News for hackers. Stuff that matters.
Re:iMac Pro 18-core Follow Up Review by NKosmatos on Saturday February 03, 2018 @06:00PM (#16300185)
Would love to see a “real world” benchmark involving Lightroom, Photoshop, Camera raw and Premier. I don’t think the target customer of the iMac pros would run CFD on them ;-)
Re:iMac Pro 18-core Follow Up Review by Recurecur on Saturday February 03, 2018 @06:05PM (#16300210)
Actually I think a good number of high-end Mac users are scientists and engineers. These benchmarks are good data for them.
Re:iMac Pro 18-core Follow Up Review by sgillen on Saturday February 03, 2018 @06:16PM (#16300270)
I haven’t seen very much of this in my experience. Most Mac using scientists and engineers I know use a MacBook for email/webrowsig etc. but when they need some serious computation they use an external cluster (which they talk to with their Mac).

That being said I do know one such scientist with the same workflow but who remoted into a Mac Pro, not sure what they are doing now.

Re:iMac Pro 18-core Follow Up Review by acdha on Saturday February 03, 2018 @07:13PM (#16300469)
Most of the scientists I know do both: interactive work to prove the idea or for moderate sized data sets, and the cluster for when they have stable code and a huge amount of data. Since the capacity of a laptop has steadily increased, the relative ratios have shifted over the years.
Re:iMac Pro 18-core Follow Up Review by danieldk on Sunday February 04, 2018 @03:13AM (#16301888)
I am from a university department with mostly high-end Mac users. But no, we do actual computation on Linux machines. Why? It's the only (reasonable) way to cram capable NVIDIA GPUs, large amounts of memory, large amounts of storage and 32/64 cores in a machine.
Re:iMac Pro 18-core Follow Up Review by PeterisP on Saturday February 03, 2018 @07:11PM (#16300463)
For me, the main use of such hardware is to allow running multiple non-lightweight VMs locally.

It would be convenient if I could run machine learning tasks locally, but that's limited to very small models (since it doesn't have an nvidia GPU and AMD hasn't yet done their homework to ensure that common software packages work), so those tasks need a remote server anyway.

Re:iMac Pro 18-core Follow Up Review by azag0 on Saturday February 03, 2018 @06:04PM (#16300204)
Came to say the same. I’ve always assumed the target audience of these machines is just video editing?

What are the requirements on running, say, TensorFlow? Isn’t it just the GPU there?

Re:iMac Pro 18-core Follow Up Review by matt4077 on Saturday February 03, 2018 @06:39PM (#16300349)
I worked in bioinformatics at a Max Planck Institute during my studies, and made good use of a Mac Pro. Some of the tools we used were interactive, such as 3D-modelling. Many other tasks, such as DNA sequence searches (BLAST), just happened to involve datasets that were too large for most notebooks, but manageable with the 32GB or RAM the Mac Pro had.

We did use a compute cluster, but using it drastically reduced the turnaround time during development in my experience.

Re:iMac Pro 18-core Follow Up Review by jrs95 on Saturday February 03, 2018 @06:32PM (#16300322)
You can't run tensorflow on it because it has an AMD GPU.
Re:iMac Pro 18-core Follow Up Review by arcanus on Saturday February 03, 2018 @08:56PM (#16300828)
Tensorflow has been supported on AMD GPUs since at least November.

https://www.anandtech.com/show/12032/amd-announces-wider-epy...

Re:iMac Pro 18-core Follow Up Review by PeterisP on Saturday February 03, 2018 @09:24PM (#16300925)
No, AMD has claimed that tensorflow works for them since 2016 IIRC, but it's not exactly correct.

The main tensorflow branch does not support AMD GPUs at all.

