http://wiki.jackslab.org/index.php?action=history&feed=atom&title=Nvidia_GPU_Architecture 2026-05-21T08:19:00Z 本wiki的该页面的版本历史 MediaWiki 1.19.2 http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18061&oldid=prev 2025-02-12T07:52:21Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 07:52的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第37行：</td> <td colspan="2" class="diff-lineno">第37行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla V100 (Volta GV100/12nm/300W/16GB/FP32 14TFlops/FP16 28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores) 2017.6</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla V100 (Volta GV100/12nm/300W/16GB/FP32 14TFlops/FP16 28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores) 2017.6</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># Tesla T4 (Turing TU104/12nm/<del class="diffchange diffchange-inline">75W</del>/16GB/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Cnout/320 Tensor Cores/40 RT Cores) 2018.9</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># Tesla T4 (Turing TU104/12nm/<ins class="diffchange diffchange-inline">70W</ins>/16GB/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Cnout/320 Tensor Cores/40 RT Cores) 2018.9</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18060&oldid=prev 2025-02-12T06:52:06Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 06:52的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第32行：</td> <td colspan="2" class="diff-lineno">第32行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3080 Ti (10240 CUDA cores, 80SMs, 320 TensorCore, 80 RTCore, 8nm, 28.3billion, 12GB, 34.10 TFLOPs / DPU: 0.5328 TFLOPs, 350W, 2021.5 $1199) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3735 RTX3080 Ti] ---&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3080 Ti (10240 CUDA cores, 80SMs, 320 TensorCore, 80 RTCore, 8nm, 28.3billion, 12GB, 34.10 TFLOPs / DPU: 0.5328 TFLOPs, 350W, 2021.5 $1199) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3735 RTX3080 Ti] ---&gt; Ampere GA102</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla K80 (Kepler GK210/28nm/300W/2x12GB/FP16 8 TFLOPS/FP32 8TFLOPS/2x2496 CUDA Cores/2x208 TMUs/2x48 ROPs/2x13 SMX Cnout) 2014.11</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla K80 (Kepler GK210/28nm/300W/2x12GB/FP16 8 TFLOPS/FP32 8TFLOPS/2x2496 CUDA Cores/2x208 TMUs/2x48 ROPs/2x13 SMX Cnout) 2014.11</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18059&oldid=prev 2025-02-12T06:51:04Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 06:51的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第32行：</td> <td colspan="2" class="diff-lineno">第32行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3080 Ti (10240 CUDA cores, 80SMs, 320 TensorCore, 80 RTCore, 8nm, 28.3billion, 12GB, 34.10 TFLOPs / DPU: 0.5328 TFLOPs, 350W, 2021.5 $1199) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3735 RTX3080 Ti] ---&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3080 Ti (10240 CUDA cores, 80SMs, 320 TensorCore, 80 RTCore, 8nm, 28.3billion, 12GB, 34.10 TFLOPs / DPU: 0.5328 TFLOPs, 350W, 2021.5 $1199) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3735 RTX3080 Ti] ---&gt; Ampere GA102</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"># Tesla K80 (Kepler GK210/28nm/300W/2x12GB/FP16 8 TFLOPS/FP32 8TFLOPS/2x2496 CUDA Cores/2x208 TMUs/2x48 ROPs/2x13 SMX Cnout) 2014.11</ins></div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla V100 (Volta GV100/12nm/300W/16GB/FP32 14TFlops/FP16 28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores) 2017.6</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla V100 (Volta GV100/12nm/300W/16GB/FP32 14TFlops/FP16 28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores) 2017.6</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># Tesla T4 (Turing TU104/12nm/75W/16GB/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM <del class="diffchange diffchange-inline">Count</del>/320 Tensor Cores/40 RT Cores) 2018.9</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># Tesla T4 (Turing TU104/12nm/75W/16GB/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM <ins class="diffchange diffchange-inline">Cnout</ins>/320 Tensor Cores/40 RT Cores) 2018.9</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18058&oldid=prev 2025-02-12T05:38:08Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 05:38的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第34行：</td> <td colspan="2" class="diff-lineno">第34行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5nm/100W/6GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla V100 (Volta GV100/12nm/300W/16GB/FP32 14TFlops/FP16 28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores) 