Re: [Beowulf] Working for DUG, new thead

Stu, I gotcha. Thanks for the clarification. Prentice On 06/20/2018 09:03 PM, Stu Midgley wrote: But the point of what I was saying is that we DID get real meaningful speedup with just a recompile...  we got enough performance to invest heavily in the technology (it made sense from a $$

Re: [Beowulf] Working for DUG, new thead

Stu, I gotcha. Thanks for the clarification. Prentice On 06/20/2018 09:03 PM, Stu Midgley wrote: But the point of what I was saying is that we DID get real meaningful speedup with just a recompile...  we got enough performance to invest heavily in the technology (it made sense from a $$

Re: [Beowulf] Working for DUG, new thead

But the point of what I was saying is that we DID get real meaningful speedup with just a recompile... we got enough performance to invest heavily in the technology (it made sense from a $$ perspective). Then, every day for years, we got more and more performance, which made the position even

Re: [Beowulf] Working for DUG, new thead

On 06/20/2018 08:36 AM, Michael Di Domenico wrote: On Tue, Jun 19, 2018 at 10:46 PM, Stu Midgley wrote: Where Intel made the mistake was to assume they could shift people from GPUs. People who have spent years writing and optimising won't shift easily... cause they have to go through that

Re: [Beowulf] Working for DUG, new thead

But what was his experience with the Phis? Was he happy with them? Do you know how much he had to rework his code to get good performance out of them? On 06/19/2018 11:57 PM, John Hearns via Beowulf wrote: This thread is going fast! Prentice Bisbal wrote: > I often wonder if that misleading

Re: [Beowulf] Working for DUG, new thead

Stu, I'm welcome to hearing other people's perspectives, especially when they have more experience in an area than I do, but I feel like your experiences below counter my statements as much as they support them. For example, this statement of yours: You can not change architecture

Re: [Beowulf] Working for DUG, new thead

On Wed, Jun 20, 2018 at 08:36:47AM -0400, Michael Di Domenico wrote: > intel could have cornered the market from gpu's if they had released > KNL first and not been so late to the market. once CUDA took hold, > game over. It seems that migrating to ROCm (HIP) from CUDA isn't all that difficult

Re: [Beowulf] Working for DUG, new thead

On Tue, Jun 19, 2018 at 10:46 PM, Stu Midgley wrote: > > Where Intel made the mistake was to assume they could shift people from > GPUs. People who have spent years writing and optimising won't shift > easily... cause they have to go through that whole process again. Getting > people back to

Re: [Beowulf] Working for DUG, new thead

I should do my research... The Celeste project is the poster child for Julia https://www.nextplatform.com/2017/11/28/julia-language-delivers-petascale-hpc-performance/ They use up to 8092 Xeon Phi nodes at NERSC with threads... The per thread runtime graph is interesting there. Only a small

Re: [Beowulf] Working for DUG, new thead

Jim Lux wrote: > I've been intrigued recently about using GPUs for signal processing kinds of things.. There's not much difference between calculating vertices of triangles and doing FIR filters. Rather than look at hardware per se, how about learning about the Julia language for this task? I

Re: [Beowulf] Working for DUG, new thead

This thread is going fast! Prentice Bisbal wrote: > I often wonder if that misleading marketing is one of the reasons why the Xeon Phi has already been canned. I know a lot of people who were excited for the Xeon Phi, but > I don't know any who ever bought the Xeon Phis once they came out. In

Re: [Beowulf] Working for DUG, new thead

we initially used them as standalone systems (ie. rsh a code onto them and run it) today we use them in offload mode (ie. the host would push memory+commands onto them and pull the results off - all via pragmas ). our last KNC systems were 2RU with 8x7120 phi's... which is a 2.1kW system. They

Re: [Beowulf] Working for DUG, new thead

We aren't after average HPC programmers... Even good compilers (Intel) are very very limited in their optimisations. We got factors of 2x and 3x by hand writing SSSE3 commands on standard Xeon's rather than let the compiler do its thing... Compiler limitations isn't particular to Phi. On Wed,

