MP 7,1 Has anyone seen a benchmark?

[danwells]

The only numbers I've seen published online are in Marquess Brownlee's YouTube video (28 cores, 384 GB RAM)...

Geekbench 5 Single-core ~1200 (very similar to all other fast Macs - iMac 8 core, iMac Pro, MBP16)
Geekbench 5 Multi-core ~21500 (about 3x as fast as the logjam of fast Macs, 1.5x as fast as 18-core iMac Pro)
Disk Speed Test - just under 3 GB/second (a tiny bit faster than iMac Pro and MBP16, the fastest other Macs)

Has anyone seen any other configuration, or any other test?

 

[defjam]

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?

 

[Polymorphic]

Probably not the only reason, but:

Sending a Mac Pro to AnandTech would no doubt result in some benchmarks against Threadripper 3000, which wouldn't portray Cascade Lake or this Mac Pro in the best light.

 

[Lobwedgephil]

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?

Just an extension of Apple marketing. No other reason.

 

[ssgbryan]

Geekbench runs too fast to make a useful metric.

And as others have pointed out, technical websites will compare the 7,1 to it's real world competition. Since it consists of obsolete technologies, it won't do very well in general bench marking.

The target audience for the 7,1 is professional video and audio houses. Those folks don't do that much in the way of reviewing.

 

[defjam]

Geekbench runs too fast to make a useful metric.

And as others have pointed out, technical websites will compare the 7,1 to it's real world competition. Since it consists of obsolete technologies, it won't do very well in general bench marking.

The target audience for the 7,1 is professional video and audio houses. Those folks don't do that much in the way of reviewing.

I just watched a video by Snazzy Labs about their AMD 7,1 Mac Pro killer. Unfortunately it was published about a month ago before the Mac Pro had been released. So I had to take their statement with a grain of salt.

However one benchmark they provided, the flash drive speed, was interesting to me as it was also mentioned by one of the YouTube bloggers who did evaluate a Mac Pro. The Snazzy Labs video stated their SSD speed ~4GB/sec whereas the YouTube blogger, Marques Brownlee, stated his observed SSD speed was ~3GB/sec. Given Apples attention and history for fast SSDs I thought that an interesting and significant difference (33% faster on the AMD system).

The Snazzy Labs video is can be found here AMD HACK MAC KILLS THE NEW MAC PRO!
The Marques Brownlee YouTube video can be found here Mac Pro Unboxing & Second Impressions: The Power is Back!

On an entirely different note it seems awfully quiet in the forum the past couple of days.

 

[MGrayson3]

And as others have pointed out, technical websites will compare the 7,1 to it's real world competition. Since it consists of obsolete technologies, it won't do very well in general bench marking.

That's the problem. They WON'T compare it to its real world competition. A gaming machine will beat it on every performance spec, but not running at 100% for a week. So they'll run it for 15 minutes and conclude that it's a total waste, and the haters will lean back and smirk.

Want an Alienware Threadripper? Get an Alienware Threadripper.

 

[defjam]

That's the problem. They WON'T compare it to its real world competition. A gaming machine will beat it on every performance spec, but not running at 100% for a week. So they'll run it for 15 minutes and conclude that it's a total waste, and the haters will lean back and smirk.

I'm especially interested in the GPU performance.

 

[Polymorphic]

I just watched a video by Snazzy Labs about their AMD 7,1 Mac Pro killer. Unfortunately it was published about a month ago before the Mac Pro had been released. So I had to take their statement with a grain of salt.

However one benchmark they provided, the flash drive speed, was interesting to me as it was also mentioned by one of the YouTube bloggers who did evaluate a Mac Pro. The Snazzy Labs video stated their SSD speed ~4GB/sec whereas the YouTube blogger, Marques Brownlee, stated his observed SSD speed was ~3GB/sec. Given Apples attention and history for fast SSDs I thought that an interesting and significant difference (33% faster on the AMD system).

The Snazzy Labs video is can be found here AMD HACK MAC KILLS THE NEW MAC PRO!
The Marques Brownlee YouTube video can be found here Mac Pro Unboxing & Second Impressions: The Power is Back!

On an entirely different note it seems awfully quiet in the forum the past couple of days.

Faster SSD speeds on Windows have been a thing for a while now. Many PCIe 3.0 NVMe SSDs can exceed 3 GB/s read speeds, and some PCIe 4.0 SSDs can do even more.

My ThinkPad P50 from 2015 has faster NVMe SSD speeds than any T2-equipped Mac I've seen.

 

[defjam]

Faster SSD speeds on Windows have been a thing for a while now. Many PCIe 3.0 NVMe SSDs can exceed 3 GB/s read speeds, and some PCIe 4.0 SSDs can do even more.

My ThinkPad P50 from 2015 has faster NVMe SSD speeds than any T2-equipped Mac I've seen.

I just thought it interesting the gap was so large. Especially given how Apple has designed their SSD drives.

 

[Pressure]

I just thought it interesting the gap was so large. Especially given how Apple has designed their SSD drives.

Please have in mind there is more to a solid state disk controller than a single performance metric. It all depends on workload as we rarely only need read or write speed but a mixture of the two. Then comes the amount of SLC cache which dictates whether or not the performance is sustained or merely long enough to benchmark well. NAND endurance (how many daily writes) is also important.

