Amd equivalent to Nvidia

[jacky123456]

hi guys in light of generaly bad support from nvidia on high sierra couple of years ago and macos mojave, i thought to change to the red team.
Can somebody tell me what AMD equivalent of a GTX 1080 is?

 

[Skymax86]

The closest to a GTX 1080 performancewise would be a RX Vega 64, i think.

 

[pastrychef]

If you are speaking of frames per seconds comparisons in gaming in Windows, yeah, the Vega 64 would probably be pretty close to a GTX 1080. That being said, a properly tuned Vega 56 can also compete quite well with with a GTX 1080. By properly tuned, I mean undervolting, overclocking, etc.

If you are talking about OpenCL performance, going by LuxMark benchmark results, either Vega will bench better than a GTX 1080.

If you are going by Geekbench 4 Metal and OpenCL results, a Vega 56 will be pretty close to a GTX 1080 while Vega 64 will probably beat it by a good margin.

If you want Mac OpenGL performance on a similar level to a GTX 1080, both Vega cards will fall short by a pretty wide margin.


That being said, OpenGL and OpenCL have both been depreciated by Apple and Metal 2 is what Apple's pushing now. In this regard, any AMD card will beat any Nvidia card because Nvidia has yet to even release Mojave drivers.

Lastly, keep in mind that Vega tuning in macOS has been made possible by the fantastic VGTab utility. It allows macOS users to adjust voltage, GPU clocks, HBM clocks, and power limits.

 

[Skymax86]

If you are speaking of frames per seconds comparisons in gaming in Windows, yeah, the Vega 64 would probably be pretty close to a GTX 1080. That being said, a properly tuned Vega 56 can also compete quite well with with a GTX 1080. By properly tuned, I mean undervolting, overclocking, etc.

If you are talking about OpenCL performance, going by LuxMark benchmark results, either Vega will bench better than a GTX 1080.

If you are going by Geekbench 4 Metal and OpenCL results, a Vega 56 will be pretty close to a GTX 1080 while Vega 64 will probably beat it by a good margin.

If you want Mac OpenGL performance on a similar level to a GTX 1080, both Vega cards will fall short by a pretty wide margin.


That being said, OpenGL and OpenCL have both been depreciated by Apple and Metal 2 is what Apple's pushing now. In this regard, any AMD card will beat any Nvidia card because Nvidia has yet to even release Mojave drivers.

Lastly, keep in mind that Vega tuning in macOS has been made possible by the fantastic VGTab utility. It allows macOS users to adjust voltage, GPU clocks, HBM clocks, and power limits.

Since I'm considering switching from my GTX 970 to a Vega 56 (probably: Sapphire Pulse or Asus Rog Strix) mainly because of MacOS (I'm using the Rig also as W10-Gaming Rig, but the 970 is still good enough) - which Vega 56 would you recommend? Are there still fan-problems with Mojave (for some Vegas) and do you know if they are fixable with something like kext patches?)

 

[pastrychef]

Since I'm considering switching from my GTX 970 to a Vega 56 (probably: Sapphire Pulse or Asus Rog Strix) mainly because of MacOS (I'm using the Rig also as W10-Gaming Rig, but the 970 is still good enough) - which Vega 56 would you recommend? Are there still fan-problems with Mojave (for some Vegas) and do you know if they are fixable with something like kext patches?)

Yes, Vega cardss still have the fan issue with Mojave.

Vega cards with reference style coolers can use the VGTab fix the fan issue.

Some Vega cards with custom style coolers are unaffected by the fan issue, some are. I don't know which do and which don't. You will have to do more research to find out which do and which don't.

I do know that the Sapphire Vega 64 Nitro suffers from the fan issue and VGTab can't fix it.

 

[chiefbeats]

@pastrychef,
Is there a guide describing what each fan parameter is actually changing or doing in VGTab? The difference between idle and target temperature is a little confusing, but I feel like I may have figured it out. Currently, my Vega 56 fan issue is considered "resolved", from my perspective, with these settings:

idle: 1500
target speed: 4936 (MAX slider)
Min: 400
Max: 4900
sensitivity: 6977 (MAX slider)
target temp: 45C
power range: 50%

I have been using LuxMark to trigger a load on the GPU to see how the fan responds in the terminal. So far, with the above settings, everything is good. The fan speeds up and slows down as expected and as needed. The biggest point is it is quiet under regular conditions (desktop - no load).

Idle Speed: When trying to think about this logically, I figured the idle speed would be under no load, of course; for example, at the desktop...not much going on.

