NVIDIA GeForce RTX 3090 versus NVIDIA Tesla T4

Comparaison des cartes vidéo NVIDIA GeForce RTX 3090 and NVIDIA Tesla T4 pour tous les caractéristiques connus dans les catégories suivants: Essentiel, Infos techniques, Sorties et ports de vidéo, Compatibilité, dimensions et exigences, Soutien API, Mémoire. Analyse du performance de référence des cartes vidéo: PassMark - G3D Mark, PassMark - G2D Mark, Geekbench - OpenCL, CompuBench 1.5 Desktop - Face Detection (mPixels/s), CompuBench 1.5 Desktop - Ocean Surface Simulation (Frames/s), CompuBench 1.5 Desktop - T-Rex (Frames/s), CompuBench 1.5 Desktop - Video Composition (Frames/s), CompuBench 1.5 Desktop - Bitcoin Mining (mHash/s), GFXBench 4.0 - Car Chase Offscreen (Frames), GFXBench 4.0 - Manhattan (Frames), GFXBench 4.0 - T-Rex (Frames), GFXBench 4.0 - Car Chase Offscreen (Fps), GFXBench 4.0 - Manhattan (Fps), GFXBench 4.0 - T-Rex (Fps), 3DMark Fire Strike - Graphics Score.

NVIDIA GeForce RTX 3090

versus

NVIDIA Tesla T4

Différences

Raisons pour considerer le NVIDIA GeForce RTX 3090

La carte vidéo est plus nouvelle: date de sortie 1 ans 11 mois plus tard
Environ 39% plus haut vitesse du noyau: 1395 MHz versus 1005 MHz
Environ 12% plus de la vitesse augmenté: 1695 MHz versus 1515 MHz
Un nouveau processus de fabrication soutient une carte vidéo plus forte, mais moins chaude: 8 nm versus 12 nm
2.5x meilleur performance en PassMark - G3D Mark: 26668 versus 10616
Environ 79% meilleur performance en PassMark - G2D Mark: 1060 versus 592
3.1x meilleur performance en Geekbench - OpenCL: 191142 versus 61276
5.4x meilleur performance en CompuBench 1.5 Desktop - Face Detection (mPixels/s): 690.619 versus 127.622
2.5x meilleur performance en CompuBench 1.5 Desktop - Ocean Surface Simulation (Frames/s): 7585.258 versus 3026.812
3.3x meilleur performance en CompuBench 1.5 Desktop - T-Rex (Frames/s): 62.614 versus 18.798
Environ 75% meilleur performance en CompuBench 1.5 Desktop - Video Composition (Frames/s): 209.424 versus 119.936
2.7x meilleur performance en CompuBench 1.5 Desktop - Bitcoin Mining (mHash/s): 2441.384 versus 910.721
2.4x meilleur performance en GFXBench 4.0 - Car Chase Offscreen (Frames): 33398 versus 14076
Environ 88% meilleur performance en GFXBench 4.0 - Manhattan (Frames): 3713 versus 1976
Environ 88% meilleur performance en GFXBench 4.0 - T-Rex (Frames): 3354 versus 1781
2.4x meilleur performance en GFXBench 4.0 - Car Chase Offscreen (Fps): 33398 versus 14076
Environ 88% meilleur performance en GFXBench 4.0 - Manhattan (Fps): 3713 versus 1976
Environ 88% meilleur performance en GFXBench 4.0 - T-Rex (Fps): 3354 versus 1781

Caractéristiques
Date de sortie	1 Sep 2020 versus 13 September 2018
Vitesse du noyau	1395 MHz versus 1005 MHz
Vitesse augmenté	1695 MHz versus 1515 MHz
Processus de fabrication	8 nm versus 12 nm
Référence
PassMark - G3D Mark	26668 versus 10616
PassMark - G2D Mark	1060 versus 592
Geekbench - OpenCL	191142 versus 61276
CompuBench 1.5 Desktop - Face Detection (mPixels/s)	690.619 versus 127.622
CompuBench 1.5 Desktop - Ocean Surface Simulation (Frames/s)	7585.258 versus 3026.812
CompuBench 1.5 Desktop - T-Rex (Frames/s)	62.614 versus 18.798
CompuBench 1.5 Desktop - Video Composition (Frames/s)	209.424 versus 119.936
CompuBench 1.5 Desktop - Bitcoin Mining (mHash/s)	2441.384 versus 910.721
GFXBench 4.0 - Car Chase Offscreen (Frames)	33398 versus 14076
GFXBench 4.0 - Manhattan (Frames)	3713 versus 1976
GFXBench 4.0 - T-Rex (Frames)	3354 versus 1781
GFXBench 4.0 - Car Chase Offscreen (Fps)	33398 versus 14076
GFXBench 4.0 - Manhattan (Fps)	3713 versus 1976
GFXBench 4.0 - T-Rex (Fps)	3354 versus 1781

Raisons pour considerer le NVIDIA Tesla T4

4.7x consummation d’énergie moyen plus bas: 75 Watt versus 350 Watt
8.2x plus de vitesse de mémoire: 10000 MHz versus 1219 MHz (19.5 Gbps effective)

