OVHcloud GPU benchmark - Llama 3

Go back to list

Ollama is a cross-platform sofware designed to run large language models (LLMs) locally on any computer. It supports a variety of AI models, including LLaMA-3, Mistral, Falcon, and more.  To assess the response speed, we employed the ollama-benchmark tool, measuring throughput in tokens per second. The specific command-line invocation used for our testing is detailed below:

model=llama3:8b  # or llama3:70b
question=81 # or 82, 83, 84 and 85'
ollama-benchmark speed --model $model --question $question_id --max-workers 1 --max_turns 1 --mirostat 0 --mirostat_eta 0.1 --mirostat_tau 5.0 --num_ctx 2048 --repeat_last_n 64 --repeat_penalty 1.1 --temperature 0.8 --seed 0 --tfs_z 1.0 --num_predict 128 --top_k 40 --top_p 0.9

 

{ "configuration": { "chart": { "type": "bar", "polar": false, "zoomType": "", "options3d": {}, "height": 550, "width": null, "margin": null, "inverted": false, "zooming": {} }, "credits": { "enabled": false }, "title": { "text": "Llama3" }, "colorAxis": null, "subtitle": { "text": "" }, "xAxis": { "title": { "text": [ "" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": [ "provider__short_name", "flavor__name", "flavor__gpu_model" ], "lineWidth": 1, "tickInterval": null, "tickWidth": 0, "tickLength": 10, "tickPixelInterval": null, "plotLines": null, "labels": { "enabled": false, "format": null, "formatter": "", "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false, "step": 0 }, "plotBands": null, "visible": true, "floor": null, "ceiling": null, "type": "linear", "min": null, "gridLineWidth": null, "gridLineColor": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "tickmarkPlacement": null, "units": null, "minRange": null }, "yAxis": { "title": { "text": [ "Token/second<br>Higher is better" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": null, "plotLines": null, "plotBands": null, "lineWidth": null, "tickInterval": null, "tickLength": 10, "floor": null, "ceiling": null, "gridLineInterpolation": null, "gridLineWidth": 1, "gridLineColor": "#CCC", "min": null, "max": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "minRange": null, "type": "linear", "tickmarkPlacement": null, "labels": { "enabled": true, "formatter": null, "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false } }, "zAxis": { "title": { "text": "Token/second<br>Higher is better" } }, "plotOptions": { "series": { "dataLabels": { "enabled": true, "format": "{series.name}", "distance": 30, "align": "left", "inside": true, "allowOverlap": false, "style": { "fontSize": "17px" } }, "showInLegend": null, "turboThreshold": 1000, "stacking": "", "groupPadding": 0, "centerInCategory": false, "findNearestPointBy": "x" } }, "navigator": { "enabled": false }, "scrollbar": { "enabled": false }, "rangeSelector": { "enabled": false, "inputEnabled": false }, "legend": { "enabled": true, "maxHeight": null, "align": "center", "verticalAlign": "bottom", "layout": "horizontal", "width": null, "margin": 12, "reversed": false }, "series": [ { "name": "AWS g5.xlarge NVIDIA A10G", "data": [ [ 0, 74.11145583602332 ] ], "grouping": false, "color": "#f7981d" }, { "name": "AWS p3.2xlarge Tesla V100-SXM2-16GB", "data": [ [ 1, 84.25722555865242 ] ], "grouping": false, "color": "#f7981d" }, { "name": "Azure Standard_NC40ads_H100_v5 NVIDIA H100 NVL", "data": [ [ 2, 191.02202188374866 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Azure Standard_NC6s_v3 Tesla V100-PCIE-16GB", "data": [ [ 3, 85.9395876676278 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Azure Standard_NV36ads_A10_v5 NVIDIA A10", "data": [ [ 4, 79.82683635781541 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Google a2-highgpu-1g NVIDIA A100-SXM4-40GB", "data": [ [ 5, 94.25855110935245 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google a2-ultragpu-1g NVIDIA A100-SXM4-80GB", "data": [ [ 6, 98.54227258468484 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google g2-standard-16 NVIDIA L4", "data": [ [ 7, 44.20125112599654 