I'm aware there are other factors. However I was addressing this one factor because the individual I responded to was commenting on it (who was also supporting your position on it).
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
You keep making the same mistake. We are not talking about the same type or amount of damage or downtime every time. People that had a bad PSU taking _everything_ with it know. Nevertheless, if we are not speaking about exotic (external or internal) RAID solutions, having the electrics of external storage fail might not always be that catastrophic; the (SATA) HDDs might still be accessible through a so-cheap/easy-to-find USB enclosure or even internally through SATA (or using another computer). Plus we are talking Apple (branded) PSUs here. Not only they are NOT the best around, they are hard to find (not everyone lives in USA where there is a repair center in the next). The AIO power/storage solution does not have any redundancy neither scales well.
$150 is about what you'd pay for Apple's keyboard and mouse.
Why does the the keyboard/mouse/monitor have to come from the Apple store (website or not).? If on a tight budget that is about the last place most would look. If budget is tight, used is an option (i.e., , e-bay ). Or just find an older friend/family that bought a new computer ( they'll probably have an 'extra' keyboard). [ even some folks coming from Hackintosh because that was all they could afford before ( e.g., some bargin basement i3 hackintosh that was cheaper than a Mac Mini. ). An inexpensive Mac is a Hackintosh deterrent; especially in countries where mainstream Mac product prices converted from US dollars put them completely out of reach for very high fractions of the population. ]
Even if they want to buy new there are plenty of budget keyboards/mice out their that aren't necessary wireless (which drives up the cost ). For example, Logitech K750 which is about $60 or Microsoft Ergo Keyboard for about $40 and $20-30 wired mouse and be done sub $100. [ All the fancy buttons may not work like they do in windows for the crossover solutions but the basic ones do unless scraping the bottom of the bottom of the barrel. The advent of the "Windows/Start" key means most keyboards have something handy to map to the 'command' key. ).
As for a "cheap" monitor.... it is called a TV (that isn't ancient. ). An HDTV with an HDMI socket is good enough on a really tight budget.
in short, with a Mini someone can "Frankenstein" something together. It may be be pretty, but it works. It won't be the 'look' that Apple wants to sell, but at the fringes compromises come into play and Apple doesn't have to make he compromises themselves.
In contrast, the MBA as the entry point is more so Apple wanting to impose control over the look of the solution. At the lower and upper extremes that is a dogma stance that gets Apple into more trouble then it helps.
I think a large part of the demographic they targeted is still there. Only to some extent folks aren't switching ( doing complete swap outs. ). For some the old PC is kept and the Mini is weaved in on a KVM (or monitor with multiple inputs ) basis. Macs are less than 10% of the folks who have been buying PCs over the last 4-5 years. 90+ % is a huge pool relative to the Mac market. Apple only has t draw in a small percentage of them for a steady stream. It is harder when most of them had a laptop previously, but that never was the market Apple was targeting with the Mini in the first play.
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Well, a number of arguments have been made for having SATA slots already, you just have to read them .
I for one am for a certain number of SATA bays, but I understand the argument for blade solutions only .
It's faster and can be designed to require less space, cooling is less of an issue, no argument there .
It also has its own set of issues though, cost, flexibility, the aforementioned design constraints of M.2, etc .
But maybe we can agree on the merits of having as many choices as reasonably possible, especially in a workstation of which there will most likely be only one version .
The tcMP is almost universally criticized for its lack of flexibility , in virtually every article I've read on it, and of course by countless persons on the internet
.We don't want more that approach again, do we now ?
[doublepost=1535617894][/doublepost]
As a side note, the old wired Apple aluminum keyboards go for cracy prices on the used market , compared to later wireless models .
Funny how Apple can make a great thing and then innovate ( arse joke here ) the life out of it .
The mistake you keep making is that by having an external drive enclosure you've now introduced additional points of failure. For example, instead of relying on the reliability of one power supply you're now relying on the reliability of two. Moving the drives outside of the system didn't remove one of the original points of failure.
The difference between your solution and ours is we can opt for yours if we want, the same cannot be said for the other direction.
You do realize that spinners run at up to 12 Gbps, so four of them would be 48 Gbps - faster than TBolt3 ?
The problem isn't the connection, it's the fact that it spins.
The HP Z4 and Z6 support two M.2 x4 slots and four 3.5" drives. The HP Z8 supports two M.2 and five 3.5".
The also support PCIe storage cards.
Last edited:
and you need an SAS card X8 pci-e 3.0You do realize that spinners run at up to 12 Gbps, so four of them would be 48 Gbps - faster than TBolt3 ?
also M.2 cards are X4 EACH.
