Vega Mackbook Benchmarks

[AvengerNX08]

Why wouldn't you trust bootcampdrivers.com? They have been great for multiple users and delivered much improved performance! I have used them for the Radeon Pro 460 with great success. Not once have I had issues with these.

 

[FrostyF]

AMD really don't care, it's an Apple computer and Apple is responsible to provide drivers

That's really not true....all component manufacturers provide their drivers too.

Looks like it can handle gaming in Windows just fine.

 

[koyoot]

That's really not true....all component manufacturers provide their drivers too.

Looks like it can handle gaming in Windows just fine.

It would still be 10-20% faster thanks to newest drivers. The drivers Vega uses are 1 year old(!).

 

[_Kiki_]

That's really not true....all component manufacturers provide their drivers too.

Looks like it can handle gaming in Windows just fine.

you can ask AMD support and they will tell you go back to the Apple, this problem isn't new, Apple using custom hardware and because of that official drivers from AMD website doesn't work, you can't buy Vega 16 or Vega 20 alone, you need to buy whole laptop, Apple is a manufacturer and they are responsible for drivers

 

[leman]

you can ask AMD support and they will tell you go back to the Apple, this problem isn't new, Apple using custom hardware and because of that official drivers from AMD website doesn't work, you can't buy Vega 16 or Vega 20 alone, you need to buy whole laptop, Apple is a manufacturer and they are responsible for drivers

The problem is that AMD treats these as custom products, even though they don't have to. This was something Nvidia was always much better at, since their normal Windows drivers always worked with the Mac GPUs. And while it might be custom hardware, its still using the same driver as any other GPU. AMD simply blacklists the regular drivers from being installed on Apple hardware. Which can be trivially circumvented by editing the driver's installer configuration.

Personally, I am quite disappointed that AMD chooses to treat products they make for Apple differently from the other Windows products. All probably because of some paranoid lawyer in the company. They could at least offer these as beta drivers with no support...

 

[koyoot]

The problem is that AMD treats these as custom products, even though they don't have to. This was something Nvidia was always much better at, since their normal Windows drivers always worked with the Mac GPUs. And while it might be custom hardware, its still using the same driver as any other GPU. AMD simply blacklists the regular drivers from being installed on Apple hardware. Which can be trivially circumvented by editing the driver's installer configuration.

Personally, I am quite disappointed that AMD chooses to treat products they make for Apple differently from the other Windows products. All probably because of some paranoid lawyer in the company. They could at least offer these as beta drivers with no support...

That is because AMD's policy is that Vendors have to provide drivers for the AMD hardware in their computers.

It was few weeks ago, that people complained about rubbish drivers for Raven Ridge Laptops. AMD rightly so claimed that it was vendors responsibility to release them. Later went saying that in upcoming weeks vendors like Lenovo will release new drivers for Raven Ridge chips.

I am however baffled - why the hell are we limited on Apple hardware to use one year old drivers?!

 

[FrostyF]

you can ask AMD support and they will tell you go back to the Apple, this problem isn't new, Apple using custom hardware and because of that official drivers from AMD website doesn't work, you can't buy Vega 16 or Vega 20 alone, you need to buy whole laptop, Apple is a manufacturer and they are responsible for drivers

Apple isn't a manufacturer. They outsource their manufacturing for everything. They're a system builder. AMD is being lazy if they don't provide drivers for their own product.

Just do a google search...they even released Linux drivers for the Vega 20 months ago.

https://www.google.ca/search?safe=o...1......0....1..gws-wiz.......0i71.OVEYHB635EY

 

[koyoot]

Apple isn't a manufacturer. They outsource their manufacturing for everything. They're a system builder. AMD is being lazy if they don't provide drivers for their own product.

Just do a google search...they even released Linux drivers for the Vega 20 months ago.

https://www.google.ca/search?safe=o...1......0....1..gws-wiz.......0i71.OVEYHB635EY

If that would be the case, why do you use drivers that are 1 year old, for Polaris 11/21 GPUs in MBPs?

Becasue Apple requested that.

