Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
SuperMicro's SuperStations are available with single or dual 28 core.
Is SM's quality still questionable as I remember it to be?
[doublepost=1527717446][/doublepost]
96 lanes in a dual Xeon-SP vs 128 in a dual EPYC. More is better, sure, but in a workstation context that 96 vs 128 doesn't make much difference. You run out of physical space and power capacity before you run out of lanes for GPU's and storage.

The most interesting SKU's in the EPYC lineup are the P variants like 7551P where the single socket retains 128 lanes. A great choice for an NVME storage server.
I'm curious how you could even saturate that many lanes. How much does a single P6000 use up?



FWIW, and just my opinion based on my own experiences using SuperMicro hardware in the past, that I'd look toward HP, Dell or even Puget Systems. Actually, Puget Systems has a fantastic articles database (they post something nearly every week) where they test their system configs against the cMP and the iMac Pro 18 core.
 
Last edited:
32 Cores....

https://www.servethehome.com/cavium-thunderx2-review-benchmarks-real-arm-server-option/

TL; DR performance is close to 28 core Intel Xeon Platinum 8180 / Epyc 7601 (except on where AVX is involved )... So there is good potential for ARM even on desktop, just not yet, its good for most common server loads ( specially those not needing AVX)] , but still far behind Intel AMD, at least ARM is now at X86's mirror. To keep in mind is those benchmark are reach at almost half the power.
Cavium-ThunderX2-SMT-Off-Linpack.jpg


Next Gen CPUs sure will run an all new instruction set on an all new architecture or PC model (its time to retire von neuman).
 
32 Cores....

https://www.servethehome.com/cavium-thunderx2-review-benchmarks-real-arm-server-option/

TL; DR performance is close to 28 core Intel Xeon Platinum 8180 / Epyc 7601 (except on where AVX is involved )...

With TDP wattage in roughly the same range too. Pricing orders of magnitude off of the Apple A-series chips also.


So there is good potential for ARM even on desktop, just not yet,

But what is really going to "buy" on a mainstream desktop system. It is not faster running software folks already own. It isn't particularly cheaper system. It want to match performance the TDP isn't that much better.

Intel (and slightly lessor degree AMD) is providing a pricing umbrella here, where the side effect of Intel "printing money" with their highest end server chips allow room for this competitor to come in. Where Intel and AMD are more competitive pricing in their product lines there won't be a relatively low barrier to entry.

This would be a particularly dubious choice chip for a single person workstation. The I/O throughput isn't up to competition. The single threaded performance is really weak. ( some non data crunching apps just are going to be dominated by single thread. ). If try to crank up broad range single threaded performance it is highly likely the TDP gap will get even smaller. ( price too to a lessor extent ).
 
  • Like
Reactions: ssgbryan
I'd like to thank the posters here for a civil and productive discussion on a hot button topic. I do hope that Apple satisfies the "Pro" user in their 2019 release. It certainly has been a long wait.
I'm very concerned on what's now their definition on a Pro user, since it is inherently wide, a PRO machine has to be powerful and flexible to adap such wide requirements, or at least top 90%, a flexible design with lots of CTO options is the obvious answer, but seems they are reluctant to this unless they clean their's errors boneyards (iMac Pro - tcMP)
Supply your own 1000 words.... ;)

angel-soft.jpg
 

Lukewarm performance on most benchmarks with no compatibility with existing software. Sounds exactly like what Pros are asking for right now.

(except on where AVX is involved )

An extremely important caveat.

So there is good potential for ARM even on desktop, just not yet

Which is exactly where we were before, and there isn't really any sign that ARM can pass up Intel on the high end yet.

A lot of these benchmarks were ARM delivering kind of mediocre performance and the author going "Well it's half the price of Xeon so 2/3 the performance isn't bad" which also kind of ignores that performance gets exponentially expensive but ok...

The exception is server oriented stuff, which again, these chips are sacrificing general performance to optimize, so that's not really special. Intel could do the same thing if they weren't trying to deliver all around performers instead of "I'm really good at encryption but not much else" chips.

There's a business case for Apple, years from now, to move the Mac Pro over to A series if they can get something competitive with Xeon, which they probably could eventually. But I'm not sure why people keep talking about ARM performance or cost. So far none of this shows competitiveness there.