Re: [Beowulf] Working for DUG, new thead

We bought KNC a long time ago and keep meaning to get them to a place where they can be used and just haven’t. Do you mount filesystems from them? We have GPFS storage, primarily, and would have to re-export it via NFS I suppose if we want the cards to use that storage. I’ve seen complaints

Re: [Beowulf] Working for DUG, new thead

On 06/19/2018 03:10 PM, Joe Landman wrote: On 6/19/18 2:47 PM, Prentice Bisbal wrote: On 06/13/2018 10:32 PM, Joe Landman wrote: I'm curious about your next gen plans, given Phi's roadmap. On 6/13/18 9:17 PM, Stu Midgley wrote: low level HPC means... lots of things.  BUT we are a huge

Re: [Beowulf] Working for DUG, new thead

On 6/19/18, 12:11 PM, "Beowulf on behalf of Joe Landman" wrote: Generally, yes. Optimizing serial code for GPUs doesn't work well. Rewriting for GPUs (e.g. taking into account the GPU data/compute flow architecture) does work well. I've been intrigued recently

Re: [Beowulf] Working for DUG, new thead

From: Beowulf on behalf of "beowulf@beowulf.org" Reply-To: John Hearns Date: Tuesday, June 19, 2018 at 11:40 AM To: "beowulf@beowulf.org" Subject: Re: [Beowulf] Working for DUG, new thead >If a sys admin position involves shell prorgramming/scripting, knowing the

Re: [Beowulf] Working for DUG, new thead

I think the boundary between a final product and the start of a project separates these two view points. Lately, I short stacks of OReilleys scattered about, off libraries and a second stack of notebooks filled with every command that really did work. And I think it is fun. Jonathan On Jun 19,

Re: [Beowulf] Working for DUG, new thead

On 6/19/18 2:47 PM, Prentice Bisbal wrote: On 06/13/2018 10:32 PM, Joe Landman wrote: I'm curious about your next gen plans, given Phi's roadmap. On 6/13/18 9:17 PM, Stu Midgley wrote: low level HPC means... lots of things.  BUT we are a huge Xeon Phi shop and need low-level programmers

Re: [Beowulf] Working for DUG, new thead

On 06/19/2018 09:47 PM, Prentice Bisbal wrote: On 06/13/2018 10:32 PM, Joe Landman wrote: I'm curious about your next gen plans, given Phi's roadmap. On 6/13/18 9:17 PM, Stu Midgley wrote: low level HPC means... lots of things.  BUT we are a huge Xeon Phi shop and need low-level

Re: [Beowulf] Working for DUG, new thead

On 06/13/2018 10:32 PM, Joe Landman wrote: I'm curious about your next gen plans, given Phi's roadmap. On 6/13/18 9:17 PM, Stu Midgley wrote: low level HPC means... lots of things.  BUT we are a huge Xeon Phi shop and need low-level programmers ie. avx512, careful cache/memory management

Re: [Beowulf] Working for DUG, new thead

>If a sys admin position involves shell prorgramming/scripting, knowing the details of a specific programming language or processor are secondary, but thinking like a >programmer is skill not everyone has or can develop.Just last week a wrote a Lua script without knowing a thing about Lua. I

Re: [Beowulf] Working for DUG, new thead

Despite the source (just kidding, Bill!) I'm going to have to support this line of questioning. HPC (and general IT), consists of systems with many different layers, and it takes the correct personality type with the good analytical skills to be able to trouble shoot things effectively. I

Re: [Beowulf] Working for DUG, new thead

Phi is dead... Long live phi... By which I mean, while the Phi as a chip is going away, its concepts live on. Massive number of cores, large vectorisation and high speed memory (and fucking high heat load - we do ~350W/socket). So, while the product code will disappear, phi lives on. For KNC I

Re: [Beowulf] Working for DUG, new thead

, and then iterate with something to get the nonlinear solution – all of this takes lots o’ cycles. From: Beowulf on behalf of Stu Midgley Reply-To: "sdm...@gmail.com" Date: Wednesday, June 13, 2018 at 7:35 PM To: "joe.land...@gmail.com" Cc: "beowulf@beowulf.org"

