What if the next Apple chip for the Mac Pro is an Apple server chip?

[bobcomer]

I have 2 Synology 3612’s and 1 24xx (can’t recall the year), fully populated, and only crashed the file system once, due to my own stupidity and a mishap with a UPS. Been rock solid for a decade now. That said, their new stance re: not supporting hard drives except for their own branded ones is really a non-starter for me, and unless the situation improves I’d avoid their big boxes from 2022 onward.

That's a good point, I don't like that policy either. I use non synology branded drives in my NAS's and haven't had a problem -- yet. The minute I do have a problem is when I'll probably start going in another direction.

 

[JouniS]

The reason I said I wouldn't buy an Apple NAS is because there's nothing in their current makeup that scream they haven't any idea at all what a NAS needs. They're far too much into soldering their RAM and SSD's to the motherboard rather than having anything upgradeable or fixable, and that's not going to work in the NAS product category, at all.

I often get the feeling that NAS manufacturers don't have any idea either, at least when it comes to the home / small business segment. Their software is bloated and unintuitive, and there are too many unnecessary features you can't hide. The next time I replace my home NAS, I'll probably just install Ubuntu on a small form factor PC. A Mac Mini with some additional slots for internal storage would do the job even better.

 

[leman]

I've never had problems like that, but I run RAID5. We have Synology at work too.

As for performance, yes, it's not blazingly fast, but comparing it to other NAS's, it's the fastest I've used. You'd have to go to a real SAN to get anything better. Anyway, it's mainly for near line storage and speed isn't the most important thing. I've never lost a byte with a Synology NAS.

I’ve since migrated to SHR and haven’t had an issue since. But yeah, performance and annoying connectivity issues persist.

The reason I said I wouldn't buy an Apple NAS is because there's nothing in their current makeup that scream they haven't any idea at all what a NAS needs. They're far too much into soldering their RAM and SSD's to the motherboard rather than having anything upgradeable or fixable, and that's not going to work in the NAS product category, at all.

The reason why I think Apple could make a good NAS is a) performance and energy consumption of their hardware is something else entirely, b) they have robust storage technologies with fusion a d AppleRAID and c) they have a proven track record caring for data integrity.

Of course, such a product would only be interesting if it had hot-swappable HDD bays with RAID support.

 

[cmaier]

That's a good point, I don't like that policy either. I use non synology branded drives in my NAS's and haven't had a problem -- yet. The minute I do have a problem is when I'll probably start going in another direction.

So far they seem to have only instituted the policy on their highest end boxes. If you don’t use their drives, then SMART is disabled. There is a workaround (you go into the box via ssh and edit a file to add your drives to the ”known” list), but it is a pain, and you have to re-do it each time you do a software update. I’m still on the version 6 branch of their software, anyway, because my old boxes can’t go to 7 and I don’t want to mix and match, but I’ve been monitoring the situation because I was getting ready to retire my 2012 boxes and replace them, and suddenly the 2019 boxes were replaced with 2022 boxes that have these bizarre limitations.

 

[leman]

I have 2 Synology 3612’s and 1 24xx (can’t recall the year), fully populated, and only crashed the file system once, due to my own stupidity and a mishap with a UPS. Been rock solid for a decade now. That said, their new stance re: not supporting hard drives except for their own branded ones is really a non-starter for me, and unless the situation improves I’d avoid their big boxes from 2022 onward.

That’s true, but then again I think that data integrity promise makes Synology drives fairly attractive. Pricing is on the high end, but competitive to other enterprise stuff.

 

[bobcomer]

I’ve since migrated to SHR and haven’t had an issue since. But yeah, performance and annoying connectivity issues persist.

I haven't ever had connectivity issues. It basically just works with everything here.

So far they seem to have only instituted the policy on their highest end boxes. If you don’t use their drives, then SMART is disabled. There is a workaround (you go into the box via ssh and edit a file to add your drives to the ”known” list), but it is a pain, and you have to re-do it each time you do a software update. I’m still on the version 6 branch of their software, anyway, because my old boxes can’t go to 7 and I don’t want to mix and match, but I’ve been monitoring the situation because I was getting ready to retire my 2012 boxes and replace them, and suddenly the 2019 boxes were replaced with 2022 boxes that have these bizarre limitations.