ARM's real benefit is that the architecture is licensable so it can be tweaked for specific situations. Apple's done a really good job tweaking it for low power usages, and Cavium does a good job tweaking it for serving and routing. But again, general use ARM cores for desktops haven't been available, and there is nothing showing that ARM has any real advantage there, which may be why no one has been willing to produce such a core.

Even Apple's A series is kind of an unknown because it's a very bursty architecture. It's been designed to handle short loads of work and not continuous loads of work. Thermally the chip is not designed to be run continuously.
 
Last edited:
  • Like
Reactions: ssgbryan
This is why we did not buy additional EPYC CPU's for workstations other than the test unit used to generate these results:

V-Ray (lower is better)
2x Xeon 8180 17.3 $10000 per chip
2x Xeon 8168 18.2 $5800 per chip
2x Xeon 6154 20.8 $3600 per chip
2x EPYC 7601 20.7 $4500 per chip

If you sit around running benchmarks that fit into EPYC's big cache all day (c-ray, etc) it hauls ass but tends to underperform with actual real world work relative to what one might expect. Enourmous improvement over prior generations though and its great to have AMD back. Looking forward to EPYC2.
 
The current CEO followed the money. He had no emotional ties to the Mac Pro... unlike the previous CEO. Virtually all of the company's engineering effort went into updating products that made them rich.
From a purely financial standpoint, the MacPro is almost a loss leader for the company. Kept in the mix grudgingly.

Behold the black line. That's how important financially the Mac Pro is to the company. Mac Pro constitutes a "single percentage point of all mac sales". iPhone killed the Mac (as I've said over & over again). Apple with its current CEO follows the money.

View attachment 760790

Based on this model, and if the current projectory is in place.. Apple could care less about MAC in the next several years.. as the tablets and the phones themselves will replace the Mac. A bad move, if you ask me.
 
Yeah, I'm in the same boat now. My Mac Pro tower is dead and I need a pro workstation and have been checking out all the components on the PC side. I just can't justify an iMac Pro. Problem is, I love the Mac OS and the stability of it. It would have been fine for all of us if Apple just let us install it legally on PC hardware, provide a few basic tools (just like they did for BootCamp), maybe even an SDK, instead of us having to fly blind and them suing startups out of business who were selling Hackintoshes. I'd even pay for the OS with no support.

Here's a memory dump from my travels lately regarding the Windows side...

If you're thinking of a Hackintosh for a pro machine, go peek at tonymacx86.com and you'll see how even experienced people can't get some hardware to work consistently if at all. The farthest they've gotten is an i9 running Sierra and High Sierra and they managed to finally get messages and thunderbolt 3 going, but not all of them. Lots of steps and testing and troubleshooting and nowhere near as stable as stock Mac OS on Apple Hardware. The i9-7980XE gets hot as hell and needs lots of cooling, especially if you overclock it.

Xeons are not working yet so don't bother. They need Apple to support the cpus in the OS first. Technically they sort of do because the W chips are Apple-slanted Xeon Golds that have been messed with a bit (lower performance from what I could find)—W in general are the OEM versions of the Xeon Scalable family.

If you do go Hackintosh, use Intel Core family (i7, i9 etc).

Also an FYI to you guys looking at those monster dual Xeon Gold configs: Higher core counts might not help you depending on what software you use and everyone agrees they are overpriced based on performance—unless you're just swinging your junk in cinebench, or you're Linus and need to transcode TB of RED footage, and even he isn't getting max cpu usage. I would only consider those if you need to do crazy 3D rendering or the former where they will blow everything else out of the water but cost you a small fortune. Graphics apps tend to be single threaded and will often be slower than an i9 or even i7. Check out this breakdown by Puget on Xeon Scalable.

One bonus is they have more PCI lanes (48 I think) and 6 channels for memory vs 4 for the Core family. So make sure to get your RAM in sixes for optimal performance—96GB 12 x 8GB, 6 per cpu. Anything above 128GB I think you need to use Load Reduce memory on most (if not all) motherboards and may take a performance hit. Supermicro supports 1-2TB, maybe 3TB on one motherboard.

