I installed HS yesterday on a test HD drive (HFS+)
When I tried to install the webdriver it was blocked from loading by the system. I could fix that however.
In the previous systems I got a PCI error in the system report.
Now the unflashed NVidia Geforce GTX 980 is correctly displayed.
But link speed is only 2.5 GT/s while the old GT 120 has 5 GT/s.
The Luxmark 3,1 error is fixed in HS. While the test failed in the previous systems, it was OK in HS.
Geekbench results were the same as in Sierra, while Cinebench was slightly lower.
SIP is enabled.
I used the security and privacy control panel to unblock the system from loading the nvidia webdriver.
I have only installed HS after the 10.13.1 update.
Ocean Wave reports following bandwidth transfer speeds:
2,79 GB/s CPU > PCIe > GPU transfer speeds
3,33 GB/s GPU > PCIe > GPU transfer speeds
12,86 GB/s GPU <>GPU=VRAM transfer speeds
89 FPS AA Off
Bet it's running at 5 GT/s. Check CudaZ if it looks like this, it's running at full speed. Look at Device to Host.
View attachment 733088
This has been discussed here in other threads.
Lou
~3GB/s, that's the 2.5GT/s speed, it seems your card is really running at PCIe 1.1 speed.
You mean use this setting to unblock the web driver?
View attachment 733115
[doublepost=1510015260][/doublepost]
Hi Flowrider, do you know if the latest web driver can only be loaded with SIP Enabled on High Sierra?
Here are my results with Cuda Z under El Capitan
https://www.dropbox.com/s/d3szhe9upc77x0l/Cuda Z results El Capitan.jpg?dl=0
The setting "Anywhere" isn't visible in the control panel under High Sierra, but I could make an exception not to block the extension from loading.
I found this additional info:
http://www.macvidcards.com/50-pcie-speed1.html
According to MacVidCards my card is running at 1,0 speed if I use OceanWaves.
But if I use Cuda Z i's running at 2,0 speed.
Anyway: the system info reports 1,0 speed (2,5 GT/s)
I tried running CUDA Z in the background and opened OceanWaves. No change. OceanWaves reports 1,0 speed.
The extension is loaded and I find it safer with the current settings in High Sierra. It makes only an exception for a small amount of applications, not all.
Flowrider, may I know how about yours? Both PCI info and OceanWave shows 5GT/s?Bet it's running at 5 GT/s. Check CudaZ if it looks like this, it's running at full speed. Look at Device to Host.
View attachment 733088
This has been discussed here in other threads.
Lou
^^^^It's also happening to me and a few others. There are other threads about it. Still there in 10.13.1.
Anyone else notice that original Nvidia Titan card is no longer supported by latest Web Driver 107?
Also, I cannot put Titan X Maxwell and Titan X Pascal in same Mac Pro or else I get a boot loop. They now need to be in separate machines in order to work with 107 web driver. Hmmm
It seems MVC posted the answer on his page.
View attachment 734554
No, this is a different boot loop. If both Maxwell and Pascal Titan X’s are installed in the Mac Pro, I can log in but 30s after logging in, I’m booted back out to the login screen. If I only have one of the Titan X’s installed, log in proceeds normally.
Also regular Titan card is no longer recognized by latest 107 driver...it was with previous driver.
How are you powering both cards at the same time? Sounds like you're drawing too much power and the machine is restarting to protect itself.
What do you mean by "no longer recognized"? Do you have SIP enabled or disabled? In High Sierra, if you run in to the Library Validation issues then it'll look like no driver was installed/enabled.
I tried running CUDA Z in the background and opened OceanWaves. No change. OceanWaves reports 1,0 speed.
The extension is loaded and I find it safer with the current settings in High Sierra. It makes only an exception for a small amount of applications, not all.