factors, and because they aren’t all in a single test. I also chose to test with Unigine because it’s what everyone else tests with and I want to make this information as relevant to as many of you as possible.

For my basic tests I’m kept the RAM in each VM at 8GB and vCPUs at 4 with a native screen resolution of 1920×1080 (16:9)@29hz.. (See NVIDIA X Server Settings screen shot.)

The table below shows the GFXBench results for each of its tests for each vGPU profile in VMware Horizon.

GFXBench test results*
Test P4-8Q P4-4Q P4-2Q P4-1Q
Car Chase 6983.29 Frames
(118.161 FPS)		 6940.41 Frames
(117.435 FPS)		 6953.41 Frames
(117.655 FPS)		 6997.82 Frames
(118.406 FPS)
1080p Car Chase Off screen 9916.48 Frames
(167.791 FPS)		 9924.63 Frames
(167.93 FPS)		 9886.87 Frames
(167.292 FPS)		 9935.8 Frames
(168.118 FPS)
1440p Manhattan 3.1.1 Off screen 8828.87 Frames
(142.401 FPS)		 8796.17 Frames
(141.874 FPS)		 8801.01 Frames
(141.952 FPS)		 8802.01 Frames
(141.968 FPS)
Manhattan 3.1 11092.4 Frames
(178.91 FPS)		 10915.6 Frames
(176.058 FPS)		 11035.8 Frames
(177.997 FPS)		 11068.1 Frames
(178.518 FPS)
1080p Manhattan 3.1 Off screen 13633. Frames
(219.889 FPS)		 13623.2 Frames
(219.73 FPS)		 13476.3 Frames
(217.359 FPS)		 13559.1 Frames
(218.696 FPS)
Manhattan 12076.2 Frames
(194.778 FPS)		 11640.7 Frames
(187.753 FPS)		 11585.2 Frames
(186.858 FPS)		 11774.7 Frames
(189.914 FPS)
1080p Manhattan Off screen 14835.3 Frames
(239.279 FPS)		 14671.1 Frames
(236.63 FPS)		 14473.8 Frames
(233.449 FPS)		 14539.1 Frames
(234.502 FPS)
T-Rex 25533.7 Frames
(455.959 FPS)		 24787.8 Frames
(442.639 FPS)		 26636.3 Frames
(476.136 FPS)		 24805.8 Frames
(442.96 FPS)
1080p T-Rex Off screen 42027 Frames
(750.482FPS)		 41957.8 Frames
(749.246 FPS)		 40691.8 Frames
(726.64 FPS)		 41538.5 Frames
(741.759 FPS)
Tessellation 21398.3 Frames
(713.276 FPS)		 21415.9 Frames
(713.862 FPS)		 21883.7 Frames
(729.457 FPS)		 22267.8 Frames
(742.259 FPS)
1080p Tessellation Off screen 88862.3 Frames
(1481.04 FPS)		 88410.1 Frames
(1473.5 FPS)		 88346.9 Frames
(1472.45 FPS)		 88181 Frames
(1469.68 FPS)
ALU 2 18045 Frames
(601.5 FPS)		 17631.8 Frames
(587.726 FPS)		 17716.9 Frames
(590.564 FPS)		 17850.1 Frames
(595.002 FPS)
1080p ALU 2 Off screen 62342.9 Frames
(1039.05 FPS)		 62537.5 Frames
(1042.29 FPS)		 62582.1 Frames
(1043.04 FPS)		 62643.6 Frames
(1044.06 FPS)
Driver Overhead 2 2450.51 Frames
(81.6837 FPS)		  2679.11Frames
(89.3036 FPS)		 2492 Frames
(83.0667 FPS)		 2600.05 Frames
(86.6682 FPS)
1080p Driver Overhead 2 Off screen 5102.23 Frames
(85.0372 FPS)		 5589.44 Frames
(93.1574 FPS)		 5194.05 Frames
(86.5675 FPS)		 5368.02 Frames
(89.4669 FPS)
Texturing 96098 MTexel/s
(63.5544 FPS)		 100066 MTexel/s
(64.4044 FPS)		 100061 MTexel/s
(64.9113 FPS)		 100385 MTexel/s
(65.2536 FPS)
1080p Texturing Off screen 99278 MTexel/s
(95.8384 FPS)		 99665 MTexel/s
(95.9874 FPS)		 99797 MTexel/s
(98.0892 FPS)		 99204 MTexel/s
(95.8811 FPS)
Render Quality 4541.54 mB PSNR
(866.644 FPS)		 4541.54 mB PSNR
(873.211 FPS)		 4541.54 mB PSNR
(1023.98 FPS)		 4541.54 mB PSNR
(977.356 FPS)
Render Quality (High Precision) 4541.54 mB PSNR
(887.633 FPS)		 4541.54 mB PSNR
(900.2 FPS)		 4541.54 mB PSNR
(1059.25 FPS)		 4541.54 mB PSNR
(1008.66 FPS)

 

Unigine Heaven was run with the settings defined in the table below.

