The least surprising thing is how full of horsehockey "Darknet Guy" was with its "insider information".
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)
- Thread starter Mago
- Start date
-
- Tags
- macpro 2016
- Sort by reaction score
The least surprising thing is how full of horsehockey "Darknet Guy" was with its "insider information".
Yep. It's a great connector but its main benefit is for mobile and non-tower use cases.
This is a ludicrous statement, bordering on the insane. Please, for the love of god, will people stop with this nonsense.
[doublepost=1560825066][/doublepost]
So, apart from everything? And a whole bunch of parts that cost a lot more?
I'm becoming very concerned that most of the people using this subforum can't even do simply addition or use the Google search bar effectively.
[doublepost=1560825123][/doublepost]
Can you please, in-depth, explain how the machine exactly as it is built & constructed, is overpriced? I'm curious to know.
Really?
Err, this CPU is 2nd from the end of the air flow; not the front. How many fans are attached to your drives (which the J2i doesn't have). Yeah the QUADRO probably do OK but you've pulled desktop GPUs with embedded fans in them. Per area heat density for them is about just as bad as the CPUs.
The 'dead zone" I was referring too was putting electronics and/or media that typically have a much lower heat footprint behind those with orders of magnitude higher footprint. Your example doesn't illustrate that in the slightest. Member of Nvidia fan boy club? Yes. Putting hottest first in flow? No. Neither do the vast majority of severs that are focused on HDD cooling. (e.g., a BackBlaze node 5 design https://www.backblaze.com/blog/cloud-storage-hardware/ ... CPU at the back end of the flow. ). The cooling design from the 80s, classic add-in-cards also have to be in the back because "form over function" demands they have to be on the edge. Their form factor drives that, not what the best solution functionally would be.
P.S. If the heat is all being primarily piped off to a different location (.e.g, loops of coolers and fewer direct flow to logic board fans ) then placement wouldn't matter as much.
[doublepost=1560829134][/doublepost]
If the Rack version goes through the effort to move the top two TB ports from the 'top' of the regular system to the 'front' of the Rack chassis that would be counter indicative that Apple wasn't interested.
Apple not using TB as a broad panacea doesn't mean they are not interested in TB. Every new Mac design from 2015-2019 has gotten TB. With Intel gen 10 Y-series the MacBook may be the last hold-out to go this Fall year. And then the Mac Line up top-to-bottom will have TB.
It's not overpriced - it's a very fair price for an extremely high end workstation. A great deal of the base price is being taken up by the power, cooling, logic and chassis to permit tremendous expandability.
The base configuration is insane if you leave it that way (it's absolutely $3000 worth of performance in a $6000 computer) - the only reason you'd buy a workstation like that is to radically expand one or more areas - whether you buy the expansion from Apple or from someone else. A few rich idiots will buy it and then leave it relatively stock, but that isn't what Apple intended.
It comes with 32 GB of RAM, but it can hold a terabyte or more.
It comes with 256 GB of drive space, but it can hold 48 terabytes of high-speed SSDs easily - use 3 PCIe slots for storage, each holding 4x 4 TB NVMe modules.
It comes with one cheap GPU, but it can hold four high-end GPUs
AND
It comes with an 8-core processor, but can hold a 28-core.
Not everybody needs 28 cores, 4 GPUs, a terabyte of RAM or 48 terabytes of high-speed SSDs.
If you don't need very high levels of one or more of these things (levels that no commodity desktop can reach), Apple didn't build the Mac Pro for you.
They have other options for you - you may not like the designs, but they have a wide range of performance choices from the MacBook on up to the iMac Pro.
They built the Mac Pro in response to a specific group of users who had extreme needs in one or more areas. It doesn't change their fundamental philosophy that sealed computers are more reliable, require less support, and are easier to write stable OS code for. They only offer expansion where nothing else will solve the problem, because they fundamentally don't like it.
The decision Apple has left us with (other than the very small minority who need the Mac Pro) is whether we want to put up with sealed systems where the (generally sensible) hardware decisions are made for us, or whether we want to put up with Windows.
A big part of why Windows is a pain is because it has to support millions of configurations, some of which are low-end junk hardware, and others of which are unstable gaming hardware. Apple, by sealing almost everything, has reduced that number from millions to hundreds (none of which are either junk or running right on the edge).