Photoshop, After Effects... hell most apps from Adobe and others have no (or have dropped) support for multi-threading—which of course is logical in an age of massive multi-core cpus being released. :rolleyes: After Effects suffers greatly even on iMac Pro the 10-core was better than 18. Certain apps like Premier suck with AMD GPUs so you'll need CUDA cards from NVIDIA to get the most out of them and there's no FCP on the Windows side. Gaming cards like to GTX 1080 ti are doing great though and can save you cash if you don't need Quadro cards for HD displays or just need to save some cash up front. You can even find 3rd-party water-cooled versions for under $1000 from companies like Corsair.

AMD Ryzen Threadrippers seem to need a bit of config and tweaks or they suck out the box for some things like AE. Intel still seems to give the best performance but Threadrippers are showing promise and give more bang for your buck at only $900-$1000 for a cpu depending on the seller. They have no dual cpu support. You need AMD Epyc cpus for that.

If you do audio and use MOTU hardware, be aware there are known USB issues with X299 chipsets. They suggested using their Thunderbolt 3 interfaces (828es etc) for best results and lowest latency. RME seems to still be at the head of the pack with their rock solid drivers—their USB drivers even outperforming Thunderbolt 3 with low latency.

The main problem I see is if Xeons are not well supported by the software you (and I) may be stuck working on consumer grade hardware that is non-ECC if we go to Windows—and Windows 10 has its own **** storm of problems to contend with. They also have no dual socket options for the Skylake chips. Only the Xeons do. Not something I'd ever want to do (non-ECC) especially on a PC with TB of data flying around that is mission critical and doing 3D sims. Just something to be aware of if you care about ECC.

I did see a board from Asus they claim can overclock the un-over-clockable Xeon Scalable family. If that's true, that might be the way for pros to go soon... if it works.

Don't go single Xeon. It's worse than using an i9 and even the motherboards are small and lack options. If you find differently, please let me know because I'm interested if it changed.

Right now, or soon if it will be a benefit to use dual Xeons, a dual 8-core Xeon Gold 6144 setup may be the ideal choice for performance in a workstation because of the high base clock and turbo speed across the cores, or even 6 or 4 to save some cash. You could always get better chips later and swap them. You don't get the crazy multithreading from 18 or more cores on one chip, but it might be the sweet spot for price and performance across very different types of software—but still not cheap at ~$3,200 or more per CPU. You can configure a ~$10,000 - ~$12,000 full setup that should (theoretically) outperform an iMac Pro if the software likes the dual CPUs and makes use of the cores.

I'm still doing research because there's scant info on the Xeon Scalable family and not many have posted test results short of Puget who is waiting to get more in-house for benchmark tests, but performance-wise for many applications, going with dual 8 core Xeon Gold builds might be best moving forward (especially if we can overclock) because they have a higher base and turbo frequency AND all the cores turbo at the same speed and have ECC. The turbo speeds are not equal across all cores on the larger ones (from what I remember) and the tests they did on Davinci Resolve saw no or worse improvement from 2 Xeons over the regular Skylakes.

As for other hardware, M.2 NVME drives are definitely the way to go if you need high bandwidth and speed. Samsung seems to be the main choice with 860/960/970 Evo and Pro for most, along with Intel Optane M.2. Some boards support up to 3 but you may sacrifice SATA lanes, so if you need a bunch of disks you'll need a solid PCI RAID controller card for more SATA ports, like an LSI MEGARAID or equivalent, even if doing JBOD. I wouldn't waste M.2 on a boot and would go SSD SATA, but for high bitrate video, active project files and scratch disk, I'm planning M.2 as well as external TB3 Sonnet Fusion drive and some HDDs for backup.

Asus and Gigabyte both make Thunderbolt 3 PCI cards which may or may not work in a motherboard you choose. Supermicro told me the Asus cards work in their Xeon boards but only have 1 port. Gigabyte cards have 2 TB3 ports and USB ports, but they haven't responded to my inquiry about using 2 cards on one motherboard for a total of 4 like the iMac Pro.

The higher-end boards will also have 1 or 2 x 10Gb/s ethernet ports, but not all. Plenty of cards from Intel and others if you need the bandwidth. Supermicro also has fiber channel options on some of their boards (I think).