Unigine Heaven 4.0 (Basic Edition) Settings
Preset Custom
API OpenGL (grayed out option)
Quality High
Tessellation Normal
Stereo 3D Disabled
Mulit-monitor Disabled
Anti-aliasing X2
Full Screen True
Resolution System

 

Testing results from Unigine Heaven

Unigine Heaven test results*
P4-8Q P4-4Q P4-2Q P4-1Q
FPS 28.3 28.2 28.1 28.2
Score 713 711 709 711
Min FPS 7.2 12.2 12.5 11.3
Max FPS 41.9 44.2 45.5 42.6
Mode 1920×1080 2xAA fullscreen 1920×1080 2xAA fullscreen 1920×1080 2xAA fullscreen 1920×1080 2xAA fullscreen

The above are the results of my testing. You probably noticed the asterisks (*) on the results. This is my caveat on these results. These aren’t the results you want to rely on for a production environment. I ran these tests once per profile, on non-HCL hardware, in a non-optimized configuration. These results may also be impacted by the fact that no other VMs were running on this host and thus consuming resources during the test. Your individual results may vary significantly. Please consider my test results as one point of data and not a complete answer to how a similar configuration will function in your environment. In short your mileage may vary.   

At the beginning of this blog I mentioned I had two labs I was using. Up to know you have heard about my P4 testing. As some of you may know I presented a session at the GPU Tech Conference with my good friend Trey Johnson on Getting started with GPUs for Linux Virtual Desktops on VMware Horizon. I ran those tests on a work lab environment. The material for that session was run on Cisco C240-M4s with the NVIDIA M60 GPUs and GRID 4.

I made one mistake that I am regretting, before the lab was repurposed I forgot to capture the full set of results from my testing. All I have are the maximum and minimums that were discussed during the session. At the same time the results do provide a good set of comparison points. In the table below are the tests results from the GTC session showing the highest and lowest tests results.

M60 with GRID 4 GFXBench Test Results (incomplete)*
Test M60-8Q M60-4Q
Texturing 44.6732 FPS 44.8432 FPS
Driver Overhead 2 61.3333 FPS 61.5149 FPS
1080p Texturing Off screen 90.7743 FPS 98.2536 FPS
1080p Tessellation Off screen 1212.87 FPS 1212.62 FPS

The same bit as above with the asterisks (*), these are single pass results from a non-optimized environment, your results may vary significantly.

These results correspond similarly to the results from the P4 GPU tests. You can see that the highest and lowest results for both the P4 GPU and the M60 are the same tests. I’ve put the relevant results in a side by side table for comparison below.

NVIDIA P4 GPU compared to M60 GPU testing with GFXBench (incomplete)*
Test P4-8Q (GRID 5) M60-8Q (GRID 4) P4-4Q (GRID 5) M60-4Q (GRID 4)
Texturing 63.5544 FPS 44.6732 FPS 64.4044 FPS 44.8432 FPS
Driver Overhead 2 81.6837 FPS 61.3333 FPS 89.3036 FPS 61.5149 FPS
1080p Texturing Off screen 95.8384 FPS 90.7743 FPS 95.9874 FPS 98.2536 FPS
1080p Tessellation Off screen 1481.04 FPS 1212.87 FPS 1473.5 FPS 1212.62 FPS

You might be getting tired of this by now… same bit as above with the asterisks (*), these are single pass results from non-optimized environments, your results may vary significantly.

As you can see from above the P4 exceeds the M60 in GFXBench tests in all but one instance (underlined in the table above). Thus showing comparable performance in single pass, non-optimized tests, between the NVIDIA P4 and NVIDIA M60 GPUs.

To put this in perspective NVIDIA basically provided half an M60 (power, slot space, etc.) in the P4 and met or exceded vGPU performance. Now if we think about what that means for servers… you can put GPUs in severs for EUC deployments that you couldn’t before (consult vendor documentation for limits and compatibility). This  makes it nice, when for instance you need to upgrade some lower end applications (for example Microsoft Office, Windows 10 (ok that’s an operating system you caught me), and the like) that take advantage of GPUs and the host for those desktops is a year or two old. Instead of a rip and replace add P4’s or P40’s to the servers (depending on support) and away you go.

I can’t remember who said it at the GPU Tech Conference this year but it really stood out to me, it went something like this. “We’ve entered a new age in the computer industry, an age where servers won’t be sold without a GPU for processing.” In my opinion, this latest release of GRID 5.0 and Pascal GPUs make vGPU based EUC accessible, for all but a couple of corner cases. Going forward adding GPUs should be a requirement for EUC deployments.