I don't think that anyone is saying that the MP7,1 should *not* have M.2 slots.
I think, like the Z-series, it should have both. Two 3.5 inch slots would allow for 24 TB of more cost effective bulk storage without adding unnecessary additional points of failure.
The Z-series also support a quad M.2 NVMe card, using an x16 slot. Four TB at up to 12 Giga Bytes per second read, 6.6 GB/sec write. http://www8.hp.com/h20195/v2/GetDocument.aspx?docname=4AA6-1667ENW https://www8.hp.com/us/en/workstations/z-turbo-drive-g3.html
One more argument in favor of a slot-box.
Last edited:
I don’t see 3.5” bays returning. Maybe a few 2.5” bays if we’re lucky. I’m kind of hoping Apple would make them hot-swappable, like Blackmagicdesign’s Multidock. From what I’ve seen of 8K video workflows in LA, that would make a lot of people happy. They use 2.5” SSD’s like cassettes.
But it doesn’t seem like something Apple would do.
But it doesn’t seem like something Apple would do.
Fair enough. I probably should also apologize for the person I was commenting to for reading that out of context. On it's own, it was just so utterly comical that I had to. That said, while reading this I've not really read many solid arguments for spinners other than "cheap", which too could be contested somewhat easily and the merits of doing other trade offs are unclear. Also, the workarounds for the lack of the spinners don't seem too hard to deal with. Personally, without knowing the trade-offs, I predict this question is the least of Apple's concerns about what they are planning. And frankly, for me and this whole office I am at now.
Since you bring up the trash can, I could argue (and this may be just an opinion) that the failure of that had little to do with the choice of internal storage; that thing had other failures and while individually one could overcome perhaps any of them, together they were just a massive pain in the arse for too many people. Re-branded consumer GPUs with high price, no upgrades to them, mediocre base configuration given the price, thermal issues, no updates year to year, etc. List goes on. Also, from back then, a lot about technology has changed so the equation is now different.
The Multidock isn't normally used on or beside/under a desktiop. It is a rack mounted device ( horizontally oriented.).
https://www.blackmagicdesign.com/products/blackmagicmultidock
" ... Each disk slot supports user replaceable socket connector boards in case a connector is damaged in heavy use installations ..."
The next Mac Pro would probably get validated when operating horizontally, but that is extremely unlikely to be the focus of the design. Apple probably isn't going to sell option rack kit to put it into standard racks.
If they go deskside and make it a bit shorter ( vertical in nominal position) then it will be much less rack hostile than the old one.
[doublepost=1535648790][/doublepost]
You do realize that spinners run at up to 12 Gbps, so four of them would be 48 Gbps - faster than TBolt3 ?
Those devices have a SAS 12Gbs connection on them. That does not mean they can actually fill that at all anymore than a normal desktop drive can fill the 3/6Gbps SATA connection it is connected to. Seagate's own spec sheet (PDF)
"...
Max. Sustained Transfer Rate OD (MB/s,MiB/s) .... Up to 261, 249 ...... Up to 261, 249
..... "
https://www.seagate.com/www-content/datasheets/pdfs/exos-x-12-DS1946-2-1712US-en_US.pdf
So that is rolling downhill with a hurricane tailwind. Put any kind of conccurent workload on them and won't see anywhere near that.
261 MB/s ---> 2.088 Gb/s. Thunderbolt for encoded disk data traffic is roughly about 32 Gbls (for x4 PCI-e v3 ) So about 6.3% of TB's bandwidth (pragmatic bandwidth. It is even smaller percentage of actual 40Gb/s. ); not even double digits. You'd need about 16 of these drives to saturate Thunderbolt v3.
Four would simply be approximately 8 Gb/s which frankly Thunderbolt v1 ( let alone v3 ) can handle.
The notion that spinning HDDs (with zero SSD assist) can easily saturate Thunderbolt 3 is pure arm flapping. It doesn't matter how "Enterprise" or expensive they get. Fill them with He or magic "juju gas" and spin them at 20K RPM, they still won't.
The SAS 12Gbe link is really for aggregate device bandwidth and latency ( multiple drives lashed together ). Some folks are using that to stuff large amounts of Flash into a classic 2.5" ( and much few instances 3.5" ) containers and run these at closer to 12Gb/s with a single device ( with parallel/concurrent data requests to drive up to 12Gb/s. ) For example could put a SAS 12Gbs 5TB SSD device and one of these 12TB both in the same storage device and could with software move the hot data from the slow spinner to the SSD. The 'hot' data would come at closer to 12Gbps but the 'cold" come at a much slower rate. SATA is inherently designed for a bunch of slow disks that share the stated SATA bandwidth; not saturate it indvidually.