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google n1-highmem-8 Intel Skylake Tesla V100 Tesla V100-SXM2-16GB", "data": [ [ 8, 88.48112862998406 ] ], "grouping": false, "color": "#55b400" }, { "name": "OVH H100-380 NVIDIA H100 PCIe", "data": [ [ 9, 120.23815120370868 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH L4-90 NVIDIA L4", "data": [ [ 10, 48.92874165631959 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH L40S-90 NVIDIA L40S", "data": [ [ 11, 115.28244713897865 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH T1-LE-45 Tesla V100-PCIE-16GB", "data": [ [ 12, 84.11657733154148 ] ], "grouping": false, "color": "#484848" } ], "drilldown": {}, "tooltip": { "enabled": true, "useHTML": false, "format": null, "headerFormat": "", "pointFormat": "<span style=\"color:{series.color}\">{series.name}</span>: <b>{point.y:.1f} token/sec</b>", "footerFormat": "", "shared": false, "outside": false, "valueDecimals": null, "split": false }, "annotations": null }, "hc_type": "chart", "id": "llama3" }
{ "configuration": { "chart": { "type": "bar", "polar": false, "zoomType": "", "options3d": {}, "height": 550, "width": null, "margin": null, "inverted": false, "zooming": {} }, "credits": { "enabled": false }, "title": { "text": "Llama3:70b" }, "colorAxis": null, "subtitle": { "text": "" }, "xAxis": { "title": { "text": [ "" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": [ "provider__short_name", "flavor__name", "flavor__gpu_model" ], "lineWidth": 1, "tickInterval": null, "tickWidth": 0, "tickLength": 10, "tickPixelInterval": null, "plotLines": null, "labels": { "enabled": false, "format": null, "formatter": "", "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false, "step": 0 }, "plotBands": null, "visible": true, "floor": null, "ceiling": null, "type": "linear", "min": null, "gridLineWidth": null, "gridLineColor": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "tickmarkPlacement": null, "units": null, "minRange": null }, "yAxis": { "title": { "text": [ "Token/second<br>Higher is better" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": null, "plotLines": null, "plotBands": null, "lineWidth": null, "tickInterval": null, "tickLength": 10, "floor": null, "ceiling": null, "gridLineInterpolation": null, "gridLineWidth": 1, "gridLineColor": "#CCC", "min": null, "max": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "minRange": null, "type": "linear", "tickmarkPlacement": null, "labels": { "enabled": true, "formatter": null, "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false } }, "zAxis": { "title": { "text": "Token/second<br>Higher is better" } }, "plotOptions": { "series": { "dataLabels": { "enabled": true, "format": "{series.name}", "distance": 30, "align": "left", "inside": true, "allowOverlap": false, "style": { "fontSize": "17px" } }, "showInLegend": null, "turboThreshold": 1000, "stacking": "", "groupPadding": 0, "centerInCategory": false, "findNearestPointBy": "x" } }, "navigator": { "enabled": false }, "scrollbar": { "enabled": false }, "rangeSelector": { "enabled": false, "inputEnabled": false }, "legend": { "enabled": true, "maxHeight": null, "align": "center", "verticalAlign": "bottom", "layout": "horizontal", "width": null, "margin": 12, "reversed": false }, "series": [ { "name": "Azure Standard_NC40ads_H100_v5 NVIDIA H100 NVL", "data": [ [ 0, 40.71168018178036 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Google a2-highgpu-1g NVIDIA A100-SXM4-40GB", "data": [ [ 1, 14.015375297515288 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google a2-ultragpu-1g NVIDIA A100-SXM4-80GB", "data": [ [ 2, 25.291842674146736 ] ], "grouping": false, "color": "#55b400" }, { "name": "OVH H100-380 NVIDIA H100 PCIe", "data": [ [ 3, 29.7484356028196 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH L40S-90 NVIDIA L40S", "data": [ [ 4, 16.64672521178475 ] ], "grouping": false, "color": "#484848" } ], "drilldown": {}, "tooltip": { "enabled": true, "useHTML": false, "format": null, "headerFormat": "", "pointFormat": "<span style=\"color:{series.color}\">{series.name}</span>: <b>{point.y:.1f} token/sec</b>", "footerFormat": "", "shared": false, "outside": false, "valueDecimals": null, "split": false }, "annotations": null }, "hc_type": "chart", "id": "llama3:70b" }