I hope this blog post was helpful. If you would like to find out more be sure to read these other great posts about NVIDIA GRID 5.0 from other NGCA members:

Permanent link to this article: https://www.wondernerd.net/my-nvidia-grid-5-0-testing/

Aug 17

Changing the GPU Virtualization Game

Today NVIDIA announced the NVIDIA GRID August 2017 release (AKA GRID 5.0), it is a major change for the virtualization industry. Over the next few days I will be publishing a few blogs about the NVIDIA GRID vPC (Virtual PC) and Quadro vDWS (Virtual Data Center Workstation). This blog covers some of what’s new with GRID 5.0 and soon I will publish some details on testing I did using an NVIDIA P4 GPU with GRID 5.0. I will also have additional blogs that go live during VMworld, around GRID vPC and Quadro vDWS. These will be part of the material I cover in my vBrownBag Tech Talk on vGPU Enabled Linux VDI (VMTN6636U)

GRID 5.0 Announcement ImageNow let’s get into the details of whats new…

The first thing that’s worth noting is that NVIDIA GRID vPC is really a software solution for datacenters. You have NVIDIA Tesla (datacenter) GPUs and then you have NVIDIA GRID software (August 2017 release as of this writing), those two combined provide a powerful datacenter solution for Virtual Desktops (VDI). This means you have the ability to get more out of your NVIDIA GPUs and can continue to use them as new GRID software is released.Pascal Series GPUs

That gets us to the first big advance with GRID 5.0. It is now possible to use both Maxwell and Pascal GPUs in your virtual environments. This means continued support for legacy Maxwell (M60, M6, and M10) GPUs as well as support for Pascal (P40, P6, P4, and P100) GPUs. Thus allowing companies to maintain their previous investments longer.

Now what could I mean by maintaining investments longer? I buy a GPU and what its capable of is based on the hardware, right? Not exactly… When Citrix announced the ability to monitor GPUs from within Citrix Director (https://www.citrix.com/blogs/2017/06/01/monitoring-of-nvidia-gpus-in-citrix-director/) that was a capability delivered by GRID software. Thus as NVIDIA adds more monitoring functionality with newer versions of GRID, virtualization vendors, 3rd party app developers, and even you can take advantage of them to monitor GPUs. If a new monitoring feature is added and your support contract is current you can upgrade your GRID software to use the new feature. That’s right the second big advancement, in my opinion, with GRID 5.0 is decoupling hardware and software functionality. NVIDIA is providing the ability to have continual new features even on older GPUs.

NVIDIA GRID 5.0 Monitoring Features

You might be thinking this is great! I’ll just go in and download GRID 5 before my license expires and I’m all set! Not so fast, NVIDIA did a fair amount of work on GRID licensing. With GRID 5.0 you have to have valid licenses. If you don’t performance is degraded. I’ve even tested it on Linux based desktops and even Linux desktops have their GPU performance capped if not licensed properly.

One enhancement delivered with the power of both GRID 5.0 and Pascal cards is the ability to run CUDA based workloads across the vGPU profiles. In other words you can build VMs with various vGPU configurations that run CUDA based code such as some Adobe applications, ArcGIS, AutoCAD, etc. You might be thinking this was possible before, and you’re right sort of. You had to use a full GPU profile like the M60-8A. With the Pascal architecture NVIDIA added preemption capabilities to the GPU. This is a fancy way of saying some applications with CUDA capabilities can share the same GPU. Possibly even bringing HPC, ML, DL, and AI into the virtual realm, which I’ll dig into in another blog.

The next cool thing that gets unlocked with GRID 5.0 is actually a feature of the Pascal cards and not a GRID feature. For those who have setup previous generations of GPUs (Maxwell) for EUC/VDI environments, you are probably familiar with having to use the gpumodeswitch command to change a Maxwell GRID GPU from compute to graphics mode. For those not familiar with it, you had to install gpumodeswitch, make the change, reboot, uninstall gpumodeswitch, reboot, install the VIB, reboot, and finally everything is in place. Guess what, you don’t have to do that any longer. The Pascal GPUs are able to switch between compute and graphics without gpumodeswitch. This reduces the reboots and steps required to have a GPU up and running. I’ll cover this during my vBrownbag TechTalk at VMworld and its accompanying blog.

It should be noted that the removal of gpumodeswitch is a feature of the Pascal cards and not something that’s part of GRID 5.0. This means if you are using the M10, M6, or M60 GPU you will still need to use gpumodeswtich before you can use GRID 5.0.

YNVIDIA P6 GPUou may have noticed above I mentioned an NVIDIA GPU that you may not have heard of. The NVIDIA Pascal P6 GPU. NVIDIA announced the P6 GPU along with GRID 5.0, this is a GPU specifically for blade server deployments. With deployments that leverage blade technologies it’s possible to provide the latest GPU technology and maintain a consistent operational model in the datacenter.  Hence, if you use rack mount server there are GPUs for it, if you prefer blade servers, there are GPUs for it.

