Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Just put in 192 gigs of the Samsung RAM, score is now:

1146, 15822
This is why people should take GeekBench scores with a grain of salt. That's a 15% increase in performance which is unlikely to be observed if most every application.
 
While I don’t doubt you, it still doesn’t explain why the 2.5 GHz 28 core Mac Pro’s single core score is so much higher than all the others with same Gen processors but higher clock speeds?

Is this a result of people who buy the 28 core machine spend more on RAM, GPU etc that may be effecting the scores somehow Or is Catalina not yet playing nice with the new machine?

Could be all of the above - I have seen some of my benchmarking scores change based on the amount of memory I had installed.
 
  • Like
Reactions: gazwas
While I don’t doubt you, it still doesn’t explain why the 2.5 GHz 28 core Mac Pro’s single core score is so much higher than all the others with same Gen processors but higher clock speeds?


Looking at the right set of clock speeds. The 28 core W-3275 Turbos up to 4.4GHz if you toss all of the other cores "out the window" to do single core drag racing. It won't be that far behind the other when down shifted into one core. ( W-3223 Turbo 4.0 (16.5MB) W-3235 Turbo 4.4 (19.25MB) )


Furthermore the L3 cache on a W-3275 is 38.5MB. For some of these single threaded drag racing benchmarks that is big enough on some of the subsegments to load all of the data in. If running it multiple times will basically load up parts of the benchmark.

Is this a result of people who buy the 28 core machine spend more on RAM, GPU etc that may be effecting the scores somehow Or is Catalina not yet playing nice with the new machine?

RAM ( and thereby file system cache size) can play a role also. But even if hold the RAM capacity the same the L3 size is changing. The all have Speed Shift and Turbo Boost Max. (i.e., can spin up quite quickly and can use the whole power envelope until the thermals make the system slow down. Both are helpful good for short duration drag racing many of these benchmarks incur. All over in a minute or so. )
 
Made to look bad right out of the gate by the predecessors, consumer gear and even laptops: this all feels way too familiar. Apple going workstation and really following in the footsteps of SGI here. ;)

In two years time the wailing will be deafening.
 
  • Like
Reactions: ssgbryan
Here are scores from two different installed processors in the same Mac Pro 7,1 with 32 GB ( 4 x 8 GB ) .

W-3223 Eight Core @ 3.5 GHz GB5 ( 64 bit ) Single Core = 1061 .
W-3223 Eight Core @ 3.5 GHz GB5 ( 64 bit ) Multi Core = 8220 .

Gold 6212U Twenty Four Core @ 2.4 GHz GB5 ( 64 bit ) Single Core = 1052 .
Gold 6212U Twenty Four Core @ 2.4 GHz GB5 ( 64 bit ) Multi Core = 15305 .

Similar issue (base clocks don't tell full story). Gold 6212U Turbos to 3.9GHz but doesn't have Turbo boost Max tech. Same huge gap in L3. 3.9GHz versus 4.0GHz isn't a large gap.

The Xeon SP Gold 'U' tagged processors are tagged toward uni-processors installs and are gifted with substantive Turbo clocks and higher TDPs than most of the rest of the Xeon SP class.
 
Last edited:
  • Like
Reactions: Flint Ironstag
Here are scores from two different installed processors in the same Mac Pro 7,1 with 32 GB ( 4 x 8 GB ) .
....
Gold 6212U Twenty Four Core @ 2.4 GHz GB5 ( 64 bit ) Single Core = 1052 .
Gold 6212U Twenty Four Core @ 2.4 GHz GB5 ( 64 bit ) Multi Core = 15305 .

The Gold 6212U System was not able to have its NVRAM refreshed . Had I been successful , that score should have been higher . System firmware is too immature for such a prototype processor installation at this time .
....

The Gold is useful as a course grain hack but only having 48 PCI-e lanes is probably going to break something. ( some slots somewhere in the system are likely 'dead' because have no bandwidth supply at all).
 
here are my Cinebench results:
Screen Shot 2019-12-24 at 10.58.23.png
 
I haven't. I'm puzzled by Apple's choice to send review units to YouTube bloggers and, as far as I can tell, none to any technical web site (such as AnandTech) who would perform such benchmarks. Apparently this Mac Pro is targeted at YouTube bloggers?

Or just sending them to customers in the targeted demographic with a NDA and not much advanced hoo-haw. Imagine that.

https://9to5mac.com/2019/12/24/new-mac-pro-animation/

As opposed to tech porn review sites that will do what with them for a week or so ?
 
Seems like these guys may have received a review unit - else I imagine they wouldn't have written an article about their experience with it?

USING THE NEW MAC PRO AND PRO DISPLAY XDR
(by Lunar Animation)

 
For comparison, here is my 2.5 Year old Core i9 7900X (10 Core CPU). It is paired with 64GB (8x8GB) 3600MHz CL17 memory:

XVtKyXs.png


I paid $640 for this CPU two years ago, quite the bargain I guess looking back. I'm kinda surprised it beats out the XEON 12 core in the $5,999 Mac Pro to be honest, especially considering its age.
 
For comparison, here is my 2.5 Year old Core i9 7900X (10 Core CPU). It is paired with 64GB (8x8GB) 3600MHz CL17 memory:

XVtKyXs.png


I paid $640 for this CPU two years ago, quite the bargain I guess looking back. I'm kinda surprised it beats out the XEON 12 core in the $5,999 Mac Pro to be honest, especially considering its age.

you bought brand new or 2nd hand? in Europe this CPU is still about $700 brand new
 
you bought brand new or 2nd hand? in Europe this CPU is still about $700 brand new

I bought it on ebay used two years ago. The motherboard I'm using actually cost more than the processor haha
 
still a good deal, which board do you have?

