38 replies to this topic

[Homy]

jeannot, on 17 November 2019 - 01:50 PM, said:

1600*900 is not adequate to test GPU performance, neither is Unigine 1. It's much too old.

I understand what you're trying to say but I think you're missing the point a bit. The point wasn't to see if 5500M is good in the most recent games at the native resolutions but how much faster 5500M is compared to previous 560X. In order to make a correct comparison they had to use the same setting as before. MBP 15" have a native resolution of 2880x1800 so instead of testing games in 1800p they tested 900p. MBP 16" has a resolution of 3072x1920. They couldn't obviously run MBP 15" with that resolution so it wouldn't be fair to run games in native resolution on both computers. I hope though we see some gaming results in the native resolution soon.

Regarding Unigine Heaven, yes it's an old engine but still heavy enough to stress almost every GPU when running at max settings. That's why almost every review site still uses it and it's recommended as a test tool in 2019. So if your computer gets good results in Ungine you can play other recent games as well. I read that Nvidia GF RTX 2080 Ti 11 GB which is the best and most expensive gaming card right now gets 200-250 fps with the same settings in Windows so I don't think Unigine Heaven is too old yet.

[ozzy]

I agree Unigine Heaven is still a worthwhile test to run, especially to compare across machines. The challenge with it on Mac OS though is it uses OpenGL, which tends to have really bad performance and isn’t what any games released in the last 2-3 years will use. For instance, you can check out details in this thread, but on the same computer on 10.3.6 a Vega 56 eGPU was 30%+ faster than an nVidia 1070 in metal games, but 20% slower  than nVidia 1070 in Heaven benchmark. Just means the OpenGL drivers can sometimes throw some screwy results in there for Unigine Heaven that aren’t representative of actual game performance.

http://www.insidemac...topic=48783&hl=

[Homy]

ozzy, on 17 November 2019 - 07:47 PM, said:

I agree Unigine Heaven is still a worthwhile test to run, especially to compare across machines. The challenge with it on Mac OS though is it uses OpenGL, which tends to have really bad performance and isn’t what any games released in the last 2-3 years will use. For instance, you can check out details in this thread, but on the same computer on 10.3.6 a Vega 56 eGPU was 30%+ faster than an nVidia 1070 in metal games, but 20% slower  than nVidia 1070 in Heaven benchmark. Just means the OpenGL drivers can sometimes throw some screwy results in there for Unigine Heaven that aren’t representative of actual game performance.

http://www.insidemac...topic=48783&hl=

Well, you're right about that it doesn't show the whole picture about the actual in-game performance and maybe cards from different manufacturers but in this case both cards are from AMD so such comparison should still be relevant. They also benchmark 5300M and 555X with Geekbench 5 Metal and 5300M is almost 65% faster. So one should of course look at different benchmarking results.  

[Homy]

New video: 5500M 8 GB 96% faster than 560X 4 GB in Shadow of the Tomb Raider 1200p medium, 25% faster than Vega 20 4 GB. Here is another gaming review.

[Frost]

A real escape key? A real keyboard? Huge SSD options?

<weeps tears of joy>

First they finally update the iPad mini with my wishlist of features to boot (to borrow a turn of phrase from tBC, the money flew out of my account so fast it imploded a window at my bank), now this.

I can't wait to try the new keyboard especially. I keep being tempted to upgrade my MacBook Pro, but every time I am, I visit the Apple Store and type on the current keyboards and it quickly cures me of any desire to own one for another 3-6 months. Put this together with ThunderMag's MagSafe USB-C and I might actually start budgeting for an upgrade.

[Sneaky Snake]

Homy, on 17 November 2019 - 01:18 PM, said:

Actually 5500M 4 GB is about 143% faster than 560X in Unigine Heaven Ultra 1600x900 according to this review.
5300M 4 GB is ca 168% faster than 555X and ca 24% faster than Vega 20 in Unigine Heaven Extreme 1600x900 according to this review.

I think iMacs will get this GPU update next year so we don't have to wait two years again for the next update. They probably will get the desktop versions.

I posted the numbers that Apple provided for Fortnite and Shadow of the Tomb Raider. It may vary game by game.

[Sneaky Snake]

Some more gaming performance numbers. These numbers are in bootcamp and are comparing the base 4GB 5300M GPU with the XPS 15 running the GTX 1650: 6:30 in the video he posts a chart

Looks like the 5300M is 10-15% ahead of the 1650 in all titles he tested at 1080p high settings. Really solid performance.

The only thing I would say is that the XPS 15 has horrendous thermal performance, so that may be lowering the 1650 numbers. My XPS 15 9570 with an i7 8750H and 1050 Ti MaxQ thermally throttles both my CPU and GPU during games and make gaming a pretty terrible experience. Thicker laptops with better cooling and the 1650 aren't really a direct competitor to the MBP16, but I would expect to see better performance numbers out of them due to them not being thermally constrained.

[Homy]

Sneaky Snake, on 20 November 2019 - 11:15 AM, said:

Some more gaming performance numbers. These numbers are in bootcamp and are comparing the base 4GB 5300M GPU with the XPS 15 running the GTX 1650:
5300M outperforms the 1650 in all titles he tested at 1080p high settings. 6:30 in the video he posts a chart

One thing I will say as an XPS 16 owner is that the thermal performance on the XPS 15 is terrible. So you may see a 1650 performing better in other laptops with better cooling. I have pretty bad thermal throttling on both the CPU and GPU while trying to game on my XPS 15 9570 (i7 8750H with 1050 Ti MaxQ)

Yes, but can it run Borderlands 3?