 

At this point I bet you’re wondering what the various vGPU profiles for the Pascal GPUs are. In the tables below I’ve assembled a list of the vGPU profiles for the P40, P6, and P4 GPUs.

 

NVIDIA GRID P40 vGPU Profiles (Link to P40 Data Sheet)
Name Max Instances FB Memory Display Heads Max X Res Max Y Res Frame Rate Limit License
P40-1Q 24 1024MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-2Q 12 2048MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-3Q 8 3072MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-4Q 6 4096MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-6Q 4 6144MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-8Q 3 8192MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-12Q 2 12288MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-24Q 1 24576MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P40-1A 8 1024MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-2A 4 2048MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-3A 2 3072MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-4A 1 4096MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-6A 4 6144MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-8A 3 8192MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-12A 2 12288MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-24A 1 24576MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P40-1B 24 1024MiB 4 2560 1600 N/A GRID-Virtual-PC,2.0;
GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0

NVIDIA GRID P6 vGPU Profiles (P6 Data Sheet)
Name Max Instances FB Memory Display Heads Max X Res Max Y Res Frame Rate Limit License
P6-1Q 16 1024MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P6-2Q 8 2048MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P6-4Q 4 4096MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P6-8Q 2 8192MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P6-16Q 1 16384MIB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P6-1A 16 1024MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P6-2A 8 2048MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P6-4A 4 4096MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P6-8A 2 8192MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P6-16A 1 16384MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P6-1B 16 1024MiB 4 2560 1600 N/A GRID-Virtual-PC,2.0;
GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0

NVIDIA GRID P4 vGPU Profiles (Link to P4 Data Sheet)
Name Max Instances FB Memory Display Heads Max X Res Max Y Res Frame Rate Limit License
P4-1Q 8 1024MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P4-2Q 4 2048MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P4-4Q 2 4096MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P4-8Q 1 8192MiB 4 4096 2160 N/A GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0
P4-1A 8 1024MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P4-2A 4 2048MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P4-4A 2 4096MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P4-8A 1 8192MiB 1 1280 1024 N/A GRID-Virtual-Apps,3.0
P4-1B 8 1024MiB 4 2560 1600 N/A GRID-Virtual-PC,2.0;
GRID-Virtual-WS,2.0;
GRID-Virtual-WS-Ext, 2.0

As you can see NVIDIA GRID 5.0 and the Pascal GPUs provide a lot of power for your virtualized GPU needs. If you are interested in discovering more about GRID 5.0 be sure to visit the NVIDIA GRID site on NVIDIA.com. In my next blog I’ll cover some of the testing I’ve had a chance to do with the NVIDIA P4 GPU and GRID 5.0.

Permanent link to this article: https://www.wondernerd.net/changing-the-gpu-virtualization-game/

May 12

GTC17 Wrap Up Report

Time for my yearly wrap up from GTC 2017. In my post I want to share the good and not so good highlights from my trip to GTC17. These are provided to give constructive feedback about both what is done right and some of the rough areas attendees (me) experienced.

GTC is fast becoming my favorite conference to attend (it used to be VMworld). The reason for this is that the conference is run by nerd for nerds. That means for the most part it’s nerd friendly. A good thing in this nerds book.

An example of this nerdvana is they don’t have loud bands and music or some other form of noise blasting from every corner of the building. It is possible to have real and valuable conversations without having to yell. I’m pleased to say I still have my voice at the end of the week.

It’s also nice that they keep the DJ in one place during the party on Wednsday evening. Which also brings me to my first set of things I didn’t particularly care for. I really enjoy the Tech Musem, but I think we are outgrowing it quickly. It was a bit cramped this year and this made it difficult to move around.

The basement is always a great place to hang out and meet new friends. But when you came up from the basement you were meet by a big mess. There was a security guard at the escalators going up to the second floor only letting a few go at a time. This created massive congestion in the only place to transition from one space to another.

Upstairs it was also problematic with part of the exhibit space being closed down, space was restricted. I’m sure there was a reason for all of this, it was just a little frustrating as an attendee.

The great thing about the party was the food. The desert forest did not disipoint they were all wonderful. The bartenders were also great and did a good job with the drinks.

Another area where GTC excels is their on site meals. The meals are flavorful and it doesn’t take forever to get them. It does have a bit of a cattle pen feel, especially the first day when it was nothing but a long line of wait staff directing people one way or another. That was remedied by the second day. Kudos for that.

I will say I found it odd that for two days we had sandwiches (Monday and Wednesday). I think there would be other alternatives to repeating almost the same menu. The south hall was cool to be in. It was very spacious. The only problem was getting to it. You had to go down a fairly steep flight of stairs to get to it. I was worried someone might fall going down them.