{ "configuration": { "chart": { "type": "bar", "polar": false, "zoomType": "", "options3d": {}, "height": 550, "width": null, "margin": null, "inverted": false, "zooming": {} }, "credits": { "enabled": false }, "title": { "text": "Llama3.1:8b-instruct-q8_0" }, "colorAxis": null, "subtitle": { "text": "" }, "xAxis": { "title": { "text": [ "" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": [ "provider__short_name", "flavor__name", "flavor__gpu_model" ], "lineWidth": 1, "tickInterval": null, "tickWidth": 0, "tickLength": 10, "tickPixelInterval": null, "plotLines": null, "labels": { "enabled": false, "format": null, "formatter": "", "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false, "step": 0 }, "plotBands": null, "visible": true, "floor": null, "ceiling": null, "type": "linear", "min": null, "gridLineWidth": null, "gridLineColor": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "tickmarkPlacement": null, "units": null, "minRange": null }, "yAxis": { "title": { "text": [ "Token/second<br>Higher is better" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": null, "plotLines": null, "plotBands": null, "lineWidth": null, "tickInterval": null, "tickLength": 10, "floor": null, "ceiling": null, "gridLineInterpolation": null, "gridLineWidth": 1, "gridLineColor": "#CCC", "min": null, "max": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "minRange": null, "type": "linear", "tickmarkPlacement": null, "labels": { "enabled": true, "formatter": null, "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false } }, "zAxis": { "title": { "text": "Token/second<br>Higher is better" } }, "plotOptions": { "series": { "dataLabels": { "enabled": true, "format": "{series.name}", "distance": 30, "align": "left", "inside": true, "allowOverlap": false, "style": { "fontSize": "17px" } }, "showInLegend": null, "turboThreshold": 1000, "stacking": "", "groupPadding": 0, "centerInCategory": false, "findNearestPointBy": "x" } }, "navigator": { "enabled": false }, "scrollbar": { "enabled": false }, "rangeSelector": { "enabled": false, "inputEnabled": false }, "legend": { "enabled": true, "maxHeight": null, "align": "center", "verticalAlign": "bottom", "layout": "horizontal", "width": null, "margin": 12, "reversed": false }, "series": [ { "name": "AWS g5.xlarge NVIDIA A10G", "data": [ [ 0, 49.45721996577291 ] ], "grouping": false, "color": "#f7981d" }, { "name": "AWS p3.2xlarge Tesla V100-SXM2-16GB", "data": [ [ 1, 61.257833654542665 ] ], "grouping": false, "color": "#f7981d" }, { "name": "Azure Standard_NC40ads_H100_v5 NVIDIA H100 NVL", "data": [ [ 2, 160.02328797391598 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Azure Standard_NC6s_v3 Tesla V100-PCIE-16GB", "data": [ [ 3, 63.16613714377257 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Google a2-highgpu-1g NVIDIA A100-SXM4-40GB", "data": [ [ 4, 82.24751543850662 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google a2-ultragpu-1g NVIDIA A100-SXM4-80GB", "data": [ [ 5, 91.5994552681905 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google g2-standard-16 NVIDIA L4", "data": [ [ 6, 27.006729680994383 ] ], "grouping": false, "color": "#55b400" }, { "name": "Google n1-highmem-8 Intel Skylake Tesla V100 Tesla V100-SXM2-16GB", "data": [ [ 7, 64.7023253930343 ] ], "grouping": false, "color": "#55b400" }, { "name": "OVH A10-45 NVIDIA A10", "data": [ [ 8, 52.23567869632677 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH H100-380 NVIDIA H100 PCIe", "data": [ [ 9, 124.8827190203083 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH L4-90 NVIDIA L4", "data": [ [ 10, 28.934930816750924 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH L40S-90 NVIDIA L40S", "data": [ [ 11, 72.03119918751457 ] ], "grouping": false, "color": "#484848" }, { "name": "OVH T1-45 Tesla V100-PCIE-16GB", "data": [ [ 12, 60.47893212079448 ] ], "grouping": false, "color": "#484848" } ], "drilldown": {}, "tooltip": { "enabled": true, "useHTML": false, "format": null, "headerFormat": "", "pointFormat": "<span style=\"color:{series.color}\">{series.name}</span>: <b>{point.y:.1f} token/sec</b>", "footerFormat": "", "shared": false, "outside": false, "valueDecimals": null, "split": false }, "annotations": null }, "hc_type": "chart", "id": "llama3.1:8b-instruct-q8_0" }