Interesting! Thanks, I'll watch for that..

 

[bobcomer]

I often get the feeling that NAS manufacturers don't have any idea either, at least when it comes to the home / small business segment. Their software is bloated and unintuitive, and there are too many unnecessary features you can't hide. The next time I replace my home NAS, I'll probably just install Ubuntu on a small form factor PC. A Mac Mini with some additional slots for internal storage would do the job even better.

Except that MacOS doesn't do RAID 5 or above, and that's necessary! You need slots for disks and you need a recoverable/redundant volume.

 

[cmaier]

That’s true, but then again I think that data integrity promise makes Synology drives fairly attractive. Pricing is on the high end, but competitive to other enterprise stuff.

But they don’t give enterprise-level support.

 

[JouniS]

Except that MacOS doesn't do RAID 5 or above, and that's necessary! You need slots for disks and you need a recoverable/redundant volume.

All that seems like a massive overkill. With the rise of working from home, the dominant NAS category is the single-user NAS that provides some additional disk space to a handful of devices used by the same person. Convenience and simplicity matter more than reliability in exceptional situations that only affect the productivity of a single person.

 

[bobcomer]

All that seems like a massive overkill.

I EXTREMELY disagree. that's the only way you'll for sure keep your data if one drive fails out right. I would never, ever, ever, buy a NAS without at least RAID5. Reliability is the #1 reason to pay enough for a NAS. I have 2 NAS's at home and I'm the only person using them.

If all you want is just an external drive, buy something cheap like an external USB drive, but in no way is that equivalent to a NAS.

 

[Joe The Dragon]

I’m wondering if what Apple is planning for their next Big Chip is really designed to their server chip. To make money on something that’s is supposed to go into a low volume product like the Mac Pro wouldn’t make financial sense in terms of R and D, unless they already have plans to make a whole lot more of them for another purpose. The M1 Ultra is a beast of a chip already and when you look at the theoretical peak performance, it’s basically an i9 12900k + W6900X hooked up to insane memory bandwidth and speed. Would be easy enough to just add two of them for the top of the line skew and call it a day.

If anybody has any server knowledge here, how many would they need to replace whatever they have now from Intel or AMD? Considering how AMD can’t keep up with demand with their Epyc Systems, would be nice for Apple to explore this market, even if for their own internal use. They are growing their services business massively and having everything running on Apple chips could yield benefits we haven’t even thought of for them as a completely vertically integrated company.

and apple will need to have things like
raid 1 or better storage with end user replacements that don't need an 2th mac to reconfigure storage.
IPMI system
VGA out
networking cards
ram on cards or some non fused MB setup

 

[JouniS]

I EXTREMELY disagree. that's the only way you'll for sure keep your data if one drive fails out right. I would never, ever, ever, buy a NAS without at least RAID5. Reliability is the #1 reason to pay enough for a NAS. I have 2 NAS's at home and I'm the only person using them.

If all you want is just an external drive, buy something cheap like an external USB drive, but in no way is that equivalent to a NAS.

The primary advantage of a home NAS is convenience. I can effortlessly share large files between computers and access them from my desk/living room/patio/wherever.

If a drive in my laptop fails, I restore the data from backups. If a drive in my desktop fails, I restore the data from backups. If a drive in my NAS fails, I restore the data from backups. With a few computers at home, this seems to be something I need to do once in a decade.

 

[bobcomer]

The primary advantage of a home NAS is convenience.

Like I said, I extremely disagree with you. That is not the primary advantage for a home NAS. Backups is the primary advantage, and that means recoverable.

We're talking VERY different levels here of need here. I need my data to always be there in case of a problem. You're just talking easy access to shared across a few PC's data, heck, you don't even need an external drive for what you want, the PC's can do it themselves.

I can effortlessly share large files between computers and access them from my desk/living room/patio/wherever.

