|
[Sponsors] |
October 18, 2021, 16:00 |
M1 Max
|
#1 |
Senior Member
Join Date: May 2012
Posts: 551
Rep Power: 16 |
400 GB/s Memory Bandwidth seems interesting. Now if we can get full Linux functionality and someone with unlimited funding that can run some OF benchmarks, that would be great. Even if this may be the best performer per Watt, I highly doubt it will be the best performer per $
And Intel's DDR5 offerings are just around the corner. At the same time AMD's 7003 series is doing great in the CFD department. Interesting. |
|
October 25, 2021, 23:03 |
|
#2 | |
Member
|
Just wondering if there is a way to install native openfoam codes on m1.
According to anandtech, both latency and bandwidth are quite impressive. https://www.anandtech.com/show/17024...mance-review/2 Acctually even a native Fluent installer would make huge difference Quote:
|
||
October 28, 2021, 21:37 |
|
#3 |
Member
Join Date: Jun 2016
Posts: 99
Rep Power: 10 |
If M1 can utilize GPU to solve matrices without mem copy, it could be even faster.
|
|
October 29, 2021, 03:49 |
|
#4 |
Member
|
I digged a little bit further, it seems that the the memory bandwidth is not as huge as I would have hoped.
No direct CFD benckmark yet, but Anandtech post the resutls of SPEC2017, the 503 subscore is around 300. https://www.anandtech.com/show/17024...mance-review/5 This is way better than a typical dual channel ddr4 but still lower than an 8-channel epyc, which scores around 350. https://www.anandtech.com/show/16594...lable-review/6 According to their website, benchmark 503 falls into the category of CFD and "simulates blast waves in three dimensional transonic transient laminar viscous flow" https://www.spec.org/cpu2017/Docs/be....bwaves_r.html The score might not have a linear relationship with Openfoam (or Fluent) performance but I think it is good reference. If we assume a m1 max performs similarly to a 1-socket epyc 7002(3), then the price is not that attracktive from performance point of view. Maybe the future mac mini (withought the cost the display, keyboard, etc) or the rumored 2-socket mac pro is the better choice. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
y+ and u+ values with low-Re RANS turbulence models: utility + testcase | florian_krause | OpenFOAM | 114 | August 23, 2023 05:37 |
Converging Diverging Nozzle with dbnsTurbFoam | Saleh Abuhanieh | OpenFOAM Running, Solving & CFD | 4 | December 13, 2019 10:26 |
Near wall treatment in k-omega SST | Arnoldinho | OpenFOAM Running, Solving & CFD | 38 | March 8, 2017 13:48 |
On the damBreak4phaseFine cases | paean | OpenFOAM Running, Solving & CFD | 0 | November 14, 2008 21:14 |
Could anybody help me see this error and give help | liugx212 | OpenFOAM Running, Solving & CFD | 3 | January 4, 2006 18:07 |