The thermal throttling is a huge problem in a professional machine. Sadly AI is seeing throttling in extremely light usage imagine how much you would loose over 8 hours.
Frankly this is not unexpected! Apples history with the word "pro" and cramming hot parts into a tight enclosure isnt good. Every day im becoming more and more convinced that Apple just doesnt understand the "PRO" market.
As a point of record i was looking at a iPad Pro in a store yesterday. Nice device but there is nothing about it that stands out as being pro. I do believe that common sense has left the building at Apple and has been replaced by marketing morons that likely have never engaged in professional work. Sad.
"Apples history with the word "pro" and cramming hot parts into a tight enclosure isnt good. Every day im becoming more and more convinced that Apple just doesnt understand the "PRO" market"
To their credit, they are hitting the restart button the Mac Pro. So we'll have to wait and see until that gets released to see if you're correct (or not).
It is more speculation at this point until we can get testing done in a way that replicates what a pro would do all day at the machine. I only brought up the concept because AI was seeing throttling in very basic and light usage.
As for the coming Mac Pro; well I've had ideas for a long time on how that should unfold but here is the problem, Apple user base is too small and too fractured for a ground swell of happiness out of any new Mac Pro design. Frankly I didn't think that the trash can was that bad of a design and frankly don't understand the lack of upgrades! The excuses coming out of Apple don't make any sense either as suitable new chips did arrive that had more performance per watt especially with regards to the GPU? All that talk about thermal limitations don't pan out if you look at Intels chip lineup, Same thing for GPU's. I'm strongly in favor of a Mac Pro with a far lower entry point price wise to pull in more users to the platform with a wide spread in possible performance improvements on up sell models. In other words Apple has to get more volume out of a Mac Pro chassis to pay for development costs. More so they need to keep revising the machine instead of these years long delays in model revamps. Same thing goes for the Mini, I'm sitting at a PC with an old Dell Model Dx0D, from 2015, that is in many ways a better Mini that Apples Mini.
Personally I think somebody at Apple got a bug up their ass with respect to the desktop and decided to ignore the whole area for years. When it became obvious that this was hurting them, they started to back pedal but had nothing in the line up for any of the markets targeted by Mini, the iMac and Mac Pro. The iMac was simply the platform they could address the quickest thus the iMac Pro. Given that I'm not sure they can even design machines anymore that will meet user needs or set them apart from generic machines running Linux. Why they haven't come out with a ARM based laptop is beyond me.
What a crackpot theory. So Craig and co are outright lying to the public, to the media, to their customers, about having painted themselves into a thermal corner and the problem with going with parallel processing when the rest of the industry didn't? And the secret truth is they just felt like desktops were a waste of time, but now for some reason, despite the very small percentage of their business it represents, are getting back into it but fabricating an elaborate lie?
Yeah you got one thing right -- you don't understand.
It seems that it took them a long time to get around to fixing the issue.
Apple isn’t even using a W-2145, let alone downclocking one.
The W-2140B is essentially a downclocked version of the W-2145. The same way the W-2150B is a downclocked W-2155.
These are straight up weaker versions of their regular counterparts.
That said, I’m confused about the comments re: thermal throttling. From the article:
In the multi-core benchmark, the 8 cores ran at 3.9 gigahertz, which seems to be the top CPU frequency when maxing out all CPU cores.
and, during 10 consecutive multi core tests:
After the second test, each additional run would cause the iMac Pro to thermal throttle when the CPU reached roughly 94 degrees celsius, which caused the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz.
The base frequency of this processor is only 3.2GHz. So multicore performance seems well beyond spec.
That doesn’t explain the single core at 3.9 but if it can do 3.6 to 3.9 with all cores, 4.2 with single core would seem to be attainable wrt thermals. There may be some further optimizations possible, trading off fan speed (which was described as inaudible and seemingly near idle) with maximum clockspeed under various load conditions.
Looks very promising so far.
Well beyond spec? What are you talking about. Turbo should be sustained. None of this looks promising.
Even the overall performance is disappointing. A 1950X scores 3100 in Cinebench R15, meanwhile, the W-2140B scored 1680.
Apple isn’t even using a W-2145, let alone downclocking one.
The W-2140B is essentially a downclocked version of the W-2145. The same way the W-2150B is a downclocked W-2155.