One of my friends with bad knees also had a problem with getting to and from the meals. There is not an easy way down to the south hall for those with disabilities. If you don’t feel good about going down or up steep steps you get a long walk to get your food and return to the conference.

I loved that beverages were out constantly. This made it easy to get something to drink. It would be nice if the cola choices were a bit broader than Sprit, Coke, and Diet Coke. The sweets wall was great again this year and many enjoyed it.

Inside the show floor, it was great having all the startups and innovators. I always learn so much talking with them and this year I only saw a single booth without anyone in it! It was great!

I do wish they would reconfigure the food/drink stations a bit in the exhibit hall. There placement made moving down the isles a little difficult. It also increased the possibility of wearing the food. I still love the drink ticket system. After having been at a few conferences where someone had a bit too much and needed a security escort out at the end of the night this seems to take care of it.

One of the things that I was miffed about was the conference t-shirts. Up until Wednesday evening, right before the exhibit hall closed, I didn’t think we got a t-shirt this year. In fact I went to the store and bought a shirt. Which it appears many others did as well (they were almost sold out of the exact same ones we got on Wednesday). It would have been nice if there were big signs or even a note in the back pack saying that shirts would be avalible following the keynote on Wednesday. It also would have saved me $10. I do think the shirts are pretty darn cool though, so it’s not horable having two of them. It reminds a lot of a Dr. Who / Futurama cross over shirt I have.

Conference materials. Bag, Drink Tickets, I am AI t-shirt (separate from registration), 6 generation t-shirt (the one that came with registration), papers, and name badge.

Another thing that left me with a similar feeling to the shirts was wrist bands for the Wednesday party. All that was said in the guide was that they would be in the exhibit hall on Wednesday evening. They didn’t say where we could find them.

I would have liked to see them use the apps notification system to push out a notice to all attendees about where to pickup wrist bands. Or it would be even cooler if they could put some of the NVIDIA technology to work. Last year they had a system that did facial recognition and enabled you to get a drink. Wouldn’t it be cool if, when you registered it snagged a picture of you and did recognition of you to let you into the party? I know there are problems with this, so it would require some thinking.

The sessions this year were outstanding. I have a few things I would change on them but they are minor. I would love to see a couple rows of tables put in either the front or the back of the room. Many people at GTC are taking notes or in my case using Twitter to take notes and share tables would make that so much easier than balancing a phone and tablet on your legs. It would also be great to mix up the sessions and bring more interaction. So in my case all the virtualization nerds hung out in room 231 the whole week and rarely ventured to other areas. At the same time it’s nice that all the sessions in your field are in the same general area. So it’s a tough call.

A few people pointed this out to me. It’s becoming imperative to add introductory courses and executive courses to GTC. AI/Deep Learning/Machine Learning are the next wave of business systems. Right now the conference addresses the “detailed how” to do these things. But as this becomes main stream professionals will want to get their feet wet and bring it into their organization. Right now there are very few sessions that speak to the simple how or even more important speak to the business decision makers. Why should executives invest in these ideas and how do they relate to their bottom line?

I think this is an awesome oppertunity for those looking to submit sessions for GTC in the coming years. Simply put make it consumable by executives and those just getting started with GPUs, regardless of application.

I would also like to see something like the vBrownBag Tech Talks at the show. Give people 15 minutes to share what they are working on with out the pressure of a full session.

Registration was simple and easy this year as in years past. The bags they provide are again of great quality and should last a long time. I saw several bags this week from previous GTCs (2014, 15, & 16) that were all still holding up well and looked like they were being used. I don’t know where the GTC events team gets them but other conferences should take note.

As a speaker I got to experience the additional fun of being a speaker. This is a great oppertunity for anyone who want to share their experiences. The speaker resources are phenomenal and they work with you on so many things. The AV teams are also fantastic making sure every thing is ready for you to speak even with only a few minutes between sessions.

GTC17 Speakers Gift

The speakers gift this year was very cool. It was a pen slide controller with laser pointer. I am looking forward to using this in presentations to come.

The app had a few more problems this year than last. It seems like on my S7 Edge every time there was an update pushed out I would have to uninstall and reinstall the app.

I wasn’t able to attend a dinner with strangers this year but I’m glad that it was back again this year. I know these dinners generate some of the best conversations of the week.

The posters were exceptional this year. The students who worked on them did a great job in developing their content and have given me so much to think about. Here is an example of one of the posters presented this year.

Social media at GTC is pretty darn good. I’d love to work with the social media team to help improve their overall impact. There are some very simple things that could be done to help increase the conversations on social channels.

Overall GTC17 was an outstanding event and all of the teams who worked so hard to put it on should be commended for their hard work. I think in the next couple of years this will be the next big conferance to attend in the IT industry. NVIDIA is well on its way to making them the best conferance at a insane scale.

