Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM | Paper Citation Record

Observation 080f7c07-e0e8-47ca-b1b9-04d6b9377184 · outbound

This paper cites cublas docs,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM cublas docs,

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:39.062564Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 3b74aec1-f5df-4baa-8530-498f67f321dd · outbound

This paper cites High bandwidth memory dram (hbm1, hbm2) jesd235d,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM High bandwidth memory dram (hbm1, hbm2) jesd235d,

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:39.047523Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 7468b991-2e9c-4c07-bbf3-33ac2f79386e · outbound

This paper cites Analyzing cuda workloads using a detailed gpu simulator,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Analyzing cuda workloads using a detailed gpu simulator,

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.705355Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 378b0234-3b45-48ff-a9de-f7567c195570 · outbound

This paper cites Moe-lightning: High-throughput moe inference on memory-constrained gpus,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Moe-lightning: High-throughput moe inference on memory-constrained gpus,

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.711044Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 444dc580-08d2-4c80-a5a2-a38ff775be57 · outbound

This paper cites Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality,

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.715953Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e90bd290-db89-4fdc-9e77-870ed6ebfafc · outbound

This paper cites Palm: Scal- ing language modeling with pathways,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Palm: Scal- ing language modeling with pathways,

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.720747Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4e7cea13-109a-4cc0-89e2-fe66b3d6da83 · outbound

This paper cites Asap7: A 7-nm finfet predictive process design kit,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Asap7: A 7-nm finfet predictive process design kit,

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.726009Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 20fef4f4-2bdd-4f8d-9c9b-1bbd8eb785b5 · outbound

This paper cites Projects – COIN-OR: Computational infrastruc- ture for operations research,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Projects – COIN-OR: Computational infrastruc- ture for operations research,

Reference 8

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.983040Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 57bb3874-8eea-4b32-99e2-08dbf80f3092 · outbound

This paper cites Deepseekmoe: Towards ultimate expert special- ization in mixture-of-experts language models,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Deepseekmoe: Towards ultimate expert special- ization in mixture-of-experts language models,

Reference 9

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.965110Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 7d215e54-60af-4e43-ac25-35cb46e632be · outbound

This paper cites The true processing in memory accelerator,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM The true processing in memory accelerator,

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.739364Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b71ffb43-2818-4c36-af5b-cf38e62eec34 · outbound

This paper cites Accelerating llm inference throughput via asynchronous kv cache prefetching,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Accelerating llm inference throughput via asynchronous kv cache prefetching,

Reference 11

Resolution: verified exact; raw_fallback, observed 2026-08-10T17:07:38.266524Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 50e9a289-18c4-43d8-9783-6937f46a02a9 · outbound

This paper cites The Llama 3 Herd of Models.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM The Llama 3 Herd of Models

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.748311Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2b136343-57ca-44f0-93c7-7c43136ff078 · outbound

This paper cites Klotski: Efficient mixture-of-expert inference via expert- aware multi-batch pipeline,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Klotski: Efficient mixture-of-expert inference via expert- aware multi-batch pipeline,

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.936738Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 235f5503-dd76-40d3-b23f-ff66dc40534a · outbound

This paper cites Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity,

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.758985Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d274dc01-2f97-413f-ab86-9d08951e2c2c · outbound

This paper cites The future of low-latency memory: Why near memory requires a new interface,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM The future of low-latency memory: Why near memory requires a new interface,

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.908916Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 3ff899c6-eaac-45f7-8bf5-314a966c18ac · outbound

This paper cites Papi: Exploiting dynamic parallelism in large language model decoding with a processing-in-memory-enabled computing system,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Papi: Exploiting dynamic parallelism in large language model decoding with a processing-in-memory-enabled computing system,

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.892099Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation b22b43ca-6b2c-4ab5-bd37-ff753a166092 · outbound

This paper cites Neupims: Npu-pim heterogeneous acceleration for batched llm inferencing,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Neupims: Npu-pim heterogeneous acceleration for batched llm inferencing,

Reference 17

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.876912Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 4d57ed89-8e91-4723-b661-49ead71d38c0 · outbound

This paper cites Lightllm: A versatile large language model for predictive light sensing,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Lightllm: A versatile large language model for predictive light sensing,

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.862011Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation d50537ad-5740-4fbb-94c1-ee347a01d742 · outbound