These are straight up weaker versions of their regular counterparts.
That said, I’m confused about the comments re: thermal throttling. From the article:
In the multi-core benchmark, the 8 cores ran at 3.9 gigahertz, which seems to be the top CPU frequency when maxing out all CPU cores.
and, during 10 consecutive multi core tests:
After the second test, each additional run would cause the iMac Pro to thermal throttle when the CPU reached roughly 94 degrees celsius, which caused the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz.
The base frequency of this processor is only 3.2GHz. So multicore performance seems well beyond spec.
That doesn’t explain the single core at 3.9 but if it can do 3.6 to 3.9 with all cores, 4.2 with single core would seem to be attainable wrt thermals. There may be some further optimizations possible, trading off fan speed (which was described as inaudible and seemingly near idle) with maximum clockspeed under various load conditions.
Looks very promising so far.
Well beyond spec? What are you talking about. Turbo should be sustained. None of this looks promising.
Even the overall performance is disappointing. A 1950X scores 3100 in Cinebench R15, meanwhile, the W-2140B scored 1680.
1) A W-2145 is a 3.7/4.5 part, the W-2140B is a 3.2/4.2 part. Are you saying Apple should also offer the faster, more expensive chip? Fine, but that would be a $5,200 or $5,300 machine, not $5,000. You don’t get something for nothing. Apple is unlikely to offer 2 different speeds of 8-core though, so don’t hold your breath.
With your logic, you might just as well say Apple should only offer the 18-core at $7,400.
2) The part is spec’ed at 3.2GHz base frequency. In 2 different multicore tests, it benched at either 3.6-3.9GHz or a solid 3.9GHz.
A 3.2GHz part running at 3.9GHz with all cores at 100% is promising. The fan seemed not to be running. As I said, with an optimization of fan speed, 4.2GHz on one core would seem to be within the thermal capacity, since it’s easily beating its 3.2GHz base speed—upon which TDP is defined.
3) Mac workstations are typically purchased by those who want to use MacOS, so your comparison of this 8-core machine to a 16-core machine, which can’t run MacOS, isn’t particularly relevant. TCO isn’t maximized by simply buying the cheapest hardware available.
No one would notice that minus 0.3 GHz in real life usage. The stock throttles more than that 0.3 GHz by the way...
“the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz”
The stock Xeon W-2145 is supposed to turbo to 4.5 GHz and will not throttle. Most desktop cooling solutions can likely sustain even higher frequencies without throttling.
Apple's solution is downclocked and still throttles.
Turbo is for single core, dude...
So your (mis)understanding "how may it throttle at 3.9 while it can Turbo at 4.5" doesn't make sense, you compare apples to oranges.
Putting a full load on just a single core, the clock speed varied from 3.94 to 3.98 GHz -- a few hundred hertz away from the max Intel Turbo Boost clock speed of 4.2GHz.
Actually, the difference between 3.98GHz and 4.2GHz is 220 megahertz, not "a few hundred hertz."
The idea of keeping the iMac Pro quiet as opposed to allowing heat to escape is quite upsetting..
Quite the opposite, I think Apple has got this aspect perfectly correct: the iMac enclosure is NOT a PRO enclosure ... it is a consumer all-in-one enclosure that sits 30cm from your head, so the last thing you want is (i) it making noise next to you, and (ii) heating up internally and damaging your new $5000 computer's internals. It is not upsetting - it is common sense for the enclosure it is in.
If you want a Pro machine, get a Mac Pro ... even with the latters constrained design, the R2D2 Mac Pro has vastly better thermal performance than the iMac Pro (and it doesn't sit next to your head) ... and the only reason it is not as fast as the iMac Pro is because Apple has not updated the Pro (choice - not design) in 4 years! 4 years!! The Mac Pro could easily be updated by Apple with new modern graphics cards, 10G ethernet, faster memory, and it would scream the pants off the iMac Pro, plus have better thermals to support the performance you seek.
Face it ... this is not a Pro model, nor a Pro enclosure. It does offer fantastic performance in a consumer enclosure - and that may be an attractive solution for a certain market segment.
Note: Intel® Turbo Boost Technology 2.0 allows the processor to operate at a power level that is higher than its TDP configuration and data sheet specified power for short durations to maximize performance.