Last edited:
And the 260MB/s is only the outer edge of platter; inner edge is half of that or 130MB/s
Obviously I know this, which is why I was pointing out that the limiting factor isn't the type of connector - it's the spinning platters.
And again, nobody is arguing that the MP7,1 shouldn't have M.2 NVMe slots.
Some use cases are more concerned with $ per TB, and TB per drive - than in GB per sec.
Money is always a major factor, Blue Tide. If you had done a TCO analysis, you would understand why spinners aren't going away anytime soon.
If you don't have much in the way of data, (and obviously you don't) then storage isn't hard to deal with. Problem is, a number of us have a lot of data - a lot more than can fit on 1 or 2 (or in my case, almost 2 dozen 1Tb) SSDs.
The lack of choice on storage (and everything else) was precisely the reason I passed on the trashcan.
I have 10Tb of data and 10Tb of backup storage. Because of the lack of storage as well as the port configuration of the trashcan, I would need to add 2 external HDD enclosures (1 for Data drives, 1 for backup, since I could not connect my external box to the trashcan - I am not backing up 8Tbs over USB 2) @ $600 each, a dock @ $200 (to replace other missing ports), and a 3rd enclosure ($100) to hold my Blu-Ray player.
That is $1,500, plus dealing with cable hell and the always fun how am I going to power all this crap. At that point, the cost of an upgrade to my Mac Pro was around $5,500. Sorry, but OSX isn't that good.
At the end of the day, the trashcan was nothing more than a dongle for FCPx.
Moving back to storage. How much is 20Tb worth of SSDs? Buying no-name 1Tb drives it comes out to $3,000. Now add the enclosures (4) to hold said SSDs ($1,600). That is $4,600. I won't bother adding the cost of connectors
OR
Purchase 4 10Tb WD Red @ $320 each ($1,280 total). 2 for Data, 2 for backup. Stuff in 4 drive bay Mac Pro. The OS and apps sit on a PCI card.
Take the $3,320 that you saved and put it back in your pocket (Or go to Dell or HP and buy a real workstation).
Why would I not simply purchase a 20TB My Book Duo Desktop RAID External Hard Drive ($900) and save even more money?
Because I have already had multiple external USB HDDs die on me (HDD was fine, PSU died). I learn from my mistakes.
And for the inevitable Why do you need that much storage? and it's corollary: Why not keep it in the cloud?
In the early days of OSX, P.T. Barnum sold Apple users the idea of Make the computer the hub of your digital lifestyle.
So we did - we stuffed our computers with all of our pictures, music, and movies onto our computers, which we could share through out the house via Apple TV. As an example, my iTunes library is about 7 or so Tb. This fall, I'll be moving that to a collection of 3Tb drives (RAID-Z1) in my Media Computer (Mac Pro 1,1 with low power quad cores).
Why not keep it in the cloud? The Cloud is like Thunderbolt - a solution in search of a problem.
Go down to the ATV forum sometime. The #1 piece of advice given is to NOT leave your iTunes purchases in the Cloud. ALWAYS download your purchases, because you never know when your purchase may disappear (due to the studio pulling it). It happens on a regular basis, and if you bought it more than 90 days ago, Apple will tell you that you are SoL. How much of your data are you willing to risk?
If you don't have much in the way of data, (and obviously you don't) then storage isn't hard to deal with. Problem is, a number of us have a lot of data - a lot more than can fit on 1 or 2 (or in my case, almost 2 dozen 1Tb) SSDs.
The lack of choice on storage (and everything else) was precisely the reason I passed on the trashcan.
I have 10Tb of data and 10Tb of backup storage. Because of the lack of storage as well as the port configuration of the trashcan, I would need to add 2 external HDD enclosures (1 for Data drives, 1 for backup, since I could not connect my external box to the trashcan - I am not backing up 8Tbs over USB 2) @ $600 each, a dock @ $200 (to replace other missing ports), and a 3rd enclosure ($100) to hold my Blu-Ray player.
That is $1,500, plus dealing with cable hell and the always fun how am I going to power all this crap. At that point, the cost of an upgrade to my Mac Pro was around $5,500. Sorry, but OSX isn't that good.
At the end of the day, the trashcan was nothing more than a dongle for FCPx.