This paper cites Decoding llm performance—prefill phase is compute bound on npu,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Decoding llm performance—prefill phase is compute bound on npu,

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.846499Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 705e2e2a-8662-45f5-9cdc-ad0240330214 · outbound

This paper cites Mixtral of Experts.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Mixtral of Experts

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.787253Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ae003cce-039d-4a6a-aa1f-6631b5566e1d · outbound

This paper cites The battle of chatbot giants: an ex- perimental comparison of chatgpt and bard,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM The battle of chatbot giants: an ex- perimental comparison of chatgpt and bard,

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.831974Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation c3281b42-77ec-45ea-bdbe-1c02dcc7cb58 · outbound

This paper cites Accel-sim: An extensible simulation framework for validated gpu modeling,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Accel-sim: An extensible simulation framework for validated gpu modeling,

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.796835Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7933b684-d142-41de-a549-86ffa7766c08 · outbound

This paper cites Sk hynix ai-specific computing memory solution: From aim device to heterogeneous aimx-xpu system for comprehensive llm inference,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Sk hynix ai-specific computing memory solution: From aim device to heterogeneous aimx-xpu system for comprehensive llm inference,

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.801418Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation afacd144-eb5c-4f26-9bca-40082e3e05fc · outbound

This paper cites Monde: Mixture of near-data experts for large-scale sparse models,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Monde: Mixture of near-data experts for large-scale sparse models,

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.797868Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation afe5b5f9-c85c-48ec-ba8e-8fde1e61f319 · outbound

This paper cites Efficient memory management for large language model serving with pagedattention,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Efficient memory management for large language model serving with pagedattention,

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.810685Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6b992dcc-1456-4ffe-9245-e8a383c00909 · outbound

This paper cites vllm: Easy, fast, and cheap llm serving for everyone.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM vllm: Easy, fast, and cheap llm serving for everyone

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.771604Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 1b98ec61-3a59-4a96-bfbc-b047e080cf63 · outbound

This paper cites A 1ynm 1.25 v 8gb, 16gb/s/pin gddr6-based accelerator-in-memory supporting 1tflops mac operation and various activation functions for deep-learning applications,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM A 1ynm 1.25 v 8gb, 16gb/s/pin gddr6-based accelerator-in-memory supporting 1tflops mac operation and various activation functions for deep-learning applications,

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.819794Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 42c4c93d-eb61-4471-be5a-5da3eb13c67a · outbound

This paper cites Hardware architecture and software stack for pim based on commercial dram technology: Industrial product,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Hardware architecture and software stack for pim based on commercial dram technology: Industrial product,

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.825298Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e8f5c68a-4265-4551-9b81-93b293924ff5 · outbound

This paper cites H2-llm: Hardware-dataflow co-exploration for heterogeneous hybrid-bonding-based low-batch llm inference,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM H2-llm: Hardware-dataflow co-exploration for heterogeneous hybrid-bonding-based low-batch llm inference,

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.736234Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation bf29786f-762f-4bd4-adeb-e0dace7d7ac5 · outbound

This paper cites Ascend: a scalable and unified architecture for ubiquitous deep neural network computing: Industry track paper,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Ascend: a scalable and unified architecture for ubiquitous deep neural network computing: Industry track paper,

Reference 30

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.834220Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d54ef2b4-5686-4447-8c92-cb43a559cf07 · outbound

This paper cites Davinci: A scalable architecture for neural network computing,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Davinci: A scalable architecture for neural network computing,

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.838777Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 883731a9-3936-4704-9b81-bae3e3063d8e · outbound

This paper cites DeepSeek-V3 Technical Report.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM DeepSeek-V3 Technical Report

Reference 32

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.843266Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 04065920-91ef-4545-b53e-51ced19f61f5 · outbound

This paper cites Ramulator 2.0: A modern, modular, and extensible dram simulator,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Ramulator 2.0: A modern, modular, and extensible dram simulator,

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.848227Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3eeb1b74-e7fd-49a2-88cd-ca2bc0b4e623 · outbound

This paper cites The design process for google’s training chips: Tpuv2 and tpuv3,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM The design process for google’s training chips: Tpuv2 and tpuv3,

Reference 34

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.852810Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 168726cf-77ee-4eea-8682-0b9035b2d1a5 · outbound

This paper cites Nvidia a100 tensor core gpu architecc ture,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Nvidia a100 tensor core gpu architecc ture,