Presumably it's safe to consider Intel, as designer and manufacturer of Xeon processors, a company that's been designing and manufacturing microprocessors for nearly fifty years, a better authority on the matter.
Apple isn’t even using a W-2145, let alone downclocking one.
The W-2140B is essentially a downclocked version of the W-2145. The same way the W-2150B is a downclocked W-2155.
These are straight up weaker versions of their regular counterparts.
That said, I’m confused about the comments re: thermal throttling. From the article:
In the multi-core benchmark, the 8 cores ran at 3.9 gigahertz, which seems to be the top CPU frequency when maxing out all CPU cores.
and, during 10 consecutive multi core tests:
After the second test, each additional run would cause the iMac Pro to thermal throttle when the CPU reached roughly 94 degrees celsius, which caused the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz.
The base frequency of this processor is only 3.2GHz. So multicore performance seems well beyond spec.
That doesn’t explain the single core at 3.9 but if it can do 3.6 to 3.9 with all cores, 4.2 with single core would seem to be attainable wrt thermals. There may be some further optimizations possible, trading off fan speed (which was described as inaudible and seemingly near idle) with maximum clockspeed under various load conditions.
Looks very promising so far.
Well beyond spec? What are you talking about. Turbo should be sustained. None of this looks promising.
Even the overall performance is disappointing. A 1950X scores 3100 in Cinebench R15, meanwhile, the W-2140B scored 1680.
1) A W-2145 is a 3.7/4.5 part, the W-2140B is a 3.2/4.2 part. Are you saying Apple should also offer the faster, more expensive chip? Fine, but that would be a $5,200 or $5,300 machine, not $5,000. You don’t get something for nothing. Apple is unlikely to offer 2 different speeds of 8-core though, so don’t hold your breath.
With your logic, you might just as well say Apple should only offer the 18-core at $7,400.
2) The part is spec’ed at 3.2GHz base frequency. In 2 different multicore tests, it benched at either 3.6-3.9GHz or a solid 3.9GHz.
A 3.2GHz part running at 3.9GHz with all cores at 100% is promising. The fan seemed not to be running. As I said, with an optimization of fan speed, 4.2GHz on one core would seem to be within the thermal capacity, since it’s easily beating its 3.2GHz base speed—upon which TDP is defined.
3) Mac workstations are typically purchased by those who want to use MacOS, so your comparison of this 8-core machine to a 16-core machine, which can’t run MacOS, isn’t particularly relevant. TCO isn’t maximized by simply buying the cheapest hardware available.
1) You're the one that's not understanding. The W-2145 is the standard model that will be used in every other Xeon W workstation. Apple's version is just a downclocked version of that chip. This has nothing to do with cost, which, by the way, likely costs more for Apple as they're not using the off the shelf variant. Unless of course they're just using the portion of the yields that couldn't become a full W-2145.
2) Nothing about that is well beyond spec. Sustaining speeds above the base clock should be expected for a desktop with a decent cooling + a PSU that draws power directly from the wall.
3) The point is that the performance as a workstation is underwhelming. Just because a macOS user has no other options is simply unfortunate.
No one would notice that minus 0.3 GHz in real life usage. The stock throttles more than that 0.3 GHz by the way...
“the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz”
The stock Xeon W-2145 is supposed to turbo to 4.5 GHz and will not throttle. Most desktop cooling solutions can likely sustain even higher frequencies without throttling.
Apple's solution is downclocked and still throttles.
Turbo is for single core, dude...
So your (mis)understanding "how may it throttle at 3.9 while it can Turbo at 4.5" doesn't make sense, you compare apples to oranges.
The W-2145 has a boost speed of 4.5 GHz, the W-2140B has a boost speed of 4.2 GHz. <-- The downclock.
The W-2140B is seen throttling down to 3.6 GHz at sustained load. <-- The throttling.
I would love to see benchmarks for Motion since it's the main app where I find performance deficiencies in my existing machine (late 2014 27" iMac).
I was expecting to see problems with 4K Final Cut Pro editing with my Nikon D5 footage but found it very smooth on both machines. I could tell the iMac Pro was faster but the slower speed of my current iMac was not different enough for it to be conspicuous. It's odd because I remember editing being stuttery in previous attempts. Maybe my LaCie 4TB Rugged RAID Thunderbolt external drive is faster than previous drives I've used.