Re: [Beowulf] Working for DUG, new thead

of Andrew Latham Date: Wednesday, June 13, 2018 at 11:28 AM Cc: "beowulf@beowulf.org" Subject: Re: [Beowulf] Working for DUG, new thead Such a broad topic. I would assume things like DHCP, TFTP, Networking, PXE and IPMI which come to mind. Troubleshooting tools, configuration management

Re: [Beowulf] Working for DUG, new thead

seismic processing is a MASSIVE user of compute resources. We run single processing steps that can take months on a 10PFlop machine... On Thu, Jun 14, 2018 at 10:25 AM Joe Landman wrote: > > > On 6/13/18 9:39 PM, Stu Midgley wrote: > > seismic processing - oil'n'gas not hard rock. > > > But

Re: [Beowulf] Working for DUG, new thead

I'm curious about your next gen plans, given Phi's roadmap. On 6/13/18 9:17 PM, Stu Midgley wrote: low level HPC means... lots of things.  BUT we are a huge Xeon Phi shop and need low-level programmers ie. avx512, careful cache/memory management (NOT openmp/compiler vectorisation etc). I

Re: [Beowulf] Working for DUG, new thead

On 6/13/18 9:39 PM, Stu Midgley wrote: seismic processing - oil'n'gas not hard rock. But ... you work with heavy metal (supercomputers that is) :D Good to see you active here, BTW! On Thu, Jun 14, 2018 at 3:32 AM Jonathan Engwall > wrote:

Re: [Beowulf] Working for DUG, new thead

64 bit words have me wondering. I have been interested in monolithic lately too. Well, I hope to get a call...and did. I think it went well. She did not find my pip3 story exciting. And thanks again ever one. On Jun 13, 2018 6:39 PM, "Stu Midgley" wrote: > seismic processing - oil'n'gas not

Re: [Beowulf] Working for DUG, new thead

seismic processing - oil'n'gas not hard rock. On Thu, Jun 14, 2018 at 3:32 AM Jonathan Engwall < engwalljonathanther...@gmail.com> wrote: > Yes. Very good idea, it is a mining company in Ausralia. > > -- Dr Stuart Midgley sdm...@gmail.com ___ Beowulf

Re: [Beowulf] Working for DUG, new thead

low level HPC means... lots of things. BUT we are a huge Xeon Phi shop and need low-level programmers ie. avx512, careful cache/memory management (NOT openmp/compiler vectorisation etc). On Thu, Jun 14, 2018 at 1:08 AM Jonathan Engwall < engwalljonathanther...@gmail.com> wrote: > John Hearne

Re: [Beowulf] Working for DUG, new thead

OR... we could answer your questions directly... Stu (who has worked at DUG for 12 years, built and designed the systems... was CTO and now Systems Architect) On Thu, Jun 14, 2018 at 3:32 AM Jonathan Engwall < engwalljonathanther...@gmail.com> wrote: > Yes. Very good idea, it is a mining

Re: [Beowulf] Working for DUG, new thead

We aren't that scary... On Thu, Jun 14, 2018 at 1:08 AM Jonathan Engwall < engwalljonathanther...@gmail.com> wrote: > John Hearne wrote: > > Stuart Midgley works for DUG? They are currently > > recruiting for an HPC manager in London... Interesting... > > Recruitment at DUG wants to call me

Re: [Beowulf] Working for DUG, new thead

Thanks for the great feedback. LART? What is that? I am doing a quick start, and just by chance I have been watching an MIT lecture on SSE. https://m.youtube.com/watch?v=Pc8DfEyAxzg This is from a YouTuber who calls himself Biscuit. Quick and beautiful with pragmas and the Mandelbrot's Set.