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.678996Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 2cbc5a3e-9369-4398-9bdb-33fc15fed075 · outbound

This paper cites Introduction to the nvidia dgx a100 system,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Introduction to the nvidia dgx a100 system,

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.663185Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 617a579e-4348-4adb-9dd0-e10c31e98954 · outbound

This paper cites Nvidia h100 tensor core gpu architecture,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Nvidia h100 tensor core gpu architecture,

Reference 37

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.647557Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation acc7508a-f626-4fb4-bf34-0cabf49df959 · outbound

This paper cites Chatgpt,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Chatgpt,

Reference 38

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.631915Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation f63776a2-4a47-46c1-8ed1-6407f5661d28 · outbound

This paper cites Gpt-oss-120b,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Gpt-oss-120b,

Reference 39

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.616441Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 498eb86c-0d7a-4e71-95d6-6352ef7bec5f · outbound

This paper cites VESPA: VIPT Enhancements for Superpage Accesses.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM VESPA: VIPT Enhancements for Superpage Accesses

Reference 40

Resolution: verified exact; local_arxiv, observed 2026-08-10T17:07:38.126461Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation c8ed4d31-0f97-4ba3-ac9b-152c07ca6a22 · outbound

This paper cites Attacc! unleashing the power of pim for batched transformer- based generative model inference,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Attacc! unleashing the power of pim for batched transformer- based generative model inference,

Reference 41

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.887482Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c7f2f9ad-fd01-4eeb-b859-ee9cb25b790f · outbound

This paper cites Pytorch: An imperative style, high-performance deep learning library,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Pytorch: An imperative style, high-performance deep learning library,

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.892413Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation adcbebf4-df7a-45c7-aab6-28ec5758aefd · outbound

This paper cites Splitwise: Efficient generative llm inference using phase splitting,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Splitwise: Efficient generative llm inference using phase splitting,

Reference 43

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.896797Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 05ba398e-7adb-45ab-ba9c-aa9d5b2a9f79 · outbound

This paper cites {DRAMA}: Exploiting{DRAM}addressing for{Cross-CPU}at- tacks,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM {DRAMA}: Exploiting{DRAM}addressing for{Cross-CPU}at- tacks,

Reference 44

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.901598Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 99cb3109-344e-4acf-91cc-a742c7736005 · outbound

This paper cites Code Llama: Open Foundation Models for Code.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Code Llama: Open Foundation Models for Code

Reference 45

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.906475Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1ca98075-9a99-47dd-82d6-f1495de28c4a · outbound

This paper cites Ramulator 2.0,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Ramulator 2.0,

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.562417Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation b8f15562-cb00-4ea5-b30e-e32804b6858b · outbound

This paper cites Khaa44801b-mc16: 8gb hbm2e flashbolt,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Khaa44801b-mc16: 8gb hbm2e flashbolt,

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.546657Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 3cd728a0-7f39-48ec-a0e5-4e529ffa946f · outbound

This paper cites Ianus: Integrated accelerator based on npu-pim unified memory system,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Ianus: Integrated accelerator based on npu-pim unified memory system,

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.515151Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 81964d28-58fe-4e09-9d97-aef75e69b67e · outbound

This paper cites Facil: Flexible dram address mapping for soc-pim cooperative on-device llm inference,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Facil: Flexible dram address mapping for soc-pim cooperative on-device llm inference,

Reference 49

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.929827Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b49f68be-cefb-4d46-b4dd-8d348a120514 · outbound

This paper cites an unresolved cited work.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Unresolved cited work

Reference 50

Resolution: unresolved; raw_fallback, observed 2026-08-10T17:07:38.489538Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation b5b64ee8-ab0d-4c2a-9d06-970ec14438ff · outbound

This paper cites Design compiler,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Design compiler,

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.475000Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation edce804e-4cf3-45d1-93c0-13f3a8e855f0 · outbound

This paper cites Qwen1.5-moe-a2.7b,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Qwen1.5-moe-a2.7b,

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.460323Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 96078f9c-7b54-41d4-b7ab-f80ab14f83f5 · outbound

This paper cites Sglang: Efficient execution of structured language model programs,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Sglang: Efficient execution of structured language model programs,

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.443975Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 910f8908-83a9-411e-8b17-1125c43caea1 · outbound