Note: Intel® Turbo Boost Technology 2.0 allows the processor to operate at a power level that is higher than its TDP configuration and data sheet specified power for short durations to maximize performance.
Presumably it's safe to consider Intel, as designer and manufacturer of Xeon processors, a company that's been designing and manufacturing microprocessors for nearly fifty years, a better authority on the matter.
That "short duration" is limited by the power and cooling, which shouldn't be an issue on a desktop workstation.
In context with you're quote, it would apply more to a laptop.
One thing that will be interesting in the better reviews will be discussion of the W-2140B/2150B that appear to be in use here.
The question is why? Obviously heat due to the constraints of the iMac form factor, but there may be other considerations, like an IGPU capability -- this was speculated upon back in October when PikerAlpha first reported it. [Sadly, his wife was hit by a car and died, and his blog is done for the foreseeable future.]
I don't think the engineering decisions made in the iMac Pro are fully understood yet -- it's premature for VRing to claim he/she has any real idea. My sense is that Apple is squeezing everything it can out of the form factor, and when you compare it like-to-like, nothing is going to come close to it for now. That said, the speed with which HP or Dell or one of the other serious players comes out with a competitor will say a lot about the emerging market for these things.
One thing that will be interesting in the better reviews will be discussion of the W-2140B/2150B that appear to be in use here.
The question is why? Obviously heat due to the constraints of the iMac form factor, but there may be other considerations, like an IGPU capability -- this was speculated upon back in October when PikerAlpha first reported it. [Sadly, his wife was hit by a car and died, and his blog is done for the foreseeable future.]
I don't think the engineering decisions made in the iMac Pro are fully understood yet -- it's premature for VRing to claim he/she has any real idea. My sense is that Apple is squeezing everything it can out of the form factor, and when you compare it like-to-like, nothing is going to come close to it for now. That said, the speed with which HP or Dell or one of the other serious players comes out with a competitor will say a lot about the emerging market for these things.
1. Product segmentation. Apple follows this religiously, where the value per dollar for each SKU is basically constant for the majority of the market, or the value in the model they want to sell most, the upsell model, is higher than the base model. Sounds like they want to sell the 10-core model.
2. The 4.2 GHz (or whatever it is as it is not clear) vs the 4.5 GHz max turbo is basically designer’s choice. The 14-core and 18-core models have max Intel advertised turbos of 4.3 GHz. If those models are demonstrated to have higher turbo frequencies, it’s definitely product segmentation as Apple wants you pay more for higher single threaded performance in these machines.
I don’t think is it number 2 as nobody are buying these machines for the best single threaded performance. They are close enough at 4 GHz max turbo. The iMac Pro is for people who can use 8 or more cores, need >64 GB RAM, 3 5K monitors, 4 TB SSDs, can use the Vega GPUs, or some combination thereof. Single threaded performance is but a small feature playing into the buying decision here, and just needs to be competitive, and they basically are.
Apple desktop hardware strategy is rather mysterious. I think they just fucked up. Whoever drove the 2013 Mac Pro and Apple branded display decisions was high enough in the management chain to drive those decisions through; and, they have to reset as the market has obviously proven those decisions drastically wrong.
They could have redesigned the 2013 Mac Pro internals to support modern components if they wanted to. They just didn’t for whatever reason. Who knows why they waited so long to reset.
Note: Intel® Turbo Boost Technology 2.0 allows the processor to operate at a power level that is higher than its TDP configuration and data sheet specified power for short durations to maximize performance.
Presumably it's safe to consider Intel, as designer and manufacturer of Xeon processors, a company that's been designing and manufacturing microprocessors for nearly fifty years, a better authority on the matter.
That "short duration" is limited by the power and cooling, which shouldn't be an issue on a desktop workstation.
In context with you're quote, it would apply more to a laptop.
These Xeons aren't for laptops but desktops have their power restraints too: even the largest of supercomputers have a finite amount of power to draw upon.
Intel® Turbo Boost Technology 2.01 accelerates processor and graphics performance for
peak loads, automatically allowing processor cores to run faster than the rated operating frequency if they’re operating below power, current, and temperature specification limits.
