1e0000a0c3c04e447f1f17773e67de29cc53146d
benchmarking_linux.md
... | ... | @@ -1,61 +0,0 @@ |
1 | -# Linux Benchmarking |
|
2 | - |
|
3 | -<!-- TOC --> |
|
4 | - - [Benchmark Summaries](#bench-summary) |
|
5 | - - [Benchmarking Systems](#bench-systems) |
|
6 | - - [Benchmarking Datasets](#bench-dataset) |
|
7 | - |
|
8 | -<!-- /TOC --> |
|
9 | - |
|
10 | -## Benchmark Summaries |
|
11 | - |
|
12 | -### Relion 2D classification |
|
13 | - |
|
14 | -System | Runtime (s) | Notes |
|
15 | ------- | ------- | ----- |
|
16 | -Workstation 1 | 712 | |
|
17 | -Workstation 2 | 398 | 2x GPU |
|
18 | -GPU Node 1 | 281 | 4x GPU |
|
19 | - |
|
20 | - |
|
21 | - |
|
22 | -## Benchmarking Systems |
|
23 | - |
|
24 | -<!-- pm-linux0 --> |
|
25 | -### Workstation 1 |
|
26 | - |
|
27 | - - 8x Xeon E5-1630 v4 |
|
28 | - - 128G RAM |
|
29 | - - 1x GeForce RTX 2080 |
|
30 | - - 1 Gb/s network |
|
31 | - |
|
32 | -<!-- haumea --> |
|
33 | -### Workstation 2 |
|
34 | - |
|
35 | - - 48x Xeon E5-268W v4 |
|
36 | - - 128G RAM |
|
37 | - - 2x GeForce GTX 1080 Ti |
|
38 | - - 10 Gb/s network |
|
39 | - |
|
40 | -<!-- sbgrid-gpu1 --> |
|
41 | -### GPU Node 1 |
|
42 | - |
|
43 | - - 64x AMD EPYC 7502 |
|
44 | - - 256G RAM |
|
45 | - - 4x RTX A5000 |
|
46 | - - 10 Gb/s network |
|
47 | - |
|
48 | -#### Notes |
|
49 | -Core counts determined via "nproc" output, and include hyperthreading where enabled. |
|
50 | - |
|
51 | - |
|
52 | - |
|
53 | -## Benchmarking Datasets |
|
54 | - |
|
55 | -### Dataset 1 |
|
56 | -Plasmodium ribosome data set from [Wong et al, 2014](http://dx.doi.org/10.7554/eLife.03080) ( [EMD 2660](https://www.ebi.ac.uk/emdb/EMD-2660) , [EMPIAR-10028](https://www.ebi.ac.uk/empiar/EMPIAR-10028/); courtesy of the [Relion Wiki](https://www3.mrc-lmb.cam.ac.uk/relion/index.php?title=Benchmarks_%26_computer_hardware). |
|
57 | - |
|
58 | -To allow for shorter runtimes during 2D classification benchmarks, input STAR files were truncated by removal of partical entries. |
|
59 | - |
|
60 | - |
|
61 | - |
recommended.md
... | ... | @@ -5,6 +5,7 @@ |
5 | 5 | - [Recommended Hardware and Benchmarking](#recommended-hardware) |
6 | 6 | - [Linux Workstations](#linux-workstations) |
7 | 7 | - [Apple Workstations](#apple-workstations) |
8 | +- [Linux Benchmarking](#linux-benchmarking) |
|
8 | 9 | |
9 | 10 | <!-- /TOC --> |
10 | 11 | |
... | ... | @@ -12,8 +13,61 @@ |
12 | 13 | |
13 | 14 | The majority of structural biology computing happens on linux; there are several vendors providing linux workstation or server configurations we've had good experience with: [ThinkMate](https://www.thinkmate.com), [Exxact](https://www.exxactcorp.com), [Microway](https://www.microway.com). Please feel free to contact us at help@sbgrid.org regarding upcoming hardware, or if you have another hardware vendor to mention. |
14 | 15 | |
15 | -System configurations can be targeted to particular compute-intenstive workflows (for example, Cryo-EM or small molecule docking). We're in the process of assembling benchmark workflows, datasets and [results](benchmarking_linux) to help guide hardware decisions. |
|
16 | +System configurations can be targeted to particular compute-intenstive workflows (for example, Cryo-EM or small molecule docking). We're in the process of assembling benchmark workflows, datasets and [results](#linux-benchmarking) to help guide hardware decisions. |
|
16 | 17 | |
17 | 18 | ## Apple Workstations |
18 | 19 | |
19 | 20 | Any Apple machine can be a capable computer for structural biology. The basic Apple educational discount is available through their [online web store](https://www.apple.com/us-edu/store), and your institution may have negotiated an arrangement with their Apple representative. SBGrid has several labs that run exclusively on Macs and OS X, though these are not advised for CryoEM data processing. |
21 | + |
|
22 | +## Linux Benchmarking |
|
23 | + |
|
24 | +#### Relion 2D classification |
|
25 | + |
|
26 | +System | Runtime (s) | Notes |
|
27 | +------ | ------- | ----- |
|
28 | +Workstation 1 | 712 | |
|
29 | +Workstation 2 | 398 | 2x GPU |
|
30 | +GPU Node 1 | 281 | 4x GPU |
|
31 | + |
|
32 | + |
|
33 | + |
|
34 | +### Benchmarking Systems |
|
35 | + |
|
36 | +<!-- pm-linux0 --> |
|
37 | +#### Workstation 1 |
|
38 | + |
|
39 | + - 8x Xeon E5-1630 v4 |
|
40 | + - 128G RAM |
|
41 | + - 1x GeForce RTX 2080 |
|
42 | + - 1 Gb/s network |
|
43 | + |
|
44 | +<!-- haumea --> |
|
45 | +#### Workstation 2 |
|
46 | + |
|
47 | + - 48x Xeon E5-268W v4 |
|
48 | + - 128G RAM |
|
49 | + - 2x GeForce GTX 1080 Ti |
|
50 | + - 10 Gb/s network |
|
51 | + |
|
52 | +<!-- sbgrid-gpu1 --> |
|
53 | +#### GPU Node 1 |
|
54 | + |
|
55 | + - 64x AMD EPYC 7502 |
|
56 | + - 256G RAM |
|
57 | + - 4x RTX A5000 |
|
58 | + - 10 Gb/s network |
|
59 | + |
|
60 | +##### Notes |
|
61 | +Core counts determined via "nproc" output, and include hyperthreading where enabled. |
|
62 | + |
|
63 | + |
|
64 | + |
|
65 | +### Benchmarking Datasets |
|
66 | + |
|
67 | +#### Dataset 1 |
|
68 | +Plasmodium ribosome data set from [Wong et al, 2014](http://dx.doi.org/10.7554/eLife.03080) ( [EMD 2660](https://www.ebi.ac.uk/emdb/EMD-2660) , [EMPIAR-10028](https://www.ebi.ac.uk/empiar/EMPIAR-10028/); courtesy of the [Relion Wiki](https://www3.mrc-lmb.cam.ac.uk/relion/index.php?title=Benchmarks_%26_computer_hardware). |
|
69 | + |
|
70 | +To allow for shorter runtimes during 2D classification benchmarks, input STAR files were truncated by removal of partical entries. |
|
71 | + |
|
72 | + |
|
73 | + |