These are also very useful if you work with a team.
These are very good. I looking at the DS1515+.
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.
Phil 'My Ass' Schiller Read Please: Our mMP whish list
- Thread starter Mago
- Start date
- Sort by reaction score
I just thought I’d cut and paste a post by hh from Ars on the off chance that Schiller does happen to come across this forum…….
-hh Seniorius Lurkius
REPLYAPR 11, 2017 10:43 PM
The diversity of comments reveal the diversity that exists in 'Pro' level use cases and their respective priorities.
Nevertheless, a few points:
* When it comes to individual tailoring, this also includes the customer trying to optimize the value that they get out of the Apple hardware. Traditionally, our customer's interests will lean towards existing 'commodity' standards, such as bog-standard SATA-3 drives because Marketplace competition keeps the price down.
But it isn't in Apple's (direct) financial interests to make it easy for us to buy 3rd party products - - although to be honest, the main reasons why we buy 3rd party is because:
a. Apple's markups are pretty outrageously high, and
b. Apple isn't particularly agile in deploying new products, and
c. Apple "never" sells incremental upgrade kits to existing systems.
So Apple has some decisions to make.
* Returning to individual tailoring, some of us will want GPU monsters. Others CPU monsters. Yet others will go for DATA (storage) monsters. And others still will Chinese Menu for B & C and not A ... the motivations for each vary, based on the workflow priorities as well as other considerations.
For example, a Small Businessman may seek a "DATA" monster simply because they don't have to collaborate with others, and a couple of extra hard drive bays means that they don't actually need to buy & manage a file server (especially a high performance one that's got more bandwidth than a 1GB Ethernet connection can handle).
Similarly, a non-graphics based data modeler will value a few really fast CPU cores, <1TB of fastest data I/O available...and not really care about GPU performance all that much.
The challenge is how to make a single flexible-enough product to minimize manufacturing complexity.
The basic answer is to sort out some test mules using the old 'cheese grater' form factor - - it has a generous motherboard for real estate, and the questions become how to bring it up to date with current specs (keep in mind that even though SATA-3 dates from 2008/2009, even the 2012 Mac Pro never got it). It really doesn't take that much in the way of space claims to provide some M.2 slots and so forth - - the Mac Pro is so far behind right now that there's simply no excuse for delays just due to wanting a tiny package ... that was the fatal flaw of the 2013, so it is the LAST thing you want to do.
And 2018 ... maybe? WRONG.
Back in 1999, the "Yikes!" G4 went from zero to on sale in nine (9) months. And that was with a much smaller, less rich Apple.
Let's have this new Mac Pro by Christmas. Yes, THIS Christmas.
It doesn't need to be super-elegant and perfect -- it merely needs to be clearly better than the Trash Can.
-hh Seniorius Lurkius
REPLYAPR 11, 2017 10:43 PM
The diversity of comments reveal the diversity that exists in 'Pro' level use cases and their respective priorities.
Nevertheless, a few points:
* When it comes to individual tailoring, this also includes the customer trying to optimize the value that they get out of the Apple hardware. Traditionally, our customer's interests will lean towards existing 'commodity' standards, such as bog-standard SATA-3 drives because Marketplace competition keeps the price down.
But it isn't in Apple's (direct) financial interests to make it easy for us to buy 3rd party products - - although to be honest, the main reasons why we buy 3rd party is because:
a. Apple's markups are pretty outrageously high, and
b. Apple isn't particularly agile in deploying new products, and
c. Apple "never" sells incremental upgrade kits to existing systems.
So Apple has some decisions to make.
* Returning to individual tailoring, some of us will want GPU monsters. Others CPU monsters. Yet others will go for DATA (storage) monsters. And others still will Chinese Menu for B & C and not A ... the motivations for each vary, based on the workflow priorities as well as other considerations.
