NEXUS

[HiRez]

Hi,

Does anyone have any news on NEXUS ?
I have registered at the beginning of October and got a “stay tuned” type of message at the end of October and nothing ever since.
Regards

[Imago]

Hi,

Does anyone have any news on NEXUS ?

I have registered at the beginning of October and got a “stay tuned” type of message at the end of October and nothing ever since.

Regards

It seems to be that some people have access to the tools which are dated with 12.02.09. A readme can be found here: Blog .

It says…

NVIDIA Code-name Nexus 1.0 Beta 1 Notes 12.02.09

Requirements

To use the Nexus Debugger for hardware debugging CUDA C or HLSL code, you will need to use two machines, a host (with Visual Studio) and a target with a supported GPU (for running your application) in order to debug remotely over a normal TCP/IP network connection.

To use only the Nexus Analyzer or Graphics Inspector, you may either use a single machine with a single supported GPU installed, or use it remotely as described for the debugger. If using a single machine, your ‘host’ and ‘target’ machine are the same physical machine. More information is available in the Nexus Knowledge Base accessible by signing in through www.nvidia.com/nexus.

Host machine:

ï‚· Operating System: Windows 7 or Vista SP1, 32 or 64-bit.

ï‚· GPU: Any DX10-capable GPU

ï‚· Visual Studio: Visual Studio 2008 with Service Pack 1 (Standard edition or above)

ï‚· Other Software: .NET framework 3.5 with Service Pack 1

ï‚· Optional: CUDA SDK, DirectX SDK (Mar 2009 or Aug 2009)

Target machine:

ï‚· Operating System: Windows 7 or Vista SP1, 32 or 64-bit.

ï‚· GPU: An NVIDIA G92 or GT200 based GPU (GeForce, Quadro, or Tesla)

ï‚· Other Software: .NET framework 3.5 with Service Pack 1

ï‚· Optional: CUDA SDK, DirectX SDK (Mar 2009 or Aug 2009)

Download and Installation steps

1. Sign-in to the NVIDIA Nexus Support site. You can get here by signing in from www.nvidia.com/nexus.

2. Go to the Downloads page.

3. Download a host installer. You will install this installer on the machine that has Visual Studio on it.

o For 32-bit: NVIDIA_Nexus_Host_Win32_Beta1_12_01_2009

o For 64-bit: NVIDIA_Nexus_Host_Win64_Beta1_12_01_2009

4. Download a monitor installer. The monitor is the debug ‘server’ you will connect to, either on another machine or on your existing machine.

o For 32-bit Monitor: NVIDIA_Nexus_Monitor_Win32_Beta1_12_01_2009

o For 64-bit Monitor: NVIDIA_Nexus_Monitor_Win64_Beta1_12_01_2009

5. Download the display driver required by Nexus.

a. 195.62 driver (32-bit): http://www.nvidia.com/object/win7_winvista…95.62_whql.html

b. 195.62 driver (64-bit): http://www.nvidia.com/object/win7_winvista…95.62_whql.html

6. Install the driver on the ‘target’ machine (that has a GPU with a G92 or GT200 chip) but do not reboot.

7. Install the Nexus Monitor installer on the same ‘target’ machine in Step 4 (that has the G92 or GT200 GPU in it).

8. Install the Nexus Host on the ‘host’ machine, where Visual Studio 2008 SP1 is installed.

9. Read the release notes by choosing Start | All Programs | NVIDIA Nexus 1.0 | Release Notes

10. Open the User Guide by choosing Start | All Programs | NVIDIA Nexus 1.0 | User Guide and follow all the steps under the “Walkthroughs” section. This is important as it will validate that you are able to debug and analyze with the samples provided.

If you would like to use your existing project with Nexus, please read “Migration from An Existing Project” in the User Guide.

Notes and Known Issues

• Samples: Nexus ships with samples located (by default) in C:\ProgramData\NVIDIA Nexus 1.0.

• Documentation: Nexus includes a User Guide located on your machine (after installation) at Start | All Programs | NVIDIA Nexus 1.0 | User Guide.

• Release Notes: Choose Start | All Programs | NVIDIA Nexus 1.0 | Release Notes to view the full release notes.

• Graphics shader debugging is only supported when the ‘host’ machine is physically a separate machine than the ‘target’ machine. Attempting to do shader debugging on the same machine as the ‘host’ machine will hang your system.