2017.6</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># Tesla V100 (Volta GV100/12nm/300W/16GB/FP32 14TFlops/FP16 28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores) 2017.6</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># <del class="diffchange diffchange-inline">TeslaT4 </del>(Turing TU104/12nm/75W/16GB/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores) 2018.9</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># <ins class="diffchange diffchange-inline">Tesla T4 </ins>(Turing TU104/12nm/75W/16GB/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores) 2018.9</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18057&oldid=prev 2025-02-12T05:37:33Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 05:37的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第32行：</td> <td colspan="2" class="diff-lineno">第32行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3080 Ti (10240 CUDA cores, 80SMs, 320 TensorCore, 80 RTCore, 8nm, 28.3billion, 12GB, 34.10 TFLOPs / DPU: 0.5328 TFLOPs, 350W, 2021.5 $1199) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3735 RTX3080 Ti] ---&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3080 Ti (10240 CUDA cores, 80SMs, 320 TensorCore, 80 RTCore, 8nm, 28.3billion, 12GB, 34.10 TFLOPs / DPU: 0.5328 TFLOPs, 350W, 2021.5 $1199) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3735 RTX3080 Ti] ---&gt; Ampere GA102</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/<del class="diffchange diffchange-inline">5 nm</del>/100W/<del class="diffchange diffchange-inline">6 GB</del>/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/<ins class="diffchange diffchange-inline">5nm</ins>/100W/<ins class="diffchange diffchange-inline">6GB</ins>/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># V100 (Volta GV100/300W/FP32 <del class="diffchange diffchange-inline">15.7TFlops</del>/FP16 <del class="diffchange diffchange-inline">125TFlops</del>) 2017</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># <ins class="diffchange diffchange-inline">Tesla </ins>V100 (Volta GV100<ins class="diffchange diffchange-inline">/12nm</ins>/300W<ins class="diffchange diffchange-inline">/16GB</ins>/FP32 <ins class="diffchange diffchange-inline">14TFlops</ins>/FP16 <ins class="diffchange diffchange-inline">28TFlops/5120 CUDA Cores/320 TMUs/128 ROPs/80 SM Count/640 Tensor Cores/40 RT Cores</ins>) 2017<ins class="diffchange diffchange-inline">.6</ins></div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># <del class="diffchange diffchange-inline">T4 </del>(Turing TU104/75W/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores) 2018.9</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># <ins class="diffchange diffchange-inline">TeslaT4 </ins>(Turing TU104<ins class="diffchange diffchange-inline">/12nm</ins>/75W<ins class="diffchange diffchange-inline">/16GB</ins>/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores) 2018.9</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18056&oldid=prev 2025-02-12T05:31:20Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 05:31的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第34行：</td> <td colspan="2" class="diff-lineno">第34行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># V100 (Volta GV100/300W/FP32 15.7TFlops/FP16 125TFlops) 2017</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># V100 (Volta GV100/300W/FP32 15.7TFlops/FP16 125TFlops) 2017</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># T4 (Turing TU104/75W/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores) &#160;</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># T4 (Turing TU104/75W/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores) <ins class="diffchange diffchange-inline">2018.9</ins></div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18055&oldid=prev 2025-02-12T05:31:02Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 05:31的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第34行：</td> <td colspan="2" class="diff-lineno">第34行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># V100 (Volta GV100/300W/FP32 15.7TFlops/FP16 125TFlops) 2017</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># V100 (Volta GV100/300W/FP32 15.7TFlops/FP16 125TFlops) 2017</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div># T4 (Turing/75W/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS) &#160;</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># T4 (Turing <ins class="diffchange diffchange-inline">TU104</ins>/75W/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS<ins class="diffchange diffchange-inline">/2560 CUDA Cores/160 TMUs/64 ROPs/40 SM Count/320 Tensor Cores/40 RT Cores</ins>) &#160;</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18054&oldid=prev 