2019 MBP - 15" - Options Assistance Please

[TommyBoy5]

Gents -

Long-time Apple person here. First Apple product I purchased was back in 1983 (Apple IIe) and I've had literally dozens since then. Actually, counting phones, watches, iPods and laptops, it's probably over a hundred!

My 12" Early 2015 MB is on its last legs. It needs yet another battery replacement and the keyboard is unresponsive for the fourth time. Rather than a new round of repairs it's time to upgrade.

I bought our son a 13" MBP last year and he loves it. Uses the Touch Bar every day and tells me every day it's the greatest.

I want to buy a new 2019 MBP - 15". It's been four years since I've had a new laptop and the 12" served me well but I need more power, more screen real estate, and more hard drive space. Apple will give me $395 trade in for this thing which isn't too shabby.

Some assistance please on recommended specs. I want the i9 8-core. I need 1Tb for the SSD but I'm a bit torn on the graphics and the processor. Should I elect to upgrade the processor it's around $180, and to go from the standard graphics to the most powerful is around $350. So we're talking around $500. I convinced myself that since this will need to last me at least 5 years I should max it out as that ends up being $100 a year. My 17" MBP lasted almost 7 years!

I want to run Windows 10 with Boot Camp and dedicate 400G to it. Honestly, 90% of the time on Windows I'll be playing games. Steam games, mostly. I'd love great performance and fast FPS. I imagine the difference between the two Vega HMD GPUs and increased processor speed would help a lot. I've tried researching it the best I could but it's difficult.

Some video editing although that's all iMovie. Some Garage Band. Basic stuff. Some photo editing. But nothing professional. Don't need Final Cut or Logic as I find the entry-level included apps to be sufficient.

But if I'm kidding myself with the difference in the two video cards and processors for what I want to do it would be nice to save the $500. Some reviews say they only kick in with multi-threading and serious video editing work. Seems like games could use the power, though - especially on the graphics upgrade.

Pretty excited about this computer. Any help is most appreciated.

(Edit: Regarding pricing - my current company has an employee discount so prices may differ from doing standard config changes on the Apple website)

 

[casperes1996]

Stay away from the CPU upgrade... Like really, stay away from it.

The difference in speed between the fastest i9 and the slower i9, is at best 2.5%. In Windows, it's 0%. You see, to run the CPUs at a high speed, Apple has not just increased heat dissipation through a better thermal interface than the last model, but they also do heavy undervaluing in macOS. That undervolting is not applied in Windows, meaning the CPU runs slower and hotter in Windows - It's not dangerous; It just won't run as fast cause it'll reach temperatures that are high enough that it slows down. So the faster chip will only be faster in macOS where the heat is managed, and even there, at most 2.5% or so.

Going for a Vega GPU can definitely make sense if gaming is important to you. A Vega in general will also produce less heat than a Polaris GPU, so it'll also allow the CPU to stretch its legs more in CPU+GPU workloads. But based on your general needs and the fact that an MB has been keeping you well served, it really is more of a case of "do you want it" than it really being something that'd benefit you. If you want it, the GPU upgrade can make sense as an entertainment add-on. The CPU will never make sense.

But if you're trying to get the GPU for the sake of "future-proofing", take that money and put it in a savings account for the next time you'll get a new MacBook Pro. That's a better future-proofing path. GPUs are still improving at a faster rate than CPUs, even with the recent boom in core-count, and Navi seems to bring major improvements; Currently only on desktops and by the end of 2020 consoles, but probably also for the next redesign of the MBP. If not an RDNA 2.0 chip by then.

I don't want to talk you out of a GPU upgrade if you want it for immediate entertainment value through boosting its game-playing abilities; Usually a mostly GPU limited task anyway - but I think you'd be satisfied with the 560X as well. If I had the budget I'd personalise pick the Vega, 20 if I could afford it but 16 would be nice enough. - But I'm sticking with my 2014... Also I'm a student so I don't have the budget . And my use-cases are also generally more demanding, though as it is now I get along alright with my 2014 MBP and iMac.

 

[TommyBoy5]

Man, what a nice response! Thank you for the detail and the great advice!

Interestingly enough, I just read a review on the two highest I9s. They had the same advice! He said as far as that 5ghz you'll never see it. He said if you maybe put the laptop in a freezer and then ran the stress test it might bump that 5Ghz for less than a second. He said never, never buy it. So I'm definitely going with the standard processor on the higher-end 15" MBP. There - I just saved $188.

Now the GPU. Same thing basically. The Vega kicks butt over the 560X - but - he played a few games on Mac OS and they were stellar even with the 560X. The price difference between the 16 and 20 is minimal - less than $100. But the bump up from the 560 is pretty steep.

I might just stick with the 560X. Probably fine for what I need.