All of the above, of course, is my personal opinion and perception. I provided it to help us all improve. If any of the GTC event teams would like clarification or more details on these please let me know. If you are a general reader and would like to add your two cents to the conversation please add it in the comments below.

Thank you to everyone who helped make GTC17 so very memorable.

Permanent link to this article: https://www.wondernerd.net/gtc17-wrap-up-report/

May 08

GTC17 Session S7349 Slides & Links

This post is about GTC Session S7349, “Getting Started with GPUs for Linux Virtual Desktops on VMware Horizon.” Attached to the post is the slide deck presented in the session and below a list of links referenced in the session.

Trey and I appreciate the attendance and hope the information was helpful in getting you on your way with deploying Linux Desktops with GPUs. We also hope these slides help those who weren’t able to attend the session as well.

Please be sure to post in the comments if you have questions about getting your Linux VDI environment working. Building the lab environment that this session is based on took me 8 – twelve hour days to configure and trouble shoot. Once I knew where the problems were I was able to repeat the process quicker each time I did it.

GTC S7349 Slide Cover

 

Links from the session:

Permanent link to this article: https://www.wondernerd.net/gtc17-session-s7349-slides-links/

May 08

Long List of Links for Configuring Linux Desktop VMs

This is a compilation of links that helped me prepare for GTC Session S7349. They aren’t in any particular order and may not be helpful as you are configuring your Linux Virtual Desktop VMs with GPU capabilities.

 

 

Permanent link to this article: https://www.wondernerd.net/long-list-of-links-for-configuring-linux-desktop-vms/

Apr 25

Error Using gpumodeswitch

This is a quick little post about something I ran into of late setting up an NVIDIA GRID M60 GPU on an ESXi 6.0 update 2 host. Though this should apply (haven’t tested) to all version of VMware ESXi as well as the M10 and M6 model GPUs (again haven’t tested).

The M60 card was installed and I had loaded the gpumodeswitch VIB on the ESXi host so I could validate that it was in graphics mode. I kept getting the following error anytime I tried to run the gpumodeswitch command:

ERROR: Read card info failed by using character device based.

The problem, as it turns out, is that passthrough was set for the card. If you get this error while you are setting up your GRID card (M60, M10, M6) , be sure to go to the hardware tab, scroll down to the NVIDIA card. If passthrough is enabled, select the card and toggle passthrough for the card. Once you’ve disabled passthrough you will need to reboot the host. Then you can run gpumodeswitch.

Remove Passthrough for GPU

Permanent link to this article: https://www.wondernerd.net/error-using-gpumodeswitch/

Apr 05

Join Me at GTC 2017

Picture of a green eye, Join me at GTC 2017

I will be attending the GPU Tech Conference (GTC) in May. This is the don’t miss GPU event of the year.

Last year was my first year attending GTC and I hope not to miss it going forward. GTC has lots of great content around VDI, Deep Learning, Analytics, Artificial Intelligence, and many other areas.  I enjoy being able to explore sessions outside of my primary area of interest to broaden my horizons.

If you would like to join me at GTC and still haven’t registered, there is still time. Here is my discount code for conference registration. This should help you save 20% on 2017 GTC conference registration.

Use code FFSGTC17

If you are already registered for GTC please join me and my friend Trey for our session S7349: Getting Started with GPUs for Linux Virtual Desktops on VMware Horizon. This session is currently scheduled for Monday, May 8th, at 1:00 PM in room 231. You can find us in the session catalog here: https://gputechconf2017.smarteventscloud.com/connect/sessionDetail.ww?SESSION_ID=110072

In this session we will walk through getting started with GPUs for Linux Virtual Desktops (VDI) in VMware Horizon. We want you empower you to take this information back, and begin your Linux with GPU VDI program. This is an introductory session, which means come with what ever knowledge you have and we will build on it.

The idea for our session came from some work Trey and I were doing where an organization wanted Linux based desktops with GPU capabilities. While we were architecting this noticed lots of little gotchas along the way. Enough to make it frustrating for someone with minimal exposure to the various technologies at play.

Not only will we review how to deploy virtual GPUs for Linux desktops, we’ll also talk about some testing and performance information we got from an environment we built. It’s great to talk about how to put something together but it’s even better when you can demonstrate the results.

It would be great if you would join us at GTC and even better if you would join us for our session.

We look forward to seeing you there…

 

Permanent link to this article: https://www.wondernerd.net/join-me-at-gtc-2017/

Mar 30

Doubly Honored

Yesterday, March 29, 2017, two community announcements were made. One was for the 2017 VMware EUC Champions program and one was for the 2017 Dell EMC Elect. I was selected for both recognition’s. I feel doubly honored to be part of both amazing groups.