2025-02-12T05:26:41Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 05:26的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第33行：</td> <td colspan="2" class="diff-lineno">第33行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"># V100 (Volta GV100/300W/FP32 15.7TFlops/FP16 125TFlops) 2017</ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"># T4 (Turing/75W/FP32 8TFlops/FP16 65TFlops/INT8 130 TOPS) </ins></div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18053&oldid=prev 2025-02-12T03:16:33Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 03:16的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第26行：</td> <td colspan="2" class="diff-lineno">第26行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3050 (2048CUDA cores, 16SMs, 64 TensorCore, 16 RTCore, 8nm, 12billion, 4GB, 4.329 TFLOPS/ DPU: 0.06765TFLOPs, 75W, 2021.5 ) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3050-mobile.c3788 NVIDIA GeForce RTX 3050 Mobile] -----&gt; Ampere GA107</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3050 (2048CUDA cores, 16SMs, 64 TensorCore, 16 RTCore, 8nm, 12billion, 4GB, 4.329 TFLOPS/ DPU: 0.06765TFLOPs, 75W, 2021.5 ) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3050-mobile.c3788 NVIDIA GeForce RTX 3050 Mobile] -----&gt; Ampere GA107</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3050 Ti (2560CUDA cores, 20SMs, 80 TensorCore, 20 RTCore, 8nm, 12billion, 4GB, 5.299 TFLOPS/ DPU: 0.08280TFLOPs, 75W, 2021.5 ) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3050-ti-mobile.c3812 NVIDIA GeForce RTX 3050 Ti Mobile] -----&gt; Ampere GA106</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3050 Ti (2560CUDA cores, 20SMs, 80 TensorCore, 20 RTCore, 8nm, 12billion, 4GB, 5.299 TFLOPS/ DPU: 0.08280TFLOPs, 75W, 2021.5 ) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3050-ti-mobile.c3812 NVIDIA GeForce RTX 3050 Ti Mobile] -----&gt; Ampere GA106</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div>#<del class="diffchange diffchange-inline">Mid-range: GTX1060 ()</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div># GTX1070 / GTX1080 (2560 CUDA cores, 20 SMs, 16nm, 7.2billion, 8GB, 8.2TFLOPs / DPU: 0.257TFLOPs, 180W, 2016.5)</div></td></tr> <tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">#High-end: </del>GTX1070 / GTX1080 (2560 CUDA cores, 20 SMs, 16nm, 7.2billion, 8GB, 8.2TFLOPs / DPU: 0.257TFLOPs, 180W, 2016.5)</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div></div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># GTX1080 Ti / TITAN X (3584 CUDA cores, 28 SMs, 16nm, 12billion, 12GB, 10TFLOPs / DPU: 0.317TFLOPs, 250W, 2016.8)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># GTX1080 Ti / TITAN X (3584 CUDA cores, 28 SMs, 16nm, 12billion, 12GB, 10TFLOPs / DPU: 0.317TFLOPs, 250W, 2016.8)</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># TITAN X (3584 CUDA cores, 28 SMs, 16nm, 12billion, 12GB, 10.97TFLOPs / DPU: 0.3429TFLOPs, 250W, 2016.8) [https://www.techpowerup.com/gpu-specs/titan-x-pascal.c2863 NVIDIA TITAN X Pascal]----&gt; Pascal GP107</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># TITAN X (3584 CUDA cores, 28 SMs, 16nm, 12billion, 12GB, 10.97TFLOPs / DPU: 0.3429TFLOPs, 250W, 2016.8) [https://www.techpowerup.com/gpu-specs/titan-x-pascal.c2863 NVIDIA TITAN X Pascal]----&gt; Pascal GP107</div></td></tr> </table>

Comcat http://wiki.jackslab.org/index.php?title=Nvidia_GPU_Architecture&diff=18052&oldid=prev 2025-02-12T03:12:34Z

<p>‎<span dir="auto"><span class="autocomment">Overview</span></span></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr valign='top'> <td colspan='2' style="background-color: white; color:black;">←上一版本</td> <td colspan='2' style="background-color: white; color:black;">2025年2月12日 (三) 03:12的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第34行：</td> <td colspan="2" class="diff-lineno">第34行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX3090 Ti (10752 CUDA cores, 84SMs, 336 TensorCore, 84 RTCore, 8nm, 28.3billion, 24GB, 40TFLOPs / DPU:0.625TFLOPs, 450W, 2022.1) [https://www.techpowerup.com/gpu-specs/geforce-rtx-3090-ti.c3829 Nvidia RTX3090 Ti][https://www.techpowerup.com/gpu-specs/geforce-rtx-3090.c3622 RTX3090] ----&gt; Ampere GA102</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div># RTX 4050 (Ada Lovelace AD107/5 nm/100W/6 GB/FP16 13.5 TFLOPS/FP32 13.5 TFLOPS/2560 CUDA cores/80 TMUs/32 ROPs/18 SM Count/120 Tensor Cores/18 RT Cores)</div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div>* https://www.techpowerup.com/gpu-specs/</div></td><td class='diff-marker'>&#160;</td><td style="background: #eee; color:black; font-size: smaller;"><div>* https://www.techpowerup.com/gpu-specs/</div></td></tr> </table>

Comcat