For example, a Small Businessman may seek a "DATA" monster simply because they don't have to collaborate with others, and a couple of extra hard drive bays means that they don't actually need to buy & manage a file server (especially a high performance one that's got more bandwidth than a 1GB Ethernet connection can handle).
Similarly, a non-graphics based data modeler will value a few really fast CPU cores, <1TB of fastest data I/O available...and not really care about GPU performance all that much.
The challenge is how to make a single flexible-enough product to minimize manufacturing complexity.
The basic answer is to sort out some test mules using the old 'cheese grater' form factor - - it has a generous motherboard for real estate, and the questions become how to bring it up to date with current specs (keep in mind that even though SATA-3 dates from 2008/2009, even the 2012 Mac Pro never got it). It really doesn't take that much in the way of space claims to provide some M.2 slots and so forth - - the Mac Pro is so far behind right now that there's simply no excuse for delays just due to wanting a tiny package ... that was the fatal flaw of the 2013, so it is the LAST thing you want to do.
And 2018 ... maybe? WRONG.
Back in 1999, the "Yikes!" G4 went from zero to on sale in nine (9) months. And that was with a much smaller, less rich Apple.
Let's have this new Mac Pro by Christmas. Yes, THIS Christmas.
It doesn't need to be super-elegant and perfect -- it merely needs to be clearly better than the Trash Can.
I use HFS+, because is the native file system.
On Windows, I use NTFS.
There are proprietary solutions with their own native optimized file systems.
Always choose the native file system.
[doublepost=1492105141][/doublepost]With the Pro Workstation market, the last thing that is needed is gimmicks and bleeding edge stuff. Think of the relationship between Redhat Enterprise Linux and Fedora Linux. Fedora is where all the very new stuff is, the same as Apple should be doing with their consumer-grade products (i.e. let the masses test it out). RHEL takes the successful stuff from Fedora and produces an enterprise-class OS. Yes it's boring, but it works. This is what I personally want from a Pro line up from Apple, and I should imagine a few others want too. How it performs is much, much more important than how it looks. As is reliability. Cost isn't a major issue either if it can be used for a few years.
Going back to this wish-list, Apple just need to take the best of breed technologies from the workstation market and put them into a case that is flexible enough for the expansion options and has sufficient cooling for a quiet and reliable operation. If it looks similar to the old Mac Pro design I couldn't care less. They should also consider a single socket i7 system in the same case for the enthusiast market - i.e. those of you who wouldn't buy an iMac but are sat in the middle between a Mini and Mac Pro.
I think the response from the market is proving to Apple that we certainly are not in a 'post PC era' as far as the professional market is concerned. That approach is fine for the consumer, but utter dribble if you need to actually get some work done.
I’m interested in cross-platform compatibility. I always (aside from boot drives) go with either ExFAT or FAT32 because I’d like to avoid being limited to only being able to use a drive with one OS. That way, if I want to view those files on a computer with a different operating system, I can. For example, if your Mac dies or is otherwise unavailable and the only other computer in the house runs Windows, you’re going to want to be able to access your media files on the Windows PC. Sure, you can write to HFS+ formatted drives on Windows and NTFS formatted drives on MacOS, but you have to install a special driver for that, and most folks don’t bother to install that driver. Aside from using a FAT variant, the only option is to have a duplicate of each media drive, where one is formatted as HFS+ and the other is formatted as NTFS.
Then you should go for ExFAT.
I never do this for media because it's never as fast as a native file system an there could be incompatibilities in backups, for instance.
But alternative of have duplicated media in two separate file systems is not feasible. You would spend most of the time duplicating and again, there could be problems during copy (string limit, different characters, and so on).
I never do this for media because it's never as fast as a native file system an there could be incompatibilities in backups, for instance.
But alternative of have duplicated media in two separate file systems is not feasible. You would spend most of the time duplicating and again, there could be problems during copy (string limit, different characters, and so on).
I'd never put data that I liked on a non-logging filesystem. (at least if I had the choice - with high capacity SD cards there's no choice)
Last edited:
I read the title of this thread out loud to some rather raised eyebrows around here - the first few words anyway.