I'm using the Rampage VI Extreme. It has some neat features, 8 Dimm slots, 10Gb networking (Although I'm using my own Intel dual-port 10Gb card instead). Can support Thunderbolt 3 with an add-in card. It has three M.2 slots, two of which can feed directly from the CPU PCIe lanes. I'm using a Samsung 970 Pro 1TB in one of those slots.

Really solid board I've been quite happy with it.
 
For comparison, here is my 2.5 Year old Core i9 7900X (10 Core CPU). It is paired with 64GB (8x8GB) 3600MHz CL17 memory:

I paid $640 for this CPU two years ago, quite the bargain I guess looking back. I'm kinda surprised it beats out the XEON 12 core in the $5,999 Mac Pro to be honest, especially considering its age.

It's not a $6000 Mac Pro: the 12 core is $7000. And the machine Bob_Stan is actually at least $8400 (AMD Pro Vega II card).

Your machine's 1080ti will beat that Mac Pro for GPU rendering as well in Cycles and other GPU render engines, for example.

All in all, the new Mac Pro is not a great choice for anyone doing any 3d rendering for their daily work.

I am aware I am comparing apples with oranges, but now compare any of the new Threadripper or Epyc CPUs which reach scores in the 16000-18000 range. Apples' CPU option maxes out at 28 only cores.

No wonder they have to rely on a separate $2000 card to accelerate video rendering to compensate for the Intel CPUs.

Ironically, this is not Apple's fault, but Intel is the one to blame for sitting on their laurels the past decade. Even the most expensive Mac Pro option (the 28 core) renders about as fast as the older AMD Threadripper 2990x (around 11000 points in CB 20). Which burns a $13000 hole in your pocket (base config!!!)! Not even mentioning the Threadripper 3970X which reaches ~17500 points in CB20.

If anything, these new Mac Pros are not meant to do any cost effective 3d rendering on.
 
It's not a $6000 Mac Pro: the 12 core is $7000. And the machine Bob_Stan is actually at least $8400 (AMD Pro Vega II card).

Your machine's 1080ti will beat that Mac Pro for GPU rendering as well in Cycles and other GPU render engines, for example.

All in all, the new Mac Pro is not a great choice for anyone doing any 3d rendering for their daily work.

I am aware I am comparing apples with oranges, but now compare any of the new Threadripper or Epyc CPUs which reach scores in the 16000-18000 range. Apples' CPU option maxes out at 28 only cores.

No wonder they have to rely on a separate $2000 card to accelerate video rendering to compensate for the Intel CPUs.

Ironically, this is not Apple's fault, but Intel is the one to blame for sitting on their laurels the past decade. Even the most expensive Mac Pro option (the 28 core) renders about as fast as the older AMD Threadripper 2990x (around 11000 points in CB 20). Which burns a $13000 hole in your pocket (base config!!!)! Not even mentioning the Threadripper 3970X which reaches ~17500 points in CB20.

If anything, these new Mac Pros are not meant to do any cost effective 3d rendering on.

Shameful isn't it? - Feels like they are preying on people who don't have the hardware knowledge. I know a lot of people trust Apple to make the best decisions on their behalf and in doing so give up some control of their devices for that curated experience.

Personally I would have liked Apple to do some kind of deal with AMD to get a workstation edition of the EPYC processors so they could still offer the very high RAM support that EPYC is known for but negotiate for higher clockspeeds like the Threadripper chips have. Just something in-between would be perfect for the Mac Pro I think.
 
Shameful isn't it? - Feels like they are preying on people who don't have the hardware knowledge. I know a lot of people trust Apple to make the best decisions on their behalf and in doing so give up some control of their devices for that curated experience.

Personally I would have liked Apple to do some kind of deal with AMD to get a workstation edition of the EPYC processors so they could still offer the very high RAM support that EPYC is known for but negotiate for higher clockspeeds like the Threadripper chips have. Just something in-between would be perfect for the Mac Pro I think.

Yes, an EPYC edition would have pushed the new MP into an acceptable proposition. I'd better understand the premium prices asked for this machine seeing that the manufacturing is great, although somewhat over-engineered.

If this machine had come out a year ~ 2 years ago, the pricing would have been tolerable for anyone invested in an OS X machine with good expandibility. In the current market with AMD mixing things up, it feels like a machine that is too little, too late, too costly.

Again, not Apple's fault, I think. Who would have thought only five years ago that AMD would make such a comeback? Apple's engineers banked on Intel, and the new MP is left with a choice of CPU options that feels like yesteryear's popular hit song. Nostalgic, but forgettable.

The new Mac Pro is just screaming its lungs out for a better CPU cost/performance ratio. I think it is quite unacceptable now. I see no 3d artist investing in this machine, unless they happen to be Mac OS X zealots.

It is a crying shame, really. Just bad luck for Apple, I guess. And bad luck for those Mac users investing in this new MP, because they are basically purchasing an outdated previous generation CPU in terms of performance, while the machine's engineering itself is excellent. The one and the other just don't match up at all.
 
Just don't say this over on the Waiting for Mac Pro 7,1 thread. The folks over there that bought one will go ballistic.

Some of us were pointing out that everything in the 7,1 would be obsolete or obsolescent on it's release date, and were pummeled for it.
 
We all knew it didn't we? I'm kinda surprised by its thermal performance though.
In terms of raw power, Intel is not the top anymore, and it's shown.
Even the consumer grade 3950x is formidable at this moment, and that price for MP 7,1 is really not suitable for real heavy work users who rely solely on the raw power of machine. I guess it has merit in the field where users really want Mac OS.

This could have been a great machine if EPYC and nVidia were used, but alas, it's not.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.