Needless to say, but all these things considered—the state of Apple, software manufacturers, scant info on existing hardware, the price spike in SSDs and GPUS—this is a really terrible time to be buying for any of us. If I had bags of cash laying around I'd buy multiple machines and use a KVM switch, because there doesn't seem to be any solution that fits my needs from either side. I just can't wait a year or 2 for Apple to come out with a pro machine that might not be very pro. So, most likely I will go the PC route, scowling as I do. Unless I decide to just join a circus and bite off chicken heads. If they were smart they'd take a financial loss and come out with an affordable Mac Pro beast with hot swappable bays and get all their pro customers back.

* briefly collapses *

Just wanted to offer my 2¢ since I've been digging through this stuff the past few weeks and I'm still unsure of what the hell to do for a workstation. If I got anything wrong, blame my injured brain from sifting through this technical web of madness. If any of you drop the hammer on a dual Xeon setup, I'd be happy to know the results.
 
Why aren't you looking at a supported config from an OEM, specced how you need it? My feet right now are resting up on a (work supplied) Z800, it's been configured from new with 4x SSD, a quadro, dual Xeons of some description etc... These machines are obviously older now, but the HP rep went through with us and got our input before ticking boxes and ordering the machines we needed...

These are production based machines so need to be under some sort of maintenance contract, if it breaks, we call HP and then send a man in a van over to fix it... Something you don't get if you build something yourself.
 
For those that would scream, "make a Hackintosh", I'd say it's dicey to base my business work on a computer that could be bricked with an Apple update that my Hakintosh doesn't support or requires going to forums to constantly read up on how to handle updates. I don't need additional unbillable work!


I agree with you however its little better on the microsoft side. My home PC is currently essentially bricked until the next round of updates fix what the last round of updates did to it (randomly powering down).

It's not a hardware issue as Linux is 100% solid on it, and the hardware was 100% solid under windows also previously.
 
Why aren't you looking at a supported config from an OEM, specced how you need it? My feet right now are resting up on a (work supplied) Z800, it's been configured from new with 4x SSD, a quadro, dual Xeons of some description etc... These machines are obviously older now, but the HP rep went through with us and got our input before ticking boxes and ordering the machines we needed...

These are production based machines so need to be under some sort of maintenance contract, if it breaks, we call HP and then send a man in a van over to fix it... Something you don't get if you build something yourself.

This is true, however I'm getting little to none to horrible feedback when inquiring with these places that offer turnkey. I'm still contacting more of them daily. I agree that a lease on a machine like this is ideal because of warranty, repair, replacement, and you don't need to drop $10k+ on the spot.

For what I do, no one place even sells everything I need—at least from those I've found and/or have reached out to—so I'd need to finance from multiple companies which is not ideal. My situation is unique though.
 
That's weird, ours are super weird configs too... maybe not as weird as yours though

Had to bend on the video cards though for some reason, at the time there was some spec of card HP wouldn't do because the length of them would stuff up the thermal management, air flow within the case and stuff... can't remember, I think this machine is 4 years old now.
 
That's weird, ours are super weird configs too... maybe not as weird as yours though

Had to bend on the video cards though for some reason, at the time there was some spec of card HP wouldn't do because the length of them would stuff up the thermal management, air flow within the case and stuff... can't remember, I think this machine is 4 years old now.

I'm surprised they didn't just give you a bigger case. I've seen mega corps limit some of their options compared to small businesses though. Small places can easily customize their cases and don't need to wade through 40 feet of red tape.

The only thing I can guess is that I want to combine so many technologies into one box they're afraid it won't work right, or I might bitch from performance suffering in one area. I'd prefer straight answers over being ignored though. :)

I'll check them out though. Would be funny if they were the only ones who gave me good info or had a good setup after all this runaround.

By the way, what are you using the workstation for?
[doublepost=1527834885][/doublepost]
What comes to my predictions that there could be an AR/VR user interface on its way to macOS, Apple has been granted some sort of a new touchpad & keyboard patent along with some VR interaction patents:

http://www.patentlyapple.com/patent...-very-mysterious-macbook-design-and-more.html

If anything, we should see new tools for creating AR/VR content.

If it's ultrasonic it may be for the new mobile technology for using high frequency vibrations to create texture sensations. That's just a guess based on some stuff I read years ago about tricking your nerve endings into thinking you're feeling fabric or stone etc.
 