[Sneaky Snake]

Homy, on 20 November 2019 - 11:20 AM, said:

Yes, but can it run Borderlands 3?

I've been following our thread on that as well

Hopefully it can. If it can do WItcher 3 at 1080p high settings at 60 fps then you'd think it should be able to run BL3 decently.

[Sneaky Snake]

I just pulled the trigger and ordered the base i7 model with the GPU upgraded to the 5500M 4GB. I'm really interested to see how it performs to to compare it with my XPS 15 9570 (work laptop)

[Sneaky Snake]

Ran a few gaming tests on the 16" macbook pro. I have the base model with everything stock except for the GPU upgraded to the 5500M 4GB.

Pretty happy with the performance results.

• Destiny 2 can hold a locked 60 fps at 1080p, medium settings. Might even try bumping this up to high
  
• COD:MW 2019 can also hold a locked 60 fps at 1080p medium settings. Going to test it out on high settings as well
  
• 3Dmark Time Spy score of 3644. Pretty much on par with a desktop 1650 (3581 avg) and around 100 points faster then a laptop 1650 (3553 avg). For comparison with an older but popular desktop GPU, the GTX 970 gets an average score of 3722 in the same benchmark

If anyone has a specific game they would like me to test just let me know. I'm going to test out some Mac games next (rise of the tomb raider, borderlands 3, DE:MK)

[Homy]

Sneaky Snake, on 29 January 2020 - 03:51 PM, said:

• Destiny 2 can hold a locked 60 fps at 1080p, medium settings. Might even try bumping this up to high
  
• COD:MW 2019 can also hold a locked 60 fps at 1080p medium settings. Going to test it out on high settings as well
  
• 3Dmark Time Spy score of 3644. Pretty much on par with a desktop 1650 (3581 avg) and around 100 points faster then a laptop 1650 (3553 avg). For comparison with an older but popular desktop GPU, the GTX 970 gets an average score of 3722 in the same benchmark

How is 1080p even possible and how does the scaling look? Officially it supports the resolutions below and not 1920x1080. Having a native screen resolution of 3072x1920 the sharpest downscaled resolution would be 1536x960?

• 2048 x 1280
  
• 1792 x 1120
  
• 1344 x 840
  
• 1152 x 720

Strange here it says:

"Supported scaled resolutions:

• 1920 by 1200
  
• 1680 by 1050
  
• 1280 by 800
  
• 1024 by 640"

Maybe the latter resolutions are for external monitors?

[Sneaky Snake]

I was hooked up to an external monitor for all of the tests, hence the 1080p.

When playing games on the laptop display I believe I would set the resolution to 3072x1920, but then set the render resolution at 0.50 scale (1536x960). Some games like COD:MW support this, which allows for menu's, HUD, text, etc. to all be at native resolution while the actual gameplay itself is rendered at the much more reasonable 1536 resolution.

[Janichsan]

By the way: apparently, the newest Macs still have the startup sound in their firmware, just disabled.

I tried the method described in that thread on my 2016 MBP, and it works. (Earlier published "hacks" had been disabled by Apple pretty soon after their discovery.)

[ipickert55]

Just wanted to throw my two cents, I've got one of these bad boys and it's great. Gaming performance has been pretty solid so far even at high resolutions and settings.

[Sneaky Snake]

Apple updated the 16" MBP today to make the 5600M an option on the unit: https://www.macrumor...5600m-graphics/

5600M is a big step up from the 5500M, so assuming that the MBP can keep it's temps somewhat in check it should be around 50%+ faster then the 5500M. Specs of the 5600M in the MBP are on par with the desktop 5600XT, which is similar performance to an RTX 2060 (desktop).

Pretty impressive graphics for the 16" MBP.

EDIT: isn't quite as powerful as I initially thought. Core count and memory speed/bandwidth are very impressive, but frequencies kept very low to keep the card with power/thermal targets. On paper performance is nearly identical to desktop 5500XT.

[Thain Esh Kelch]

But can the chassis cope thermal wise with 700$ worth of better GPU?

[Sneaky Snake]

Looks like it has the same 50W TDP as the current 5500M which my 16" MBP can adequately cool (it is hot and loud under full load, but it is able to sustain loads for hours without thermal throttling).

Full specs here on AMD's website: https://www.amd.com/...le-5000m-series

Those specs put it pretty much identical performance with a desktop 5500XT, with perhaps slightly more performance in certain tasks due to the 5600M having much higher memory bandwidth. Looks like they had to keep the clockspeeds really low to maintain power and thermal efficiency, but still a nice 32% Tflop improvement over the 5500M. I would expect games to perform 25% better at least.

For comparison with desktop cards, this would put the 5600M faster then the RX580/GTX1060/1650Super and slightly behind the GTX 1660.

[Homy]

Sneaky Snake, on 15 June 2020 - 01:37 PM, said:

Looks like it has the same 50W TDP as the current 5500M which my 16" MBP can adequately cool (it is hot and loud under full load, but it is able to sustain loads for hours without thermal throttling).

Full specs here on AMD's website: https://www.amd.com/...le-5000m-series

Those specs put it pretty much identical performance with a desktop 5500XT, with perhaps slightly more performance in certain tasks due to the 5600M having much higher memory bandwidth. Looks like they had to keep the clockspeeds really low to maintain power and thermal efficiency, but still a nice 32% Tflop improvement over the 5500M. I would expect games to perform 25% better at least.

For comparison with desktop cards, this would put the 5600M faster then the RX580/GTX1060/1650Super and slightly behind the GTX 1660.

Well, it doesn't matter much if you can't pay $700 for a mediocre GPU. Typical Apple price!