Thank you again for such a detailed reply.

 

[casperes1996]

Interestingly enough, I just read a review on the two highest I9s. They had the same advice! He said as far as that 5ghz you'll never see it. He said if you maybe put the laptop in a freezer and then ran the stress test it might bump that 5Ghz for less than a second. He said never, never buy it. So I'm definitely going with the standard processor on the higher-end 15" MBP. There - I just saved $188.

To be fair, maybe 80% of reviewers also misunderstand the purpose of Turbo-boosts. They were never meant to be sustained, even just a little. The idea is that if your CPU is rated to run at heat and power X. If it's been running at X-30 for a while, you can for a short wile run it at X+Y before it hits X and then only run it at X from thereon out. - It's less about power user applications, and more about increasing responsiveness for small tings. - But at the scales the i9s operate in, it's largely irrelevant for that too. Of course that's turbo-max. The full turbo range just gives flexibility for conditions which is generally just a good thing.

Now the GPU. Same thing basically. The Vega kicks butt over the 560X - but - he played a few games on Mac OS and they were stellar even with the 560X. The price difference between the 16 and 20 is minimal - less than $100. But the bump up from the 560 is pretty steep.

I might just stick with the 560X. Probably fine for what I need.

Yeah exactly. I'd probably advice the Vega 20 to someone already willing to go Vega 16, but the 560X should be just great as well. I think you'll be perfectly happy with it.

 

[pallymore]

avoid the CPU upgrade. the differences are really too small.

the Vega GPU is nice to have - some tests claim it helps the CPU to run a tiny bit cooler however I wouldn’t spend 300+ just for a little improved CPU temps.

If you mostly game on a secondary monitor, save that money and get an external GPU enclosure instead - after you decide that the 560X is not enough. You will get much better results with a mid-tier external GPU and you may upgrade it in the future.

 

[am2am]

Not sure about CPU upgrade - some reviewers claim only higher 8 core (2.4) has undervolting implemented = lower temp and better results under heavy prolonged load.

 

[hajime]

To be fair, maybe 80% of reviewers also misunderstand the purpose of Turbo-boosts. They were never meant to be sustained, even just a little. The idea is that if your CPU is rated to run at heat and power X. If it's been running at X-30 for a while, you can for a short wile run it at X+Y before it hits X and then only run it at X from thereon out. - It's less about power user applications, and more about increasing responsiveness for small tings. - But at the scales the i9s operate in, it's largely irrelevant for that too. Of course that's turbo-max. The full turbo range just gives flexibility for conditions which is generally just a good thing.



Yeah exactly. I'd probably advice the Vega 20 to someone already willing to go Vega 16, but the 560X should be just great as well. I think you'll be perfectly happy with it.

Given everything the same, will a 15" MBP 2019 with Vega 20 runs quieter and have better battery life than one with 560X?
[doublepost=1560615775][/doublepost]

Not sure about CPU upgrade - some reviewers claim only higher 8 core (2.4) has undervolting implemented = lower temp and better results under heavy prolonged load.

Any reference? If I am buying, I would buy i9 8 cores. If both 2.3 and 2.4GHz have undervolting implementation, I just get the 2.3 version.

 

[casperes1996]

Given everything the same, will a 15" MBP 2019 with Vega 20 runs quieter and have better battery life than one with 560X?

On GPU workloads, yes. With regular usage that doesn't really send many tasks to the GPU, and typically run entirely on the iGPU anyway - no difference.

 

[pshifrin]

On GPU workloads, yes. With regular usage that doesn't really send many tasks to the GPU, and typically run entirely on the iGPU anyway - no difference.

What if you aren't doing heavy GPU related tasks most of the time on a 15 but you are always plugged into a 4k screen when at your desk. Still no reason for the Vega?

 

[casperes1996]

What of you aren't doing heavy GPU related tasks most of the time on a 15 but you are always plugged into a 4k screen when at your desk. Still no reason for the Vega?

Would it have a minor effect, probably, yes. But not worth the price gap I'd say.

 

[Thysanoptera]

A Vega in general will also produce less heat than a Polaris GPU, so it'll also allow the CPU to stretch its legs more in CPU+GPU workloads.

Vega 20 is 50W GPU, rx560 is 35W. I'm really getting tired of that Vega is cooler and less power hungry nonsense.

 

[casperes1996]

Vega 20 is 50W GPU, rx560 is 35W. I'm really getting tired of that Vega is cooler and less power hungry nonsense.

Vega may have a higher power ceiling, but at a given workload, Vega is more efficient. If they are both locked to a certain frame rate at a given workload or similar, the Vega will produce less heat.