Dell EMC Elect 2017For those who don’t know about the Dell EMC Elect it is the social advocacy group for Dell EMC. Social Advocacy means, we are active on social media platforms talking about the product, solutions, and things happening at an organization. It doesn’t mean we are experts or know everything about every product, but we happily share what we know.

Being part of the Elect is a special honor for me. I was part of the 2012 Dell TechCenter Rockstars which was the Dell social advocacy program and I have worked closely with the EMC Elect on several things. It’s also special to me with this being the first year that both the Rockstars and Elect are combined. This makes it even more special to me and a great honor to be part of this select group of 153 individuals.

If you’re interested in reading more about the Elect you can visit the announcement here.

I’m also ecstatic to be a returning member of the VMware EUC Champions. I’ve been a member of this prestigious group since its inception. This group of champions is focused on VMware End User Computing.

For those not familiar with EUC the general idea is, that it’s no longer necessary to have a desktop or laptop to work. It’s possible to turn these desktops into virtual machines and move them into a data center. With them in the data center I can then access them from a smart phone, tablet, or any other device.

2017 VMware EUC Champion Logo - MetalI’ve been working on EUC for about seven years. I’ve found it important to share with others what I’ve learned and I hope to continue to do this for a long time to come.

I have also learned an immense amount from my friends that are part of the EUC Champions. Their knowledge of EUC has helped me go well beyond what I could do alone. So to all my friends returning or moving on from the EUC Champions thank you.

If you are interested in learning more about the VMware EUC Champions you can visit the blog post here. You can also leave a comment at the end of the post if you are interested in becoming an EUC Champion.

If you are interested in working towards being a social champion and aren’t sure how to get involved or how to get started, send me a message. I might not be the best person for what you want to champion but I’m more than happy to help you get started.

Permanent link to this article: https://www.wondernerd.net/doubly-honored/

Mar 23

This Space NOT For #Rent

Not For Rent Sign. Black background with red border. Not is scribbled in along with Don't bother in the call blank.Of late I have been hearing about companies strongly ‘asking’ their employees to publish corporate content on their personal social media feeds like Twitter, Facebook, and Linked-In. I’m not a fan of letting employers spam my friends. If I want to share something it needs to be something I’m interested in and something my followers are interested in. Thus I decided to share some thoughts about companies asking for social media shares.

Since I will be talking about some of the social media practices at Dell Technologies I’d like to say up front, that I am an employee of Dell Technologies. (#IWorkForDell) This statement becomes important here in a bit and you’ll see why.

Reasons For A Share

So why do employers find it so tempting to ask employees to promote corporate messages on their employees social media feeds?

My thoughts are:

  • Many are enticed by Nielsen research from April, 2010 that has shown “[….] 92 percent of consumers around the world say they trust earned media, such as recommendations from friend and family, above all other forms of advertising […]”
  • The other tempting thing, from my perspective, is the ready made “share of voice” that is gained when an employee shares on social media.

In both of these cases this puts the message in front of more consumers, and general wisdom is the more people you reach the better the chance that someone will be interested and become a customer. It’s a very constrained and telling way to look at it.

If you will, a company can try to have hundreds of social media accounts to reach thousands of potential customers or they can ask employees to do it for them. This makes logical business sense but it gets a bit more messy when you start digging into it.

The Mess

Some of the mess happens when you start looking at the following:

  • Delivering the whole message
  • Consistency
  • Private accounts
  • Fairness
  • Employee personal brand
  • FTC requirements

Let’s dig into some of these points, starting with delivering the whole message. If an employee is being asked to broadcast a canned message on their social media account, the first question I always ask is, will the person broadcasting the message be able to have a follow on conversation or know who can. If all they are doing is parroting the message what good is it if someone is actually interested in the message and engages them on it. They won’t be able to have an intelligent conversation about it.

This is a loose loose type of social interaction. The employee’s personal brand is damaged because his or her followers are not able to trust him or her further on such things. And the company’s message is lost because the follow on messaging couldn’t be delivered. A lost opportunity!

When an employee sends out a message on their personal social media account, they may customize the content of that message. This may create an entirely new and unintended message. Thus unintentionally lessening the impact of or completely turning off viewers of the message. As any marketer knows, if the message isn’t consistent, it can create problems down the road and there is no guaranteed way to insure a consistent message from private accounts.

When employee’s own the social media account it’s a private account. In some instances employee’s may have multiple accounts with the same social media platform. If they have multiple accounts then there must be a reason for it. Most likely to reach different audiences. Where does the companies message fit in those accounts?

Asking an employee to spread a message across multiple accounts consumes more of their time. In rare cases consuming so much that they are only focused on pushing out the message across all of their accounts. This can place an undue strain on the employee, as they may not have access to social media tools that can speed messaging on their personal social media accounts.