The base configuration is insane if you leave it that way (it's absolutely $3000 worth of performance in a $6000 computer) - the only reason you'd buy a workstation like that is to radically expand one or more areas - whether you buy the expansion from Apple or from someone else. A few rich idiots will buy it and then leave it relatively stock, but that isn't what Apple intended.
It comes with 32 GB of RAM, but it can hold a terabyte or more.
It comes with 256 GB of drive space, but it can hold 48 terabytes of high-speed SSDs easily - use 3 PCIe slots for storage, each holding 4x 4 TB NVMe modules.
It comes with one cheap GPU, but it can hold four high-end GPUs
AND
It comes with an 8-core processor, but can hold a 28-core.
Not everybody needs 28 cores, 4 GPUs, a terabyte of RAM or 48 terabytes of high-speed SSDs.
If you don't need very high levels of one or more of these things (levels that no commodity desktop can reach), Apple didn't build the Mac Pro for you.
They have other options for you - you may not like the designs, but they have a wide range of performance choices from the MacBook on up to the iMac Pro.
They built the Mac Pro in response to a specific group of users who had extreme needs in one or more areas. It doesn't change their fundamental philosophy that sealed computers are more reliable, require less support, and are easier to write stable OS code for. They only offer expansion where nothing else will solve the problem, because they fundamentally don't like it.
The decision Apple has left us with (other than the very small minority who need the Mac Pro) is whether we want to put up with sealed systems where the (generally sensible) hardware decisions are made for us, or whether we want to put up with Windows.
A big part of why Windows is a pain is because it has to support millions of configurations, some of which are low-end junk hardware, and others of which are unstable gaming hardware. Apple, by sealing almost everything, has reduced that number from millions to hundreds (none of which are either junk or running right on the edge).
I'll go back to the 6,1 for a moment and refer to your last sentence. This is a computer that had well documented over heating and failing GPUs. Is this something covered by the term junk, or by running right on the edge.
Man what a ludicrous post. Apple have, are doing and will continue to release junk that is running right on the edge. All manufacturers do. That single paragraph in your post discredits all the rest.
Also, someone buying it and leaving it stock is an idiot?
As far as I can see, the real problem is that there's a market for a lower end xeon / core i9 machine, with the basic idea of the 2019 mac pro in its design...
In the past, (secondhand) cMPs sufficed as xMac substitutes, but that's going away. And no, a mac mini with an eGPU isn't good enough, and no, an iMac with its built-in screen, isn't good enough. If Apple offered iMac price with MPX bay instead of a display, maybe a 128gb ram ceiling - totally competitive with any number of the Pro non-Xeon workstations I've been looking at of late (Puget, Boxx etc).
- io on a PCI card (toot my own horn, I predicted an all-in on PCI cards, including the basic io being on a removable card, way back when the amigos conference first happened).
- a single MPX bay including the power feeds so you can have one mpx, or a normal gpu with a second free slot.
In the past, (secondhand) cMPs sufficed as xMac substitutes, but that's going away. And no, a mac mini with an eGPU isn't good enough, and no, an iMac with its built-in screen, isn't good enough. If Apple offered iMac price with MPX bay instead of a display, maybe a 128gb ram ceiling - totally competitive with any number of the Pro non-Xeon workstations I've been looking at of late (Puget, Boxx etc).
Last edited:
Why isn’t a Mac mini with eGPU good enough? You’re getting up to 90% of the performance for under half the cost.
As far as I know there are only a few Adobe rgb monitors, there are no ProPhoto monitors, they don't exist. P3 is very close to Adobe, Adobe goes into the greens more, P3 goes into the reds more, but fairly close:
P3 is close to an expanded sRGB, which is the smaller space below::
I see two to three disadvantages to eGPUs:
- expensive because of the enclosure
- performance penalty compared to PCIe
- stability / support issues (unsure about that)
- user upgradable GPU because of the enclosure
- performance penalty of TB3 compared with PCI-E 2.0 16x max. 20% https://translate.google.com/translate?hl=en&sl=de&tl=en&u=https://www.computerbase.de/2011-08/test-grafikkarten-mit-pcie/4/&sandbox=1
- stability / support issues - not yet with my eGPU (knock on wood). The vulnerability of TB3 is the connector. With a stationary Mac Mini you don't have to reconnect/connect the eGPU again and again as with a Laptop.
