Report your NVIDIA Web Driver and CUDA issues and I'll pass them on to the developers

[pastrychef]

Of course, iMac13,2 is an Nvidia SMBIOS. Seems it has that prerequisite along with a pre-Skylake CPU. Consistent with what I've seen.

Yeah. I've read about other people trying 10.13.4 beta. Never helps. Seems the issue is in the driver itself or in something new it is doing with macOS . . .

I highly doubt it has anything to do with the SMBIOS of my HP. I'm willing to bet that if I changed it to iMac18,3, there would be still no lag. I'm confident the issue is just with Skylake and newer.

 

[ReddestDream]

I highly doubt it has anything to do with the SMBIOS of my HP. I'm willing to bet that if I changed it to iMac18,3, there would be still no lag. I'm confident the issue is just with Skylake and newer.

I'm not a betting person, but I'd love to hear the results if you do try it.

I'm not convinced SMBIOS is irrelevant, but I do agree that CPU/Chipset/Arch are part of the equation.

 

[pastrychef]

I'm not a betting person, but I'd love to hear the results if you do try it.

I'm not convinced SMBIOS is irrelevant, but I do agree that CPU/Chipset/Arch are part of the equation.

Confirmed. I just tested my HP with iMac18,3 SMBIOS. Smooth as silk. The issue has absolutely nothing to do with choice of system definition.

 

[ReddestDream]

Confirmed. I just tested my HP with iMac18,3 SMBIOS. Smooth as silk. The issue has absolutely nothing to do with choice of system definition.

You made my day.

Thanks so much! I'll update my ticket on NvidiaGraphicsFixup to reflect this.

 

[ggeorge]

Well enabling Gsync causes a hard lock on my computer that I have to restart to fix. Sometimes I've had it work for as much as 20-30 min, but that's usually after a clean installation of the OS. More typically it happens within a minute or so of logging in, sometimes only a few seconds.

I have a 1080ti, but I've gotten it with a 980ti as well. other people have had the same issue. At least on other person had a 34" ultrawide like I have.

 

[ReddestDream]

Confirmed. I just tested my HP with iMac18,3 SMBIOS. Smooth as silk. The issue has absolutely nothing to do with choice of system definition.

Though the magic of Hopper Disassembler and the strings command, I have discovered a new check in 156's GeForceWeb.kext for "is_removable" that is not present in 104. I wonder if we rename that string kinda like Piker did here:

https://pikeralpha.wordpress.com/2015/11/23/patching-applegraphicsdevicepolicy-kext/

We might be able to make that check/property fail if it is causing an issue . . .

Would make sense since the lag is also occurring on eGPUs, which are really in fact removable . . .

 

[pastrychef]

Though the magic of Hopper Disassembler and the strings command, I have discovered a new check in 156's GeForceWeb.kext for "is_removable" that is not present in 104. I wonder if we rename that string kinda like Piker did here:

https://pikeralpha.wordpress.com/2015/11/23/patching-applegraphicsdevicepolicy-kext/

We might be able to make that check/property fail if it is causing an issue . . .

Would make sense since the lag is also occurring on eGPUs, which are really in fact removable . . .

Sounds like you are on to something! Unfortunately, I wouldn't even know where to begin trying to make the check fail...

 

[ReddestDream]

Sounds like you are on to something! Unfortunately, I wouldn't even know where to begin trying to make the check fail...

I do. I just can't test my stuff ATM. If you would test this for me, it would help. Thanks!

<key>KernelAndKextPatches</key>
    <dict>
        <key>KextsToPatch</key>
        <array>
            <dict>
                <key>Comment</key>
                <string>This is a horrible idea.</string>
                <key>Disabled</key>
                <false/>
                <key>Find</key>
                <data>
                aXNfcmVtb3ZhYmxl
                </data>
                <key>Name</key>
                <string>GeForceWeb</string>
                <key>Replace</key>
                <data>
                aXNucmVtb3ZhYmxl
                </data>
            </dict>
        </array>
    </dict>

As an image in Clover Configurator: https://imgur.com/a/hYaAh

Just be careful. I'm not sure if this will work at all and it may break booting, but it is unlikely.

What this does is just rename "is_removable" to "isnremovable" in hopes that it now it won't match anything and will fail without any need to nop out big chunks of code. Small and elegant is the hope.

 

[pastrychef]

I do. I just can't test my stuff ATM. If you would test this for me, it would help. Thanks!

<key>KernelAndKextPatches</key>
    <dict>
        <key>KextsToPatch</key>
        <array>
            <dict>
                <key>Comment</key>
                <string>This is a horrible idea.</string>
                <key>Disabled</key>
                <false/>
                <key>Find</key>
                <data>
                aXNfcmVtb3ZhYmxl
                </data>
                <key>Name</key>
                <string>GeForceWeb</string>
                <key>Replace</key>
                <data>
                aXNucmVtb3ZhYmxl
                </data>
            </dict>
        </array>
    </dict>

As an image in Clover Configurator: https://imgur.com/a/hYaAh

Just be careful. I'm not sure if this will work at all and it may break booting, but it is unlikely.

What this does is just rename "is_removable" to "isnremovable" in hopes that it now it won't match anything and will fail without any need to nop out big chunks of code. Small and elegant is the hope.

No difference. Still lagging like crazy...

 

[ReddestDream]

Ah. Okay. I'll keep looking . . .