Personal accounts also pose another problem, they are private accounts. That means a corporate message could wind up right next to the post about an employee’s child loosing their first tooth or a recipe for egg drop soup. These are great if the employer is a dentist or a restaurant, but not so much if you work in just about any other industry.

This should also make everyone ponder, is content like that fair to the reader, the employee, and the employer? My personal response is that it is not. It’s not fair. It’s not fair to the grandma who hardly knows how to use the social media platform, Nor is it fair to the employee who is disrupting their personal brand. And it most certainly it isn’t helping the company because neither a lost tooth or soup are their target market.

Fairness extends beyond just the consideration of content placement. Employers must also consider if they are being fair and consistent with all employees. If the request is only made of a few ‘highly social’ employees, does it speak poorly of the other employees? Does it have the potential to create a discriminatory environment in the workplace? All of this may well create a very cumbersome HR mess.

‘Asking’ employees to share company posts can also be disruptive to the personal brand the employee is trying to create. For example if the employee is working on creating a personal brand focused on their underwater basket weaving abilities, an unrelated post on hydroponics can be disruptive to the employees brand. It also means that their effort to post the message is wasted. It’s not helping the employee grow their brand readership, nor their credibility. The company also looses out, the reach may be greater but it isn’t resulting in additional customers. In many cases this is a net loss to the company.

That gets us to the last bullet in my list of why ‘asking’ employees to share content creates problems. Not just anecdotal problems but, legally ones for both employee and employer. It’s not well known outside of social media marketing groups that the Federal Trade Commission (FTC) expects social media posts to clearly state bias. It’s a fancy way of saying if someone is posting something on a personal account their interest should be disclosed. The FTC views that as “Native Advertising.” See: Native Advertising: A Guide for Business

In researching this blog post I ran across an interesting story in Ad Week. The long and short of it was that the FTC and an advertising agency reached a settlement after employees were encouraged  to promote a project on their personal social media accounts without disclosing their bias. The FTC has said “Agencies asking their employees to spread the word must instruct those employees about their responsibilities to disclose their relationship.” See: The FTC’s Endorsement Guidelines: What People Are Asking – Employee Endorsements

There is a lot more to read on the FTC’s policies on bias. I recommend checking out the FTC website and a website called “All About Advertising Law” by Venable, LLP who has a post titled Your Social Media Policy, the FTC and the DOL.  Both of these resources have more material than can be covered in this blog post.

All of the above are reasons why companies should be extremely careful asking employees to share material on social media. In all fairness though I do post about some of the things I’m working on that may be of interest to my social media followers and fits my brand. I just don’t post everything blindly.

I’m also happy to be part of Dell Technologies when it comes to social media. Dell Technologies has a really good social media training program for employees. They also have a well defined social media policy (you can read it here http://dell.com/socialmediapolicy). It’s also great having leaders that are active on social media. You probably know who they are.

All that said. I’m not a fan of blindly tweeting company content just to tweet content. In fact reading this you can probably guess for me to share on my social networks it has to be retentive to my communities. I can sum this up very simply by paraphrasing 1 Corinthians, Chapter 13, Verse 1, I want to love what I share; I don’t want to be  “a resounding gong or a clashing cymbal.”

What are your thoughts on this? Share them below.

Permanent link to this article: https://www.wondernerd.net/this-space-not-for-rent/

Dec 10

Good Sounding Virtual Desktops

You’re prepping an image for use with a VMware Horizon deployment. You get all your applications tweaked and life is good. You optimized your image with the VMware OS Optimization Tool Fling. You go and test it, it appears to work great. On to the next task, oh wait you didn’t hear the familiar sound of logging in or out of the virtual desktop.

You check the speakers are hooked up. The sound on your system isn’t muted… what the heck?

speaker with sound bars and a no symbol across itI ran into this during a POC this last week. Here’s my quick tips on addressing this issue…

Start out validating that it’s not an ID:10t error and the speakers aren’t muted.

Then its time to check the registry. If you aren’t versed in working in the registry you probably shouldn’t be mucking around in it. If you are familiar with the registry visit the VMware KB 1004839 on enabling sound for RDP.

Find your OS in the list and follow the instructions for enabling the audio/video playback redirection.

It’s worth noting that the guide says its for RDP based connections. From my experience this also works for PCoIP connections as well.

This works great, you hear sound and life is good… Till those pesky end users start complaining that the sound and video is off when they watch their favorite Youtube clips. Now what?

Have no fear, there’s a solution for that.

Teradici has a virtual audio driver for PCoIP. It’s actually harder to login than it is to install. Download the driver, unzip it, and follow the instructions to install it.

When you test you should have sound that’s in sync with your video and your user’s can return to their regularly scheduled cat videos. You’re their IT hero.

Permanent link to this article: https://www.wondernerd.net/good-sounding-virtual-desktops/