Micron 6500 ION SSD: Turn AI with 256 Accelerators

These outcomes demonstrate how effectively the Micron 9400 NVMe SSD performs in the use case of an AI server as a local cache

[{"selector":"#anim-f8d277bf-bed5-4ded-be83-c49afcd92478 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(-35.56249991239381%, 0, 0) translate(-25%, 0%) scale(1.5)","translate3d(0%, 0, 0) translate(0%, 0%) scale(1)"]},"delay":0,"duration":2000,"fill":"forwards"}] [{"selector":"#anim-226436f7-fe9c-402c-8458-788439bde703","keyframes":{"transform":["translate3d(122.02797%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .8, 1)","fill":"both"}] [{"selector":"#anim-bb77140a-5a90-4e6e-8bd8-f33bf5411201","keyframes":{"transform":["rotateZ(180deg)","rotateZ(0deg)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .5, 1)","fill":"forwards"}]

MLPerf Storage AI workload was chosen to be tested for SC23 on a WEKA storage cluster that was powered by a 30TB Micron 6500 ION SSD

[{"selector":"#anim-f2963287-1dfb-440f-8267-6647611703f5 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(18.42570735557823%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-eddead84-b929-494e-bd1b-ba9545f94ace","keyframes":{"transform":["translate3d(-128.73563%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-492a020c-2865-49c6-b903-a3b3f6cca537","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-d1ef3d0d-b483-4795-ba9c-9a24c920fd43","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

MLPerf Storage aims to tackle several issues related to the storage workload of AI training systems, including the limited size of available datasets and the high expense of AI accelerators

[{"selector":"#anim-ba9f986e-e409-4856-bf1f-be1cad5cc4cb [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(31.919642747431954%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-3c2595c5-9112-4a63-b485-effd61038715","keyframes":[{"offset":0,"transform":"translate3d(0, -145.82748%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.29,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.45,"transform":"translate3d(0, -41.006687376%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.61,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.71,"transform":"translate3d(0, -13.941107088%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.8,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.85,"transform":"translate3d(0, -5.235206532%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.92,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.96,"transform":"translate3d(0, -2.274908688%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":1,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"}],"delay":0,"duration":600,"fill":"both"}]

Changes are also being made to the MLPerf Storage benchmark from version 0.5 to the upcoming version for the first 2024 release

[{"selector":"#anim-e3aad535-e1cc-490a-af42-7fee899d43a5 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate(3.552713678800501e-15%, 0%) scale(1.5)","translate(0%, 0%) scale(1)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"forwards"}] [{"selector":"#anim-f2f0cbf1-1459-4b5f-b265-5f91a3a72faf","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-3e73cc5e-a81a-49cd-9bd4-824dfcf9eb33","keyframes":{"transform":["translate3d(120.54793%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}]

In the 0.5 version, the MLPerf Storage benchmark simulates NVIDIA V100 accelerators. There are sixteen V100 accelerators on the NVIDIA DGX-2 server

[{"selector":"#anim-5f1b0d13-016f-498a-b147-a4e903afe8c8 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(70.42338691730446%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-629282be-e29c-4c0e-a416-47419b0bd52e","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-c1866303-6957-4813-a77c-36f82a02b564","keyframes":{"transform":["translate3d(-119.59459%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}]

With 16 V100 GPUs per system, this performance is equivalent to 16 NVIDIA DGX-2 systems, which is an astonishingly large number of AI systems powered by a six-node WEKA cluster

[{"selector":"#anim-903908e9-9ca6-44e4-8c4c-88d5763f86c7 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(30.448419955711216%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-e0023b22-b694-48cf-aea1-7ed5ea7383a0","keyframes":{"transform":["translate3d(-120.06803%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-913c1abe-e59b-4c90-908b-11e45fa24a1b","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-0fb7f353-6335-404d-8b47-720f4987ac66","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

Future-focused AI training servers, such as the H100 / H200 (PCIe Gen5) and X100 (PCIe Gen6) models, are expected to push extremely high throughput

[{"selector":"#anim-7819fccb-e167-4117-82f8-a289ed17fdcd [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(35.34967836769406%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-caf10e69-3dfc-4c26-9ae2-30466efe195a","keyframes":[{"offset":0,"transform":"translate3d(0, -150.39959%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.29,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.45,"transform":"translate3d(0, -42.292364708%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.61,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.71,"transform":"translate3d(0, -14.378200804%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.8,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.85,"transform":"translate3d(0, -5.3993452809999996%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":0.92,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"},{"offset":0.96,"transform":"translate3d(0, -2.3462336039999996%, 0)","easing":"cubic-bezier(.5, 0, 1, 1)"},{"offset":1,"transform":"translate3d(0, 0%, 0)","easing":"cubic-bezier(0, 0, .5, 1)"}],"delay":0,"duration":600,"fill":"both"}]

Micron 6500 ION SSD and WEKA storage offer the ideal balance of scalability, performance, and capacity for your AI workloads For More Details Govindhtech.com

[{"selector":"#anim-e0ec972c-0dc4-4aa8-a78b-7882e0f22c0b [data-leaf-element=\"true\"]","keyframes":{"transform":["translate(0%, 0%) scale(1.5)","translate(0%, 0%) scale(1)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"forwards"}]