{ "configuration": { "chart": { "type": "bar", "polar": false, "zoomType": "", "options3d": {}, "height": 550, "width": null, "margin": null, "inverted": false, "zooming": {} }, "credits": { "enabled": false }, "title": { "text": "Llama3.1:70b-instruct-q8_0" }, "colorAxis": null, "subtitle": { "text": "" }, "xAxis": { "title": { "text": [ "" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": [ "provider__short_name", "flavor__name", "flavor__gpu_model" ], "lineWidth": 1, "tickInterval": null, "tickWidth": 0, "tickLength": 10, "tickPixelInterval": null, "plotLines": null, "labels": { "enabled": false, "format": null, "formatter": "", "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false, "step": 0 }, "plotBands": null, "visible": true, "floor": null, "ceiling": null, "type": "linear", "min": null, "gridLineWidth": null, "gridLineColor": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "tickmarkPlacement": null, "units": null, "minRange": null }, "yAxis": { "title": { "text": [ "Token/second<br>Higher is better" ], "useHTML": false, "style": { "color": "#666666" } }, "categories": null, "plotLines": null, "plotBands": null, "lineWidth": null, "tickInterval": null, "tickLength": 10, "floor": null, "ceiling": null, "gridLineInterpolation": null, "gridLineWidth": 1, "gridLineColor": "#CCC", "min": null, "max": null, "minorTickInterval": null, "minorTickWidth": 0, "minTickInterval": null, "startOnTick": true, "endOnTick": null, "minRange": null, "type": "linear", "tickmarkPlacement": null, "labels": { "enabled": true, "formatter": null, "style": { "color": "#666666", "cursor": "default", "fontSize": "11px" }, "useHTML": false } }, "zAxis": { "title": { "text": "Token/second<br>Higher is better" } }, "plotOptions": { "series": { "dataLabels": { "enabled": true, "format": "{series.name}", "distance": 30, "align": "left", "inside": true, "allowOverlap": false, "style": { "fontSize": "17px" } }, "showInLegend": null, "turboThreshold": 1000, "stacking": "", "groupPadding": 0, "centerInCategory": false, "findNearestPointBy": "x" } }, "navigator": { "enabled": false }, "scrollbar": { "enabled": false }, "rangeSelector": { "enabled": false, "inputEnabled": false }, "legend": { "enabled": true, "maxHeight": null, "align": "center", "verticalAlign": "bottom", "layout": "horizontal", "width": null, "margin": 12, "reversed": false }, "series": [ { "name": "Azure Standard_NC40ads_H100_v5 NVIDIA H100 NVL", "data": [ [ 0, 27.38188940629102 ] ], "grouping": false, "color": "#55b3ff" }, { "name": "Google a2-ultragpu-1g NVIDIA A100-SXM4-80GB", "data": [ [ 1, 17.472856372685843 ] ], "grouping": false, "color": "#55b400" }, { "name": "OVH H100-380 NVIDIA H100 PCIe", "data": [ [ 2, 21.132689666982362 ] ], "grouping": false, "color": "#484848" } ], "drilldown": {}, "tooltip": { "enabled": true, "useHTML": false, "format": null, "headerFormat": "", "pointFormat": "<span style=\"color:{series.color}\">{series.name}</span>: <b>{point.y:.1f} token/sec</b>", "footerFormat": "", "shared": false, "outside": false, "valueDecimals": null, "split": false }, "annotations": null }, "hc_type": "chart", "id": "llama3.1:70b-instruct-q8_0" }

Cloud Mercato's observations:

  • Without at least 40GB of VRAM, it's impossible for most VMs to run llama3 70b.
  • 80GB GPUs and especially the H100 are the best options for running large llama3
  • No matter the size of the model H100 are always the best performer