Loading showdown arena...

经典硬件选型痛点与实测案例对比

选卡用户常常纠结于：是买显存较小的新架构显卡，还是买显存充裕的老旗舰？以下是四大最受关注对决的实测工程结论：

RTX 3060 12G vs RTX 4060 8G

胜出: RTX 3060 12G
测试基准模型: DeepSeek-R1-Distill-Qwen-14B

The classic $300 dilemma. The older RTX 3060 12G fits the entire 14B model in VRAM, delivering ~24 tok/s. The newer RTX 4060 8G runs out of memory, spills layers to DDR4 RAM over PCIe, and crawls at ~1.8 tok/s.

进入对决竞技场

RTX 3090 24G vs RTX 4070 Ti Super 16G

胜出: RTX 3090 24G
测试基准模型: Qwen 2.5 Coder 32B (Q4_K_M)

Used flagship vs modern upper-midrange. The RTX 3090 with 24GB VRAM loads the complete 32B coding model with zero offload. The 4070 Ti Super with 16GB suffers memory deficits, throttling heavily on complex coding tasks.

进入对决竞技场

RTX 5090 32G vs RTX 4090 24G

胜出: RTX 5090 32G
测试基准模型: DeepSeek-R1-Distill-Llama-70B

Blackwell GDDR7 (1,792 GB/s) vs Ada GDDR6X (1,008 GB/s). The RTX 5090's 32GB frame buffer allows running 70B models at IQ4_XS with 40+ tok/s, while the 4090 requires layer offload or heavy quantizations.

进入对决竞技场

RTX 4090 24G vs Apple M3 Max 36G

胜出: Context-dependent
测试基准模型: DeepSeek-R1-Distill-Qwen-32B

The RTX 4090 delivers unmatched raw generation speed (60+ tok/s) thanks to 1,008 GB/s bandwidth. The M3 Max offers a larger 36GB memory pool at lower power consumption, handling larger context windows quietly.

进入对决竞技场

本地大模型装机选卡的三大铁律

1. 显存容量是一票否决制（非生即死）一旦模型所需显存超过显卡容量，多余层数必须跨 PCIe 总线溢出至系统内存。PCIe 4.0 x16 仅有 31.5 GB/s 带宽，远低于显卡 1,000+ GB/s，生成速度直接崩溃暴跌 95%。
2. 显存带宽唯一决定每秒生成速度单用户自回归解码时，每输出一个 Token 都必须从显存中完整读取一次权重。算力并非瓶颈，显存物理带宽才是 100% 的速度决定者。
3. KV 缓存线性激增（MLA 除外）标准 GQA 模型长文本需要额外消耗 4GB 至 12GB 显存。而 DeepSeek MLA 将 KV 压缩 80% 以上，让 16GB/24GB 显卡也能轻松承载 64k 长文本。

显卡横向对比常见高频疑问解答

关于显存带宽、PCIe 延迟惩罚与架构选型的深度工程分析：

Why does the RTX 3060 12GB outperform the RTX 4060 8GB in local AI?
Local AI inference requires the entire model weights and KV cache to fit in VRAM. For 14B models like DeepSeek-R1 14B, total VRAM demand is ~11.9 GB. The RTX 3060's 12GB frame buffer holds the model completely, running at ~28 tok/s. The RTX 4060 has only 8GB, forcing 4GB of layers to offload across the PCIe bus to DDR4/DDR5 system RAM, slowing inference down by 90% to ~1.8 tok/s.
Should I buy a used RTX 3090 24GB or a new RTX 4070 Ti Super 16GB for AI?
For AI workloads, the RTX 3090 24GB is decisively superior. The extra 8GB of VRAM allows running 32B models (like Qwen 2.5 Coder 32B and DeepSeek R1 32B) and 70B models at IQ3 quantizations with zero offloading. The RTX 4070 Ti Super is limited to 16GB, restricting it to 14B models or short-context 32B models.
How much faster is the RTX 5090 GDDR7 compared to the RTX 4090 in AI inference?
The RTX 5090 features GDDR7 memory delivering 1,792 GB/s bandwidth compared to 1,008 GB/s on the RTX 4090 (a 78% increase). In batch=1 autoregressive token generation, throughput scales directly with bandwidth, increasing generation speeds from ~42 tok/s on the 4090 to ~65+ tok/s on the 5090 for 32B models.
Does memory bandwidth or CUDA core count matter more for LLM tokens per second?
Memory bandwidth matters far more. In single-user (batch=1) autoregressive decoding, every single token requires reading all model weights from VRAM once. Since compute utilization during decoding is very low (typically under 20%), the memory bus is the 100% bottleneck. CUDA cores and TFLOPS primarily accelerate the prompt prefill phase (TTFT), not token generation speed.