I'm a bit disappointed to hear about the thermal throttling which I was sure would happen in such a smallish case as the iMac Pro is sporting. The idea of keeping the iMac Pro quiet as opposed to allowing heat to escape is quite upsetting. Personally, I'll take the noise instead of throttling because I'm not in an office setting. I had always somewhat expected this to happen. It still won't stop me from buying an iMac Pro because I won't be taxing it nearly as much and as often but I think it's going to force many potential buyers to take a pass on it. I don't think Apple will ever attempt to beat the Windows platform and I find that rather sad how no one at Apple has the pride to butt-hurt Windows desktops.
I assure everyone I have no intention of abandoning Apple as a user or shareholder but I continue to think Apple could do a lot better in terms of performance desktop design with the amount of money they have at their disposal. I know for certain it takes a large case with plenty of fans to stop thermal throttling and I only hoped Apple had found some perfect solution. They didn't find that perfect solution and that's the end of that. Jony Ive might be some hotshot designer but he sure can't beat the laws of thermodynamics. Would it really have hurt Apple that much if they went with a slightly larger case with larger fans for the iMac Pro? I'm going to try my best to get a 10-core that won't have to work as hard. It will still be more powerful than any CTO iMac and will surely last me for more than five satisfying years.
1. Yes, it would have hurt to change the case design from the iMac. By using the same enclosure, they can use the iMac assembly line and tooling. The internal design also looks easier and cheaper to build than the 2013 Mac Pro.
2. Apple has repeatedly said they are also offering a modular and upgradeable desktop Mac in the future. They’d like the customers for this machine to wait, but obviously you shouldn’t if you need to buy now. Just get a PC that’ll meet your needs. Apple loses a customer. They know that just as much as you.
3. Apple trades performance for less noise all the time. The iMac has to sit on a desk where you are working. So, less noise is a feature. Desktops can be put in the closet or farther away with some sound isolation, so more noise with headless machines is something that doesn’t hurt.
4. You are arguing over 10% single threaded performance. Not sure that is noticeable. Getting software that can utilize 8 to 10 cores on the other hand, will be a big win. Customers have to know that their workflows can utilize the cores or GPU, or they need 3 5K monitors.
Note: Intel® Turbo Boost Technology 2.0 allows the processor to operate at a power level that is higher than its TDP configuration and data sheet specified power for short durations to maximize performance.
Presumably it's safe to consider Intel, as designer and manufacturer of Xeon processors, a company that's been designing and manufacturing microprocessors for nearly fifty years, a better authority on the matter.
That "short duration" is limited by the power and cooling, which shouldn't be an issue on a desktop workstation.
In context with you're quote, it would apply more to a laptop.
These Xeons aren't for laptops but desktops have their power restraints too: even the largest of supercomputers have a finite amount of power to draw upon.
Intel® Turbo Boost Technology 2.01 accelerates processor and graphics performance for
peak loads, automatically allowing processor cores to run faster than the rated operating frequency if they’re operating below power, current, and temperature specification limits.
b) that your understanding of how Xeon processors operate is better than Intel's, the company that designed and manufactured them.
The "Turbo Boost" link is a general link that clearly states temperature, current and power as aspects that determine the boost. These are not constrained on most workstation systems. Both cooling and the power supply will be more than adequate to maintain long term boost performance. Point being, you're wrong in your interpretation.
No one would notice that minus 0.3 GHz in real life usage. The stock throttles more than that 0.3 GHz by the way...
“the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz”
The stock Xeon W-2145 is supposed to turbo to 4.5 GHz and will not throttle. Most desktop cooling solutions can likely sustain even higher frequencies without throttling.
Apple's solution is downclocked and still throttles.
Turbo is for single core, dude...
So your (mis)understanding "how may it throttle at 3.9 while it can Turbo at 4.5" doesn't make sense, you compare apples to oranges.
The W-2145 has a boost speed of 4.5 GHz, the W-2140B has a boost speed of 4.2 GHz. <-- The downclock.
The W-2140B is seen throttling down to 3.6 GHz at sustained load. <-- The throttling.
Stop flooding the forum with meaningless numbers. You said “turbo must be sustained” above and that’s enough. You have transcended even Intel and I wish you a good Turbo life. I’m done with your posts.