Target Speed: I believe the target speed is related to the target temperature. I believe when the target temperature is reached, the fan will increase to the target speed as needed based on how high the temperature goes above the target temperature. For example, with having the target speed set to the MAX value, the fan will potentially increase to the target speed if needed to maintain the temperature around the target temperature. If you were to reduce this target speed value, I'm assuming the speed would not increase above this value (going to experiment some more on this).

Sensitivity: The sensitivity seems to work best on the MAX value. The fan seems to increase and decrease the quickest at this setting.

Target Temp: The target temp determines at what temperature the fan starts to increase above the idle value.

Power Range: I'm not sure what this is. Is this related to the GPU or the fan? Would reducing this value result in better performance as the card would not be able to reduce power as much to maintain temperature control? Or is the opposite true? Is this true at all?

I'm looking forward to any clarification you may have.

 

[pastrychef]

@pastrychef,
Is there a guide describing what each fan parameter is actually changing or doing in VGTab? The difference between idle and target temperature is a little confusing, but I feel like I may have figured it out. Currently, my Vega 56 fan issue is considered "resolved", from my perspective, with these settings:

idle: 1500
target speed: 4936 (MAX slider)
Min: 400
Max: 4900
sensitivity: 6977 (MAX slider)
target temp: 45C
power range: 50%

I have been using LuxMark to trigger a load on the GPU to see how the fan responds in the terminal. So far, with the above settings, everything is good. The fan speeds up and slows down as expected and as needed. The biggest point is it is quiet under regular conditions (desktop - no load).

Idle Speed: When trying to think about this logically, I figured the idle speed would be under no load, of course; for example, at the desktop...not much going on.

Target Speed: I believe the target speed is related to the target temperature. I believe when the target temperature is reached, the fan will increase to the target speed as needed based on how high the temperature goes above the target temperature. For example, with having the target speed set to the MAX value, the fan will potentially increase to the target speed if needed to maintain the temperature around the target temperature. If you were to reduce this target speed value, I'm assuming the speed would not increase above this value (going to experiment some more on this).

Sensitivity: The sensitivity seems to work best on the MAX value. The fan seems to increase and decrease the quickest at this setting.

Target Temp: The target temp determines at what temperature the fan starts to increase above the idle value.

Power Range: I'm not sure what this is. Is this related to the GPU or the fan? Would reducing this value result in better performance as the card would not be able to reduce power as much to maintain temperature control? Or is the opposite true? Is this true at all?

I'm looking forward to any clarification you may have.

I am unaware of any guides or tutorials on using VGTab.

This is what I've figured out...
Idle Speed = RPM of the fan when there is low load on the GPU. I have mine set to 800.
Target Speed = RPM of the fan when the GPU is under load. I have mine set to 4900.
Minimum Speed = Not sure what this does. I left it at default setting.
Maximum Speed = Not sure what this does. I left it at default setting.
Sensitivity = I don't know what this is and left it at the default setting.
Target Temp = Define what the highest temp the GPU should be before the fan starts to ramp up. I have mine set to 55.
Power Adjustment Range = I believe that this setting was supposed to allow us to set the power target, however, in my trial and error testing, it didn't seem to do anything. I just left mine at the default.

 

[chiefbeats]

I am unaware of any guides or tutorials on using VGTab.

This is what I've figured out...
Idle Speed = RPM of the fan when there is low load on the GPU. I have mine set to 800.
Target Speed = RPM of the fan when the GPU is under load. I have mine set to 4900.
Minimum Speed = Not sure what this does. I left it at default setting.
Maximum Speed = Not sure what this does. I left it at default setting.
Sensitivity = I don't know what this is and left it at the default setting.
Target Temp = Define what the highest temp the GPU should be before the fan starts to ramp up. I have mine set to 55.
Power Adjustment Range = I believe that this setting was supposed to allow us to set the power target, however, in my trial and error testing, it didn't seem to do anything. I just left mine at the default.

For target speed, here is what I believe: RPM upper limit of the fan when the target temperature is exceeded by some value. I disagree with it being based on being under load, because when all the values are set to default, and temperature keeps climbing, the fan speed never changed much (2400). I never waited to see if the fan sped up after exceeding 75C in worries of damaging the card; I don't have a lot of experience with GPUs and temperatures, etc. When reducing the target temperature to 45 and increasing the target speed to 4900, for example, the fan started speeding up around 55C, in comparison, so I think it's safe to assume a target temp of 75C would likely result in fan speed responding around 85C or so. It's definitely linked to the target temperature. Basing it on load is not the whole story because loads vary in time and intensity, of course, so being under load may not result in exceeding the target temp. Agreed on all others, however! Thanks for the info.