Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Will you get this PCIe 3 version of the 2019 Mac Pro or wait for the next update?


  • Total voters
    67
Based on what's said on this forum regarding 7.1, why would PIC-e 4.0 matter? For the target audience, Mac Pro is just buy then make profit machine. Do enjoy the first batch of 7.1, and upgrade if necessary when the newer one comes.

For me, I will be waiting until Apple comes out with sub $4,000 headless mac. Until then, no.
 
Based on what's said on this forum regarding 7.1, why would PIC-e 4.0 matter? For the target audience, Mac Pro is just buy then make profit machine. Do enjoy the first batch of 7.1, and upgrade if necessary when the newer one comes.

For me, I will be waiting until Apple comes out with sub $4,000 headless mac. Until then, no.

It matters because it’s literally double the bandwidth of pci 3
 
I’m moving to Ubuntu for CUDA support.
Hoping that AMD release a 64 core threadripper by years end.
2x RTX2080Ti and I’m all good.

I’m a scientist and Apple no longer offer professional machines for us anymore. Too much focus on media production to the detriment of other professional industries.
 
  • Like
Reactions: ssgbryan
That's very true - the MP and much else that Apple does is meant for pros in one industry. They always have been, though - Apple has always loved media-specific features. There was a time when the G5 was enough faster per dollar than contemporary Intel chips at some scientific tasks that Apple came out ahead on FLOPS per dollar even after accounting for all the money they spent wrapping things that scientists generally didn't need around the G5. It also helped that the only other vendor for the same architecture was IBM, who charged a much higher premium than Apple did.

Ever since Apple switched to Intel, they've given up any chance of a FLOPS/$ lead. They have always been, and will always be, a relatively high cost box builder, so unless their chip has a massive advantage, their whole box will be on the expensive side.

It depends to some extent on what you want in your box - if you need massive memory support, ECC and a super high-grade power supply, we still don't know how the Mac Pro will come out against HP Z8s and the like. It may end up cheaper, depending on upgrade prices (neither the base Z8 nor the base Mac Pro is a good deal for most scientific applications). If Apple has worthwhile configurations out there cheaper than HP does, it could still end up a winner.

If your job fits in the maximum RAM of a Dell XPS or a generic Core i9/Ryzen/Threadripper box, you don't need ECC, and you don't care about workstation-level parts quality, any white box PC or a Dell below the Precision line wlll offer much higher FLOPS/$. Xeons and everything surrounding them are priced higher than consumer chips, and cheap parts make a huge difference compared to the workstation stuff Apple uses.

Of course, anything that takes advantage of NVidia's GPU architecture will run faster on something that supports that architecture...
 
It matters because it’s literally double the bandwidth of pci 3

For Apple's demo for the Audio DAW market of 5 Avid HDX PCI-e cards that makes zero material impact at all.
For the Audio space PCI-e v4 doesn't do much substantive at all to the direct process and only minor impact to some storage read/write variations.

for Apple demos for Video work if for every one second of data transfer the data spends 6+ seconds on the card being used in calculations then there won't be much impact at all. You'd go from 1 second to .5 and the overall time would change from 7 seconds (of only transfer one way) to 6.5 ( a -7% change ). Round trip 8 to 7 ( -13% change ).

Something like Afterburner would make a difference between it is bandwidth capped ( so only can go to three 8k screens) and the data up and return time is extremely short. if Afterburner is priced reasonably and the PCI-e v4 version was equally reasonable then that would be a major impact. If it is one of these megabucks cards that a small digital number of Mac Pro buyers acquire then it is a now so big of a deal.