I can do that too, but that's not its primary function. (either at home or work) Here, all the data I'm working on right now is on the PC I'm using -- for speed, then it gets backed up to the NAS and maybe shared to another PC if that makes sense.

If a drive in my laptop fails, I restore the data from backups. If a drive in my desktop fails, I restore the data from backups. If a drive in my NAS fails, I restore the data from backups.

The NAS is where the backups are stored here. I don't have a big enough backup device to backup my primary NAS (~24TB), that's the reason for RAID 5. But all the critical backups are also on either my other NAS, which is only 10TB, or on a USB drive.

With a few computers at home, this seems to be something I need to do once in a decade.

When the need is greatest! (murphy's law)

 

[kinghuang]

I wonder how Apple Silicon compares to AWS Graviton3? M1 Ultra has way more memory bandwidth, but Graviton3 has more cores. Both are manufactured on the TSMC 5 nm process.

 

[JouniS]

The NAS is where the backups are stored here. I don't have a big enough backup device to backup my primary NAS (~24TB), that's the reason for RAID 5. But all the critical backups are also on either my other NAS, which is only 10TB, or on a USB drive.

My full backups are only ~1 TB. Most of the data is either not worth backing up, or it's someone else's data and they are responsible for the backups. In my backup scheme, the NAS is an intermediate location where I collect data from various sources before backing it up on USB drives.

 

[deconstruct60]

I wonder how Apple Silicon compares to AWS Graviton3? M1 Ultra has way more memory bandwidth, but Graviton3 has more cores. Both are manufactured on the TSMC 5 nm process.

Does the Ultra CPU cores have way more bandwidth? Apple puts a bandwidth data throttle on the CPU cores so that the GPU/NPU/ProRes decode don't run into bandwidth throughput issues.

The M1 Max CPU bandwidth tops out at around 240GB/s

Apple's M1 Pro, M1 Max SoCs Investigated: New Performance and Efficiency Heights

www.anandtech.com

The Graviton doesn't have a GPU cohort on same memory bus problem and tops out at 300GB/s (with always on memory encryption on. )

AWS Goes Wide And Deep With Graviton3 Server Chip

It is always an exciting time when there is a new compute engine coming into the market, and interest is particularly keen with any new Arm server chip

www.nextplatform.com

The graviton has 32 PCI-e v5. lanes to connect to any 3rd party GPU want to add to the mix. The Ultra has something fewer than x8 PCI-e v4 .

Graviton 3 runs at about 100W so it is lower power than M1 Ultra too. More CPU cores , more CPU core bandwidth , lower power. Amazon made different design choices and got to a different outcome. They do not have a good 'single user at a time' SoC.


From same NextPlatform article

"... According to the report in SemiAnalysis, which is presumably based on a presentation given at re:Invent 2021, the 64 cores on the Graviton3 chip are on one chiplet, and two PCI-Express 5.0 controllers have a chiplet each and four DDR5 memory controllers have a chiplet each, for a total of seven chiplets. These are linked together using a 55 micron microbump technology, and the Graviton3 package is soldered directly to a motherboard rather than put into a socket. ..."

There are 8 DDR5 memory channels. So have to fully populate all the DIMM slots to get max memory throughput. But likely can end up with a having memory capacity than the M1 Ultra when do so (if don't go cheap on the DIMM capacity size ). AWS allows provisioning 1TB instances on one of these processors.

Amazon has Nitro DPU that also likely whip badly any DPU could remotely try to connect to an M1 Ultra box.

If the M1 Ultra can leverage an app that pulls some computation into AMX/NPU/GPU cores then probably has some traction. But running 1000's of Apache images? Apple doesn't have much. Apple wasn't trying to build something for that space either. But the point is why should they even bother to? Graviton 3 and the next gen Ampere on 5nm will be deployed widely in 2022. If Apple needed ARM powered web services they can just rent/buy an Arm server chip that is already better than what they got with the top of the line M1 offering.