P.S. The drivers you quoted, are for Vega 20 chip. Which is COMPLETELY DIFFERENT chip from Vega in MBP. Vega 20 has 4096 GCN cores, 32 GB HBM2, and is made on 7 nm process.

What you are looking for are Vega 12 drivers.

 

[FrostyF]

If that would be the case, why do you use drivers that are 1 year old, for Polaris 11/21 GPUs in MBPs?

Becasue Apple requested that.

P.S. The drivers you quoted, are for Vega 20 chip. Which is COMPLETELY DIFFERENT chip from Vega in MBP. Vega 20 has 4096 GCN cores, 32 GB HBM2, and is made on 7 nm process.

What you are looking for are Vega 12 drivers.

Damn, I didn't know you had insider knowledge as to what Apple did or do not request from their suppliers.

 

[FrostyF]

Anyone still thinking about wasting money on the i9?

https://browser.geekbench.com/v4/compute/search?dir=desc&q=AMD+Radeon+Pro+Vega+20&sort=score

i7 has now scored in the top 4 results (84817)

 

[leman]

Anyone still thinking about wasting money on the i9?

So what? You are testing the GPU, ideally, CPU speed shouldn't even matter there.

 

[koyoot]

So what? You are testing the GPU, ideally, CPU speed shouldn't even matter there.

If the software is properly written, that is correct.

But Geekbench is not properly written as we can on the pictures.

Scores can be affected by RAM pool and CPU clock speeds.

What that tells us, really?

Damn, I didn't know you had insider knowledge as to what Apple did or do not request from their suppliers.

I will repeat my question: if what you have said would be the case, why is Radeon Pro 560X, and all other GPUs in Apple computers using this chip(Polaris 11/21), THE ONLY GPU in the world, using Polaris 11/21 chip, to work on 1 year old drivers, on Apple computers, both under MacOS, and Windows 10/Bootcamp?

Why AMD can provide drivers for this GPU to every single platform there is, but not for Apple?

 

[leman]

Scores can be affected by RAM pool and CPU clock speeds.

Is it so though? Did anyone grab all the reported benchmark data and do proper stats on it with multivariate analysis? I randomly clicked on approx. 10 benchmarks from random pages, both highest and lowest scores. *All* of them had 32GB RAM. I have a suspicion that people who run benchmarks (and there are very few of them, as evident from only 186 entries) tend to be obsessed by specs and would get 32GB RAM even if its completely pointless for them.

 

[FrostyF]

Is it so though? Did anyone grab all the reported benchmark data and do proper stats on it with multivariate analysis? I randomly clicked on approx. 10 benchmarks from random pages, both highest and lowest scores. *All* of them had 32GB RAM. I have a suspicion that people who run benchmarks (and there are very few of them, as evident from only 186 entries) tend to be obsessed by specs and would get 32GB RAM even if its completely pointless for them.

I already posted detailed information about the Geekbench scores a few days ago.

https://forums.macrumors.com/threads/vega-mackbook-benchmarks.2156142/page-8#post-26843711

 

[leman]

I already posted detailed information about the Geekbench scores a few days ago.

Yeah, I read your posts and all I remember are generalisations of kind "top 8 scores are i7" (well, not anymore). If you have a post with statistical analysis of all the scores, maybe I did miss it. If you could provide a link, I'd be very interested.

 

[FrostyF]

Yeah, I read your posts and all I remember are generalisations of kind "top 8 scores are i7" (well, not anymore). If you have a post with statistical analysis of all the scores, maybe I did miss it. If you could provide a link, I'd be very interested.

"i7/16GB top score was 72287"

"i7/32GB top score was 84817"

It's not rocket surgery to see the differences.

 

[koyoot]

Really meaningful performance differences measure of this GPU is actually gaming.

Comparing Vega 20 on i7/16GB vs i9/32GB in games would show you that... there is no difference 😉.

 

[FrostyF]

Really meaningful performance differences measure of this GPU is actually gaming.