Supply your own 1000 words.... ;)

I went at it from another angle but with the same material.
I sent Apple a threatening letter stating if they did not give us the mMP we wanted I would TP their HQ!
WOW! That place is huge! My arm is tired and I think I'm going to need help! :p
HQ TP.png
 
This is why we did not buy additional EPYC CPU's for workstations other than the test unit used to generate these results:

V-Ray (lower is better)
2x Xeon 8180 17.3 $10000 per chip
2x Xeon 8168 18.2 $5800 per chip
2x Xeon 6154 20.8 $3600 per chip
2x EPYC 7601 20.7 $4500 per chip


The 7601 why? Because it is the most expensive one? The 7551 is slower ( same core count but 200Mhz base clock drop ), but it is also about $1000 less expensive. So it probably comes in a bit slower than Xeon 6154 at about $100-200 less in cost. For a buy of 50-100 scale out servers that is $5,000-20,000 more money left over. It could mean just more severs. Both of these Xeon SP and Epyc are more server oriented offerings. There are some relatively limited single user corner cases where they are a better fit than the Xeon W ( or Threadripper) but for a broad customer range workstation they aren't good fit.

Same "scale out" advantage for the ARM chip, that's why they'll sell. If can beat the AMD and Intel offerings by $100-400 then when buy in significant numbers the savings can go to buying even more whole servers. In the context of embarrassingly parallel workloads usually amount to better overall infrastructure throughput.
[doublepost=1527868886][/doublepost]
...
Asus and Gigabyte both make Thunderbolt 3 PCI cards which may or may not work in a motherboard you choose. Supermicro told me the Asus cards work in their Xeon boards but only have 1 port. Gigabyte cards have 2 TB3 ports and USB ports, but they haven't responded to my inquiry about using 2 cards on one motherboard for a total of 4 like the iMac Pro.

Technically those aren't PCI-e cards ( obviously not strict PCI cards). Thunderbolt (TB) nominally needs three inputs: PCI-e , GPIO, and DisplayPort. Corner cases can drop the last (e.g., some of these add in TB cards will "happen to work" with the DisplayPort not connected. ). However without both of the first two it won't work at all. All of the "Thunderbolt capable" motherboards i've ever seen only have one (and only one) GPIO header to plug the expansion board into.

Two cards isn't likely going to happen any time soon. An extremely significantly large number of boards have no GPIO->TB header at all which is indicative of the relative importance the board makers put on it. So it is a feat to get to one. Getting two would be on a "trip to Mars' priority level for most of them.



this is a really terrible time to be buying for any of us. If I had bags of cash laying around I'd buy multiple machines and use a KVM switch, ..... If they were smart they'd take a financial loss and come out with an affordable Mac Pro beast with hot swappable bays and get all their pro customers back.

Implicitly here you are smarter because you don't blow money, but (explicit) Apple would be smarter if they did. Running around 'buying' customers (in and of itself) doesn't make Apple smart. The previous Mac Pro and vast majority of the major workstation vendors ( Dell, HP, Lenovo) don't have hot swap HDDs as mainstream configurations.

Apple doesn't need specific people; they need customers ( as group of people. ). As long as they have a group that is willing to collectively spend at least as much money as the old group; they don't need the same exact set of people. If Macs sales volume , revenue , and profits were down over the last 7-8 years then perhaps the Mac Pro folks who have drifted away would be critical. They haven't.

"Pro" is folks doing work for significant compensation. "Pro" is not some set of system features. Features on a physical system doesn't make someone a "Pro". This who "pro" thing tends to drift into the swamp when "pro" becomes a euphemism for "my favorite features" as opposed to actually doing something.
 
Last edited:
  • Like
Reactions: barmann
This is the same Steve Jobs (in roughly same time frame ) that said ....

" .. If I were running Apple, I would milk the Macintosh for all it's worth — and get busy on the next great thing. The PC wars are over. Done. Microsoft won a long time ago. ... "

I don't see the " It is all Tim Cook's" notion if he is simply executing the plan that Jobs laid out. The slow down (and or complete dis-investment) of the Mac Pro started back in 2008-9. That's why there was nothing in 2011-2012 (not even video card speed bumps). There are practical lead times to getting new systems out. ( again a substantive factor as to why there was nothing in 2017-2018 for the Mac Pro ).