No one would notice that minus 0.3 GHz in real life usage. The stock throttles more than that 0.3 GHz by the way...
“the clock speed to drop from 3.9GHz to about 3.6GHz for a second or two. This allowed the CPU to drop below 92 degrees, and the clock speed to rise back to the maximum 3.9GHz”
The stock Xeon W-2145 is supposed to turbo to 4.5 GHz and will not throttle. Most desktop cooling solutions can likely sustain even higher frequencies without throttling.
Apple's solution is downclocked and still throttles.
Turbo is for single core, dude...
So your (mis)understanding "how may it throttle at 3.9 while it can Turbo at 4.5" doesn't make sense, you compare apples to oranges.
The W-2145 has a boost speed of 4.5 GHz, the W-2140B has a boost speed of 4.2 GHz. <-- The downclock.
The W-2140B is seen throttling down to 3.6 GHz at sustained load. <-- The throttling.
Stop flooding the forum with meaningless numbers. You said “turbo must be sustained” above and that’s enough. You have transcended even Intel and I wish you a good Turbo life. I’m done with your posts.
"Should" is what I said, and yes, it should be sustained with an adequate cooling solution and power supply found in a reputable desktop workstation.
So you're saying you don't understand what the numbers mean and because it proves you wrong you dismiss my posts. Yikes!
Personally I think somebody at Apple got a bug up their ass with respect to the desktop and decided to ignore the whole area for years.
That would have been Steve Jobs.
Jobs died in 2011, two years before Apple replaced the fully upgradable, very ventilated, sufficiently fan-cooled, dongle-free, and never under-clocked Mac Pro (the last truly Pro workstation they built). That was also the last year that they upgraded notebook RAM to 16GB, which has not increased since then. Adding insult to injury, listening to Phil Schiller tell us that the same amount of RAM is faster when we're all running out of RAM is like someone telling me I need a faster motorcyle to drive 3 of my friends to the beach. This is 100% Tim, Phil and Jony. Craig is the hero we need.
Comments
These are straight up weaker versions of their regular counterparts. Well beyond spec? What are you talking about. Turbo should be sustained. None of this looks promising.
Even the overall performance is disappointing. A 1950X scores 3100 in Cinebench R15, meanwhile, the W-2140B scored 1680.
With your logic, you might just as well say Apple should only offer the 18-core at $7,400.
2) The part is spec’ed at 3.2GHz base frequency. In 2 different multicore tests, it benched at either 3.6-3.9GHz or a solid 3.9GHz.
A 3.2GHz part running at 3.9GHz with all cores at 100% is promising. The fan seemed not to be running. As I said, with an optimization of fan speed, 4.2GHz on one core would seem to be within the thermal capacity, since it’s easily beating its 3.2GHz base speed—upon which TDP is defined.
3) Mac workstations are typically purchased by those who want to use MacOS, so your comparison of this 8-core machine to a 16-core machine, which can’t run MacOS, isn’t particularly relevant. TCO isn’t maximized by simply buying the cheapest hardware available.
So your (mis)understanding "how may it throttle at 3.9 while it can Turbo at 4.5" doesn't make sense, you compare apples to oranges.
Actually, the difference between 3.98GHz and 4.2GHz is 220 megahertz, not "a few hundred hertz."
If you want a Pro machine, get a Mac Pro ... even with the latters constrained design, the R2D2 Mac Pro has vastly better thermal performance than the iMac Pro (and it doesn't sit next to your head) ... and the only reason it is not as fast as the iMac Pro is because Apple has not updated the Pro (choice - not design) in 4 years! 4 years!! The Mac Pro could easily be updated by Apple with new modern graphics cards, 10G ethernet, faster memory, and it would scream the pants off the iMac Pro, plus have better thermals to support the performance you seek.
Face it ... this is not a Pro model, nor a Pro enclosure. It does offer fantastic performance in a consumer enclosure - and that may be an attractive solution for a certain market segment.
from Intel:
Presumably it's safe to consider Intel, as designer and manufacturer of Xeon processors, a company that's been designing and manufacturing microprocessors for nearly fifty years, a better authority on the matter.
2) Nothing about that is well beyond spec. Sustaining speeds above the base clock should be expected for a desktop with a decent cooling + a PSU that draws power directly from the wall.