Moving back to storage. How much is 20Tb worth of SSDs? Buying no-name 1Tb drives it comes out to $3,000. Now add the enclosures (4) to hold said SSDs ($1,600). That is $4,600. I won't bother adding the cost of connectors
OR
Purchase 4 10Tb WD Red @ $320 each ($1,280 total). 2 for Data, 2 for backup. Stuff in 4 drive bay Mac Pro. The OS and apps sit on a PCI card.
Take the $3,320 that you saved and put it back in your pocket (Or go to Dell or HP and buy a real workstation).
Why would I not simply purchase a 20TB My Book Duo Desktop RAID External Hard Drive ($900) and save even more money?
Because I have already had multiple external USB HDDs die on me (HDD was fine, PSU died). I learn from my mistakes.
And for the inevitable Why do you need that much storage? and it's corollary: Why not keep it in the cloud?
In the early days of OSX, P.T. Barnum sold Apple users the idea of Make the computer the hub of your digital lifestyle.
So we did - we stuffed our computers with all of our pictures, music, and movies onto our computers, which we could share through out the house via Apple TV. As an example, my iTunes library is about 7 or so Tb. This fall, I'll be moving that to a collection of 3Tb drives (RAID-Z1) in my Media Computer (Mac Pro 1,1 with low power quad cores).
Why not keep it in the cloud? The Cloud is like Thunderbolt - a solution in search of a problem.
Go down to the ATV forum sometime. The #1 piece of advice given is to NOT leave your iTunes purchases in the Cloud. ALWAYS download your purchases, because you never know when your purchase may disappear (due to the studio pulling it). It happens on a regular basis, and if you bought it more than 90 days ago, Apple will tell you that you are SoL. How much of your data are you willing to risk?
You (and others) borrow a thinking from somewhere else (RAID0 ? ) that clearly does not apply here. We are not talking about fault tolerance of just one thing. CPU/GPU/Storage are different components that all need power. You assume that 2 PSUs are more likely to fail than one but these would handle different components, e.g. if the external PSU goes down it does not affect the rest of the system, just the storage. AND a single (proprietary=difficult to replace in the first place) PSU would probably last longer if it has to deal with just the m/b instead of everything, wouldn't it?
I am not against internal storage but if you are talking about multiple 3.5/5.25 internal bays to put magnetic HDDs with TBs inside the main system case you are doing something wrong.
I'm not sure who you are arguing with. I haven't seen a single person in this thread argue that there is no longer a role for spinning platters when it comes to Pro or Prosumer data storage. There's clearly a strong case to be made for the economics of platter storage in a wide variety of workflows. It seems like everyone here agrees with this.
What I have seen is a lot of people who believe that there's no longer a compelling case to be made for locating those platters as internal storage in a workstation. A decade ago it was a technical necessity because external storage paid a performance penalty over internal. That's clearly no longer the situation. The downsides no longer outweigh the upsides when it comes to external hard disks and the market is moving in that direction.
Only the smallest shops are going to be satisfied with just four bays (cut that in half, logically, if you're mirroring the data with RAID10). I'd much rather have a box in another room with 10Gbit ethernet for shared access, hot swap bays, and RAID/filesystem options that aren't limited by my OS choice. Synology and QNAP and the rest can sell you a solution that's far more flexible, reliable, and useful than internal bays.
I have no idea why you thought this was worth mentioning in this thread. Literally nobody has said this here.
Sure, but this connector is incompatible with SSDs. I mean, not technically. But if anyone is making SAS SSDs I haven't heard of it. And they certainly wouldn't be cheap, which blows out the "I want to buy a cheap SATA SSD" argument.
Even on a 2012 Mac Pro, that's going to require a PCIe card, which is fine. I have nothing against PCIe cards. The new Mac Pro should support a fiber channel PCIe card or whatever operating at a speed that can connect to those drives.
But again, any pro worth their salt would be running those in an array to bump up the speeds, and then be mirroring to guard against one drive dropping throwing off the whole array. That usually gets well past what you can do internally in a tower. I've seen a lot of systems but never seen anyone serious about hard drives run a setup like that internally. 200 mb/s isn't nearly fast enough for a lot of use cases, even 400 mb/s if you RAID'd is iffy. Maybe as deep storage or archive, that's ok, but for archiving it's almost better to not have that internal to your machine. You don't want a single point of failure taking out your current work and your archive as well.
[doublepost=1535657465][/doublepost]
^ Yep
Again, I think this is about price. People want cheaper Mac Pro configurations, and they want to buy cheaper upgrades. SATA SSDs are probably on the verge of death in the pro market, but we're even still going around in circles about those.