That isn't likely to change in 2023 when Apple has M2/M3 and the other folks have also iterated. The x86_64 folks are not in "Rip van Winkle" mode either (Epyc 5/Genoa , Xeon SP Gen 4). ( On an easily multithreaded, workstation , CPU-focused-scalable workload the Threadripper W5995 smokes the M1 Ultra also. Probably not accident that AMD wasn't 'scared' to release that one the same day Apple showed their cards. )

 

[bobcomer]

My full backups are only ~1 TB. Most of the data is either not worth backing up, or it's someone else's data and they are responsible for the backups. In my backup scheme, the NAS is an intermediate location where I collect data from various sources before backing it up on USB drives.

NAS sounds like overkill price-wise for you, but I agree, that's a nice use of it for you. Sorry I'm so negative, but given how companies name products, there are definitely different levels in play, and backup of large amounts of data is my biggest need.

 

[JouniS]

NAS sounds like overkill price-wise for you, but I agree, that's a nice use of it for you. Sorry I'm so negative, but given how companies name products, there are definitely different levels in play, and backup of large amounts of data is my biggest need.

The NAS itself was a trivial expense. I spent much more money on the 2x2 TB SSDs in it, because I was tired of hearing random HDD noises from my desk. But that was years ago, and today even 8 TB would be quite affordable.

 

[quarkysg]

The M1 Max CPU bandwidth tops out at around 240GB/s

The Graviton doesn't have a GPU cohort on same memory bus problem and tops out at 300GB/s (with always on memory encryption on. )

You seem to be comparing measured bandwidth with theoretical bandwidth.

The M1 Max have 400GB/s theoretical memory bandwidth going into it's caches, while one M1 Max CPU core can consume up to 200GB/s of those bandwidth, as measured by Anandtech.

With 2.6GHz clock speed per core, I don't think the Graviton core can push that many bits compared to the M1 Max CPU core running at 3.2GHz clock per core. I don't think the Graviton core's L1 cache line will be larger compared to the M1 L1 cache.

 

[Xiao_Xi]

TrendForce predicts that the penetration rate of ARM-based servers will reach 22% by 2025.

Press Center - ARM-based Server Penetration Rate to Reach 22% by 2025 with Cloud Data Centers Leading the Way, Says TrendForce | TrendForce - Market research, price trend of DRAM, NAND Flash, LEDs, TFT-LCD and green energy, PV

According to TrendForce research, corporate demand for digital transformation including artificial intelligence and high-performance computing has accelerated in recent years, which has led to increasing adoption of cloud computing. In order to improve service flexibility, the world's major...

trendforce.com

Bad news for developers using Docker on Apple Silicon! ?

 

[deconstruct60]

TrendForce predicts that the penetration rate of ARM-based servers will reach 22% by 2025.

Press Center - ARM-based Server Penetration Rate to Reach 22% by 2025 with Cloud Data Centers Leading the Way, Says TrendForce | TrendForce - Market research, price trend of DRAM, NAND Flash, LEDs, TFT-LCD and green energy, PV

According to TrendForce research, corporate demand for digital transformation including artificial intelligence and high-performance computing has accelerated in recent years, which has led to increasing adoption of cloud computing. In order to improve service flexibility, the world's major...

trendforce.com

Bad news for developers using Docker on Apple Silicon! ?

Bad news? A Docker Linux image in a virtual machine image doesn't work on a Apple Silicon system?
Developing on/for a VM image instance is not a big problem.

More Arm data center servers just means there were be more mixed flow docker images for Arm in the future. ( There are already many. The numbers will just get bigger over time. )

As a docker deployment target ... yeah nobody is going to be waiting around for a macOS only solution as the default "Arm" path.


"... TrendForce also observed that AWS's deployment of ARM-based processors in 2021 reached 15% of overall server deployment and will exceed 20% in 2022. ... . If testing is successful, these projects are expected to start mass introduction in 2025. ... "

So really the forecast is really just with AWS mainly pushing. If Ampere and Nvidia really deliver in 2022-2024 that is probably a low-ball estimate of share. Depends in part where those two come in on pricing. I have not clue why they think other viable data center ARM cpus packages won't show up until 2025. That is some lame analysis. It will be sooner than that. Ampere already has product and has a pipleline of next generation versions in progress. (Oracle Cloud has nodes you can engage now. ) Nvidia's Grace is delivering in first half 2023.