 

[Thysanoptera]

Vega may have a higher power ceiling, but at a given workload, Vega is more efficient. If they are both locked to a certain frame rate at a given workload or similar, the Vega will produce less heat.

You're describing something that happens once in a blue moon. There is no way to reliably limit Vega to be below 35W. "In general" it will be running at 50W with fans blazing, much hotter than Polaris.

 

[casperes1996]

You're describing something that happens once in a blue moon. There is no way to reliably limit Vega to be below 35W. "In general" it will be running at 50W with fans blazing, much hotter than Polaris.

It’s true that you cannot reliably limit it. But I’d argue that in most circumstances it will stay under Polaris. Both based on the empirical evidence provided all around, and based on the fact that general monitor refreshing for desktop usage locks at 60, GPGPU tasks will finish faster and does go back to cool sleep faster, and I’m sure we’ve all seen the empirical evidence for video editing tasks. For gaming workloads, sure, the Vega will probably run at its TDP pretty consistently.

But regardless of all of this, neither of us really recommend going for the Vega anywhere under these circumstances, so it’s under the current context largely irrelevant.

 

[Thysanoptera]

It’s true that you cannot reliably limit it. But I’d argue that in most circumstances it will stay under Polaris. Both based on the empirical evidence provided all around, and based on the fact that general monitor refreshing for desktop usage locks at 60, GPGPU tasks will finish faster and does go back to cool sleep faster, and I’m sure we’ve all seen the empirical evidence for video editing tasks. For gaming workloads, sure, the Vega will probably run at its TDP pretty consistently.

But regardless of all of this, neither of us really recommend going for the Vega anywhere under these circumstances, so it’s under the current context largely irrelevant.

Just because it will finish a task quicker does not make it cooler. By this logic I should use my RTX2080Ti as air conditioner.

 

[casperes1996]

Just because it will finish a task quicker does not make it cooler. By this logic I should use my RTX2080Ti as air conditioner.

Actually that was Intel's strategy for reducing heat and power consumption for a while

The idea isn't that it'll be cooler while it's running, but that if you can complete a task before you really start heating up, you can go back to sleep and never hit high temperatures. Or if you do get to heat up, at the very least still go back to sleep and let the cool down process start sooner.

 

[Thysanoptera]

Actually that was Intel's strategy for reducing heat and power consumption for a while

The idea isn't that it'll be cooler while it's running, but that if you can complete a task before you really start heating up, you can go back to sleep and never hit high temperatures. Or if you do get to heat up, at the very least still go back to sleep and let the cool down process start sooner.

You do realize that at high power states the CPU requires exponentially more power due to loses, doing something the way you described will result in higher energy usage for a given task. It just doesn't make sense, simply running in between the whole time (rather than cycling) is the best solution

You're trying too hard, the problem is people repeat such information and then are surprised.

I'll tell you how the whole "Vega is cooler" started: RX was getting in high power state when connected to external monitor. Vega did not. Apple fixed this on RX couple of months ago, but the perception remains.

 

[casperes1996]

You do realize that at high power states the CPU requires exponentially more power due to loses, doing something the way you described will result in higher energy usage for a given task. It just doesn't make sense, simply running in between the whole time (rather than cycling) is the best solution

You're trying too hard, the problem is people repeat such information and then are surprised.

I'll tell you how the whole "Vega is cooler" started: RX was getting in high power state when connected to external monitor. Vega did not. Apple fixed this on RX couple of months ago, but the perception remains.

https://en.wikichip.org/wiki/race-to-sleep



I won't dispute your general point about the power states when connected to monitors. I have no insights on that.
But the testing I've seen was without any connected monitors, and in a lot of cases the Vega equipped model was cooler. - Could be lucky chips of course, I am working with a test size of like three samples.

 

[Thysanoptera]

https://en.wikichip.org/wiki/race-to-sleep



I won't dispute your general point about the power states when connected to monitors. I have no insights on that.
But the testing I've seen was without any connected monitors, and in a lot of cases the Vega equipped model was cooler. - Could be lucky chips of course, I am working with a test size of like three samples.

Race-to-sleep was the only way to reduce power consumption of a CPU 14 years ago, otherwise it would be in high power state regardless whether the CPU was doing something or not, it had nothing to do with active temperature control, it was purely a power saving technique. And the CPU was going to max frequency because it couldn't go any lower, it was max or off, nothing in between.

That's how Vega 20 looks like compared to RX555 doing the same task - Heaven benchmark. Yes, it will have almost double the score. But it won't be cooler:

 

[casperes1996]

That's how Vega 20 looks like compared to RX555 doing the same task - Heaven benchmark. Yes, it will have almost double the score. But it won't be cooler:

On this one I am willing to relent that I perhaps have incorrect information. But as mentioned I have three sources, including Max Yuriev, where the device was not connected to a monitor, reporting lower temps on the Vega. That seems interesting if the monitor connection was the central issue, pushing Polaris to its highest power state.
But chip lottery is entirely possible as an explanation I support. - Or just test specific things.