Last edited:
because it's not 90% of the performance in the things I want a GPU to do. Thunderbolt chokes on lower total screen resolutions than most GPUs can support, and it's significantly worse performance for realtime 3D environments, which is the difference between viable, and vomiting, for VR.
eGPU is OK for non-display purposes. For anything involving a display card doing displaying, it's a secondrate option.
[doublepost=1560867216][/doublepost]
Dell claims 100% Adobe RGB for their 27" 4k FALD display. Benq, EIZO & NEC are all up around 97-99%
$6,000 gets you:
8 cores in 2019.
Polaris video card in 2019.
256 Gb SSD storage in 2019.
PCIe 3.0 in 2019.
The case design is a Sir Idiot Boy I just learned about the Laws of Thermodynamics case design.
The case design is expensive, because they won't be selling very many of these - people that buy them will be getting high core count Xeons, and that will drive the price into 5 figures.
Now, if they used the exact same case design and added a motherboard with 4 PCIe slots and 1 MPX module (which will take up 2 PCIe slots), they could make a Mac for the rest of us.
They could sell that for the same price as a tcmp, and they would sell enough to drive down the cost of development, not only for the case design, but also sell enough to justify 3rd parties building MPX modules.
[doublepost=1560871628][/doublepost]
Max 6 cores/64Gb of ram. And it throttles under load - the same problem with the iMac.
Yes, of course, ProPhoto is only a working colour space, an excellent one.In general, it is ideal that your working space is ProPhoto RGB when you edit a RAW photo.
I do not think that there are only a few Adobe RGB monitors, every company with a good reputation has some offerings, of course at the high end with this colour space.
P3 is better than sRGB and a step up but still inferior, a bit, than AdobeRGB. If I remember well AdobeRGB is better for printing environments, P3 is better for digital movies.
so, it depends on your workflow and what are you using this monitor for, of course it is possible to use a P3 display for printing too.
Anyway, I'm really curious to check the reviews of this $6000 display regarding calibration.
I've heard this a lot (and don't own either so can't do my own tests), do they actually thermal throttle or are they just thermally constrained?
ie: do they throttle to below base clock under normal/heavy load or is it just that they can't sustain boost clocks?
The former would be an actual 'not fit for purpose' issue, where as the latter is just a disappointing lack of forethought in design and intent but not actually a fault as such...
Don't most eGPUs allow for direct-connection of an external monitor? I have seen pictures of Blackmagic eGPUs using the MacBook Pro's display, but the BM does have HDMI and DisplayPort connectors for external 4K and 5K displays.
Yes. They are mostly a graphics card in a box, with whatever outputs the particular graphics card may have.
So we should judge that it's not overpriced ... but is insane at the same time?
Yes, it's expandable, but in such a highend workstation (and we don't use HP or dell stuff here, mostly produced by someone else, I have no idea who) because I see no logos, this level of slots is not unknown in my realm.
You just said it's 'insane' for the base config: can you clarify... that it is indeed overpriced then, UNLESS you plan to expand it? Expand it with what though? Most industry professionals aren't likely to fill up all of the slots that I can think of, unless you have three afterburner cards.. if software can even use it, or whatever.
We haven't seen configuration pricing, only base pricing, but my guesstimate would be that a CPU upgrade would cost another 1.3-1.6k more to 12core. Maybe even 2k.
Whether it's overpriced or not depends on who you are asking.
And it's too early to ask - since we only have a price for the base configuration.
Make it stop...we've just barely started the PCIe 4.0 transition...I feel like this is PCIe's USB moment...
Somewhere, in a darkened conference room,
"So we're calling it 3.1 Gen2, no wait...that's already taken. USB 3.2, that's taken too...well, wait, USB 3.2x2!!!"
"Why not 4X4, that sounds more robust...no?"
"Well screw it, just call it USB 4.0, or should we call it Thunderbolt 4?"
"We're not sure yet? Why? Who makes the final decision?"