There is no intrinsic need that every single major cloud vendor has to roll their own from 'parts' before they can start doing deployments.

 

[ADGrant]

TrendForce predicts that the penetration rate of ARM-based servers will reach 22% by 2025.

Press Center - ARM-based Server Penetration Rate to Reach 22% by 2025 with Cloud Data Centers Leading the Way, Says TrendForce | TrendForce - Market research, price trend of DRAM, NAND Flash, LEDs, TFT-LCD and green energy, PV

According to TrendForce research, corporate demand for digital transformation including artificial intelligence and high-performance computing has accelerated in recent years, which has led to increasing adoption of cloud computing. In order to improve service flexibility, the world's major...

trendforce.com

Bad news for developers using Docker on Apple Silicon! ?

Why?

 

[orionquest]

I’m wondering if what Apple is planning for their next Big Chip is really designed to their server chip. To make money on something that’s is supposed to go into a low volume product like the Mac Pro wouldn’t make financial sense in terms of R and D, unless they already have plans to make a whole lot more of them for another purpose. The M1 Ultra is a beast of a chip already and when you look at the theoretical peak performance, it’s basically an i9 12900k + W6900X hooked up to insane memory bandwidth and speed. Would be easy enough to just add two of them for the top of the line skew and call it a day.

If anybody has any server knowledge here, how many would they need to replace whatever they have now from Intel or AMD? Considering how AMD can’t keep up with demand with their Epyc Systems, would be nice for Apple to explore this market, even if for their own internal use. They are growing their services business massively and having everything running on Apple chips could yield benefits we haven’t even thought of for them as a completely vertically integrated company.

Is there a next big chip? I thought the Ultra was the last chip in the series. But the MacPro is the last in the Mac line up to receive the update. I suspect the next Mac Pro will only come with the Ultra, probably higher available RAM configs. Where things get really interesting is if they will offer expansion of some kind and for what. Since everything else is Unified memory, would they allow GPU card(s)? Are any 3rd parties even interested in making GPU's for Apples market since they are pretty much locked out of the rest of the line up and eGPU is not an option anymore.

Either way you are going to need a Hollywood type budget to purchase one if Apple is going to allow any kind of expansion.

 

[orionquest]

I often get the feeling that NAS manufacturers don't have any idea either, at least when it comes to the home / small business segment. Their software is bloated and unintuitive, and there are too many unnecessary features you can't hide. The next time I replace my home NAS, I'll probably just install Ubuntu on a small form factor PC. A Mac Mini with some additional slots for internal storage would do the job even better.

NAS is just a server, what do you need to do once it's setup other then monitor it. Set and forget basically. It's not like a workstation where you are infront of it constantly. Never felt setting up file sharing and other services were that difficult Synology has this pretty well figured out and is very Mac like it's operation with a easy to use GUI.

I've been a long time user of a Synology 1515+ it's been great. I doubt Apple could produce something to compete with this hardware as Apple is all about selling closed systems and having you buy into more "services" once you reach the edges of your storage limits. With a NAS the whole purpose is to give you control, that is not the Apple way.

 

[mr_roboto]

Is there a next big chip? I thought the Ultra was the last chip in the series. But the MacPro is the last in the Mac line up to receive the update. I suspect the next Mac Pro will only come with the Ultra, probably higher available RAM configs.

1. Apple said M1 Ultra is the last member of the M1 family
2. M1 Ultra can't match the current Intel Mac Pro maximum RAM capacity (by a lot - a factor of 12!)
3. M1 Ultra can't match the current IMP on GPU expansion
4. M1 Ultra can't match the current IMP on general purpose PCIe expansion

All these things add up to Apple planning on shipping the first Apple Silicon Mac Pro with a M2 family chip. It will probably be in a new category beyond the current M1 Ultra and next-gen M2 Ultra, with its own marketing name (M2 Extreme?).