If Apple didn't have card-to-card Infinity Fabric on their Vega II cards then PCI-e v4 would make a bigger difference also. ( but they do so it is substantively blunted. in this context also. )


Folks moving out of relatively ancient x8 PCI-e v2 are already seeing a double in bandwidith in the Mac Pro's x8 PCI-e v3 slots. If that context the majority of new buyers for the next 12 months is coming from their speed increase is all too real just with this. ( and those moving up with high sunk cost x8 PCI-e v2 (or v3 ! ) cards won't see a lick of difference between v3 or v4 slots since the card is the bottleneck. )
 
For Apple's demo for the Audio DAW market of 5 Avid HDX PCI-e cards that makes zero material impact at all.
For the Audio space PCI-e v4 doesn't do much substantive at all to the direct process and only minor impact to some storage read/write variations.

for Apple demos for Video work if for every one second of data transfer the data spends 6+ seconds on the card being used in calculations then there won't be much impact at all. You'd go from 1 second to .5 and the overall time would change from 7 seconds (of only transfer one way) to 6.5 ( a -7% change ). Round trip 8 to 7 ( -13% change ).

Something like Afterburner would make a difference between it is bandwidth capped ( so only can go to three 8k screens) and the data up and return time is extremely short. if Afterburner is priced reasonably and the PCI-e v4 version was equally reasonable then that would be a major impact. If it is one of these megabucks cards that a small digital number of Mac Pro buyers acquire then it is a now so big of a deal.

If Apple didn't have card-to-card Infinity Fabric on their Vega II cards then PCI-e v4 would make a bigger difference also. ( but they do so it is substantively blunted. in this context also. )


Folks moving out of relatively ancient x8 PCI-e v2 are already seeing a double in bandwidith in the Mac Pro's x8 PCI-e v3 slots. If that context the majority of new buyers for the next 12 months is coming from their speed increase is all too real just with this. ( and those moving up with high sunk cost x8 PCI-e v2 (or v3 ! ) cards won't see a lick of difference between v3 or v4 slots since the card is the bottleneck. )

For 4 slot nvme cards it makes a difference. And for a WiFi card I won’t. So yea, for some cards it will and some it won’t. Also, engineers have ways for coming up with new uses to saturate and use computer capacity. So thanks for the unwilling confirmation.
 
For 4 slot nvme cards it makes a difference.

Only v4 ones. And how many of those do folks have now versus the number of v3 ones? Over time the demand will build. But huge, necessary demand now or in the next 12-18 months... not very likely at all.

There are lots of folks getting real work down now with v3 cards stuffed in v2 MP 2009-2012 models. That work will simply just go faster in a MP 2019 without necessarily having to buy a new NVMe card at all. That's pretty good having to plop down a minimum $6K to get to the MP 2019.

Waiting another 1-2 years in a v2 MP 2009-2012 because the MP doesn't have a v4 card that you don't know is very substantial amount of time to sit and wait because the "next year will be better". The same thing will be true the next year after that. And the year after that . ...


PCI-e v4 is not 2019 for the Macintosh product line up. It is highly likely not 2020 for vast majority of the Mac line up; including the Mac Pro. Folks would are in the "hold my breath until v4" mode that is fine if really don't have much pressing work to do for the next two years. For those who do have work then no v4 isn't a deal breaker and it is a distraction.


Also, engineers have ways for coming up with new uses to saturate and use computer capacity. So thanks for the unwilling confirmation.

How many folks ran out and bought Power9 systems because the sky was going to fall if they didn't get to PCI-e v4 'right now' over the last two years?

It isn't millions of folks. Or even several hundreds of thousands.
 
Only v4 ones. And how many of those do folks have now versus the number of v3 ones? Over time the demand will build. But huge, necessary demand now or in the next 12-18 months... not very likely at all.

There are lots of folks getting real work down now with v3 cards stuffed in v2 MP 2009-2012 models. That work will simply just go faster in a MP 2019 without necessarily having to buy a new NVMe card at all. That's pretty good having to plop down a minimum $6K to get to the MP 2019.

Waiting another 1-2 years in a v2 MP 2009-2012 because the MP doesn't have a v4 card that you don't know is very substantial amount of time to sit and wait because the "next year will be better". The same thing will be true the next year after that. And the year after that . ...


PCI-e v4 is not 2019 for the Macintosh product line up. It is highly likely not 2020 for vast majority of the Mac line up; including the Mac Pro. Folks would are in the "hold my breath until v4" mode that is fine if really don't have much pressing work to do for the next two years. For those who do have work then no v4 isn't a deal breaker and it is a distraction.