"Maybe we should we ask Intel, Apple, the Shriners...somebody else, first?"
"Nah, let's just issue a Press Release?!?"
"I don't think that's a good idea..."
"Press release issued! All done!"
"Sir...a Mr. Cook is on the phone for you. Shall I send it to your office or the conference room?"
"Uh....."
https://appleinsider.com/articles/1...le-the-bandwidth-of-pcie-50-to-256gbs-in-2021
https://www.businesswire.com/news/home/20190618005945/en/
Somewhere, in a darkened conference room,
"So we're calling it 3.1 Gen2, no wait...that's already taken. USB 3.2, that's taken too...well, wait, USB 3.2x2!!!"
"Why not 4X4, that sounds more robust...no?"
"Well screw it, just call it USB 4.0, or should we call it Thunderbolt 4?"
"We're not sure yet? Why? Who makes the final decision?"
"Maybe we should we ask Intel, Apple, the Shriners...somebody else, first?"
"Nah, let's just issue a Press Release?!?"
"I don't think that's a good idea..."
"Press release issued! All done!"
"Sir...a Mr. Cook is on the phone for you. Shall I send it to your office or the conference room?"
"Uh....."
https://appleinsider.com/articles/1...le-the-bandwidth-of-pcie-50-to-256gbs-in-2021
https://www.businesswire.com/news/home/20190618005945/en/
Tell me if you are silly enough to buy an 8 core machine, what the hell would you need 7 PCIe slots for? Makes no sense
Are we still barking the same story of mp price? Shouldn’t we thank apple for mp? After more than 6 years?
Not really the same.
There are essentially switching to a new technology.
"... Instead, the PCI-SIG is going to upend the signaling technology entirely, moving from the Non-Return-to-Zero (NRZ) tech used since the beginning, and to Pulse-Amplitude Modulation 4 (PAM4). ... "
https://www.anandtech.com/show/1455...-again-pcie-60-announced-spec-to-land-in-2021
This really isn't the same thing flogged harder and faster. And there are other couplings that are helping driving this.
"... Furthermore in the last few years their efforts have taken on an increased level of importance as well, as other major interconnect standards are building off of PCIe. CCIX, Intel’s CXL, and other interfaces have all extended PCIe, and will in turn benefit from PCIe improvements. So PCIe speed boosts serve as the core of building ever-faster (and more interconnected) systems. ... "
The notion that PCI-e v6 is going to quickly roll down to mainstream desktop process for end user cards and/or upgrades that is probably missing the point. PCI-e v6 will probably get pushed into a role closer to what QPI or Infinity Fabric plays. Coupling major system components and not expansion gadget for "John Smith" shopping in the discount aisle at Frys or newegg.
PCI-e v5 is already in the range of faster bandwidth than the QPI link coupling the two CPU packages in the old 2009-2010 era Mac Pros. PCI-e v6 is step beyond that. At 128GB/s that's faster than NVLink v1 and reasonably close to NVLink v2.
As mentioned in the article.
"... , because of the additional signal states a PAM4 signal itself is more fragile than a NRZ signal. And this means that along with PAM4, for the first time in PCIe’s history the standard is also getting Forward Error Correction (FEC) ..."
FEC isn't going to save bad , sloppy implementations . Where I would expect to see PCI-e v6 first is in intra package multiple chip modules and not as "standard form factor" slots. This probably doesn't have a big impact on path Mac Pro design is on at this point for more than several years.
An eGPU is (as far as I can tell from researching it) effectively constrained by Thunderbolt - the connection between the computer and the eGPU, is narrower than the connection between the eGPU (or motherboard PCI slot) and the card. For example the RX580 Blackmagic can drive 2 displays total, whereas an RX580 in a PCI slot can drive 4x4K displays or a 5/8K dual-cable and two 4Ks.
I've read one description that the RX 5xx series can theoretically drive 6x5k displays.
Point being, you can't feed the card enough data over Thunderbolt, to reach it's maximum ability to use that data to drive display tasks.
Ok, but your points do not exactly refute my concerns. If PCIe is user upgradeable, why have the workaround of an eGPU? 20% performance penalty is 20% too much. Or so you pay 20% less for a card when you use it via TB?