This paper cites Introducing the next generation of claude.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Introducing the next generation of claude

Reference 54

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.426565Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 5a9f9032-2d6b-4cd1-a3fd-95e6b61f826e · outbound

This paper cites Upmem software development kit (sdk),.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Upmem software development kit (sdk),

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.410712Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation ac10e1cc-307d-4650-b64f-292aa15c45a8 · outbound

This paper cites BLOOM: A 176B-Parameter Open-Access Multilingual Language Model.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM BLOOM: A 176B-Parameter Open-Access Multilingual Language Model

Reference 56

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.962298Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 16927598-8583-4e6b-9871-b9556d42bcbb · outbound

This paper cites Pim gpt a hybrid process in memory accelerator for autoregressive transformers,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Pim gpt a hybrid process in memory accelerator for autoregressive transformers,

Reference 57

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.395249Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation bb2bf72f-8ddb-4c6a-8ea8-7a31bf9af2e5 · outbound

This paper cites Waitgpt: Monitoring and steering conversational llm agent in data analysis with on-the-fly code visualization,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Waitgpt: Monitoring and steering conversational llm agent in data analysis with on-the-fly code visualization,

Reference 58

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.379994Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 99b76cda-51fc-4b0d-99ae-85fb04ce4241 · outbound

This paper cites Patterns behind Chaos: Forecasting Data Movement for Efficient Large-Scale MoE LLM Inference.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Patterns behind Chaos: Forecasting Data Movement for Efficient Large-Scale MoE LLM Inference

Reference 59

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.977874Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5f496768-78ad-464e-b06d-01e9d5510997 · outbound

This paper cites Duplex: A device for large language models with mixture of experts, grouped query attention, and continuous batching,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Duplex: A device for large language models with mixture of experts, grouped query attention, and continuous batching,

Reference 60

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.364518Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 1a943f7c-7f6a-461c-a0d9-e69183ee3b60 · outbound

This paper cites Llmcompass: Enabling efficient hardware design for large language model inference,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Llmcompass: Enabling efficient hardware design for large language model inference,

Reference 61

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:37.987438Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 561cccd9-80e0-41b3-b6fd-2de7b77a3ceb · outbound

This paper cites Um-pim: Dram-based pim with uniform & shared memory space,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Um-pim: Dram-based pim with uniform & shared memory space,

Reference 62

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.339889Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 89f41714-2435-494d-a2d4-fbf1fe817cec · outbound

This paper cites Lmsys-chat-1m: A large-scale real-world llm conversation dataset,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Lmsys-chat-1m: A large-scale real-world llm conversation dataset,

Reference 63

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.324063Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 71c53d55-488d-4640-9f8b-b233e79affa2 · outbound

This paper cites HybriMoE: Hybrid CPU-GPU Scheduling and Cache Management for Efficient MoE Inference.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM HybriMoE: Hybrid CPU-GPU Scheduling and Cache Management for Efficient MoE Inference

Reference 64

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:38.001641Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e4237319-436f-4a50-9577-d0bae12978e3 · outbound

This paper cites {DistServe}: Disaggregating prefill and decoding for goodput-optimized large language model serving,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM {DistServe}: Disaggregating prefill and decoding for goodput-optimized large language model serving,

Reference 65

Resolution: unresolved; no resolver link, observed 2026-08-10T17:07:38.006403Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 62ce0ff7-c17c-4158-8e2c-e5372e6186df · outbound

This paper cites Mixture-of-experts with expert choice routing,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Mixture-of-experts with expert choice routing,

Reference 66

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.297303Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation f3e9a925-87d8-4d46-b866-a74baf7971b1 · outbound

This paper cites A comprehensive analysis of superpage management mechanisms and policies,.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM A comprehensive analysis of superpage management mechanisms and policies,

Reference 67

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.281956Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 93da23a9-244c-4f81-b22e-cd7f9d2d58e7 · outbound

This paper cites Available: https://semiconductor.samsung.com/dram/ hbm/hbm2e-flashbolt/khaa44801b-mc16/.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM Available: https://semiconductor.samsung.com/dram/ hbm/hbm2e-flashbolt/khaa44801b-mc16/

Reference 2020

Resolution: verified fuzzy; raw_fallback, observed 2026-08-10T17:07:38.531333Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Coverage vector

Reference resolution

Outbound references

Pith citing papers