Race-to-sleep was the only way to reduce power consumption of a CPU 14 years ago, otherwise it would be in high power state regardless whether the CPU was doing something or not, it had nothing to do with active temperature control, it was purely a power saving technique. And the CPU was going to max frequency because it couldn't go any lower, it was max or off, nothing in between.

This one I'll argue on a bit more though, because you're entirely correct that the higher up the frequency curve, past a certain point, the less efficient things get, exponentially so. You're entirely right that for effeciency there is a frequency sweet spot that should not be overshot if power consumption is a valued factor.

But race to sleep still has benefits for heat management in certain scenarios, because the heatsink has a temperature capacity. If the entire system is cold, a quick burst of heat through a very intense workload run very fast may not run for long enough to heat up the capacity of the heatsink, so fans will not kick in and the extremely low heat production of an entirely power gated core or chip has a much greater delta up to the frequency sweet spot, than the efficiency delta between the sweet spot and the maximum tolerance frequency.

Race to sleep isn't a good choice for longer running tasks, but on a cold system for tasks that are like 4 seconds or less - preferably a good deal less, it can still be a good option. Which is why turbo boost max and XFR Velocity Boost can go quite a bit beyond the sweet spot

 

[Thysanoptera]

On this one I am willing to relent that I perhaps have incorrect information. But as mentioned I have three sources, including Max Yuriev, where the device was not connected to a monitor, reporting lower temps on the Vega.

Max Yuriev (and multiple others) forgot about a crucial component of temperature equation - fan speed. Apple sets the temperature target to similar if not identical values, since Vega produces a lot more heat the fans (in auto) speed up a lot faster leading to a temporary situation where you could have Vega temperatures lower than RX. What is missed is that the fans are couple thousand rpm higher on Vega. That Heaven benchmark - that was with fans forced to max, to match what was happening on Vega. In auto, on the first run, which lasts couple of minutes, the fans on RX barely moved from idle position.

Race to sleep isn't a good choice for longer running tasks, but on a cold system for tasks that are like 4 seconds or less - preferably a good deal less, it can still be a good option. Which is why turbo boost max and XFR Velocity Boost can go quite a bit beyond the sweet spot

Race-to-sleep wasn't a choice, that's the main point you're missing. It was the only power-saving technique available at the time. To reduce the frequency you had to power down the machine, flip a jumper on motherboard or desolder the oscillator.

 

[casperes1996]

Race-to-sleep wasn't a choice, that's the main point you're missing. It was the only power-saving technique available at the time. To reduce the frequency you had to power down the machine, flip a jumper on motherboard or desolder the oscillator.

But it's a choice now, and still remains used under certain, admittedly very specific, scenarios.

 

[Thysanoptera]

But it's a choice now, and still remains used under certain, admittedly very specific, scenarios.

Yes, but this is not some kind of separate mode operating in this binary mode you presented, like in the past. If anything, this is still the main principle of CPU operation, drop to idle as soon as possible. But modern CPUs dynamically scale frequency and voltage, going far above what in the past would be considered reasonable, using much more sophisticated algorithms.

 

[casperes1996]

Yes, but this is not some kind of separate mode operating in this binary mode you presented, like in the past. If anything, this is still the main principle of CPU operation, drop to idle as soon as possible. But modern CPUs dynamically scale frequency and voltage, going far above what in the past would be considered reasonable, using much more sophisticated algorithms.

Uhm... Pretty sure that's also what I said?

 

[aimbdd]

Man, what a nice response! Thank you for the detail and the great advice!

Interestingly enough, I just read a review on the two highest I9s. They had the same advice! He said as far as that 5ghz you'll never see it. He said if you maybe put the laptop in a freezer and then ran the stress test it might bump that 5Ghz for less than a second. He said never, never buy it. So I'm definitely going with the standard processor on the higher-end 15" MBP. There - I just saved $188.

Now the GPU. Same thing basically. The Vega kicks butt over the 560X - but - he played a few games on Mac OS and they were stellar even with the 560X. The price difference between the 16 and 20 is minimal - less than $100. But the bump up from the 560 is pretty steep.

I might just stick with the 560X. Probably fine for what I need.

Thank you again for such a detailed reply.

I’m not sure why no one is really suggestions it, but I would definitely go with the 16 or 20 in relation to gaming. (Unless you plan on or could us a external gpu setup) from benchmarks I’ve seen the Vega 20 is 30-70 percent faster than the 560x! Definitely not inconsequential for gaming. I would asleep recommend against the upgraded cpu. Not worth the cost.