3) The point is that the performance as a workstation is underwhelming. Just because a macOS user has no other options is simply unfortunate.
The W-2140B is seen throttling down to 3.6 GHz at sustained load. <-- The throttling.
I was expecting to see problems with 4K Final Cut Pro editing with my Nikon D5 footage but found it very smooth on both machines. I could tell the iMac Pro was faster but the slower speed of my current iMac was not different enough for it to be conspicuous. It's odd because I remember editing being stuttery in previous attempts. Maybe my LaCie 4TB Rugged RAID Thunderbolt external drive is faster than previous drives I've used.
In context with you're quote, it would apply more to a laptop.
The question is why? Obviously heat due to the constraints of the iMac form factor, but there may be other considerations, like an IGPU capability -- this was speculated upon back in October when PikerAlpha first reported it. [Sadly, his wife was hit by a car and died, and his blog is done for the foreseeable future.]
I don't think the engineering decisions made in the iMac Pro are fully understood yet -- it's premature for VRing to claim he/she has any real idea. My sense is that Apple is squeezing everything it can out of the form factor, and when you compare it like-to-like, nothing is going to come close to it for now. That said, the speed with which HP or Dell or one of the other serious players comes out with a competitor will say a lot about the emerging market for these things.
1. Product segmentation. Apple follows this religiously, where the value per dollar for each SKU is basically constant for the majority of the market, or the value in the model they want to sell most, the upsell model, is higher than the base model. Sounds like they want to sell the 10-core model.
2. The 4.2 GHz (or whatever it is as it is not clear) vs the 4.5 GHz max turbo is basically designer’s choice. The 14-core and 18-core models have max Intel advertised turbos of 4.3 GHz. If those models are demonstrated to have higher turbo frequencies, it’s definitely product segmentation as Apple wants you pay more for higher single threaded performance in these machines.
I don’t think is it number 2 as nobody are buying these machines for the best single threaded performance. They are close enough at 4 GHz max turbo. The iMac Pro is for people who can use 8 or more cores, need >64 GB RAM, 3 5K monitors, 4 TB SSDs, can use the Vega GPUs, or some combination thereof. Single threaded performance is but a small feature playing into the buying decision here, and just needs to be competitive, and they basically are.
Apple desktop hardware strategy is rather mysterious. I think they just fucked up. Whoever drove the 2013 Mac Pro and Apple branded display decisions was high enough in the management chain to drive those decisions through; and, they have to reset as the market has obviously proven those decisions drastically wrong.
They could have redesigned the 2013 Mac Pro internals to support modern components if they wanted to. They just didn’t for whatever reason. Who knows why they waited so long to reset.
These Xeons aren't for laptops but desktops have their power restraints too: even the largest of supercomputers have a finite amount of power to draw upon.
So you're telling us that:
a) you've not read the link I provided to Intel's explanation as to how they intend their Turbo Boost technology to work:
here it is directly in case you didn't realise I had embedded it:
https://www.intel.com/content/www/us/en/architecture-and-technology/turbo-boost/turbo-boost-technology.html
b) that your understanding of how Xeon processors operate is better than Intel's, the company that designed and manufactured them.
1. Yes, it would have hurt to change the case design from the iMac. By using the same enclosure, they can use the iMac assembly line and tooling. The internal design also looks easier and cheaper to build than the 2013 Mac Pro.
2. Apple has repeatedly said they are also offering a modular and upgradeable desktop Mac in the future. They’d like the customers for this machine to wait, but obviously you shouldn’t if you need to buy now. Just get a PC that’ll meet your needs. Apple loses a customer. They know that just as much as you.
3. Apple trades performance for less noise all the time. The iMac has to sit on a desk where you are working. So, less noise is a feature. Desktops can be put in the closet or farther away with some sound isolation, so more noise with headless machines is something that doesn’t hurt.
4. You are arguing over 10% single threaded performance. Not sure that is noticeable. Getting software that can utilize 8 to 10 cores on the other hand, will be a big win. Customers have to know that their workflows can utilize the cores or GPU, or they need 3 5K monitors.
So you're saying you don't understand what the numbers mean and because it proves you wrong you dismiss my posts. Yikes!