There's a bunch of experimental code by AMD that possibly works in certain conditions, but is neither properly supported nor widely used. For example, there's hiptensorflow which seems to be a TF port stuck on 1.0.1 (TF is now up to 1.5, with many major releases and key changes), not being actively developed (the last commit seems to be at the end of 2017) and core parts of it (ROCm) don't work in OS X (tensorflow as such runs just fine on OSX, just CPU-only, which is a major performance limitation compared to any machine with a good nvidia gpu), and thus are not that relevant for discussing Apple systems. There are also some OpenCL port efforts, none of which are stable, up to date, and actually usable to get the expected performance improvements. All these projects are interesting, promising, difficult, and their developers deserve all the best, but calling them "Tensorflow has been supported" is an extreme exaggeration, PR vaporware.

Proper support is not there yet. I believe that there are likely ways to jury-rig some code together and run it on AMD hardware, but that's not comparable to the maturity level of these tools on nvidia (essentially, run the install, and it works). AMD still has to do their homework and fulfil their promises. They've spent far too long saying that the software support is coming soon, and now their claims have to be taken with a grain of salt. They can hope that the community will take over effective development a year or two after they do that and people in the ML scene start widely using their hardware, but not earlier. Currently, even if you have a nice AMD GPU, it's cheaper to just buy another high-end GPU from nvidia rather than spend a week of engineer-time to possibly maybe get their shoddy software to do what you need.

Re:iMac Pro 18-core Follow Up Review by danieldk on Sunday February 04, 2018 @03:22AM (#16301903)
Sorry, this is not just a box that you tick off. The CUDA code paths in Tensorflow have been optimized and stabilized for years. Why would anyone drop 5000-10000 Euro on a Tesla-class GPU and get a subpar experience when you can just buy NVIDIA?

AMD has a lot more work to do in this area before people will seriously consider them in GPU computing.

Re:iMac Pro 18-core Follow Up Review by petercooper on Saturday February 03, 2018 @06:10PM (#16300233)
I think we'll see a ton of them pretty soon as content creator types like Marques Brownlee/iJustine/Casey Neistat get their hands on them :)
Re:iMac Pro 18-core Follow Up Review by ams6110 on Saturday February 03, 2018 @06:15PM (#16300265)
Are those software packages written to take advantage of that many cores? Since they all date back 10 years or more, I would guess not.
Re:iMac Pro 18-core Follow Up Review by dboreham on Saturday February 03, 2018 @07:40PM (#16300567)
10 years ago was 2008 : pretty much every machine had multiple cores then, except perhaps low end laptops.

20 years ago in 1998 SMP (before we decided the word "core" sounded nice) was quite common, certainly in machines used by developers and graphics designers.

30 years ago in 1988 SMP in microcomputers was uncommon, but it was very common in minicomputers and mainframes.

Whether software has been written for SMP scaling is quite another thing. Plenty of software written today isn't.

Re:iMac Pro 18-core Follow Up Review by free652 on Saturday February 03, 2018 @08:19PM (#16300693)
My first dual cpu motherboard: https://en.wikipedia.org/wiki/ABIT_BP6

So that's 1999, 19 years ago. You're absolutely right.

Re:iMac Pro 18-core Follow Up Review by JudasGoat on Saturday February 03, 2018 @10:55PM (#16301227)
Yes, mine too. Neat trick they pulled on Intel, allowing non multi processing Celeron's to do MP.
Re:iMac Pro 18-core Follow Up Review by wmf on Saturday February 03, 2018 @06:51PM (#16300386)
Often "old" software like Linpack will be rewritten under the same name. Presumably Intel's implementation is tuned for their latest processors.
Re:iMac Pro 18-core Follow Up Review by gumby on Saturday February 03, 2018 @06:27PM (#16300310)
And you would guess wrong. LINPACK is a lot older than 10years too :-)...more like 35
Re:iMac Pro 18-core Follow Up Review by ams6110 on Saturday February 03, 2018 @06:55PM (#16300406)
Sure but LINPACK has always been for scientific and supercomputing applications. Photoshop is a prosumer PC software application.
Re:iMac Pro 18-core Follow Up Review by acdha on Saturday February 03, 2018 @07:20PM (#16300505)
Photoshop is heavily used by pros (i.e. people for whom time = money) and in scientific or medical imaging. It’s had SMP for a long time (I thought going back to the 90s) and has often been one of the first mainstream adopters of new things like SIMD because it’s a measurable boost for many operations.