How many folks ran out and bought Power9 systems because the sky was going to fall if they didn't get to PCI-e v4 'right now' over the last two years?

It isn't millions of folks. Or even several hundreds of thousands.

Wow the boards are just released and the entire industry didn’t update everything overnight.

Like everything else, it will be updated in upcoming iterations. Considering it’s backwards compatible with pci 3 there is no reason not to, and we’ll start seeing upgraded products, likely, with updated iterations of the individual products.
 
I'm getting one as soon as it's available, along with the new Pro Display XDR, but I'm going to try renting instead of buying. There are a couple of companies in the UK who offer this, and I'm thinking it's a better option than buying as you can return the equipment at the end of the lease (two or three years) and swap it for the latest model, including the display. I can't really think of a downside to this as I never sell my old computers and just leave them lying around or give them away.
 
I'm getting one as soon as it's available, along with the new Pro Display XDR, but I'm going to try renting instead of buying. There are a couple of companies in the UK who offer this, and I'm thinking it's a better option than buying as you can return the equipment at the end of the lease (two or three years) and swap it for the latest model, including the display. I can't really think of a downside to this as I never sell my old computers and just leave them lying around or give them away.

The downsides depend upon context and several factors.

If you have the previous generation 'lying around' then you have 'cold' backups or possibly a small farm to throw peak load batch jobs to. If you lease then they goes away at the end of a lease.

Also depends upon the rental rate , business tax rate (and deduction schedule for capital equipment costs. ), and available borrowing rate (and/or available investment cash).

The rental companies have their borrowing rate and a profit margin stuffed into their pricing. If can match or beat their effective rate with a lower borrowing rate then can acquire the backups/farm alluded to above at much lower cost.

It can also have a downside if spending more if the margins and/revenue flow aren't substantive enough cover this. If a $6K screen brings zero increase in real revenue there is a downside. For some narrow workload areas it can make a difference. However, for a large number of ones it won't; even when scoped down to the classic Mac Pro user space. Additionally, if Apple doesn't have a evolutionary better screen ready in 2-3 years , then it is a total downside. You'd likely have to "buy out' the XDR at the end of the lease and that will be higher than just paying for in the first place.

There are rumors that Apple is going to roll out microLED displays and then next discrete monitor could be fantastic in several years. I wouldn't bet the farm on Apple delivering on that in 2-3 year cycle. This Mac Pro 2019 is years late of where it should have been. The iMac Pro will likely stumble into 2020. Probably no microLED on the Watch this year. For a substantive new Mac Pro in 2-3 years there is a decent chance given Apple could make some upgrades to the motherboard and release. Apple sat on the 30" and Thunderbolt Display Docking station for way longer than 3 years. ( 6 and 8 years respectively ). if they revised this in 4 years that would be them greatly picking up in speed. ( something that was more mainstream 3 years is closer to their cycle, but this isn't mainstream at all. They'll have to 'top' this to directly replace it. ). Minus a major change in backlght there isn't much of a "major change" Apple can make in the display to something that is going to be "big leap" better. ( I wouldn't bet on 'more pixels' solely driving them forward on this product in the intermediate term. )


The XDR i'd wait to see what the quality is once at volume production. It is also a "nobody else is using" backlight for which longevity isn't tested. The downside here is get an early long term lease on a display where the hype was higher than the long term performance. The exit terms on the lease could be a substantive downside. On a relatively short term lease ( months to 'try out' on several projects) could have some upside even though it is a bit more money before buying.
 
Probably the current iteration, but not "NOW NOW NOW." As others have pointed out, it's likely that there won't be a new iteration for quite some time. But I'm likely to wait at least 6 months after the new model starts shipping to benefit from the real world experiences of users, many shared on this forum, especially those with similar video editing needs. Being so configurable, there will be lots of choices to find the sweet spot for the various options that'll provide best bang for the buck for video editing (or whatever your uses require). And if there are any serious 1,1 issues, I would guess they'd be evident after about a half year out in the wild.

Bottom line, I'll be a bit behind the leading edge and certainly not on the bleeding edge.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.