If you want capacity, an external box is going to do a far more serious job of it with dedicated hardware. I mean, internally you're going to do what? Software RAID using CoreStorage?
It's not that I don't doubt there are people who want internal storage. I just don't think that's who Apple is building this machine for. All those use cases sound Prosumer, not Pro. And the days of the Prosumer Mac Pro I think are gone.
[doublepost=1535657644][/doublepost]
If my Mac Pro goes down, I grab my MacBook Pro, and because my storage is available on my network, and I keep working. I at least have access to my Mac Pro's full backups just in case whatever I was working on was only on my Mac Pro. I'm working slower than I would on my Mac Pro, but I'm still working.
If my storage is inside my Mac Pro, and my Mac Pro goes down.... then I'm just stuck.
You keep talking about single point of failure like it's a good thing. It's not. It's literally a single point of failure to derail my whole workflow.
Last edited:
If it's properly sized, it should last a long time regardless of how many components are connected. If it's loaded to near capacity, it's more likely to die young.
And nobody's mentioned that the most likely failures are accidents - cable disconnects (TBolt cables don't have latches, power bricks seldom have latches, power strips with on/off rocker switches can be shut off,...).
Crazy argument - you can connect external drives to the MP7,1 if you prefer. A couple of empty 3.5" drive bays would have few, if any, downsides.
It is also likely what is provisioned in the nominal set up. Back in the 2005-2009 era Intel provisioned SAS in the workstation chipsets. Now it doesn't. ( Isn't in the SP server ones either after quickly checking those C620 series docs). SAS would be a either logic board add in ( which Apple hasn't done in previous Mac Pros. ) or add-in-card solution ( which Apple dropped new development on long before the 2010-12. )
So with SAS, "off the table" , SATA is relatively stuck in time. Probably won't move to 12Gbps either in the mainstream. ( chipset support points to SAS and SATA diverging a bit in scope. )
SAS solutions are higher $/GB than SATA. The bulk of Mac Pro user base doesn't even want to pay for SAS let alone higher for spinners. For PCI-e based solutions they have. Apple probably won't be catering to the folks who want SAS.
Similarly probably not 3.5" either. People have been willing to pay for PCI-e SSD like performance. That is probably who they are going to look toward. It would be more reasonable if they allowed a 2.5" SATA vector for SSDs. First, because the customer base has largely shown they want to buy them. ( many will have them and want to move, but new options with higher capacities will fall into the price range of drives commonly bought with Mac Pros in the past over time. ). Second, they can backstop the transition a bit with allowing in 2.5" for those are still price sensitive. ( no the transfer from old to new system won't be well matched but that probably where the rest of Mac line is making a "transition gap" solutions too. )
but yes Apple should have at least one slot so folks can put some add-in-card out for SAS/SATA if they want to gang up 10+ drives to "beat" thunderbolt. That's the point the number of HDDs needed now to swamp the nominal I/O ports is so high that the internal volume required to match that is much different than 12-16 years ago when Apple come up with the "4 drive sled" design criteria. At that point 4 drives (short stroked and 'rolling down hill in a hurricane' ) were a much higher percentage of the usual "PCI" slot and completely swamped the nominal I/O port options. It is over a decade later and technology has changed. The next Mac Pro is probably going to change with it.
Last edited:
The NMP maxed out the pci-e lanes so they only had 1 PCI-e SSD.
If an new mac pro shows up there should be the pci-e lanes to do 2 with 1 cpu and 2 video cards + 2 TB 3.0 buses.
If an new mac pro shows up there should be the pci-e lanes to do 2 with 1 cpu and 2 video cards + 2 TB 3.0 buses.
Except if pros have already migrated to external/network storage solutions, who's left to put in internal 3.5" drives? Again, 4 is kind of the absolute bare minimum to get stripping + mirroring, and even as we've mentioned, that's pretty minimal because you don't get any sort of dedicated hardware and you have a single point of failure.
This still seems to be circling around some Prosumer case that I don't think Apple is going to target or care about. This is like "I have a big iTunes library and I want to put it on a single big spinner" which again, is not really the Mac Pro. Even if you have a footage library, one HD, or even a risky stripped dual HD config is not going to get you much in the way of transfer speeds.
No, you pull the drives and stick them in your alternate Mac Pro and continue working.
So, given your MP / MBP scenario...how do you continue working when the external enclosures power supply fails?
[doublepost=1535661110][/doublepost]
Why do you care? The cMP permits internal and external storage. Your use case is unaffected. Contrast this to the nMP where the internal option is not available. Why do you feel the latter is a better option than the former?