Comparing Vega 20 on i7/16GB vs i9/32GB in games would show you that... there is no difference 😉.

Got anything to prove that other than your theory?

 

[koyoot]

Got anything to prove that other than your theory?

Have you ever looked at gaming benchmarks on GPUs that are slower than RX 470(from which Vega Pro 20 is slower)?

You are ALWAYS GPU bound with them. Adding RAM, or CPU power to this level will not make any difference.

 

[leman]

"i7/16GB top score was 72287"

"i7/32GB top score was 84817"

It's not rocket surgery to see the differences.

And both the highest and lowest scores are i9/32GB. Your point is totally moot. You can't make any systematic conclusions by just looking at few top scores. All the things you "see" could be just random stuff. As you have noted yourself, there are many more 32Gb configs in those benchmark entries. So 32Gb has a higher chance to have a higher score simply because its more frequent. You need to look at multiple (ideally, all available) 32GB vs. 16GB entries and do a variance analysis to see if they are likely different. Its basic statistics 101.

 

[FrostyF]

Have you ever looked at gaming benchmarks on GPUs that are slower than RX 470(from which Vega Pro 20 is slower)?

You are ALWAYS GPU bound with them. Adding RAM, or CPU power to this level will not make any difference.

I don't really care about other results, do you have anything that proves there's no difference in gaming between the i7/16GB and i9/32GB like you claimed? *crickets*
[doublepost=1543438430][/doublepost]

And both the highest and lowest scores are i9/32GB. Your point is totally moot. You can't make any systematic conclusions by just looking at few top scores. All the things you "see" could be just random stuff. As you have noted yourself, there are many more 32Gb configs in those benchmark entries. So 32Gb has a higher chance to have a higher score simply because its more frequent. You need to look at multiple (ideally, all available) 32GB vs. 16GB entries and do a variance analysis to see if they are likely different. Its basic statistics 101.

The top scores are most important, because you can run the test with a hundred other things open to produce a low score. There's lots of data on the 555/560 models between 16 and 32GB RAM. You're welcome to go do your own analysis and post the results. Instead you're more interested in discrediting people who put in a little effort.

 

[IdentityCrisis]

Anyone still thinking about wasting money on the i9?

https://browser.geekbench.com/v4/compute/search?dir=desc&q=AMD+Radeon+Pro+Vega+20&sort=score

i7 has now scored in the top 4 results (84817)

View attachment 807144

A waste? Definitely not. Only a couple hundred more to get a bigger boost from the i9.

 

[FrostyF]

A waste? Definitely not. Only a couple hundred more to get a bigger boost from the i9.

It's been pointed many times that the % performance difference for the price shows the upgrade is not worth it.

 

[IdentityCrisis]

I don't really care about other results, do you have anything that proves there's no difference in gaming between the i7/16GB and i9/32GB like you claimed? *crickets*
[doublepost=1543438430][/doublepost]

The top scores are most important, because you can run the test with a hundred other things open to produce a low score. There's lots of data on the 555/560 models between 16 and 32GB RAM. You're welcome to go do your own analysis and post the results. Instead you're more interested in discrediting people who put in a little effort.

Why do you continue to be so agressive and rude in your replies to people? I’m constantly seeing you attack people and most of the time you have been wrong here and on reddit.
[doublepost=1543439017][/doublepost]

It's been pointed many times that the % performance difference for the price shows the upgrade is not worth it.

In the results I’ve seen, the difference is noticeable. And benchmarks won’t show it well. When you actually use the processor like video rendering and stuff then it is worth it.

 

[FrostyF]

Why do you continue to be so agressive and rude in your replies to people? I’m constantly seeing you attack people and most of the time you have been wrong here and on reddit.

Why are you attacking the way I respond to people? I'm offended. I've also made thousands of comments on reddit; you'll have to be more specific as to which ones were wrong.
[doublepost=1543439397][/doublepost]

In the results I’ve seen, the difference is noticeable. And benchmarks won’t show it well. When you actually use the processor like video rendering and stuff then it is worth it.

Which specific results are those that aren't using benchmarks?