DeepSeek-V4.1-Flash (764B) on 4×CMP 170HX: PP4 + shadow KV + EXL3 2bpw + DSpark + 512k ctx + native vision — patches, recipes, acceptance bench, honest findings
☆15Sep 29, 2026Updated last month
Alternatives and similar repositories for dsv41-flash-pp4-170hx
Users that are interested in dsv41-flash-pp4-170hx are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- glm-5.2 on too many 64 / 80 GB A100's☆22Sep 9, 2026Updated 2 months ago
- dsv41-flash-pp5-170hx &pp6 Supports the original (full/unquantized) model. Prefill 7,000 t/s, decode 45-75-125 t/s. dspark. Requires 200 …☆13Updated this week
- DeepSeek-V4.1-Flash on 4× NVIDIA CMP 170HX with vLLM: up to 297 tok/s for one user, up to 868 tok/s at eight, 262K context, one-command c…☆31Nov 2, 2026Updated last week
- 双 SM80 / CMP 170HX 的 Qwen3.8-Flash-Next 社区 vLLM：NVIDIA NVFP4、GDS、TEP2、MTP6，支持 1～10 并发，词表与 PCIe 通信优化。中英文部署与可构建源码。☆28Oct 15, 2026Updated 3 weeks ago
- 限制问题区域显存 62906MiB / 65024MiB 支持驱动 610.57.04 ，修改显卡显示名称为NVIDIA Tesla A100☆12Aug 30, 2026Updated 2 months ago
- GPU virtual machines on DigitalOcean Gradient AI • AdGet to production fast with high-performance AMD and NVIDIA GPUs you can spin up in seconds. The definition of operational simplicity.
- GLM-5.3-Flash on 4× NVIDIA CMP 170HX with vLLM: up to 482.6 tok/s single-user, up to 943.7 tok/s at 8 users, 262K context, TP4 or PP4, on…☆90Oct 26, 2026Updated 2 weeks ago
- GA100 voltage glitch resources☆136Updated this week
- Linux P2P unlock and performance fixes for NVIDIA CMP 170HX☆39Sep 22, 2026Updated last month
- Qwen3.8-Flash-Next (W4A16) on 2x CMP 170HX: 110 tok/s decode, 7K prefill, 800 tok/s @ 23 concurrent - runs on 38 GB home RAM, no pinned m…☆20Sep 18, 2026Updated last month
- Community-tested recipes, diagnostics, and benchmarks for CMP 170HX AI workloads☆99Updated this week
- A tool to unlobotomize your NVIDIA card!☆841Oct 28, 2026Updated 2 weeks ago
- Unlock tool for the NVIDIA CMP 170HX (GA100) mining card. Restores full A100 compute throughput by exploiting the Falcon BootROM .fwsigna…☆57Jul 30, 2026Updated 3 months ago
- 170hx-pve-gen2-controller☆11Aug 20, 2026Updated 2 months ago
- some experiments about 170hx 10g version.☆11Oct 19, 2026Updated 3 weeks ago
- End-to-end encrypted cloud storage - Proton Drive • AdSpecial offer: 40% Off Yearly / 80% Off First Month. Protect your most important files, photos, and documents from prying eyes.
- 适配泰安S8030显卡机自动调节系统风扇组件☆13Dec 13, 2024Updated last year
- Mixed-precision acceleration patch for MiniMax H3 inference on NVIDIA V100☆40Sep 6, 2026Updated 2 months ago
- NInfer SM80 port for NVIDIA CMP 170HX☆29Aug 22, 2026Updated 2 months ago
- 在四卡A100上适配Sglang部署deepseek-v4-flash正式版模型☆15Aug 2, 2026Updated 3 months ago
- CMP 170HX technical wiki: silicon, firmware, the community unlock, procedures and open problems. 55 pages.☆78Aug 22, 2026Updated 2 months ago
- Single-card Intel Gaudi2 vLLM source snapshot: FP8 Qwen, non-greedy DFlash2/MTP, GDN kernels and host-side optimizations.☆18Updated this week
- ☆30Oct 20, 2026Updated 3 weeks ago
- Async PLE SSD offload for Qwen3.8 Flash Next on a 64 GiB SM80 GPU: vLLM patch, setup guide, and measured PP/TG results.☆23Sep 24, 2026Updated last month
- Qwen3.8-27B INT8 + DFlash2 on a rented CMP 170HX — failure analysis (private GHCR image prevented benchmark)☆17Aug 30, 2026Updated 2 months ago
- Managed Kubernetes at scale on DigitalOcean • AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- forked from vllm-project/flash-attention☆74May 9, 2026Updated 6 months ago
- vLLM for NVIDIA CMP 170HX (Ampere sm_80): fast, repeatable model serving on one card or several☆28Nov 2, 2026Updated last week
- cpp write language detect model☆11Sep 22, 2021Updated 5 years ago
- 解锁自有 NVIDIA CMP 170HX / 90HX 的算力限制。Linux x86-64，需要 root。☆62Aug 6, 2026Updated 3 months ago
- Flash Attention v2 for Pascal (P100, GTX 1080) and Volta (V100) GPUs — bringing O(N) memory attention to hardware the official flash-attn…☆30Nov 3, 2026Updated last week
- Tuning and qualification harness for the NVIDIA CMP 170HX (GA100). Measures, gates and recovers clock/voltage settings, and treats a comp…☆98Sep 28, 2026Updated last month
- ComfyUI MiniMax H3 inference optimizations for NVIDIA Tesla V100 (SM70)☆33Sep 29, 2026Updated last month
- Dogfooding vLLM backport for older GPUs.☆362Updated this week
- Android免Root安裝Docker☆23Oct 2, 2025Updated last year
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- A cmpunlocker fork with working p2p on SYS-4028GR-TRT2☆27Aug 8, 2026Updated 3 months ago
- The most comprehensive community resource for the NVIDIA CMP 170HX.☆137Jul 30, 2026Updated 3 months ago
- Use OpenCode in WeChat☆23Mar 22, 2026Updated 7 months ago
- The AI Architect for your codebase☆28Apr 10, 2026Updated 7 months ago
- A ComfyUI custom node enabling **Flash Attention 1** on legacy NVIDIA GPUs (Tesla V100, T4) that lack Compute Capability 8.0+ required by…☆34Feb 9, 2026Updated 9 months ago
- A monitor of resources for DGX Spark☆26Feb 13, 2026Updated 9 months ago
- ☆23Dec 15, 2016Updated 9 years ago
Are these results useful?