• Note: Hardware source debugging (either CUDA kernels or HLSL shaders) is only supported when your ‘host’ machine is physically separate machine than your target machine and they remotely communicate over TCP/IP. Information on experimental single-machine Debugger support (for machines with 2 GPUs) for CUDA C debugging only is available in the Nexus Knowledge Base (sign-in through www.nvidia.com/nexus)

• Note: If using Vista, you can check if you have SP1 installed by opening the Control Panel, clicking on the System icon, and looking for the words “Service Pack 1” in the “Windows edition” section.

• Note: Installer: Some commands are executed that bring up the command prompt during installation.

• Known issue (Stability): Analysis: Starting an Analysis session while stopped at a breakpoint causes the system to crash.

• Known issue (Stability): Under some circumstances, using Stop Debugging (Shift + F5) when halted at a breakpoint in a CUDA C kernel will cause your target machine to hang. The workaround is to remove all breakpoints, and hit Continue (F5) before stopping your debug session.

If you are registered for the developer program… the only thing you can do is to request a new account and to wait…

Why am I unable to log-in to my account?

Nexus Beta users: If you have signed up for the Nexus Beta, your account will only be active once you have received an invitation to the Beta. You will receive a notification e-mail at the e-mail address you used to sign up for the Beta titled “Welcome to the Nexus Beta Program” once your account has been activated.

Other Products / Nexus Early Access users: Contact us or your NVIDIA account representative.

You can also e-mail us at devtoolssupport@nvidia.com as well, and we will help correct any account or login problems you are having.

Note: The account for the tools site is not the same as your NVIDIA registered developer account so make sure you are using the right set of credentials when logging in.

The last mail I received was on the 31th october:

Hi folks -

Thanks again for registering for the NVIDIA Nexus Beta Program. We are currently putting the finishing touches on our Beta build, and will begin to enable accounts once it is ready for download. You will receive another e-mail once your account has been enabled.

In the meantime, we have recently posted the Nexus Webinar, an in-depth 20 minute video tutorial on the Nexus web page: www.nvidia.com/nexus. Take a look!

Regards,

Kumar Iyer

Product Manager, Nexus

NVIDIA

P.S. One other note: due to an overwhelming number of developer sign-ups, we will be staging invitations over the next couple of weeks to make sure our systems can handle the support load. We will be working to get everyone signed-up access as soon as we can.

Does someone know how long we have to wait? I also need it to boost up some development process… and it would also enhance a fun-factor while developing with cuda ;) Perhaps some information from the support? :rolleyes:

[Imago]

Does someone know how long we have to wait? I also need it to boost up some development process… and it would also enhance a fun-factor while developing with cuda ;) Perhaps some information from the support? :rolleyes:

There is also a new forum here called “Nexus”… http://forums.nvidia.com/index.php?showforum=191 … it is empty for now :)

[_PM]

Got access 4 days ago but had very little time to play with.

Biggest (and saddest) news is:

To use the Nexus Debugger for hardware debugging CUDA C or HLSL code, you will need to use two machines, a host (with Visual Studio) and a target with a supported GPU (for running your application) in order to debug remotely over a normal TCP/IP network connection.

[Imago]

I don’t think this is really bad. Are there any other people who already have access to nexus?

[E.D_Riedijk]

That is/will not be true for machines with more than 1 GPU as far as I read. I was actually very happy that it was possible to use 2 1GPU machines to do debugging, as we have MB’s that only support 1 GPU in our standard PC’s ;)

[HiRez]

I was hoping that the 2 machine requirement could be circumvented by using a loopback TCP/IP connection to the localhost (i.e. 127.0.0.1).
Is it an established fact that this cannot be achieved ?
Regards

[E.D_Riedijk]

I was hoping that the 2 machine requirement could be circumvented by using a loopback TCP/IP connection to the localhost (i.e. 127.0.0.1).

Is it an established fact that this cannot be achieved ?

Regards

I don’t know, but that would not work anyway when you only have 1 card. And with 2 cards it is either already working, or it will be later on.

[spinjob]

I was hoping that the 2 machine requirement could be circumvented by using a loopback TCP/IP connection to the localhost (i.e. 127.0.0.1).

Is it an established fact that this cannot be achieved ?

Regards

I recall hearing an instructor on a Nexus demo (webinar?) addressing a class and mentioning that the computers they were using had Parallels (see http://www.parallels.com/products/desktop/pd4wl/) running on them. Is this a valid way of only using a single PC with Nexus? I don’t yet have Nexus so I can’t say whether this will work. Anyone?