In many ways Jobs was in the marketing guys camp. Jobs was not running around Apple with a soldering iron and grinding out code. It is Jobs philosophy that Apple should do a limited number of narrowly focused products in the areas where they complete in (i.e., Apple should offer a relatively fixed number of Macs. ).

There is almost nothing that Apple has said (via Jobs or Cook ) where they convey they have any notion that they have some kind of monopoly power in the space that the Macs operate in. So the key point of relevancy to his quote in the video is what in the context of Macs and even more so in the context of Mac Pro ( where the percentage of the overall workstation market is even less for over approximately two decades ) ?

Yes but it hasn't happened the so called post PC era that started in 2010 still hasnt evolved because they have specifically crippled the functionality of those products AKA IOS. If you want to do any pro work you need a computer.

Its also a fact that you cant do everything on a touch interface and a mouse and keyboard are also needed. They just wont allow it to happen.

Ipad and mac sales arent too dissimilar.

Like anything there is also the age old tier of product. Having a top tier expensive machine with the best tech available is to many a show that they are invested in the product line and also something to inspire too. I dont care who you are when you get going in the Pro world you dont start with a 5k desktop its probably a hand me down or an entry level product and as you improve your skill and earn more money you aspire to own a machine that makes your workflow more productive and a machine that can make your ideas come to life.

Now you look at the line, the macbook is a great machine crippled with one port. The macbook pro 2 port is no faster than the macbook really within 5% then the higher watt CPU in the 13" 4 port macbook pro again is within 10% but significantly more expensive. All have dual cores which are all perform similarly all with integrated graphics so the £1300-2100 line is a pretty crap place to be and very difficult to decide. There have never been as many what to buy threads on this forum as there has been since 2016.

Then there is the reliability, the crap keyboard and the difficulty of port selection... its all well and good having forward thinking ports but why not leave a couple to help the transition especially one USB 3 and SD card.

Then there is the 15" which starts at £2300 and is the only decent mobile power house as it has dedicated graphics and a quad. But what if you want a bigger screen but portable and dont need the power? Apple offers nothing, your stuck with 12" and you still have to pay £1300 for the pleasure.

Then there is the iMac, non of the build to order machines are suitable for decent work, the fact they all have spinning drives yet the whole portable lineup doesnt... then there is the fact the i5 is the only CPU that runs adequately within the thermal envelope and the i7 the only hyperthreaded CPU yet all the mobile chips are hyperthreaded... It runs hot, the fan goes like mad and it throttles... When there is about 40% free space there is no excuse for such crap cooling.

Then we have the mac pro which was a joke on launch and at £3899 for an 8 core with 256gb ssd...

Then the iMac pro which is a parts bin special that was rushed to market and didnt have the proper testing or support in place and has been a joke for a lot of people.

A great machine but many dont want an AIO

That 5k display is also a pain although the graphics options are better its still not a smooth experience for a lot of pixel based programs. Refreshing a 50mp image at 60fps through a 14mp display when you have other adjustments rendering too just not a good experience. You should be able to buy a headless mac.

The whole line is a joke really there isnt one product that really shines or has no compromises.

The fact most programs are now multi platform cloud based and offer the same experience across OS's there is less need to spend more on hardware that isnt as reliable as it was, more expensive and on the windows side you can tailor a machine to your needs for the software you need rather than just dealing with what apples deems suitable.
 
Implicitly here you are smarter because you don't blow money, but (explicit) Apple would be smarter if they did. Running around 'buying' customers (in and of itself) doesn't make Apple smart.

Because comparing a company like Apple, to a consumer like beaker7 ( my assumption ) is an apples to apples comparison, as you seem to elude to in your retort ...
 
This is the same Steve Jobs (in roughly same time frame ) that said ....

" .. If I were running Apple, I would milk the Macintosh for all it's worth — and get busy on the next great thing. The PC wars are over. Done. Microsoft won a long time ago. ... "

Steve said this because Apple was a competitor and he was pissed at them. He was throwing shade at the company that fired him. He had also just gotten his clock cleaned by Windows NT again, and had moved on to building solutions for Windows at NeXT.

Unsurprisingly when he was in charge of Apple again he had a very different opinion.

People lean on this remark way too much.
 
