corpus.blog
42,805 blogs · 5,350,595 posts Claim your blog
Record · corpus.blog/posts/01a0f1d7-9cc3-704c-8ee9-bacb9486558c

How GLM5.3 Sparse Attention Affects HBM Memory Usage

newsletter.semianalysis.com · published 28 September 2026

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at newsletter.semianalysis.com Wayback capture
1blogs citing
First cited
28 Sept 2026
Most recent
28 Sept 2026
Rank this month
not ranked
External links
44
Words captured
3,302

Cited by

Blog
In the post
Date
inferencex.semianalysis.com
How GLM5.3 Sparse Attention Affects HBM Memory Usage Read at the source ↗
28 Sept 2026

Links from this post

Link
Host
https://substackcdn.com/image/fetch/$s_!1i4c!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4cd044d2-888a-4569-97ea-5354b07e4f62_1600x900.png
substackcdn.com
https://www.lmsys.org/blog/2026-04-10-sglang-hisparse
lmsys.org
https://substackcdn.com/image/fetch/$s_!yJNz!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe5373e7c-4b2c-4741-8213-ec54052779cd_1126x174.png
substackcdn.com
https://arxiv.org/abs/2403.19708
arxiv.org
https://inferencex.semianalysis.com/inference/glm-5-3?g_model=GLM-5.2&g_rundate=2026-09-28&i_seq=agentic-traces&i_prec=fp4%2Cfp8&i_pctl=p90&i_best=0&i_optimal=1&i_metric=y_costh&i_xmode=interactivity
inferencex.semianalysis.com
https://substackcdn.com/image/fetch/$s_!YLxK!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcd794226-78c0-4143-8b48-f632a507f676_2700x1900.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!I0ht!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5d4056ef-c9ad-424f-b93a-a67ca9574aac_2700x2160.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!591h!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb2a4b342-0525-4274-af86-3e6d5e9f7fa8_2700x1800.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!XMNu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc49af533-cf97-4c90-8c33-0d5bee12e029_2048x1772.png
substackcdn.com
https://inferencex.semianalysis.com/inference/agentic/440962
inferencex.semianalysis.com
https://inferencex.semianalysis.com/inference/agentic/440961
inferencex.semianalysis.com
https://inferencex.semianalysis.com/inference/agentic/440959
inferencex.semianalysis.com
https://github.com/SemiAnalysisAI/InferenceX/actions/runs/33683520699/attempts/1
github.com
https://github.com/vllm-project/vllm-project.github.io/blob/3be7020c6da40428eca107387d7faa8e58bd7074/_posts/2026-07-23-glm-5.2-nvfp4-b300-pd.md
github.com
https://github.com/ROCm/ATOM/pull/1552
github.com
https://substackcdn.com/image/fetch/$s_!3tYc!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F22959305-42e4-48ef-afb2-913f9474f339_1582x1450.png
substackcdn.com
https://arxiv.org/abs/2512.02556
arxiv.org
https://substackcdn.com/image/fetch/$s_!-FZf!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc3abe0c5-92c8-4912-ad2b-1b5ba3753580_1404x682.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!n_SD!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F375d0869-c668-4145-bf66-3b019701c834_2048x2048.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!kk6W!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff496f5be-6eff-4a54-ae7e-8d2555bc6716_1898x1265.png
substackcdn.com
https://github.com/vllm-project/vllm/pull/48770
github.com
https://substackcdn.com/image/fetch/$s_!qpBS!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc0857f2d-5a6a-45a5-accd-3ec332737582_1258x204.png
substackcdn.com
https://arxiv.org/abs/2602.15763
arxiv.org
https://github.com/deepseek-ai/FlashMLA/blob/main/docs/20250422-new-kernel-deep-dive.md?utm_source=chatgpt.com
github.com
https://www.mthreads.com/news/341
mthreads.com
https://substackcdn.com/image/fetch/$s_!gPFQ!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1afc9d13-969e-449a-91d1-cd3b310c779a_944x654.png
substackcdn.com
https://arxiv.org/abs/2603.12201
arxiv.org
https://substackcdn.com/image/fetch/$s_!0XK_!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F88c2f29b-e00a-4240-92a0-2a1c30c7ee15_1258x264.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!J1l1!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff70d8106-0784-4721-b55f-e2285ce91f6d_2048x629.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!-nRK!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb2665b55-e60e-4d73-ac0f-68b93e205460_2048x1432.png
substackcdn.com
https://z.ai/blog/glm-5.2
z.ai
https://substackcdn.com/image/fetch/$s_!gNw1!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Feab874c6-4338-4a58-b394-8de2794966dd_2048x472.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!VIZ-!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fdb5823b7-0ac8-4326-af4f-5e3b3829f100_1654x484.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!7JkY!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F855b6009-e0d8-4e01-bab5-d6abfdd5849b_2048x479.png
substackcdn.com
https://arxiv.org/abs/2607.07508
arxiv.org
https://arxiv.org/abs/1707.06347
arxiv.org
https://substackcdn.com/image/fetch/$s_!1MnE!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F29bbbf45-36e4-470f-be0c-092db886317d_1255x669.png
substackcdn.com
https://www.swordart-online.net/
swordart-online.net
https://substackcdn.com/image/fetch/$s_!YbQd!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb83b1151-c583-4cb6-bd89-a247fb7f4f92_335x597.png
substackcdn.com
https://github.com/THUDM/slime
github.com
https://substackcdn.com/image/fetch/$s_!u9lu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F749e2de3-ef84-4776-864e-846243d30d9e_2048x1039.png
substackcdn.com
https://z.ai/blog/glm-5.3
z.ai
https://github.com/THUDM/slime/pull/1538
github.com
https://github.com/THUDM/slime/blob/main/slime/backends/megatron_utils/actor.py
github.com
Elsewhere on newsletter.semianalysis.com (2)