Re: [Beowulf] Working for DUG, new thead

Add to that "Please identify the situations that justify using a LART?" https://en.wiktionary.org/wiki/LART -- Doug > One of my standard interview questions is to say ok, you start on Monday > and you're placed in charge of a web/db server, tell me what you do your > first week. > > What I

Re: [Beowulf] Working for DUG, new thead

Yes. Very good idea, it is a mining company in Ausralia. On Jun 13, 2018 12:23 PM, "Ryan Novosielski" wrote: > > On Jun 13, 2018, at 1:07 PM, Jonathan Engwall < > engwalljonathanther...@gmail.com> wrote: > > > > John Hearne wrote: > > > Stuart Midgley works for DUG? They are currently > > >

Re: [Beowulf] Working for DUG, new thead

> On Jun 13, 2018, at 1:07 PM, Jonathan Engwall > wrote: > > John Hearne wrote: > > Stuart Midgley works for DUG? They are currently > > recruiting for an HPC manager in London... Interesting... > > Recruitment at DUG wants to call me about Low Level HPC. I have at least > until 6pm. > I am

Re: [Beowulf] Working for DUG, new thead

I thought it was for sysadmin, not developer. Disregard most of what I said. Bill On 06/13/2018 02:49 PM, Fred Youhanaie wrote: On 13/06/18 18:07, Jonathan Engwall wrote: John Hearne wrote:  > Stuart Midgley works for DUG?  They are currently  > recruiting for an HPC manager in London...

Re: [Beowulf] Working for DUG, new thead

Bill's question re. "the cluster is slow"is fantastic. That covers people skills in addition to technical skills. ps. best of luck. You never know who you might end up working with ;-) On 13 June 2018 at 20:46, Andrew Latham wrote: > Bill's question is good and I have heard it many times.

Re: [Beowulf] Working for DUG, new thead

On 13/06/18 18:07, Jonathan Engwall wrote: John Hearne wrote: > Stuart Midgley works for DUG?  They are currently > recruiting for an HPC manager in London... Interesting... Recruitment at DUG wants to call me about Low Level HPC. I have at least until 6pm. I am excited but also terrified.

Re: [Beowulf] Working for DUG, new thead

Bill's question is good and I have heard it many times. Practice answering that. Related examples: Tell me how you would setup and install 100 new servers? How do you describe to onsite support staff (smart hands) to move a network cable to another port on the same system? At 2am you are trouble

Re: [Beowulf] Working for DUG, new thead

One of my standard interview questions is to say ok, you start on Monday and you're placed in charge of a web/db server, tell me what you do your first week. What I want to hear is security, backups, log checking, monitoring, performance, functionality, etc., but most of all I want to know

Re: [Beowulf] Working for DUG, new thead

Jonathan, if you have taken the interest to join this list then there is no need to be terrified. I have learned that people who are enthusiastic are quite rare. Also interviews are a two way street - this is your opportunity to find out about the role. Hopefully you will be enthused about it and

Re: [Beowulf] Working for DUG, new thead

Such a broad topic. I would assume things like DHCP, TFTP, Networking, PXE and IPMI which come to mind. Troubleshooting tools, configuration management, version control, monitoring, issue tracking and many other processes are at play. I would love to hear any interview questions which could cover

Re: [Beowulf] Working for DUG, new thead

s/linux/cabling/g On 06/13/2018 02:04 PM, Prentice Bisbal wrote: I would think that "low-level" means "close to the hardware" (troubleshooting drivers, low-level hardware debugging, etc. ) , and that junior-level would mean "basic Linux system admin skills", but they're in Australia, so they

Re: [Beowulf] Working for DUG, new thead

I would think that "low-level" means "close to the hardware" (troubleshooting drivers, low-level hardware debugging, etc. ) , and that junior-level would mean "basic Linux system admin skills", but they're in Australia, so they might use these terms differently. Prentice On 06/13/2018 01:53

Re: [Beowulf] Working for DUG, new thead

linux, mostly On 06/13/2018 01:07 PM, Jonathan Engwall wrote: John Hearne wrote: > Stuart Midgley works for DUG?  They are currently > recruiting for an HPC manager in London... Interesting... Recruitment at DUG wants to call me about Low Level HPC. I have at least until 6pm. I am excited