Thermal Design Power (TDP)	75 Watt versus 350 Watt
Vitesse de mémoire	10000 MHz versus 1219 MHz (19.5 Gbps effective)

Comparer les références

GPU 1: NVIDIA GeForce RTX 3090
GPU 2: NVIDIA Tesla T4

PassMark - G3D Mark

GPU 1

GPU 2

26668

10616

PassMark - G2D Mark

GPU 1

GPU 2

1060

592

Geekbench - OpenCL

GPU 1

GPU 2

191142

61276

CompuBench 1.5 Desktop - Face Detection (mPixels/s)

GPU 1

GPU 2

690.619

127.622

CompuBench 1.5 Desktop - Ocean Surface Simulation (Frames/s)

GPU 1

GPU 2

7585.258

3026.812

CompuBench 1.5 Desktop - T-Rex (Frames/s)

GPU 1

GPU 2

62.614

18.798

CompuBench 1.5 Desktop - Video Composition (Frames/s)

GPU 1

GPU 2

209.424

119.936

CompuBench 1.5 Desktop - Bitcoin Mining (mHash/s)

GPU 1

GPU 2

2441.384

910.721

GFXBench 4.0 - Car Chase Offscreen (Frames)

GPU 1

GPU 2

33398

14076

GFXBench 4.0 - Manhattan (Frames)

GPU 1

GPU 2

3713

1976

GFXBench 4.0 - T-Rex (Frames)

GPU 1

GPU 2

3354

1781

GFXBench 4.0 - Car Chase Offscreen (Fps)

GPU 1

GPU 2

33398

14076

GFXBench 4.0 - Manhattan (Fps)

GPU 1

GPU 2

3713

1976

GFXBench 4.0 - T-Rex (Fps)

GPU 1

GPU 2

3354

1781

Nom	NVIDIA GeForce RTX 3090	NVIDIA Tesla T4
PassMark - G3D Mark	26668	10616
PassMark - G2D Mark	1060	592
Geekbench - OpenCL	191142	61276
CompuBench 1.5 Desktop - Face Detection (mPixels/s)	690.619	127.622
CompuBench 1.5 Desktop - Ocean Surface Simulation (Frames/s)	7585.258	3026.812
CompuBench 1.5 Desktop - T-Rex (Frames/s)	62.614	18.798
CompuBench 1.5 Desktop - Video Composition (Frames/s)	209.424	119.936
CompuBench 1.5 Desktop - Bitcoin Mining (mHash/s)	2441.384	910.721
GFXBench 4.0 - Car Chase Offscreen (Frames)	33398	14076
GFXBench 4.0 - Manhattan (Frames)	3713	1976
GFXBench 4.0 - T-Rex (Frames)	3354	1781
GFXBench 4.0 - Car Chase Offscreen (Fps)	33398	14076
GFXBench 4.0 - Manhattan (Fps)	3713	1976
GFXBench 4.0 - T-Rex (Fps)	3354	1781
3DMark Fire Strike - Graphics Score	5118

Comparer les caractéristiques

	NVIDIA GeForce RTX 3090	NVIDIA Tesla T4
Essentiel
Architecture	Ampere	Turing
Nom de code	GA102	TU104
Date de sortie	1 Sep 2020	13 September 2018
Prix de sortie (MSRP)	$1499
Position dans l’évaluation de la performance	46	270
Genre	Desktop	Workstation
Infos techniques
Vitesse augmenté	1695 MHz	1515 MHz
Vitesse du noyau	1395 MHz	1005 MHz
Processus de fabrication	8 nm	12 nm
Peak Double Precision (FP64) Performance	556.0 GFLOPS (1:64)
Peak Half Precision (FP16) Performance	35.58 TFLOPS (1:1)
Peak Single Precision (FP32) Performance	35.58 TFLOPS
Pipelines	10496
Pixel fill rate	189.8 GPixel/s
Taux de remplissage de la texture	556.0 GTexel/s
Thermal Design Power (TDP)	350 Watt	75 Watt
Compte de transistor	28300 million	13,600 million
Sorties et ports de vidéo
Connecteurs d’écran	1x HDMI, 3x DisplayPort	No outputs
Compatibilité, dimensions et exigences
Hauteur	138 mm (5.4 inches)
Interface	PCIe 4.0 x16	PCIe 3.0 x16
Longeur	313 mm (12.3 inches)
Énergie du systeme recommandé (PSU)	750 Watt
Connecteurs d’énergie supplementaires	1x 12-pin	None
Largeur	Triple-slot
Soutien API
DirectX	12.2	12.0 (12_1)
OpenCL	2.0
OpenGL	4.6	4.6
Shader Model	6.5
Vulkan
Mémoire
RAM maximale	24 GB
Bande passante de la mémoire	936.2 GB/s
Largeur du bus mémoire	384 bit
Vitesse de mémoire	1219 MHz (19.5 Gbps effective)	10000 MHz
Genre de mémoire	GDDR6X