The 7601 why? Because it is the most expensive one? The 7551 is slower ( same core count but 200Mhz base clock drop ), but it is also about $1000 less expensive. So it probably comes in a bit slower than Xeon 6154 at about $100-200 less in cost. For a buy of 50-100 scale out servers that is $5,000-20,000 more money left over. It could mean just more severs. Both of these Xeon SP and Epyc are more server oriented offerings. There are some relatively limited single user corner cases where they are a better fit than the Xeon W ( or Threadripper) but for a broad customer range workstation they aren't good fit.

We chose the 7601 for testing because we need the most all-around performance possible in a 1 or 2 socket tower workstation and wanted to give AMD a try - and it is nearly impossible to find well done, real-world benchmarks on the web from reputable sites.

We also tested 1950x and the 18 core i9. We have a few artists running the i9's clocked around 4.4 ghz and those are a nice option. The 1950x is a terrific value bang for buck but when you're more concerned with bang than buck they fall short.

The ones who do more heavy lifting are on mostly dual 6154 systems and a few 8168's with Titan V GPUs.

I'm sure you'll tell me that we'd be better off with something slower and less expensive.
 
Last edited:
This is the same Steve Jobs (in roughly same time frame ) that said ....

" .. If I were running Apple, I would milk the Macintosh for all it's worth — and get busy on the next great thing. The PC wars are over. Done. Microsoft won a long time ago. ... "

I don't see the " It is all Tim Cook's" notion if he is simply executing the plan that Jobs laid out. The slow down (and or complete dis-investment) of the Mac Pro started back in 2008-9. That's why there was nothing in 2011-2012 (not even video card speed bumps). There are practical lead times to getting new systems out. ( again a substantive factor as to why there was nothing in 2017-2018 for the Mac Pro ).

In many ways Jobs was in the marketing guys camp. Jobs was not running around Apple with a soldering iron and grinding out code. It is Jobs philosophy that Apple should do a limited number of narrowly focused products in the areas where they complete in (i.e., Apple should offer a relatively fixed number of Macs. ).

There is almost nothing that Apple has said (via Jobs or Cook ) where they convey they have any notion that they have some kind of monopoly power in the space that the Macs operate in. So the key point of relevancy to his quote in the video is what in the context of Macs and even more so in the context of Mac Pro ( where the percentage of the overall workstation market is even less for over approximately two decades ) ?

Well said indeed .

Jobs has also famously declared the iPad to be the future of computing .
As in the immediate future , which hasn't quite worked out . In this future or any of them . ;)
I find that even more noteworthy than Phil the Shill's 'can't innovate my bubbly corporate behind' remark .

Yet Jobs also moved the Macs to Intel and helped to make the cMP happen, warts and all, when it was the smart thing to do .
[doublepost=1527878862][/doublepost]
We chose the 7601 for testing because we need the most all-around performance possible in a 1 or 2 socket tower workstation and wanted to give AMD a try - and it is nearly impossible to find well done, real-world benchmarks on the web from reputable sites.

We also tested 1950x and the 18 core i9. We have a few artists running the i9's clocked around 4.4 ghz and those are a nice option. The 1950x is a terrific value bang for buck but when you're more concerned with bang than buck they fall short.

The ones who do more heavy lifting are on mostly dual 6154 systems and a few 8168's with Titan V GPUs.

I'm sure you'll tell me that we'd be better off with something slower and less expensive.

Fair enough, but how could Apple fit an all bang, mucho buck product into the Mac line - if such an approach would mean the end of more affordable models of the same basic design ?

Apart from the fact that they never even tried to be top dog , and 3rd party program support for OSX is sparse in many fields .
 
  • Like
Reactions: ssgbryan
Jobs has also famously declared the iPad to be the future of computing .
As in the immediate future , which hasn't quite worked out . In this future or any of them . ;)
I find that even more noteworthy than Phil the Shill's 'can't innovate my bubbly corporate behind' remark .

Yet Jobs also moved the Macs to Intel and helped to make the cMP happen, warts and all, when it was the smart thing to do.

Jobs is a salesperson. If he was selling calculators he'd tell you calculators were the future of computing. People hang on his every word like there was some sort of master plan, when there wasn't. He was a guy trying to sell computers, and doing a good job at it.

He also said Apple would never ever do a cell phone at the exact same time they had two parallel prototypes in the lab. He said that as a sales tactic, not as a truthful statement.
 
Last edited by a moderator:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.