 

[danwells]

Apple has a very good SSD controller, and we see the same one on Mac Pro, iMac and recent MBPs (Mac Mini as well)?

 

[defjam]

Please have in mind there is more to a solid state disk controller than a single performance metric. It all depends on workload as we rarely only need read or write speed but a mixture of the two. Then comes the amount of SLC cache which dictates whether or not the performance is sustained or merely long enough to benchmark well. NAND endurance (how many daily writes) is also important.

I agree 100% which is why I am interested in seeing more benchmarks. However the Snazzy Labs benchmark was performed with Black Maginc which, in these forums, has always been held up as the gold standard of disk benchmarks (which I do not agree with). I can't recall what the YouTube blogger used for their benchmark...I'll have to go back and look.

Regardless I want to see more than a single test.

 

[danwells]

Marques Brownlee used BlackMagic Disk Speed Test

 

[fokmik]

 

[Digital_Sousaphone]

That's the problem. They WON'T compare it to its real world competition. A gaming machine will beat it on every performance spec, but not running at 100% for a week. So they'll run it for 15 minutes and conclude that it's a total waste, and the haters will lean back and smirk.

Want an Alienware Threadripper? Get an Alienware Threadripper.

And it's real world competition trounces it to smithereens. Not exactly sure what your point is?

 

[Pressure]

MacOS only supports up to 64-threads anyway.

 

[smbu2000]

I just watched a video by Snazzy Labs about their AMD 7,1 Mac Pro killer. Unfortunately it was published about a month ago before the Mac Pro had been released. So I had to take their statement with a grain of salt.

However one benchmark they provided, the flash drive speed, was interesting to me as it was also mentioned by one of the YouTube bloggers who did evaluate a Mac Pro. The Snazzy Labs video stated their SSD speed ~4GB/sec whereas the YouTube blogger, Marques Brownlee, stated his observed SSD speed was ~3GB/sec. Given Apples attention and history for fast SSDs I thought that an interesting and significant difference (33% faster on the AMD system).

The Snazzy Labs video is can be found here AMD HACK MAC KILLS THE NEW MAC PRO!
The Marques Brownlee YouTube video can be found here Mac Pro Unboxing & Second Impressions: The Power is Back!

On an entirely different note it seems awfully quiet in the forum the past couple of days.

All Intel cpus are limited to pcie 3.0 bandwidth which tops out at 3500MB/s. If used on an X570/TR4 motherboard then the latest Ryzen/Threadripper 3000 models can support pcie 4.0 bandwidth which tops out at 7000MB/s. Pcie 4.0 nvme drives are still very new so speeds are still “slower“, 4000-4500 range, right now. Still much faster than pcie 3.0 nvme drives.

 

[thisisnotmyname]

All Intel cpus are limited to pcie 3.0 bandwidth which tops out at 3500MB/s. If used on an X570/TR4 motherboard then the latest Ryzen/Threadripper 3000 models can support pcie 4.0 bandwidth which tops out at 7000MB/s. Pcie 4.0 nvme drives are still very new so speeds are still “slower“, 4000-4500 range, right now. Still much faster than pcie 3.0 nvme drives.

no. PCIe v3 is 1GB/s per lane = 16GB/s on each x16 slot.

edit to add: maybe you meant to imply on x4 after overhead? In any case, RAID together on an x16 card, plenty of bandwidth.

 

[Truefan31]

Impressive

 

[smbu2000]

no. PCIe v3 is 1GB/s per lane = 16GB/s on each x16 slot.

edit to add: maybe you meant to imply on x4 after overhead? In any case, RAID together on an x16 card, plenty of bandwidth.

Yes, I was referring to the typical M.2 X4 slots found on most motherboards. Intel based boards using 3.0 x4 M.2 slots and the 4.0 x4 M.2 slots on AMD X570/TRX40 motherboards. Needing to use more (slower) PCIe lanes to get the same bandwidth does negate some of the benefits of having more PCIe lanes in the first place. A weak spot of the Intel based cpus compared to Threadripper.

 

[chfilm]

https://blog.vincentlaforet.com/2019/12/10/macpro/

 

[vel0city]

https://blog.vincentlaforet.com/2019/12/10/macpro/

Brilliant, thank you for this.

 

[chfilm]

Brilliant, thank you for this.

Yea, was happy to discover this just now, finally a professional who's not a youtuber talking about the thing, also in a more real life config at 16, 2xVega and 192gb of ram.

 

[defjam]

https://blog.vincentlaforet.com/2019/12/10/macpro/

What I find interesting about these reviews is the substandard level of performance Mac users have tolerated from Apple over the past several years. This reviewer states:

For more than a decade, the top of the line still cameras and motion picture cameras have brought almost all computers to their knees, and this is the first machine that I’ve worked with that can stay ahead of the ever-increasing onslaught of pixels and Terabytes of data.​

Systems just as or more capable than the 2019 Mac Pro have existed for years, they're just not Apple systems. Everyone who has used the 2019 Mac Pro appears to be in awe as to its capabilities, as if Apple has done something no other workstation vendors has done before (granted MPX and After Burner are two things which are unique to the 2019 Mac Pro).

I guess when you've been saddled with 2013 technology for so long anything new has got to be exciting.

All this attention to this new Mac Pro demonstrate the "nay sayers" of the 2013 Mac Pro, Mini, and iMac Pro as not being substitutes for a full fledge workstation were right. We've been vindicated!