Dual CPU socket: Cannon Lake Xeon E5-2699 v6 (64+ cores / 128+ threads)
Four PCIe: 16, 16, 16, 16 (96+ total PCIe lanes)
Twenty Four: DDR4 DIMM slots (6-channel)
Four PCIe: 16, 16, 16, 16 (96+ total PCIe lanes)
Twenty Four: DDR4 DIMM slots (6-channel)
I am very pessimistic to say the least. Apple's given no indication other than recent announcements that they are willing to back-peddle. I just get the feeling that the whole "modularity" buzzword may not mean "standard parts," especially with PCIe slots. Here's to hoping though, I was wrong about them discontinuing the line, hopefully I'm wrong about this.
I don't disagree that they may be thinking something different but could you imagine if this thread and that one at Ars went viral, all over the Mac forums and also mentioned by Gruber, Rene et al in podcasts and interviews?
At least then they’d have some idea what the vocal ones want. Does it mean they’ll keep heading due north at full speed, probably. But maybe with enough noise in the right places they’ll go 15 degrees off their usual course.
Last edited:
I have been extremely happy using a pair of DS1511+ units with DX510 expansion boxes since I bought them over six years ago. Performance is comparable to locally attached storage & the DSM operating system is very straightforward to use. It's also good that Synology continue to support these units with the latest version of DSM long after they have been superseded.
This makes me remember an interesting take on "modularity".
For a while, probably 12 years ago or so, Dell had an option for a system that could have I think 4 PCI slots in a mid-tower. They also had a taller tower that had 6 or 7 slots.
The motherboard was the same in both. On the bottom edge of the motherboard was an edge connector (looked like the big PCIe card connector. On the taller tower, an "extension" with the extra PCI slots plugged into the edge connector.
For the mMP Apple could do something similar - even have a single-socket motherboard with RAM and slots, and an extender with the second socket, the RAM for the second socket, and the PCIe slots for the second socket.
One of the reasons I sold mine was because it would cost $1K to upgrade the Flash Drive. SO I wouldn't mind seeing a m.2 SATA option. Also I would of loved the Power Button in front and a few USB in the front too.
I liked the quietness of the 2013 when surfing the web and liked the small form factor but I am thinking they will need to go bigger to make it anything worth using.
I liked the quietness of the 2013 when surfing the web and liked the small form factor but I am thinking they will need to go bigger to make it anything worth using.
Keep jony ive away from the hardware specs. He can design the buttons and handles but keep him away from hardware specs.
Last edited:
This seems highly detached from the reality that the core problem it isn't just the Mac Pro. The Mac mini is comatose. The iMac looks like it close to comatose. The MBA ... again comatose. The MacBook a "version 1.0" design that also seriiously needs an iteration. The MBA and iMac are bigger threats to Apple's overall Mac business if they screw those up then Mac Pro. ( The Mac Pro has hiccuped and finacially Apple has taken no deep hit. ).
Apple clean up all of that up plus a Mac Pro by Christmas? That is just setting grossly unrealistic expectations. The only way t o do that would be to push some much larger impact Mac products out of the way and pragmatically certify and ship a hackintosh.
If Apple doesn't have a Xeon E5 v5 ( Skylake ) board work in flight already they are probably at ground zero on that. And that is spinning up at Intel. Other folks are shooting for Fall but they have been working on those since last Fall.
It is bad timing at this point to hit the "reset' button as E5 v4 was more than stable to launch one and 2nd half of 2017 is a less stable target.
Yiles shipped fast because it was basically the G3 "blue and white" system bumped to G4. Same case, mostly same parts ( same RAM , slots , etc. ). All the initial G4 did was tweak the CPU socket and firmware for the G4. That is largely it. In contrast, there is no Thunderbolt on MP 5,1. No PCIe SSD slot. The SATA set up is a decade out of date. it is a 10 year old design. Not a 10 month old design.
I don't disagree fundamentally with any of your arguments - but I notice the dissonance with the rest of the industry.
HPE/HP/Dell/Lenovo/Supermicro usually have new systems ready to ship on the day that Intel announces a new CPU. The Pascal GPU cards could be plugged into most existing systems the day they were available. (At least those systems which could accept and power a double-wide GPU.) Look at all of the motherboards available for the Ryzen launch - agile engineering can produce stuff pretty quickly when you don't reinvent the wheel with every system. (That is, "when you don't design a new proprietary non-interchangeable board".)
If these vendors don't have systems ready on day one, it's usually because there are ramp-up issues and nobody has sufficient CPUs/GPUs at the start.
I read Apple's plans as saying "it will take us 12 months to 18 months to engineer proprietary non-interchangeable modules based on these parts" - and almost nobody is taking them to task on that.
Apple could have the mMP available by the end of the year if they'd tell Jony Ive to build a beautiful case to hold an ATX motherboard. Then build an ATX form factor board, or work with one of the top tier mobo manufacturers to produce a minor variant of one of their many boards. Or strike a deal with HP to glue an "
" to a couple of Z-series models.It's not setting "unrealistic expectations" to have a new Mac Pro by Saturnalia. It's apologizing for Apple's stubborn and arrogant inability to address the needs of some people who want to continue to be Apple customers.
Most of your posts show a lot of insight, but I'm not sure about these comments.
Of course a new pro Apple tower would have T-Bolt ports. And I hope that it would support 12 Gbps SAS and SATA drives, as well as NVMe. Of course it would have some M.2 or other next generation connectors for IO. You seem to be assuming that using a "new cheese grater" as the model for the mMP means that it would have SATA II 3.5" drive sleds. That's absurd.
NVMe is over-hyped most of the time. The amazing synthetic benchmark numbers simply don't translate into noticeable performance improvements for typical workstation users.
For the single-user workstation, AHCI PCIe drives often perform close to NVMe drives, and SAS/SATA drives are much cheaper per TB if you don't need 10,000 user database performance.
In March, I bought 480TB of SAS arrays for $65K. I won't even try to estimate how many servers I'd need to buy to 480 TB of NVMe, let alone the price of 480 TB of NVME drives.
Note: I do have three systems with 8 TB NVMe RAID-0 arrays - because those systems run with the thousands of active threads that can generate the massively parallel IO loads where NVMe shines.
Last edited:
Dear Phil "My Ass" Schiller, as if you are going to read this, or care,
Incorporate replaceable GPU's across the Mac and MacBook line, and stop soldering ram to the logic board, *******s.
Make a DYI Mac™ Logic Board with Intel X299 chipset, M.2, USB-C, 8 ram slots, and at least 4 full length PCI-E slots. For a Retail price of around $799. Let us use any non-xeon cpu, power supply, case and other things we want.
I realize this will vulture the other Macs in the line, so lock the MacOS to the Serial Number and kill the Hackintoch market.
Give us an enthusiast Firmware, so we can overclock and tweak.
Support VT-d in the MacOS, so we can utilize PCI Passthrough in our virtual machines.
Thanks Phil, you ass is great, BTW.......
Incorporate replaceable GPU's across the Mac and MacBook line, and stop soldering ram to the logic board, *******s.
Make a DYI Mac™ Logic Board with Intel X299 chipset, M.2, USB-C, 8 ram slots, and at least 4 full length PCI-E slots. For a Retail price of around $799. Let us use any non-xeon cpu, power supply, case and other things we want.
I realize this will vulture the other Macs in the line, so lock the MacOS to the Serial Number and kill the Hackintoch market.
Give us an enthusiast Firmware, so we can overclock and tweak.
Support VT-d in the MacOS, so we can utilize PCI Passthrough in our virtual machines.
Thanks Phil, you ass is great, BTW.......
Last edited: