GPU Compute Throughput Timeline

Fig 1. FP32 throughput by launch year

Reading three decades on one scale

Published single-precision throughput offers one consistent, if limited, way to trace how programmable GPU arithmetic expanded. Plotting launch year against FP32 GFLOPS places products from different vendors and market eras in the same frame. The vertical axis is logarithmic because the range spans orders of magnitude; a linear axis would compress early products into an unreadable strip. The 747 variants plotted variants run from 2001 through 2026, with 488 variants from Nvidia, 247 variants from AMD, and 12 variants from Intel.

The unit is GPU variants. Closely related configurations can carry different clocks or enabled resources and therefore different theoretical figures. Keeping them separate preserves the source grain. The chart omits rows without both a launch year and a published FP32 value rather than treating missing throughput as zero. Zero would describe a measured capability; an unknown describes absent evidence, and confusing them would manufacture a trend.

What theoretical throughput means

FP32 GFLOPS is an arithmetic ceiling derived from resources, clock behavior, and the operations counted per cycle. It is not a benchmark and cannot rank gaming experience across architectures. Memory bandwidth, cache, scheduling, instruction mix, drivers, power limits, and workload occupancy all influence realized performance. Architectural changes can also make equal headline throughput behave differently. The plot should therefore be read as the growth of a published design capacity, not as a universal performance ladder.

Its strongest signal is structural. Long periods of gradual improvement are interrupted by denser process nodes, wider parallel designs, or clock changes; product tiers fan out within the same launch year. Cross-vendor series overlap instead of forming one clean succession. Those patterns are hard to see on isolated specification pages. This complete server-rendered frame makes the history inspectable without JavaScript, while the table retains the exact variants behind every point. Readers can use the figure to form questions, then use the rows and source revision to reproduce the set rather than relying on a decorative trend line.

Catalog variants

GPU variants in this editorial dataset
Vendor	Variant	Year	VRAM	Memory	FP32 GFLOPS	TDP	Connector evidence	Node	Transistors M
nvidia	XGPU(Xbox)	2001	64 MB	DDR	13.9	—	not recorded by source	150 nm	—
amd	Stream Processor (R580)	2006	1,024 MB	GDDR3	375	165 W	not recorded by source	—	—
nvidia	GeForce 8800 GTS 640	2006	640 MB	GDDR3	228.096	143 W	not recorded by source	90 nm	681
nvidia	GeForce 8800 GTX	2006	768 MB	GDDR3	345.6	155 W	not recorded by source	90 nm	681
amd	FireGL V3600	2007	256 MB	GDDR3	144	—	not recorded by source	65 nm	—
amd	FireGL V5600	2007	512 MB	GDDR4	192	—	not recorded by source	65 nm	—
amd	FireGL V7600	2007	512 MB	GDDR3	288	—	not recorded by source	80 nm	—
amd	FireGL V8600	2007	1,024 MB	GDDR4	440.32	—	not recorded by source	80 nm	—
amd	FireGL V8650	2007	2,048 MB	GDDR4	440.32	—	not recorded by source	80 nm	—
nvidia	GeForce 8800 GTS 320	2007	320 MB	GDDR3	228.096	143 W	not recorded by source	90 nm	681
nvidia	Quadro FX 4600	2007	768 MB	GDDR3	345	134 W	not recorded by source	90 nm	—
nvidia	Quadro FX 4600 SDI	2007	768 MB	GDDR3	345	154 W	not recorded by source	90 nm	—
nvidia	Quadro FX 5600	2007	1,536 MB	GDDR3	518.4	171 W	not recorded by source	90 nm	—
nvidia	GeForce 8300 GS	2007	128 MB	DDR2	14.688	40 W	connector absence confirmed by source	80 nm	210
nvidia	GeForce 8400 GS PCI Rev.2	2007	512 MB	DDR3	39.36	40 W	connector absence confirmed by source	40 nm	260
nvidia	GeForce 8500 GT	2007	—	GDDR3	29.376	30 W	connector absence confirmed by source	80 nm	210
nvidia	GeForce 8600 GS	2007	512 MB	DDR2	38.08	47 W	connector absence confirmed by source	80 nm	289
nvidia	GeForce 8600 GT Mac Edition	2007	128 MB	GDDR3	92.8	47 W	connector absence confirmed by source	80 nm	289
nvidia	C870 GPU Computing Module	2007	1,536 MB	GDDR3	345.6	170.9 W	not recorded by source	—	—
nvidia	D870 Deskside Computer	2007	—	GDDR3	691.2	520 W	not recorded by source	—	—
nvidia	GeForce 8800 Ultra	2007	768 MB	GDDR3	387.072	171 W	not recorded by source	90 nm	681
nvidia	S870 GPU Computing Server	2007	—	GDDR3	1,382	—	not recorded by source	—	—
nvidia	Quadro FX 360M	2007	256 MB	DDR2	38.4	17 W	not recorded by source	80 nm	—
nvidia	Quadro NVS 130M	2007	—	DDR2	19.2	10 W	not recorded by source	80 nm	—
nvidia	Quadro NVS 135M	2007	—	GDDR3	38.4	10 W	not recorded by source	80 nm	—
nvidia	Quadro NVS 140M	2007	—	GDDR3	38.4	10 W	not recorded by source	80 nm	—
amd	Radeon HD 2900 XT	2007	512 MB	GDDR3	475.5	215 W	not recorded by source	80 nm	720
amd	Radeon HD 2900 XT	2007	1,024 MB	GDDR4	475.5	215 W	not recorded by source	80 nm	720
amd	Radeon E2400 (RV610)	2007	—	GDDR3	48	—	not recorded by source	—	—
nvidia	Quadro FX 1600M	2007	512 MB	GDDR3	120	—	not recorded by source	80 nm	—
nvidia	Quadro FX 570M	2007	512 MB	GDDR3	91.2	45 W	not recorded by source	80 nm	—
nvidia	Quadro NVS 320M	2007	—	GDDR3	110.4	20 W	not recorded by source	65 nm	—
amd	Radeon HD 2350	2007	—	DDR2	42	20 W	not recorded by source	65 nm	180
amd	Radeon HD 2400 PRO	2007	256 MB	DDR2	42	20 W	not recorded by source	65 nm	180
amd	Radeon HD 2400 PRO	2007	512 MB	DDR2	42	20 W	not recorded by source	65 nm	180
amd	Radeon HD 2400 PRO	2007	128 MB	DDR2	42	20 W	not recorded by source	65 nm	180
amd	Radeon HD 2400 XT	2007	256 MB	DDR2 GDDR3	52	25 W	not recorded by source	65 nm	180
amd	Radeon HD 2600 PRO	2007	256 MB	DDR2	144	35 W	not recorded by source	65 nm	390
amd	Radeon HD 2600 PRO	2007	512 MB	GDDR3	144	35 W	not recorded by source	65 nm	390
amd	Radeon HD 2600 XT	2007	256 MB	GDDR3	192	45 W	not recorded by source	65 nm	390
amd	Radeon HD 2600 XT	2007	512 MB	GDDR4	192	50 W	not recorded by source	65 nm	390
nvidia	Quadro FX 1700	2007	512 MB	DDR2	88.32	42 W	not recorded by source	80 nm	—
nvidia	Quadro FX 370	2007	256 MB	DDR2	34.56	35 W	not recorded by source	80 nm	—
nvidia	Quadro FX 470	2007	—	DDR2	67.2	30 W	not recorded by source	65 nm	—
nvidia	Quadro FX 570	2007	256 MB	DDR2	44.1	38 W	not recorded by source	80 nm	—
nvidia	GeForce 8600 GTS Mac Edition	2007	512 MB	GDDR3	92.8	75 W	not recorded by source	80 nm	289
amd	Radeon HD 2900 PRO	2007	—	GDDR3 GDDR4	384	200 W	not recorded by source	80 nm	720
nvidia	NVS 290	2007	256 MB	DDR2	44.16	21 W	not recorded by source	80 nm	—
nvidia	GeForce 8800 GT	2007	512 MB	GDDR3	336	125 W	not recorded by source	65 nm	754
amd	Radeon HD 2900 GT	2007	512 MB	GDDR3	288.5	150 W	not recorded by source	80 nm	720
amd	Radeon HD 2900 GT	2007	256 MB	GDDR3	288.5	150 W	not recorded by source	80 nm	720
amd	FireStream 9170 (RV670)	2007	2,048 MB	GDDR3	512	105 W	not recorded by source	—	—
amd	Radeon HD 3850	2007	—	GDDR3 GDDR4	427.5	75 W	not recorded by source	55 nm	666
amd	Radeon HD 3870	2007	1,024 MB	GDDR4	497.3	106 W	not recorded by source	55 nm	666
amd	Radeon HD 3870	2007	512 MB	GDDR3	497.3	106 W	not recorded by source	55 nm	666
nvidia	GeForce 8800 GTS Core 112	2007	640 MB	GDDR3	268.8	150 W	not recorded by source	90 nm	681
nvidia	GeForce 8400	2007	256 MB	DDR2	20.8	25 W	connector absence confirmed by source	65 nm	210
nvidia	GeForce 8400 GS PCI	2007	512 MB	DDR2	22.4	25 W	connector absence confirmed by source	65 nm	210
nvidia	GeForce 8400 GS Rev. 2	2007	512 MB	DDR2	22.4	25 W	connector absence confirmed by source	65 nm	210
nvidia	GeForce 8800 GTS 512	2007	512 MB	GDDR3	416	135 W	not recorded by source	65 nm	754
nvidia	Quadro FX 3700	2008	512 MB	GDDR3	420	78 W	not recorded by source	65 nm	—
nvidia	Quadro VX 200	2008	512 MB	GDDR3	324	75 W	not recorded by source	65 nm	—
amd	Radeon HD 3450	2008	—	DDR2	48	25 W	not recorded by source	55 nm	181
amd	Radeon HD 3470	2008	256 MB	DDR2	64	30 W	not recorded by source	55 nm	181
amd	Radeon HD 3470	2008	512 MB	GDDR3	64	30 W	not recorded by source	55 nm	181
amd	Radeon HD 3650	2008	—	DDR2 GDDR3 GDDR4	174	65 W	not recorded by source	55 nm	378
nvidia	GeForce 8800 GT Mac Edition	2008	512 MB	GDDR3	336	125 W	not recorded by source	65 nm	754
amd	Radeon HD 3830	2008	256 MB	GDDR3	427.5	75 W	not recorded by source	55 nm	666
nvidia	GeForce 8100 + nForce 720a	2008	—	System shared DDR2 or DDR3	38.4	40 W	connector absence confirmed by source	80 nm	210
nvidia	GeForce 8300	2008	—	System shared DDR2 or DDR3	48	40 W	connector absence confirmed by source	80 nm	210
nvidia	S1070 GPU Computing Server "400 configuration"	2008	—	GDDR3	2,488	800 W	not recorded by source	—	—
nvidia	S1070 GPU Computing Server "500 configuration"	2008	—	GDDR3	2,765	800 W	not recorded by source	—	—
nvidia	S1075 GPU Computing Server	2008	—	GDDR3	2,765	—	not recorded by source	—	—
amd	FireStream 9250 (RV770)	2008	1,024 MB	GDDR3	1,000	150 W	not recorded by source	—	—
nvidia	GeForce GTX 260	2008	896 MB	GDDR3	476.928	182 W	not recorded by source	65 nm	1,400
nvidia	GeForce GTX 280	2008	1,024 MB	GDDR3	622.08	236 W	not recorded by source	65 nm	1,400
amd	Radeon HD 4850	2008	512 MB	GDDR3	1,000	110 W	not recorded by source	55 nm	956
amd	Radeon HD 4850	2008	2,048 MB	GDDR5	1,000	110 W	not recorded by source	55 nm	956
amd	Radeon HD 4850	2008	1,024 MB	GDDR4	1,000	110 W	not recorded by source	55 nm	956
amd	Radeon HD 4870	2008	512 MB	GDDR5	1,200	150 W	not recorded by source	55 nm	956
amd	Radeon HD 4870	2008	2,048 MB	GDDR5	1,200	150 W	not recorded by source	55 nm	956
amd	Radeon HD 4870	2008	1,024 MB	GDDR5	1,200	150 W	not recorded by source	55 nm	956
nvidia	GeForce GTX 260 Rev. 2	2008	896 MB	GDDR3	476.928	182 W	not recorded by source	55 nm	1,400
nvidia	Quadro Plex 2200 D2 Visual Computing System	2008	—	GDDR3	1,244	—	not recorded by source	—	—
nvidia	Quadro Plex 2200 S4 Visual Computing System	2008	—	GDDR3	2,488	1,200 W	not recorded by source	—	—
nvidia	GeForce 8400 SE	2008	—	DDR2	29.376	50 W	connector absence confirmed by source	80 nm	210
amd	Radeon HD 3550	2008	512 MB	DDR2	47.52	30 W	not recorded by source	55 nm	181
amd	FirePro 3D V3700	2008	256 MB	GDDR3	64	32 W	not recorded by source	55 nm	181
amd	FirePro 3D V5700	2008	512 MB	GDDR3	448	58 W	not recorded by source	55 nm	514
amd	Radeon HD 4870 X2	2008	—	GDDR5	2,400	286 W	not recorded by source	55 nm	—
nvidia	Quadro FX 2700M	2008	512 MB	GDDR3	190.8	65 W	not recorded by source	65 nm	—
nvidia	Quadro FX 3700M	2008	1,024 MB	GDDR3	528	75 W	not recorded by source	65 nm	—
nvidia	Quadro FX 3800M	2008	1,024 MB	GDDR3	648.192	100 W	not recorded by source	55 nm	—
nvidia	Quadro FX 770M	2008	512 MB	GDDR3	119	35 W	not recorded by source	65 nm	—
nvidia	Quadro FX 370M	2008	256 MB	GDDR3	33.6	20 W	not recorded by source	65 nm	—
nvidia	Quadro NVS 150M	2008	—	GDDR3	31.2	10 W	not recorded by source	65 nm	—
nvidia	Quadro NVS 160M	2008	256 MB	GDDR3	34.8	12 W	not recorded by source	65 nm	—
amd	Radeon HD 4670	2008	—	DDR2 GDDR3 GDDR4	480	59 W	not recorded by source	55 nm	514
amd	FirePro 3D V3750	2008	256 MB	GDDR3	352	48 W	not recorded by source	55 nm	514
amd	FirePro 3D V8700	2008	1,024 MB	GDDR5	1,200	151 W	not recorded by source	55 nm	956
nvidia	GeForce GTX 260 Core 216	2008	896 MB	GDDR3	536.544	182 W	not recorded by source	65 nm	1,400
amd	Radeon HD 4350	2008	—	DDR2 DDR3	92	20 W	not recorded by source	55 nm	242
amd	Radeon HD 4550	2008	—	DDR2 DDR3	96	25 W	not recorded by source	55 nm	242
nvidia	Quadro FX 1700M	2008	512 MB	GDDR3	148.8	50 W	not recorded by source	65 nm	—
amd	Radeon HD 3730	2008	1,024 MB	DDR2	173.3	65 W	not recorded by source	55 nm	378
amd	Radeon HD 3730	2008	512 MB	DDR2	173.3	65 W	not recorded by source	55 nm	378
amd	Radeon HD 3750	2008	512 MB	GDDR3	191	65 W	not recorded by source	55 nm	378
amd	Radeon HD 4830	2008	1,024 MB	GDDR4	736	95 W	not recorded by source	55 nm	956
amd	Radeon HD 4830	2008	512 MB	GDDR3	736	95 W	not recorded by source	55 nm	956
nvidia	Quadro FX 370 LP	2008	256 MB	DDR2	25.92	25 W	not recorded by source	80 nm	—
amd	Radeon HD 4850 X2	2008	—	GDDR3	2,000	250 W	not recorded by source	55 nm	—
nvidia	Quadro CX	2008	1,536 MB	GDDR3	693.504	150 W	not recorded by source	55 nm	—
nvidia	Quadro FX 4800	2008	1,536 MB	GDDR3	693.504	150 W	not recorded by source	55 nm	—
nvidia	Quadro FX 5800	2008	4,096 MB	GDDR3	878.4	189 W	not recorded by source	55 nm	—
amd	FireStream 9270 (RV770)	2008	2,048 MB	GDDR5	1,200	160 W	not recorded by source	—	—
amd	Radeon HD 4570	2008	1,024 MB	DDR2	104	25 W	not recorded by source	55 nm	242
nvidia	GeForce GTX 260 Core 216 Rev. 2	2008	896 MB	GDDR3	536.544	171 W	not recorded by source	55 nm	1,400
nvidia	GeForce GTX 275	2009	896 MB	GDDR3	673.92	219 W	not recorded by source	55 nm	1,400
nvidia	GeForce GTX 285	2009	1,024 MB	GDDR3	708.48	204 W	not recorded by source	55 nm	1,400
nvidia	GeForce GTX 285 Mac Edition	2009	1,024 MB	GDDR3	708.48	204 W	not recorded by source	55 nm	1,400
nvidia	GeForce GTS 250	2009	1,024 MB	GDDR3	414.72	150 W	not recorded by source	55 nm	754
nvidia	GeForce GTS 250	2009	1,024 MB	GDDR3	387.072	150 W	not recorded by source	55 nm	754
amd	FirePro 3D V7750	2009	1,024 MB	GDDR3	512	76 W	not recorded by source	55 nm	514
nvidia	Quadro FX 1800	2009	768 MB	GDDR3	264	59 W	not recorded by source	65 nm	—
nvidia	Quadro FX 380	2009	256 MB	GDDR3	52.8	34 W	not recorded by source	65 nm	—
nvidia	Quadro FX 3800	2009	1,024 MB	GDDR3	691.2	108 W	not recorded by source	55 nm	—
amd	Radeon HD 4890	2009	1,024 MB	GDDR5	1,360	190 W	not recorded by source	55 nm	959
amd	Radeon HD 4890	2009	2,048 MB	GDDR5	1,360	190 W	not recorded by source	55 nm	959
nvidia	C1060 GPU Computing Module	2009	4,096 MB	GDDR3	622.1	187.8 W	not recorded by source	—	—
nvidia	Quadro FX 580	2009	512 MB	GDDR3	108	40 W	not recorded by source	65 nm	—
nvidia	GeForce GT 230 OEM	2009	1,536 MB	DDR2	240	75 W	connector absence confirmed by source	55 nm	754
amd	Radeon HD 4770	2009	512 MB	GDDR5	960	80 W	not recorded by source	40 nm	826
amd	Radeon HD 3410	2009	256 MB	DDR2	41.52	20 W	not recorded by source	65 nm	180
nvidia	NVS 295	2009	256 MB	GDDR3	31.2	23 W	not recorded by source	65 nm	—
nvidia	GeForce G210 OEM	2009	512 MB	DDR2	43.2	30.5 W	connector absence confirmed by source	55 nm	314
amd	Radeon E4690 (RV730)	2009	512 MB	GDDR3	384	—	not recorded by source	—	—
nvidia	Quadro FX 1800M	2009	1,024 MB	GDDR3 GDDR5	233.28	45 W	not recorded by source	40 nm	—
nvidia	GeForce GTS 240 OEM	2009	1,024 MB	GDDR3	362.88	120 W	not recorded by source	55 nm	754
amd	FirePro 3D V8750	2009	2,048 MB	GDDR5	1,200	154 W	not recorded by source	55 nm	956
nvidia	GeForce G210 OEM Rev. 2	2009	128 MB	DDR2	44.864	30.5 W	connector absence confirmed by source	40 nm	260
nvidia	GeForce 210 OEM	2009	1,024 MB	DDR2	35.2	30.5 W	connector absence confirmed by source	40 nm	486
amd	Radeon HD 4750	2009	512 MB	GDDR5	934.4	80 W	not recorded by source	40 nm	826
amd	Radeon HD 4860	2009	1,024 MB	GDDR5	896	130 W	not recorded by source	55 nm	959
amd	Radeon HD 4860	2009	512 MB	GDDR5	896	130 W	not recorded by source	55 nm	959
amd	Radeon HD 5870	2009	2,048 MB	GDDR5	2,720	228 W	not recorded by source	40 nm	2,154
amd	Radeon HD 5870	2009	1,024 MB	GDDR5	2,720	188 W	not recorded by source	40 nm	2,154
amd	Radeon HD 3610	2009	1,024 MB	DDR2	142.6	35 W	not recorded by source	65 nm	390
amd	Radeon HD 3610	2009	512 MB	DDR2	142.6	35 W	not recorded by source	65 nm	390
amd	Radeon HD 5850	2009	1,024 MB	GDDR5	2,088	151 W	not recorded by source	40 nm	2,154
amd	Radeon HD 5850	2009	2,048 MB	GDDR5	2,088	151 W	not recorded by source	40 nm	2,154
nvidia	GeForce 210	2009	512 MB	DDR3	39.36	30.5 W	connector absence confirmed by source	40 nm	260
nvidia	GeForce 210 PCI	2009	512 MB	DDR2	35.2	30.5 W	connector absence confirmed by source	40 nm	486
nvidia	GeForce 210 Rev. 2	2009	1,024 MB	DDR3	39.36	30.5 W	connector absence confirmed by source	40 nm	260
nvidia	GeForce GT 220	2009	—	DDR2 GDDR3	130.56	58 W	connector absence confirmed by source	40 nm	486
amd	Radeon HD 5750	2009	1,024 MB	GDDR5	1,008	86 W	not recorded by source	40 nm	1,040
amd	Radeon HD 5750	2009	512 MB	GDDR5	1,008	86 W	not recorded by source	40 nm	1,040
amd	Radeon HD 5770	2009	1,024 MB	GDDR5	1,360	108 W	not recorded by source	40 nm	1,040
amd	Radeon HD 5770	2009	512 MB	GDDR5	1,360	108 W	not recorded by source	40 nm	1,040
nvidia	GeForce GT 240	2009	1,024 MB	GDDR5	257.28	69 W	connector absence confirmed by source	40 nm	727
amd	Radeon HD 5970	2009	—	GDDR5	4,640	294 W	not recorded by source	40 nm	—
nvidia	GeForce 205 OEM	2009	512 MB	DDR2	44.864	30.5 W	connector absence confirmed by source	40 nm	260
nvidia	Quadro FX 2800M	2009	1,024 MB	GDDR3	360	75 W	not recorded by source	55 nm	—
nvidia	Quadro FX 380 LP	2009	512 MB	GDDR3	67.296	28 W	not recorded by source	40 nm	—
nvidia	GeForce GTX 260 OEM	2009	1,792 MB	GDDR3	414.72	150 W	not recorded by source	65 nm	1,400
nvidia	NVS 2100M	2010	—	GDDR3	59.04	14 W	not recorded by source	40 nm	—
nvidia	NVS 3100M	2010	—	GDDR3	70.56	14 W	not recorded by source	40 nm	—
nvidia	NVS 4200M	2010	—	GDDR3	155.52	25 W	not recorded by source	40 nm	—
nvidia	Quadro FX 380M	2010	512 MB	GDDR3	73.44	25 W	not recorded by source	40 nm	—
nvidia	Quadro FX 880M	2010	1,024 MB	GDDR3	174.24	35 W	not recorded by source	40 nm	—
amd	Radeon HD 5670	2010	—	GDDR3 GDDR5	620	64 W	not recorded by source	40 nm	627
nvidia	GeForce GT 220	2010	1,024 MB	GDDR3	144	58 W	connector absence confirmed by source	65 nm	505
amd	Radeon HD 5450	2010	—	DDR2 DDR3	104	19.1 W	not recorded by source	40 nm	292
amd	Radeon HD 5550	2010	1,024 MB	GDDR3	352	39 W	not recorded by source	40 nm	627
amd	Radeon HD 5550	2010	2,048 MB	GDDR5	352	39 W	not recorded by source	40 nm	627
amd	Radeon HD 5550	2010	512 MB	DDR2	352	39 W	not recorded by source	40 nm	627
amd	Radeon HD 5570	2010	—	DDR2 GDDR3 GDDR5	520	39 W	not recorded by source	40 nm	627
amd	Radeon HD 5830	2010	1,024 MB	GDDR5	1,792	175 W	not recorded by source	40 nm	2,154
amd	Radeon HD 5870 Eyefinity Edition	2010	2,048 MB	GDDR5	2,720	228 W	not recorded by source	40 nm	2,154
nvidia	GeForce GTX 470	2010	1,280 MB	GDDR5	1,088.6	215 W	not recorded by source	—	3,100
nvidia	GeForce GTX 480	2010	1,536 MB	GDDR5	1,345	250 W	not recorded by source	—	3,100
amd	FirePro 3D V8800	2010	2,048 MB	GDDR5	2,640	208 W	not recorded by source	40 nm	2,154
amd	FirePro 3D V3800	2010	512 MB	GDDR3	520	43 W	not recorded by source	40 nm	627
amd	FirePro 3D V4800	2010	1,024 MB	GDDR5	620	69 W	not recorded by source	40 nm	672
amd	FirePro 3D V5800	2010	1,024 MB	GDDR5	1,120	74 W	not recorded by source	40 nm	1,040
amd	FirePro 3D V7800	2010	2,048 MB	GDDR5	2,016	138 W	not recorded by source	40 nm	2,154
amd	FirePro RG220	2010	512 MB	GDDR3 for GPU, RDRAM for PCoIP	80	35 W	not recorded by source	55 nm	—
nvidia	GeForce GTX 465	2010	1,024 MB	GDDR5	855.4	200 W	not recorded by source	—	3,100
amd	FireStream 9350 (Cypress XT)	2010	2,048 MB	GDDR5	2,016	150 W	not recorded by source	—	—
amd	FireStream 9370 (Cypress XT)	2010	4,096 MB	GDDR5	2,640	225 W	not recorded by source	—	—
amd	Radeon HD 3570	2010	512 MB	DDR2	63.68	30 W	not recorded by source	55 nm	181
nvidia	GeForce 8400 GS Rev. 3	2010	512 MB	DDR2	39.36	25 W	connector absence confirmed by source	40 nm	260
nvidia	GeForce GTX 460	2010	768 MB	GDDR5	907.2	160 W	not recorded by source	—	1,950
nvidia	GeForce GTX 460	2010	768 MB	GDDR5	907.2	160 W	not recorded by source	—	1,950
nvidia	GeForce GTX 460	2010	1,024 MB	GDDR5	907.2	160 W	not recorded by source	—	1,950
nvidia	Quadro 5000M	2010	2,048 MB	GDDR5	518.4	100 W	not recorded by source	40 nm	—
nvidia	GeForce GT 420 OEM	2010	1,024 MB	DDR3	134.4	50 W	connector absence confirmed by source	—	585
amd	FirePro 3D V9800	2010	4,096 MB	GDDR5	2,720	225 W	not recorded by source	40 nm	2,154
nvidia	GeForce GTS 450	2010	1,024 MB	GDDR5	601.3	106 W	not recorded by source	—	1,170
nvidia	GeForce GTS 450 OEM	2010	512 MB	GDDR5	451	106 W	not recorded by source	—	1,170
nvidia	GeForce GT 430	2010	512 MB	DDR3	268.8	49 W	connector absence confirmed by source	—	585
nvidia	GeForce GT 430 OEM	2010	2,048 MB	DDR3	268.8	49 W	connector absence confirmed by source	—	585
nvidia	GeForce GT 440 OEM	2010	1,536 MB	DDR3	466.6	65 W	connector absence confirmed by source	—	1,170
nvidia	GeForce GTS 450 OEM	2010	1,536 MB	GDDR5	455	106 W	not recorded by source	—	1,170
nvidia	GeForce GTX 460 OEM	2010	1,024 MB	GDDR5	873.6	150 W	not recorded by source	—	1,950
amd	Radeon HD 6870 (Barts XT)	2010	1,024 MB	GDDR5	2,016	151 W	not recorded by source	—	1,700
nvidia	Quadro 4000	2010	2,048 MB	GDDR5	486.4	142 W	not recorded by source	40 nm	—
nvidia	GeForce GTX 580	2010	1,536 MB	GDDR5	1,581.056	244 W	not recorded by source	—	3,000
nvidia	GeForce GTX 460 SE	2010	1,024 MB	GDDR5	748.8	150 W	not recorded by source	—	1,950
nvidia	GeForce GTX 460 SE V2	2010	768 MB	GDDR5	748.8	150 W	not recorded by source	—	1,950
nvidia	GeForce GTX 570	2010	1,280 MB	GDDR5	1,405.44	219 W	not recorded by source	—	3,000
nvidia	Quadro 6000	2010	6,144 MB	GDDR5	1,028.608	204 W	not recorded by source	40 nm	—
nvidia	Quadro 600	2010	1,024 MB	DDR3	245.76	40 W	not recorded by source	40 nm	—
nvidia	Quadro 2000	2010	1,024 MB	GDDR5	480	62 W	not recorded by source	40 nm	—
nvidia	NVS 300	2011	512 MB	DDR3	67.3	17.5 W	not recorded by source	40 nm	—
nvidia	Quadro 2000M	2011	2,048 MB	DDR3	422.4	55 W	not recorded by source	40 nm	—
amd	Radeon HD 6750 (Juniper Pro)	2011	512 MB	GDDR5	1,008	86 W	not recorded by source	—	1,040
nvidia	GeForce GTX 560 Ti	2011	1,024 MB	GDDR5	1,236.36	170 W	not recorded by source	—	1,950
nvidia	GeForce GT 440 Mac Edition	2011	1,024 MB	DDR3	311	65 W	connector absence confirmed by source	—	585
nvidia	NVS 5100M	2011	—	GDDR3	174.24	35 W	not recorded by source	40 nm	—
nvidia	Quadro 3000M	2011	2,048 MB	GDDR5	432	75 W	not recorded by source	40 nm	—
nvidia	Quadro 4000M	2011	2,048 MB	GDDR5	638.4	100 W	not recorded by source	40 nm	—
nvidia	Quadro 500M	2011	1,024 MB	DDR3	268.8	35 W	not recorded by source	40 nm	—
nvidia	Quadro 5010M	2011	4,096 MB	GDDR5	691.2	100 W	not recorded by source	40 nm	—
nvidia	Quadro 5000	2011	2,560 MB	GDDR5	722.304	152 W	not recorded by source	40 nm	—
amd	Radeon HD 6990 (Antilles XT)	2011	—	GDDR5	5,099	375 W	not recorded by source	—	—
nvidia	GeForce GTX 560 Ti OEM	2011	1,024 MB	GDDR5	1,236.36	170 W	not recorded by source	—	1,950
nvidia	GeForce GTS 450 Rev. 2	2011	1,024 MB	GDDR5	601.3	106 W	not recorded by source	—	1,170
nvidia	GeForce GTX 550 Ti	2011	1,024 MB	GDDR5	691.2	116 W	not recorded by source	—	1,170
nvidia	GeForce GTX 560 Ti OEM	2011	1,280 MB	GDDR5	1,030.656	210 W	not recorded by source	—	3,000
amd	Radeon E6460 (Caicos)	2011	512 MB	GDDR5	192	—	not recorded by source	—	—
nvidia	Quadro 400	2011	512 MB	DDR3	108	32 W	not recorded by source	40 nm	—
nvidia	GeForce GT 520	2011	1,024 MB	DDR3	155.52	29 W	connector absence confirmed by source	—	292
amd	Radeon HD 6570 (Turks Pro)	2011	2,048 MB	DDR3 GDDR5	624	60 W	not recorded by source	—	716
amd	Radeon HD 6770 (Juniper XT)	2011	512 MB	GDDR5	1,360	108 W	not recorded by source	—	1,040
amd	Radeon E6760 (Turks)	2011	1,024 MB	GDDR5	576	—	not recorded by source	—	—
amd	Radeon HD 5610	2011	1,024 MB	GDDR3	520	39 W	not recorded by source	40 nm	627
nvidia	GeForce GT 545 OEM	2011	1,024 MB	GDDR5	501.408	105 W	not recorded by source	—	1,170
nvidia	GeForce GTX 555 OEM	2011	1,024 MB	GDDR5	847.872	150 W	not recorded by source	—	1,950
nvidia	GeForce GTX 560	2011	1,024 MB	GDDR5	1,088.64	150 W	not recorded by source	—	1,950
amd	FirePro V5900	2011	2,048 MB	GDDR5	610	—	not recorded by source	40 nm	—
amd	FirePro V7900	2011	2,048 MB	GDDR5	1,860	—	not recorded by source	40 nm	—
nvidia	GeForce GTX 580 Rev. 2	2011	1,536 MB	GDDR5	1,581.056	244 W	not recorded by source	—	3,000
nvidia	C2050 GPU Computing Module	2011	3,072 MB	GDDR5	1,030	247 W	not recorded by source	—	—
nvidia	C2070 GPU Computing Module	2011	6,144 MB	GDDR5	1,030	247 W	not recorded by source	—	—
nvidia	C2075 GPU Computing Module	2011	6,144 MB	GDDR5	1,030	225 W	not recorded by source	—	—
nvidia	M2050 GPU Computing Module	2011	3,072 MB	GDDR5	1,030	225 W	not recorded by source	—	—
nvidia	M2070/M2070Q GPU Computing Module	2011	6,144 MB	GDDR5	1,030	225 W	not recorded by source	—	—
nvidia	M2090 GPU Computing Module	2011	6,144 MB	GDDR5	1,331	225 W	not recorded by source	—	—
nvidia	S2050 GPU Computing Server	2011	—	GDDR5	4,122	900 W	not recorded by source	—	—
nvidia	S2070 GPU Computing Server	2011	—	GDDR5	4,122	900 W	not recorded by source	—	—
nvidia	GeForce GTX 460 V2	2011	1,024 MB	GDDR5	1,046.3	160 W	not recorded by source	—	1,950
nvidia	GeForce GTX 460 V2 E.S.	2011	1,280 MB	GDDR5	1,046.3	160 W	not recorded by source	—	1,950
amd	FirePro V4900	2011	1,024 MB	GDDR5	768	—	not recorded by source	40 nm	—
amd	Radeon HD 4580	2011	512 MB	GDDR3	191	65 W	not recorded by source	55 nm	378
nvidia	GeForce GTX 560 OEM	2011	1,280 MB	GDDR5	847.872	150 W	not recorded by source	—	3,000
nvidia	GeForce GTX 560 Ti 448	2011	1,280 MB	GDDR5	1,311.744	210 W	not recorded by source	—	3,000
amd	Radeon HD 6930 (Cayman CE)	2011	1,024 MB	GDDR5	1,920	186 W	not recorded by source	—	2,640
amd	Radeon HD 7970 (Tahiti XT)	2011	6,144 MB	GDDR5	3,788.8	250 W	not recorded by source	—	4,313
amd	Radeon HD 7970 (Tahiti XT)	2011	3,072 MB	GDDR5	3,788.8	250 W	not recorded by source	—	4,313
amd	Radeon HD 7350 (Cedar)	2012	—	DDR2 DDR3	104	19.1 W	not recorded by source	—	292
amd	Radeon HD 7450 (Caicos)	2012	1,024 MB	DDR3	240	18 W	not recorded by source	—	370
amd	Radeon HD 7450 (Caicos)	2012	512 MB	DDR3	200	18 W	not recorded by source	—	370
amd	Radeon HD 7470 (Caicos)	2012	512 MB	DDR3 GDDR5	200	27 W	not recorded by source	—	370
amd	Radeon HD 7470 (Caicos)	2012	1,024 MB	DDR3 GDDR5	248	27 W	not recorded by source	—	370
amd	Radeon HD 7570 (Turks Pro-L)	2012	512 MB	DDR3 GDDR5	624	44 W	not recorded by source	—	716
amd	Radeon HD 7570 (Turks Pro-L)	2012	1,024 MB	DDR3 GDDR5	624	60 W	not recorded by source	—	716
amd	Radeon HD 7670 (Turks XT)	2012	512 MB	GDDR5	768	66 W	not recorded by source	—	716
amd	Radeon HD 7670 (Turks XT)	2012	1,024 MB	GDDR5	768	66 W	not recorded by source	—	716
amd	Radeon HD 7950 (Tahiti Pro)	2012	3,072 MB	GDDR5	2,867.2	200 W	not recorded by source	—	4,313
amd	Radeon HD 7770 GHz Edition (Cape Verde XT)	2012	1,024 MB	GDDR5	1,280	80 W	not recorded by source	—	1,500
amd	Radeon HD 7770 GHz Edition (Cape Verde XT)	2012	2,048 MB	GDDR5	1,280	80 W	not recorded by source	—	1,500
amd	FirePro V3900	2012	1,024 MB	GDDR3	624	50 W	not recorded by source	40 nm	—
nvidia	GeForce GTX 560 SE	2012	1,024 MB	GDDR5	847.872	150 W	not recorded by source	—	1,950
amd	Radeon HD 7850 (Pitcairn Pro)	2012	1,024 MB	GDDR5	1,761.28	130 W	not recorded by source	—	2,800
amd	Radeon HD 7850 (Pitcairn Pro)	2012	2,048 MB	GDDR5	1,761.28	130 W	not recorded by source	—	2,800
amd	Radeon HD 7870 GHz Edition (Pitcairn XT)	2012	2,048 MB	GDDR5	2,560	175 W	not recorded by source	—	2,800
nvidia	GeForce GTX 680	2012	2,048 MB	GDDR5	3,090.432	195 W	not recorded by source	28 nm	3,540
nvidia	GeForce 605 OEM	2012	—	DDR3	100.416	25 W	connector absence confirmed by source	40 nm	292
nvidia	GeForce GT 610	2012	512 MB	DDR3	155.52	29 W	connector absence confirmed by source	40 nm	292
nvidia	GeForce GT 620 OEM	2012	—	DDR3	155.52	30 W	connector absence confirmed by source	40 nm	292
nvidia	GeForce GT 630 OEM	2012	—	DDR3	336	50 W	connector absence confirmed by source	28 nm	1,270
nvidia	GeForce GT 640 OEM	2012	1,024 MB	GDDR5	612.096	50 W	connector absence confirmed by source	28 nm	1,270
nvidia	GeForce GT 640 OEM	2012	—	DDR3	612.096	50 W	connector absence confirmed by source	28 nm	1,270
nvidia	GeForce GT 640 OEM Rebrand	2012	1,536 MB	DDR3	414.72	75 W	connector absence confirmed by source	40 nm	1,170
nvidia	GeForce GT 645 OEM	2012	—	GDDR5	893.952	102 W	not recorded by source	40 nm	1,950
nvidia	K10 GPU accelerator	2012	—	GDDR5	4,577	225 W	not recorded by source	—	—
nvidia	Quadro 7000	2012	6,144 MB	GDDR5	1,332	204 W	not recorded by source	40 nm	—
nvidia	GeForce GTX 670	2012	2,048 MB	GDDR5	2,459.52	170 W	not recorded by source	28 nm	3,540
nvidia	GeForce 615 OEM	2012	1,024 MB	DDR3	126.72	49 W	connector absence confirmed by source	40 nm	585
nvidia	GeForce GT 620	2012	1,024 MB	DDR3	268.8	49 W	connector absence confirmed by source	40 nm	585
nvidia	GeForce GT 630	2012	1,024 MB	DDR3	311.04	65 W	connector absence confirmed by source	40 nm	585
amd	Radeon HD 7970 GHz Edition (Tahiti XT2)	2012	3,072 MB	GDDR5	4,096	250 W	not recorded by source	—	4,313
amd	Radeon HD 7970 GHz Edition (Tahiti XT2)	2012	6,144 MB	GDDR5	4,300.8	250 W	not recorded by source	—	4,313
nvidia	NVS 5200M	2012	—	GDDR3	240	35 W	not recorded by source	—	—
nvidia	NVS 5400M	2012	—	GDDR3	253.44	35 W	not recorded by source	—	—
nvidia	Quadro K1000M	2012	2,048 MB	DDR3	326.4	45 W	not recorded by source	28 nm	—
nvidia	Quadro K2000M	2012	2,048 MB	DDR3	572.16	55 W	not recorded by source	28 nm	—
nvidia	Quadro K3000M	2012	2,048 MB	GDDR5	753.41	75 W	not recorded by source	28 nm	—
nvidia	Quadro K4000M	2012	4,096 MB	GDDR5	1,152	100 W	not recorded by source	28 nm	—
nvidia	Quadro K500M	2012	1,024 MB	DDR3	326.4	35 W	not recorded by source	28 nm	—
nvidia	GeForce GT 640	2012	2,048 MB	DDR3	692.736	65 W	connector absence confirmed by source	28 nm	1,270
amd	FirePro W5000	2012	2,048 MB	GDDR5	1,267.2	—	not recorded by source	28 nm	—
amd	FirePro W600	2012	2,048 MB	GDDR5	768	75 W	not recorded by source	28 nm	—
amd	FirePro W7000	2012	4,096 MB	GDDR5	2,432	—	not recorded by source	28 nm	—
amd	FirePro W8000	2012	4,096 MB	GDDR5	3,225.6	—	not recorded by source	28 nm	—
amd	FirePro W9000	2012	6,144 MB	GDDR5	3,993.6	274 W	not recorded by source	28 nm	—
nvidia	NVS 310	2012	512 MB	DDR3	100.4	19.5 W	not recorded by source	40 nm	—
nvidia	GeForce GTS 450 Rev. 3	2012	1,024 MB	DDR3	451	106 W	not recorded by source	—	1,170
nvidia	Quadro 410	2012	512 MB	DDR3	271.1	38 W	not recorded by source	28 nm	—
nvidia	Quadro K5000M	2012	4,096 MB	GDDR5	1,897.73	100 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 660 Ti	2012	2,048 MB	GDDR5	2,459.52	150 W	not recorded by source	28 nm	3,540
nvidia	Quadro K5000	2012	4,096 MB	GDDR5	2,169	122 W	not recorded by source	28 nm	—
nvidia	GeForce GT 520 OEM	2012	1,024 MB	DDR3	134.592	29 W	connector absence confirmed by source	—	292
nvidia	GeForce GTX 660 OEM	2012	—	GDDR5	1,896.192	130 W	not recorded by source	28 nm	3,540
amd	FirePro S7000	2012	4,096 MB	GDDR5	2,432	—	not recorded by source	28 nm	—
amd	FirePro S9000	2012	6,144 MB	GDDR5	3,225.6	—	not recorded by source	28 nm	—
nvidia	GeForce GTX 650	2012	1,024 MB	GDDR5	812.554	65 W	not recorded by source	28 nm	1,270
nvidia	GeForce GTX 660	2012	2,048 MB	GDDR5	1,505.28	140 W	not recorded by source	28 nm	2,540
nvidia	GeForce GT 610 OEM	2012	1,024 MB	DDR3	115.2	29 W	connector absence confirmed by source	40 nm	292
nvidia	GeForce GTX 650 Ti	2012	1,024 MB	GDDR5	1,425.408	110 W	not recorded by source	28 nm	2,540
nvidia	NVS 510	2012	2,048 MB	DDR3	306	35 W	not recorded by source	28 nm	—
amd	FirePro S10000	2012	—	GDDR5	5,913.6	—	not recorded by source	28 nm	—
amd	FirePro S10000 passive	2012	—	GDDR5	5,913.6	—	not recorded by source	28 nm	—
nvidia	K20 GPU accelerator	2012	5,120 MB	GDDR5	3,524	225 W	not recorded by source	—	—
nvidia	K20X GPU accelerator	2012	6,144 MB	GDDR5	3,935	235 W	not recorded by source	—	—
amd	Radeon HD 8350 (Cedar)	2013	—	DDR2 DDR3	104	—	not recorded by source	—	292
amd	Radeon HD 8450 (Caicos)	2013	512 MB	DDR3	200	—	not recorded by source	—	370
amd	Radeon HD 8470 (Caicos)	2013	1,024 MB	GDDR5	240	—	not recorded by source	—	370
amd	Radeon HD 8570 (Oland)	2013	2,048 MB	DDR3 GDDR5	560	—	not recorded by source	—	950
amd	Radeon HD 8670 (Oland)	2013	2,048 MB	GDDR5	768	—	not recorded by source	—	950
amd	Radeon HD 8760 (Cape Verde XT)	2013	2,048 MB	GDDR5	1,280	—	not recorded by source	—	1,500
amd	Radeon HD 8870 (Pitcairn XT)	2013	2,048 MB	GDDR5	2,560	—	not recorded by source	—	2,800
amd	Radeon HD 7510 (Turks LE)	2013	1,024 MB	DDR3	416	—	not recorded by source	—	716
nvidia	GeForce GT 625 OEM	2013	1,024 MB	DDR3	167.808	29 W	connector absence confirmed by source	40 nm	292
nvidia	GeForce GTX TITAN	2013	6,144 MB	GDDR5	4,494.3	250 W	not recorded by source	—	7,080
amd	FirePro R5000	2013	2,048 MB	GDDR5	1,267.2	—	not recorded by source	28 nm	—
amd	Radeon HD 7790 (Bonaire XT)	2013	2,048 MB	GDDR5	1,792	85 W	not recorded by source	—	2,080
amd	Radeon HD 7790 (Bonaire XT)	2013	1,024 MB	GDDR5	1,792	85 W	not recorded by source	—	2,080
nvidia	Quadro K2000	2013	2,048 MB	GDDR5	732.7	51 W	not recorded by source	28 nm	—
nvidia	Quadro K2000D	2013	2,048 MB	GDDR5	732.7	51 W	not recorded by source	28 nm	—
nvidia	Quadro K4000	2013	3,072 MB	GDDR5	1,245	80 W	not recorded by source	28 nm	—
nvidia	Quadro K600	2013	1,024 MB	DDR3	336.4	41 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 650 Ti Boost	2013	2,048 MB	GDDR5	1,505.28	134 W	not recorded by source	28 nm	2,540
nvidia	GeForce GTX 650 Ti OEM	2013	2,048 MB	GDDR5	1,425.408	110 W	not recorded by source	28 nm	2,540
amd	Radeon HD 7730 (Cape Verde LE)	2013	1,024 MB	DDR3 GDDR5	614.4	47 W	not recorded by source	—	1,500
amd	Radeon Sky 500	2013	4,096 MB	GDDR5	2,432	150 W	not recorded by source	28 nm	—
amd	Radeon Sky 700	2013	6,144 MB	GDDR5	3,225.6	225 W	not recorded by source	28 nm	—
amd	Radeon Sky 900	2013	—	GDDR5	5,913.6	300 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 680 Mac Edition	2013	2,048 MB	GDDR5	3,090.432	195 W	not recorded by source	28 nm	3,540
nvidia	GeForce GTX 645 OEM	2013	1,024 MB	GDDR5	949.248	65 W	connector absence confirmed by source	28 nm	2,540
nvidia	GeForce GTX 780	2013	3,072 MB	GDDR5	3,976.7	250 W	not recorded by source	—	7,080
nvidia	GeForce GT 630 Rev. 2	2013	2,048 MB	DDR3	692.736	25 W	connector absence confirmed by source	28 nm	1,020
nvidia	GeForce GT 640 Rev. 2	2013	1,024 MB	GDDR5	803.328	49 W	connector absence confirmed by source	28 nm	1,020
nvidia	GeForce GTX 770	2013	2,048 MB	GDDR5	3,213.2	230 W	not recorded by source	—	3,540
nvidia	GeForce GTX 760	2013	2,048 MB	GDDR5	2,258.2	170 W	not recorded by source	—	3,540
nvidia	GeForce GTX 760 OEM	2013	1,536 MB	GDDR5	1,896.2	130 W	not recorded by source	—	3,540
amd	Radeon HD 8490 (Caicos)	2013	1,024 MB	DDR3L GDDR5	280	—	not recorded by source	—	370
nvidia	Quadro K1100M	2013	2,048 MB	GDDR5	549.89	45 W	not recorded by source	28 nm	—
nvidia	Quadro K2100M	2013	2,048 MB	GDDR5	753.41	55 W	not recorded by source	28 nm	—
nvidia	Quadro K3100M	2013	4,096 MB	GDDR5	1,044.48	75 W	not recorded by source	28 nm	—
nvidia	Quadro K4100M	2013	4,096 MB	GDDR5	1,626.624	100 W	not recorded by source	28 nm	—
nvidia	Quadro K5100M	2013	8,192 MB	GDDR5	2,368.51	100 W	not recorded by source	28 nm	—
nvidia	Quadro K510M	2013	1,024 MB	GDDR5	326.4	30 W	not recorded by source	28 nm	—
nvidia	Quadro K6000	2013	12,288 MB	GDDR5	5,196	225 W	not recorded by source	28 nm	—
nvidia	Quadro K610M	2013	1,024 MB	GDDR5	376.32	30 W	not recorded by source	28 nm	—
amd	Radeon R7 240 (Oland Pro)	2013	2,048 MB	DDR3 GDDR5	467.2	30 W	not recorded by source	—	1,040
amd	Radeon R7 250 (Oland XT)	2013	1,024 MB	DDR3 GDDR5	768	75 W	not recorded by source	—	1,040
amd	Radeon R7 260X (Bonaire XTX)	2013	1,024 MB	GDDR5	1,971.2	115 W	not recorded by source	—	2,080
amd	Radeon R9 270X (Pitcairn XT)	2013	2,048 MB	GDDR5	2,560	180 W	not recorded by source	—	2,800
amd	Radeon HD 8770 (Bonaire XT)	2013	2,048 MB	GDDR5	1,792	—	not recorded by source	—	2,080
amd	Radeon HD 8730 (Cape Verde LE)	2013	1,024 MB	GDDR5	614.4	—	not recorded by source	—	1,500
nvidia	GeForce GTX 780 Rev. 2	2013	6,144 MB	GDDR5	3,976.7	250 W	not recorded by source	—	7,080
nvidia	GeForce GTX 780 Rev. 2	2013	3,072 MB	GDDR5	3,976.7	250 W	not recorded by source	—	7,080
nvidia	GeForce GTX 760 Ti OEM	2013	2,048 MB	GDDR5	2,459.5	170 W	not recorded by source	—	3,540
nvidia	GeForce GT 635 OEM	2013	1,024 MB	DDR3	742.656	35 W	connector absence confirmed by source	28 nm	1,020
amd	Radeon R5 240 (Oland)	2013	1,024 MB	DDR3 GDDR3	560.6	30 W	not recorded by source	—	1,040
nvidia	GeForce GTX 780 Ti	2013	3,072 MB	GDDR5	5,040	250 W	not recorded by source	—	7,080
nvidia	GeForce GTX 760 X2	2013	—	GDDR5	4,635.6	250 W	not recorded by source	—	—
nvidia	GeForce GTX 650	2013	1,024 MB	GDDR5	812.554	65 W	not recorded by source	28 nm	2,540
amd	Radeon R7 250E (Cape Verde Pro)	2013	1,024 MB	GDDR5	819.2	55 W	not recorded by source	—	1,500
amd	FirePro D300	2014	2,048 MB	GDDR5	2,176	150 W	not recorded by source	28 nm	—
amd	FirePro D500	2014	3,072 MB	GDDR5	2,227	274 W	not recorded by source	28 nm	—
amd	FirePro D700	2014	6,144 MB	GDDR5	3,482	274 W	not recorded by source	28 nm	—
amd	Radeon R7 250X (Cape Verde XT)	2014	1,024 MB	GDDR5	1,280	95 W	not recorded by source	—	1,500
nvidia	GeForce GTX 745 OEM	2014	4,096 MB	DDR3	793.3	55 W	connector absence confirmed by source	—	1,870
nvidia	GeForce GTX 750	2014	1,024 MB	GDDR5	1,044.4	55 W	connector absence confirmed by source	—	1,870
nvidia	GeForce GTX 750 Ti	2014	2,048 MB	GDDR5	1,305.6	60 W	connector absence confirmed by source	—	1,870
nvidia	GeForce GTX 750 Ti OEM	2014	2,048 MB	GDDR5	1,983.2	60 W	connector absence confirmed by source	—	2,540
nvidia	GeForce GTX TITAN Black	2014	6,144 MB	GDDR5	5,120.6	250 W	not recorded by source	—	7,080
amd	FirePro W9100	2014	—	GDDR5	5,237.8	275 W	not recorded by source	28 nm	—
nvidia	GeForce GT 705 OEM	2014	1,024 MB	DDR3	167.8	29 W	connector absence confirmed by source	40 nm	292
nvidia	GeForce GT 710	2014	1,024 MB	DDR3	366.3	19 W	connector absence confirmed by source	—	1,020
nvidia	GeForce GT 710 OEM	2014	1,024 MB	DDR3	306	—	connector absence confirmed by source	—	1,020
amd	Radeon R5 230 (Caicos Pro)	2014	1,024 MB	DDR3	200	19 W	not recorded by source	—	370
nvidia	GeForce GTX TITAN Z	2014	—	GDDR5	8,121.6	375 W	not recorded by source	—	—
nvidia	GeForce GT 740	2014	1,024 MB	GDDR5	762.6	64 W	not recorded by source	—	1,270
nvidia	GeForce GT 730	2014	1,024 MB	GDDR5	692.7	38 W	connector absence confirmed by source	—	1,020
amd	FirePro W8100	2014	8,192 MB	GDDR5	4,218.9	220 W	not recorded by source	28 nm	—
nvidia	Quadro K2200M	2014	2,048 MB	GDDR5	1,472	65 W	not recorded by source	28 nm	—
nvidia	Quadro K2200	2014	4,096 MB	GDDR5	1,338.9	68 W	not recorded by source	28 nm	—
nvidia	Quadro K420	2014	—	DDR3	299.52	41 W	not recorded by source	28 nm	—
nvidia	Quadro K4200	2014	4,096 MB	GDDR5	2,096.64	105 W	not recorded by source	28 nm	—
nvidia	Quadro K5200	2014	8,192 MB	GDDR5	2,995.2	150 W	not recorded by source	28 nm	—
nvidia	Quadro K620	2014	2,048 MB	DDR3	768	45 W	not recorded by source	28 nm	—
amd	FirePro S9050	2014	12,288 MB	GDDR5	3,225.6	—	not recorded by source	28 nm	—
amd	FirePro S9150	2014	16,384 MB	GDDR5	5,070	—	not recorded by source	28 nm	—
amd	FirePro S4000x	2014	2,048 MB	GDDR5	992	—	not recorded by source	28 nm	—
amd	FirePro W2100	2014	2,048 MB	DDR3	403.2	—	not recorded by source	28 nm	—
amd	FirePro W4100	2014	2,048 MB	GDDR5	645.1	—	not recorded by source	28 nm	—
amd	FirePro W5100	2014	4,096 MB	GDDR5	1,430	—	not recorded by source	28 nm	—
amd	FirePro W7100	2014	8,192 MB	GDDR5	3,297	—	not recorded by source	28 nm	—
nvidia	GeForce GTX 660 Rev. 2	2014	2,048 MB	GDDR5	1,505.28	140 W	not recorded by source	28 nm	3,540
nvidia	GeForce GTX 980	2014	4,096 MB	GDDR5	4,616.19	165 W	not recorded by source	—	5,200
nvidia	GeForce GTX 970	2014	4,096 MB	GDDR5	3,494.4	148 W	not recorded by source	—	5,200
nvidia	GeForce GT 720	2014	1,024 MB	DDR3	306	19 W	connector absence confirmed by source	—	1,020
amd	FirePro S9100	2014	12,288 MB	GDDR5	4,219	—	not recorded by source	28 nm	—
nvidia	GeForce GTX 960	2015	2,048 MB	GDDR5	2,308.1	120 W	not recorded by source	—	2,940
nvidia	Quadro K1200	2015	4,096 MB	GDDR5	1,083	45 W	not recorded by source	28 nm	—
nvidia	GeForce GTX TITAN X	2015	12,288 MB	GDDR5	6,144	250 W	not recorded by source	—	8,000
nvidia	Quadro M6000	2015	24,576 MB	GDDR5	6,070	250 W	not recorded by source	28 nm	—
nvidia	Quadro M6000	2015	12,288 MB	GDDR5	6,070	250 W	not recorded by source	28 nm	—
nvidia	GeForce GT 730 OEM	2015	1,024 MB	GDDR5	692.7	64 W	not recorded by source	—	1,270
nvidia	GeForce GT 740 OEM	2015	1,024 MB	GDDR5	772.6	65 W	connector absence confirmed by source	—	2,540
amd	Radeon R5 330 (Oland Pro)	2015	1,024 MB	DDR3	547.2	30 W	not recorded by source	—	1,040
amd	Radeon R5 340 (Oland XT)	2015	1,024 MB	DDR3 GDDR5	633.6	75 W	not recorded by source	—	1,040
amd	Radeon R7 350 (Oland XT)	2015	1,024 MB	DDR3 GDDR5	768	75 W	not recorded by source	—	1,040
amd	Radeon R9 370 (Pitcairn Pro)	2015	2,048 MB	GDDR5	1,945.6	150 W	not recorded by source	—	2,800
amd	Radeon R7 370 (Pitcairn Pro)	2015	2,048 MB	GDDR5	1,996.8	110 W	not recorded by source	—	2,800
nvidia	GeForce GTX 980 Ti	2015	6,144 MB	GDDR5	5,632	250 W	not recorded by source	—	8,000
nvidia	Quadro M4000	2015	8,192 MB	GDDR5	2,662.4	120 W	not recorded by source	28 nm	—
nvidia	Quadro M5000	2015	8,192 MB	GDDR5	4,300.8	150 W	not recorded by source	28 nm	—
amd	FirePro S9170	2015	32,768 MB	GDDR5	5,240	—	not recorded by source	28 nm	—
amd	Radeon R9 370X (Pitcairn XT)	2015	2,048 MB	GDDR5	2,560	185 W	not recorded by source	—	2,800
nvidia	Quadro M1000M	2015	2,048 MB	GDDR5	1,017	40 W	not recorded by source	28 nm	—
nvidia	Quadro M3000M	2015	4,096 MB	GDDR5	2,150	75 W	not recorded by source	28 nm	—
nvidia	Quadro M4000M	2015	4,096 MB	GDDR5	2,496	100 W	not recorded by source	28 nm	—
nvidia	Quadro M5000M	2015	8,192 MB	GDDR5	2,955.3	100 W	not recorded by source	28 nm	—
nvidia	Quadro M600M	2015	2,048 MB	GDDR5	672.8	30 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 950	2015	2,048 MB	GDDR5	1,572.86	90 W	not recorded by source	—	2,940
amd	Radeon E6465 (Caicos)	2015	2,048 MB	GDDR5	192	20 W	not recorded by source	—	—
amd	Radeon E8870 (Saturn)	2015	4,096 MB	GDDR5	1,536	75 W	not recorded by source	—	—
amd	Radeon E8950 (Amethyst)	2015	8,192 MB	GDDR5	3,011	95 W	not recorded by source	—	—
nvidia	NVS 810	2015	—	DDR3	1,058	68 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 750	2015	2,048 MB	GDDR5	1,113.1	60 W	connector absence confirmed by source	—	2,940
nvidia	GeForce GTX 960 OEM	2015	3,072 MB	GDDR5	2,365.44	—	connector absence confirmed by source	—	5,200
nvidia	GeForce GTX 960 OEM	2015	4,096 MB	GDDR5	2,408.45	—	connector absence confirmed by source	—	2,940
amd	FirePro W4300	2015	4,096 MB	GDDR5	1,428.5	—	not recorded by source	28 nm	—
nvidia	Quadro M2000M	2015	4,096 MB	GDDR5	1,405	55 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 950 OEM	2016	2,048 MB	GDDR5	1,918.98	—	connector absence confirmed by source	—	2,940
amd	FirePro S7150	2016	8,192 MB	GDDR5	3,763	150 W	not recorded by source	28 nm	—
amd	FirePro S7150 X2	2016	16,384 MB	GDDR5	7,540	265 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 950 Low Power	2016	2,048 MB	GDDR5	1,575.94	75 W	connector absence confirmed by source	—	2,940
amd	FirePro S9300 x2	2016	—	HBM	13,900	300 W	not recorded by source	28 nm	—
nvidia	GeForce GT 610 OEM	2016	2,048 MB	DDR3	134.592	29 W	connector absence confirmed by source	40 nm	585
nvidia	Quadro M2000	2016	4,096 MB	GDDR5	1,812.5	75 W	not recorded by source	28 nm	—
nvidia	Quadro M5500 Mobile	2016	8,192 MB	GDDR5	4,669	150 W	not recorded by source	28 nm	—
nvidia	Quadro M500M	2016	2,048 MB	DDR3	863.2	25 W	not recorded by source	28 nm	—
nvidia	M10 GPU accelerator	2016	—	GDDR5	5,289	225 W	not recorded by source	—	—
amd	FirePro S7100X	2016	8,192 MB	GDDR5	2,961	100 W	not recorded by source	28 nm	—
nvidia	GeForce GTX 1080	2016	8,192 MB	GDDR5X	8,227.84	180 W	not recorded by source	—	7,200
nvidia	GeForce GTX 1070	2016	8,192 MB	GDDR5	5,783.04	150 W	not recorded by source	—	7,200
amd	Radeon R5 430 (Oland Pro)	2016	1,024 MB	DDR3 GDDR5	560	50 W	not recorded by source	—	1,040
amd	Radeon R7 450 (Cape Verde Pro)	2016	2,048 MB	GDDR5	1,075	65 W	not recorded by source	—	1,500
nvidia	GeForce GTX 1060 Rev. 2	2016	6,144 MB	GDDR5	3,855.36	120 W	not recorded by source	—	4,400
nvidia	TITAN X Pascal	2016	12,288 MB	GDDR5X	10,157.06	250 W	not recorded by source	—	11,800
amd	Radeon RX 470 (Ellesmere Pro)	2016	4,096 MB	GDDR5	3,793	120 W	not recorded by source	—	5,700
nvidia	GeForce GTX 1070 (Notebook)	2016	8,192 MB	GDDR5	5,906	115 W	not recorded by source	16 nm	7,200
nvidia	GeForce GTX 1080 (Notebook)	2016	8,192 MB	GDDR5X	7,967	150 W	not recorded by source	16 nm	7,200
nvidia	GeForce GTX 1060	2016	3,072 MB	GDDR5	3,469.82	120 W	not recorded by source	—	4,400
amd	Radeon E9260 (Polaris11)	2016	4,096 MB	GDDR5	2,500	—	not recorded by source	—	—
amd	Radeon E9550 (Polaris10)	2016	8,192 MB	GDDR5	5,728	95 W	not recorded by source	—	—
nvidia	Quadro GP100	2016	16,384 MB	HBM2	10,336	235 W	not recorded by source	16 nm	—
nvidia	Quadro P5000	2016	16,384 MB	GDDR5X	8,873	180 W	not recorded by source	16 nm	—
nvidia	Quadro P6000	2016	24,576 MB	GDDR5X	10,882	250 W	not recorded by source	16 nm	—
nvidia	GeForce GTX 1050	2016	2,048 MB	GDDR5	1,733.12	75 W	connector absence confirmed by source	—	3,300
nvidia	GeForce GTX 1050 Ti	2016	4,096 MB	GDDR5	1,982.98	75 W	connector absence confirmed by source	—	3,300
amd	Radeon Pro 450 (Polaris 11)	2016	2,048 MB	GDDR5	1,024	35 W	not recorded by source	—	—
amd	Radeon Pro 455 (Polaris 11)	2016	2,048 MB	GDDR5	1,313	35 W	not recorded by source	—	—
nvidia	GeForce GTX 1060	2016	3,072 MB	GDDR5	3,469.82	120 W	not recorded by source	—	7,200
nvidia	GeForce GTX 1050 (Notebook)	2017	4,096 MB	GDDR5	1,733	53 W	not recorded by source	14 nm	3,300
nvidia	GeForce GTX 1050 Ti (Notebook)	2017	4,096 MB	GDDR5	2,293	64 W	not recorded by source	14 nm	3,300
nvidia	Quadro M1200 Mobile	2017	4,096 MB	GDDR5	1,400	45 W	not recorded by source	28 nm	—
nvidia	Quadro M2200 Mobile	2017	4,096 MB	GDDR5	2,100	55 W	not recorded by source	28 nm	—
nvidia	Quadro M520 Mobile	2017	1,024 MB	GDDR5	840	25 W	not recorded by source	28 nm	—
nvidia	Quadro M620 Mobile	2017	2,048 MB	GDDR5	1,000	30 W	not recorded by source	28 nm	—
nvidia	Quadro P3000 Mobile	2017	6,144 MB	GDDR5	3,098	75 W	not recorded by source	16 nm	—
nvidia	Quadro P4000 Mobile	2017	8,192 MB	GDDR5	4,398	100 W	not recorded by source	16 nm	—
nvidia	Quadro P5000 Mobile	2017	16,384 MB	GDDR5	6,197	100 W	not recorded by source	16 nm	—
nvidia	Quadro P2000	2017	5,120 MB	GDDR5	3,010	75 W	not recorded by source	16 nm	—
nvidia	Quadro P2000 Mobile	2017	4,096 MB	GDDR5	2,400	50 W	not recorded by source	14 nm	—
nvidia	Quadro P4000	2017	8,192 MB	GDDR5	5,300	105 W	not recorded by source	16 nm	—
nvidia	Quadro P1000	2017	4,096 MB	GDDR5	1,894	47 W	not recorded by source	14 nm	—
nvidia	Quadro P1000 Mobile	2017	4,096 MB	GDDR5	1,600	40 W	not recorded by source	14 nm	—
nvidia	Quadro P400	2017	2,048 MB	GDDR5	641	30 W	not recorded by source	14 nm	—
nvidia	Quadro P600	2017	2,048 MB	GDDR5	1,195	40 W	not recorded by source	14 nm	—
nvidia	Quadro P600 Mobile	2017	2,048 MB	GDDR5	1,200	25 W	not recorded by source	14 nm	—
nvidia	GeForce GTX 1080 Ti	2017	11,264 MB	GDDR5X	10,615.81	250 W	not recorded by source	—	11,800
nvidia	P6 GPU accelerator	2017	16,384 MB	GDDR5	6,169	90 W	not recorded by source	—	—
nvidia	TITAN Xp	2017	12,288 MB	GDDR5X	10,790.4	250 W	not recorded by source	—	11,800
amd	Radeon 520 (Banks)	2017	1,024 MB	DDR3 GDDR5	659.2	—	not recorded by source	—	690
amd	Radeon RX 540 (Polaris 12)	2017	2,048 MB	GDDR5	1,150.976	—	not recorded by source	—	—
amd	Radeon RX 570 (Polaris 20)	2017	4,096 MB	GDDR5	4,784	150 W	not recorded by source	—	—
amd	Radeon RX 550 (Polaris 12)	2017	2,048 MB	GDDR5	1,126.4	50 W	not recorded by source	—	—
amd	Radeon RX 550 640SP (Polaris 11)	2017	2,048 MB	GDDR5	1,304.32	60 W	not recorded by source	—	—
nvidia	GeForce GTX 1060 Rev. 2	2017	6,144 MB	GDDR5	3,855.36	120 W	not recorded by source	—	4,400
nvidia	GeForce GTX 1080	2017	8,192 MB	GDDR5X	8,227.84	180 W	not recorded by source	—	7,200
nvidia	GeForce GTX 1060 Max-Q	2017	6,144 MB	GDDR5	3,789	80 W	not recorded by source	16 nm	4,400
nvidia	GeForce GTX 1060 Max-Q	2017	6,144 MB	GDDR5	3,789	80 W	not recorded by source	16 nm	4,400
nvidia	GeForce GTX 1070 Max-Q	2017	8,192 MB	GDDR5	5,648	—	not recorded by source	16 nm	7,200
nvidia	GeForce GTX 1080 Max-Q	2017	8,192 MB	GDDR5X	7,516	—	not recorded by source	16 nm	7,200
nvidia	V100 GPU accelerator (mezzanine)	2017	—	HBM2	14,900	300 W	not recorded by source	—	—
nvidia	GeForce GT 1030	2017	2,048 MB	GDDR5	943.1	30 W	connector absence confirmed by source	—	1,800
nvidia	GeForceMX150 (Pascal)	2017	2,048 MB	GDDR5	1,127	25 W	not recorded by source	—	1,800
nvidia	GeForceMX150 (Pascal)	2017	2,048 MB	GDDR5	719.6	10 W	not recorded by source	—	1,800
amd	Radeon Pro 555 (Polaris 21)	2017	2,048 MB	GDDR5	1,306	50 W	not recorded by source	—	—
amd	Radeon Pro 560 (Polaris 21)	2017	4,096 MB	GDDR5	1,858	50 W	not recorded by source	—	—
amd	Radeon Pro 575 (Polaris 20)	2017	4,096 MB	GDDR5	4,489	120 W	not recorded by source	—	—
amd	Radeon Instinct MI8 (Fiji)	2017	4,096 MB	HBM	8,192	175 W	not recorded by source	—	—
amd	Radeon RX 560 (Polaris 21)	2017	2,048 MB	GDDR5	1,953	80 W	not recorded by source	—	—
amd	Radeon RX 560D (Polaris 21)	2017	2,048 MB	GDDR5	1,953	65 W	not recorded by source	—	—
nvidia	GeForce GT 720 OEM	2017	1,024 MB	DDR3	381.3	50 W	connector absence confirmed by source	—	1,270
nvidia	GeForce GTX 1070 Ti	2017	8,192 MB	GDDR5	7,816.45	180 W	not recorded by source	—	7,200
nvidia	GeForceMX130 (Maxwell)	2017	2,048 MB	DDR3 GDDR5	861.7	30 W	not recorded by source	28 nm	—
nvidia	Nvidia Titan V	2017	12,288 MB	HBM2	12,290	250 W	not recorded by source	—	—
nvidia	GeForce GTX 1060 OEM	2017	5,120 MB	GDDR5	3,855.36	120 W	not recorded by source	—	4,400
nvidia	Quadro P500 Mobile	2018	2,048 MB	GDDR5	750	18 W	not recorded by source	14 nm	—
nvidia	Quadro P620	2018	2,048 MB	GDDR5	1,490	40 W	not recorded by source	14 nm	—
nvidia	Quadro P3200 Mobile	2018	6,144 MB	GDDR5	5,530	75 W	not recorded by source	16 nm	—
nvidia	Quadro P4200 Max-Q	2018	8,192 MB	GDDR5	6,820	100 W	not recorded by source	16 nm	—
nvidia	Quadro P4200 Mobile	2018	8,192 MB	GDDR5	7,345	100 W	not recorded by source	16 nm	—
nvidia	Quadro P5200 Max-Q	2018	16,384 MB	GDDR5	7,578	100 W	not recorded by source	16 nm	—
nvidia	Quadro P5200 Mobile	2018	16,384 MB	GDDR5	9,006	100 W	not recorded by source	16 nm	—
nvidia	GeForce GTX 1060 6 GB	2018	6,144 MB	GDDR5	3,855.36	120 W	not recorded by source	—	7,200
nvidia	GeForce GT 1030	2018	2,048 MB	DDR4	884.74	20 W	connector absence confirmed by source	—	1,800
nvidia	Quadro GV100	2018	32,768 MB	HBM2	14,800	250 W	not recorded by source	12 nm	—
nvidia	V100 GPU accelerator (PCIe FHHL card)	2018	16,384 MB	HBM2	13,210	250 W	not recorded by source	—	—
amd	Radeon RX 550X (Polaris 12)	2018	2,048 MB	GDDR5	1,126.4	50 W	not recorded by source	—	—
amd	Radeon RX 550X 640SP (Polaris 11)	2018	2,048 MB	GDDR5	1,304.32	60 W	not recorded by source	—	—
nvidia	GeForce GTX 1050	2018	3,072 MB	GDDR5	2,138.11	75 W	connector absence confirmed by source	—	3,300
nvidia	Nvidia Titan V (CEO Edition)	2018	32,768 MB	HBM2	12,290	250 W	not recorded by source	—	—
amd	Radeon Pro 555X (Polaris 21)	2018	4,096 MB	GDDR5	1,393	50 W	not recorded by source	—	—
amd	Radeon Pro 560X (Polaris 21)	2018	4,096 MB	GDDR5	2,056	75 W	not recorded by source	—	—
nvidia	Quadro RTX 5000	2018	16,384 MB	GDDR6	11,150	—	not recorded by source	12 nm	—
nvidia	Quadro RTX 6000	2018	24,576 MB	GDDR6	16,310	—	not recorded by source	12 nm	—
nvidia	Quadro RTX 8000	2018	49,152 MB	GDDR6	16,310	—	not recorded by source	12 nm	—
nvidia	T4 GPU accelerator (PCIe card)	2018	16,384 MB	GDDR6	8,100	70 W	not recorded by source	—	—
nvidia	GeForce RTX 2080	2018	8,192 MB	GDDR6	8,920	215 W	not recorded by source	—	13,600
nvidia	GeForce RTX 2080 Ti	2018	11,264 MB	GDDR6	11,750	250 W	not recorded by source	—	18,600
nvidia	GeForce GT 1030	2018	2,048 MB	GDDR5	812.54	65 W	not recorded by source	—	1,270
amd	Radeon RX 580 2048SP (Polaris 20)	2018	4,096 MB	GDDR5	4,784	150 W	not recorded by source	—	—
nvidia	GeForce RTX 2070	2018	8,192 MB	GDDR6	6,497	175 W	not recorded by source	—	10,800
nvidia	GeForce GTX 1060 6 GB GDDR5X	2018	6,144 MB	GDDR5X	3,855.36	120 W	not recorded by source	—	7,200
nvidia	Quadro RTX 4000	2018	8,192 MB	GDDR6	7,119	—	not recorded by source	12 nm	—
nvidia	GeForce GTX 1070 GDDR5X	2018	8,192 MB	GDDR5X	5,783.04	150 W	not recorded by source	—	7,200
nvidia	Titan RTX	2018	24,576 MB	GDDR6	12,442	280 W	not recorded by source	—	18,600
nvidia	GeForce RTX 2060	2019	6,144 MB	GDDR6	5,242	160 W	not recorded by source	—	10,800
nvidia	GeForce RTX 2080	2019	8,192 MB	GDDR6	9,362	150 W	not recorded by source	—	13,600
nvidia	GeForce RTX 2080 Max-Q	2019	8,192 MB	GDDR6	6,447	80 W	not recorded by source	—	13,600
nvidia	GeForce GTX 1660 Ti	2019	6,144 MB	GDDR6	4,608	120 W	not recorded by source	—	6,600
amd	Radeon RX 560 XT (Polaris 10/20)	2019	4,096 MB	GDDR5	3,487	150 W	not recorded by source	—	—
nvidia	GeForce GTX 1660	2019	6,144 MB	GDDR5	4,308	120 W	not recorded by source	—	6,600
amd	Radeon Pro 575X (Polaris 20)	2019	4,096 MB	GDDR5	4,489	150 W	not recorded by source	—	—
amd	Radeon Pro Vega 48 (Vega 10)	2019	8,192 MB	HBM2	7,373	—	not recorded by source	—	—
nvidia	GeForce GTX 1650	2019	4,096 MB	GDDR5	2,661	75 W	connector absence confirmed by source	—	4,700
nvidia	GeForce GTX 1650 (Laptop)	2019	4,096 MB	GDDR5	3,195	50 W	not recorded by source	—	4,700
nvidia	GeForce GTX 1650 Max-Q	2019	4,096 MB	GDDR5	2,550	30 W	not recorded by source	—	4,700
nvidia	GeForce GTX 1660 Ti (Laptop)	2019	6,144 MB	GDDR6	4,884	80 W	not recorded by source	—	6,600
nvidia	GeForce GTX 1660 Ti Max-Q	2019	6,144 MB	GDDR6	4,101	60 W	not recorded by source	—	6,600
nvidia	Quadro RTX 3000 Max-Q	2019	6,144 MB	GDDR6	5,600	60 W	not recorded by source	12 nm	—
nvidia	Quadro RTX 3000 Mobile	2019	6,144 MB	GDDR6	6,359	—	not recorded by source	12 nm	—
nvidia	Quadro RTX 4000 Max-Q	2019	8,192 MB	GDDR6	7,066	80 W	not recorded by source	12 nm	—
nvidia	Quadro RTX 4000 Mobile	2019	8,192 MB	GDDR6	7,987	110 W	not recorded by source	12 nm	—
nvidia	Quadro RTX 5000 Max-Q	2019	16,384 MB	GDDR6	8,294	80 W	not recorded by source	12 nm	—
nvidia	Quadro RTX 5000 Mobile	2019	16,384 MB	GDDR6	9,492	110 W	not recorded by source	12 nm	—
nvidia	Quadro T1000 Mobile	2019	4,096 MB	GDDR5	2,607	—	not recorded by source	12 nm	—
nvidia	Quadro T2000 Max-Q	2019	4,096 MB	GDDR5	2,857	40 W	not recorded by source	12 nm	—
nvidia	Quadro T2000 Mobile	2019	4,096 MB	GDDR5	3,656	60 W	not recorded by source	12 nm	—
nvidia	Quadro P2200	2019	5,120 MB	GDDR5X	3,822	75 W	not recorded by source	16 nm	—
nvidia	GeForce RTX 2060 SUPER	2019	8,192 MB	GDDR6	6,397	175 W	not recorded by source	—	10,800
nvidia	GeForce RTX 2070 SUPER	2019	8,192 MB	GDDR6	8,218	215 W	not recorded by source	—	13,600
nvidia	GeForce RTX 2080 SUPER	2019	8,192 MB	GDDR6	10,138	250 W	not recorded by source	—	13,600
amd	Radeon 610 (Banks)	2019	2,048 MB	GDDR5	659.2	50 W	not recorded by source	—	—
amd	Radeon 620 (Polaris 24)	2019	2,048 MB	DDR3	560.6	50 W	not recorded by source	—	—
amd	Radeon 630 (Polaris 23)	2019	2,048 MB	GDDR5	1,108	50 W	not recorded by source	—	—
amd	Radeon RX 640 (Polaris 23)	2019	2,048 MB	GDDR5	1,385	50 W	not recorded by source	—	—
nvidia	Quadro RTX 6000 Mobile	2019	24,576 MB	GDDR6	13,410	200 W	not recorded by source	12 nm	—
amd	Radeon E9390 (Polaris10)	2019	8,192 MB	GDDR5	3,892	75 W	not recorded by source	—	—
amd	Radeon E9560 (Polaris10)	2019	8,192 MB	GDDR5	5,728	130 W	not recorded by source	—	—
nvidia	GeForce GTX 1660 Super	2019	6,144 MB	GDDR6	4,308	125 W	not recorded by source	—	6,600
amd	Radeon Pro 5500M (Navi 14)	2019	4,096 MB	GDDR6	3,072	50 W	not recorded by source	—	—
amd	Radeon Pro 5500M (Navi 14)	2019	8,192 MB	GDDR6	4,454	50 W	not recorded by source	—	—
nvidia	GeForce GTX 1650 Super	2019	4,096 MB	GDDR6	3,917	100 W	not recorded by source	—	6,600
amd	Radeon Pro W5500X (Navi 14)	2020	8,192 MB	GDDR6	5,600	130 W	not recorded by source	—	—
nvidia	GeForce RTX 2060	2020	6,144 MB	GDDR6	5,242	160 W	not recorded by source	—	13,600
nvidia	GeForce GTX 1650	2020	4,096 MB	GDDR6	2,527	75 W	connector absence confirmed by source	—	4,700
nvidia	GeForce GTX 1650 Ti	2020	4,096 MB	GDDR6	3,041	55 W	not recorded by source	—	4,700
nvidia	GeForce GTX 1650 Ti Max-Q	2020	4,096 MB	GDDR6	2,458	35 W	not recorded by source	—	4,700
nvidia	GeForce RTX 2070 Super	2020	8,192 MB	GDDR6	7,066	115 W	not recorded by source	—	13,600
nvidia	GeForce RTX 2070 Super Max-Q	2020	8,192 MB	GDDR6	5,530	80 W	not recorded by source	—	13,600
nvidia	GeForce RTX 2080 Super	2020	8,192 MB	GDDR6	9,585	150 W	not recorded by source	—	13,600
nvidia	GeForce RTX 2080 Super Max-Q	2020	8,192 MB	GDDR6	6,636	80 W	not recorded by source	—	13,600
nvidia	A100 GPU accelerator (PCIe card)	2020	—	HBM2	19,500	250 W	not recorded by source	—	—
nvidia	GeForce GTX 1650	2020	4,096 MB	GDDR6	2,527	90 W	not recorded by source	—	10,800
nvidia	GeForce GTX 1650	2020	4,096 MB	GDDR6	2,527	80 W	connector absence confirmed by source	—	6,600
nvidia	GeForceMX450 (Turing)	2020	2,048 MB	GDDR5	1,290	—	not recorded by source	—	4,700
nvidia	GeForceMX450 (Turing)	2020	2,048 MB	GDDR6	1,290	—	not recorded by source	—	4,700
amd	Radeon E9565 (Polaris10)	2020	16,384 MB	GDDR5	5,728	135 W	not recorded by source	—	—
nvidia	GeForce RTX 3080	2020	10,240 MB	GDDR6X	25,068	320 W	not recorded by source	—	28,300
nvidia	GeForce RTX 3090	2020	24,576 MB	GDDR6X	29,284	350 W	not recorded by source	—	28,300
nvidia	A40 GPU accelerator (PCIe card)	2020	49,152 MB	GDDR6	37,420	300 W	not recorded by source	—	—
nvidia	RTX A6000	2020	49,152 MB	GDDR6	38,710	300 W	not recorded by source	8 nm	—
nvidia	GeForce RTX 3070	2020	8,192 MB	GDDR6	17,664	220 W	not recorded by source	—	17,400
nvidia	GeForce RTX 3060 Ti	2020	8,192 MB	GDDR6	13,716	200 W	not recorded by source	—	17,400
nvidia	Quadro T500 Mobile	2020	2,048 MB	GDDR5	2,796	25 W	not recorded by source	12 nm	—
nvidia	Quadro T500 Mobile	2020	4,096 MB	GDDR5	3,037	18 W	not recorded by source	12 nm	—
nvidia	GeForce GT 1010 OEM	2021	2,048 MB	DDR4	590.82	20 W	connector absence confirmed by source	—	1,800
nvidia	GeForce GT 1010 OEM	2021	2,048 MB	GDDR5	628.74	30 W	connector absence confirmed by source	—	1,800
nvidia	GeForce RTX 3060 12 GB	2021	12,288 MB	GDDR6	9,462	170 W	not recorded by source	—	12,000
nvidia	GeForce RTX 3060 OEM	2021	6,144 MB	GDDR6	12,395	185 W	not recorded by source	—	17,400
nvidia	A10 GPU accelerator (PCIe card)	2021	24,576 MB	GDDR6	31,240	150 W	not recorded by source	—	—
nvidia	A30 GPU accelerator (PCIe card)	2021	24,576 MB	HBM2	10,320	165 W	not recorded by source	—	—
nvidia	NVIDIA T600	2021	4,096 MB	GDDR6	1,700	40 W	not recorded by source	—	—
nvidia	Quadro T1200 Mobile	2021	4,096 MB	GDDR5	3,656	18 W	not recorded by source	12 nm	—
nvidia	Quadro T600 Mobile	2021	4,096 MB	GDDR5	2,527	40 W	not recorded by source	12 nm	—
nvidia	RTX A2000 Mobile	2021	4,096 MB	GDDR6	8,637	95 W	not recorded by source	—	—
nvidia	RTX A3000 Mobile	2021	6,144 MB	GDDR6	10,080	70 W	not recorded by source	—	—
nvidia	RTX A4000	2021	16,384 MB	GDDR6	19,170	140 W	not recorded by source	8 nm	—
nvidia	RTX A4000 Mobile	2021	8,192 MB	GDDR6	17,200	140 W	not recorded by source	—	—
nvidia	RTX A5000	2021	24,576 MB	GDDR6	27,770	230 W	not recorded by source	8 nm	—
nvidia	RTX A5000 Mobile	2021	16,384 MB	GDDR6	19,350	150 W	not recorded by source	—	—
nvidia	NVIDIA T1000	2021	4,096 MB	GDDR6	2,500	50 W	not recorded by source	—	—
nvidia	NVIDIA T1000	2021	8,192 MB	GDDR6	2,500	50 W	not recorded by source	—	—
nvidia	NVIDIA T400	2021	2,048 MB	GDDR6	1,090	30 W	not recorded by source	—	—
nvidia	NVIDIA T400	2021	4,096 MB	GDDR6	1,090	30 W	not recorded by source	—	—
amd	Radeon RX 6600M (Navi 23)	2021	8,192 MB	GDDR6	7,800	100 W	not recorded by source	—	—
amd	Radeon RX 6700M (Navi 22)	2021	10,240 MB	GDDR6	10,605	135 W	not recorded by source	—	—
amd	Radeon RX 6800M (Navi 22)	2021	12,288 MB	GDDR6	11,780	—	not recorded by source	—	—
nvidia	GeForce RTX 3080 Ti	2021	12,288 MB	GDDR6X	27,955	350 W	not recorded by source	—	28,300
nvidia	GeForce RTX 3070 Ti	2021	8,192 MB	GDDR6X	19,354	290 W	not recorded by source	—	17,400
nvidia	RTX A2000	2021	12,288 MB	GDDR6	7,987	70 W	not recorded by source	8 nm	—
nvidia	RTX A2000	2021	6,144 MB	GDDR6	7,987	70 W	not recorded by source	8 nm	—
nvidia	GeForce RTX 3060 12 GB	2021	12,288 MB	GDDR6	9,462	170 W	not recorded by source	—	17,400
nvidia	A2 GPU accelerator (PCIe card)	2021	16,384 MB	GDDR6	4,531	—	not recorded by source	—	—
nvidia	RTX A4500	2021	20,480 MB	GDDR6	23,660	200 W	not recorded by source	8 nm	—
nvidia	GeForce RTX 2060	2021	12,288 MB	GDDR6	6,397	184 W	not recorded by source	—	10,800
nvidia	GeForce RTX 3080 Ti 20 GB	2022	20,480 MB	GDDR6X	27,341	350 W	not recorded by source	—	28,300
amd	Radeon RX 6300M (Navi 24)	2022	2,048 MB	GDDR6	3,130	25 W	not recorded by source	—	—
amd	Radeon RX 6500M (Navi 24)	2022	4,096 MB	GDDR6	4,980	50 W	not recorded by source	—	—
amd	Radeon RX 6600S (Navi 23)	2022	4,096 MB	GDDR6	7,810	80 W	not recorded by source	—	—
amd	Radeon RX 6650M (Navi 23)	2022	8,192 MB	GDDR6	8,850	120 W	not recorded by source	—	—
amd	Radeon RX 6650M XT (Navi 23)	2022	8,192 MB	GDDR6	9,970	120 W	not recorded by source	—	—
amd	Radeon RX 6700S (Navi 23)	2022	8,192 MB	GDDR6	7,920	80 W	not recorded by source	—	—
amd	Radeon RX 6800S (Navi 23)	2022	8,192 MB	GDDR6	9,220	100 W	not recorded by source	—	—
amd	Radeon RX 6850M XT (Navi 22)	2022	12,288 MB	GDDR6	13,209	165 W	not recorded by source	—	—
nvidia	GeForce RTX 3050	2022	8,192 MB	GDDR6	7,946	130 W	not recorded by source	—	12,000
nvidia	GeForce RTX 3080 12 GB	2022	12,288 MB	GDDR6X	22,579	350 W	not recorded by source	—	28,300
amd	Radeon Pro W6300M (Navi 24)	2022	2,048 MB	GDDR6	3,401	25 W	not recorded by source	—	—
amd	Radeon Pro W6500M (Navi 24)	2022	4,096 MB	GDDR6	5,239	50 W	not recorded by source	—	—
nvidia	GeForce RTX 3050 OEM	2022	8,192 MB	GDDR6	6,981	130 W	not recorded by source	—	12,000
nvidia	GeForce RTX 3060 Ti	2022	8,192 MB	GDDR6	13,716	200 W	not recorded by source	—	22,000
amd	Radeon Pro W6600X (Navi 23)	2022	8,192 MB	GDDR6	9,837	120 W	not recorded by source	—	—
nvidia	H100 GPU accelerator (PCIe card)	2022	81,920 MB	HBM2E	51,200	350 W	not recorded by source	—	—
nvidia	H100 GPU accelerator (SXM card)	2022	—	HBM3	66,900	700 W	not recorded by source	—	—
nvidia	RTX A3000 Mobile 12GB	2022	12,288 MB	GDDR6	11,800	115 W	not recorded by source	—	—
nvidia	RTX A4500 Mobile	2022	16,384 MB	GDDR6	17,660	140 W	not recorded by source	—	—
nvidia	RTX A500 Mobile	2022	4,096 MB	GDDR6	6,296	60 W	not recorded by source	—	—
nvidia	RTX A5500	2022	24,576 MB	GDDR6	34,100	230 W	not recorded by source	8 nm	—
nvidia	RTX A5500 Mobile	2022	16,384 MB	GDDR6	22,270	165 W	not recorded by source	—	—
nvidia	GeForce RTX 3090 Ti	2022	24,576 MB	GDDR6X	33,546	450 W	not recorded by source	—	28,300
nvidia	RTX A1000 Mobile	2022	4,096 MB	GDDR6	4,669	60 W	not recorded by source	—	—
nvidia	RTX A1000 Mobile 6GB	2022	6,144 MB	GDDR6	5,837	95 W	not recorded by source	—	—
nvidia	Quadro T550 Mobile	2022	4,096 MB	GDDR5	3,410	23 W	not recorded by source	12 nm	—
nvidia	GeForce GTX 1630	2022	4,096 MB	GDDR6	1,782	75 W	connector absence confirmed by source	—	4,700
nvidia	Geforce GTX 1630	2022	4,096 MB	GDDR6	1,828	75 W	not recorded by source	—	4,700
intel	Arc Pro A30M	2022	4,096 MB	GDDR6	4,200	50 W	not recorded by source	—	7,200
intel	Arc Pro A40	2022	6,144 MB	GDDR6	5,020	50 W	not recorded by source	—	7,200
intel	Arc Pro A50	2022	6,144 MB	GDDR6	5,020	75 W	not recorded by source	—	7,200
intel	Arc 3 A310	2022	4,096 MB	GDDR6	3,072	75 W	not recorded by source	—	7,200
nvidia	GeForce RTX 3060	2022	8,192 MB	GDDR6	9,462	195 W	not recorded by source	—	17,400
nvidia	GeForce RTX 3060	2022	8,192 MB	GDDR6	9,462	170 W	not recorded by source	—	12,000
nvidia	GeForce RTX 4090	2022	24,576 MB	GDDR6X	73,236	450 W	not recorded by source	—	76,300
nvidia	L40 GPU accelerator	2022	49,152 MB	GDDR6	90,520	300 W	not recorded by source	—	—
nvidia	GeForce RTX 3060 Ti GDDR6X	2022	8,192 MB	GDDR6X	13,716	225 W	not recorded by source	—	17,400
nvidia	GeForce RTX 3070 Ti	2022	8,192 MB	GDDR6X	19,354	290 W	not recorded by source	—	28,300
nvidia	GeForce RTX 4080	2022	16,384 MB	GDDR6X	42,900	320 W	not recorded by source	—	45,900
nvidia	RTX 6000 Ada Generation	2022	49,152 MB	GDDR6	91,060	300 W	not recorded by source	—	—
nvidia	GeForce RTX 3050	2022	8,192 MB	GDDR6	7,946	115 W	not recorded by source	—	8,700
amd	Radeon RX 6450M (Navi 24)	2023	4,096 MB	GDDR6	3,780	50 W	not recorded by source	—	—
amd	Radeon RX 6550M (Navi 24)	2023	4,096 MB	GDDR6	5,800	80 W	not recorded by source	—	—
amd	Radeon RX 6550S (Navi 24)	2023	4,096 MB	GDDR6	4,900	50 W	not recorded by source	—	—
nvidia	GeForce RTX 4070 Ti	2023	12,288 MB	GDDR6X	35,482	285 W	not recorded by source	—	35,800
nvidia	H800 GPU accelerator (SXM card)	2023	81,920 MB	HBM3	59,300	700 W	not recorded by source	—	—
nvidia	L4 GPU accelerator	2023	24,576 MB	GDDR6	30,300	72 W	not recorded by source	—	—
nvidia	RTX 4000 SFF Ada Generation	2023	20,480 MB	GDDR6	19,230	70 W	not recorded by source	—	—
nvidia	GeForce RTX 4070	2023	12,288 MB	GDDR6X	22,610	200 W	not recorded by source	—	35,800
nvidia	GeForce RTX 4060 Ti	2023	8,192 MB	GDDR6	20,106	160 W	not recorded by source	—	22,900
intel	Arc Pro A60	2023	12,288 MB	GDDR6	10,040	130 W	not recorded by source	—	—
intel	Arc Pro A60M	2023	8,192 MB	GDDR6	9,420	95 W	not recorded by source	—	—
nvidia	GeForce RTX 4060	2023	8,192 MB	GDDR6	11,244	115 W	not recorded by source	—	18,900
nvidia	GeForce RTX 4060 Ti 16 GB	2023	16,384 MB	GDDR6	20,106	165 W	not recorded by source	—	22,900
nvidia	RTX 4000 Ada Generation	2023	20,480 MB	GDDR6	26,730	130 W	not recorded by source	—	—
nvidia	RTX 4500 Ada Generation	2023	24,576 MB	GDDR6	39,630	210 W	not recorded by source	—	—
nvidia	RTX 5000 Ada Generation	2023	32,768 MB	GDDR6	65,280	250 W	not recorded by source	—	—
intel	Arc 5 A580	2023	8,192 MB	GDDR6	10,445	175 W	not recorded by source	—	21,700
nvidia	GeForce RTX 4090 D	2023	24,576 MB	GDDR6X	66,540	425 W	not recorded by source	—	76,300
nvidia	RTX 5880 Ada Generation	2024	49,152 MB	GDDR6	69,270	285 W	not recorded by source	—	—
nvidia	GeForce RTX 4070 Super	2024	12,288 MB	GDDR6X	28,385	220 W	not recorded by source	—	35,800
nvidia	GeForce RTX 4070 Ti Super	2024	16,384 MB	GDDR6X	39,537	285 W	not recorded by source	—	45,900
nvidia	GeForce RTX 4080 Super	2024	16,384 MB	GDDR6X	47,002	320 W	not recorded by source	—	45,900
nvidia	GeForce RTX 3050 6 GB	2024	6,144 MB	GDDR6	4,802	70 W	not recorded by source	—	8,700
nvidia	RTX 2000 Ada Generation	2024	16,384 MB	GDDR6	12,000	70 W	not recorded by source	—	—
nvidia	GeForce RTX 4070	2024	12,288 MB	GDDR6X	22,610	200 W	not recorded by source	—	45,900
nvidia	GeForce RTX 4060	2024	8,192 MB	GDDR6	11,244	115 W	not recorded by source	—	22,900
nvidia	GeForce RTX 4060 Ti	2024	8,192 MB	GDDR6	20,106	160 W	not recorded by source	—	35,800
nvidia	RTX A1000	2024	8,192 MB	GDDR6	6,737	50 W	not recorded by source	8 nm	—
nvidia	RTX A400	2024	4,096 MB	GDDR6	2,706	50 W	not recorded by source	8 nm	—
nvidia	GeForce RTX 4070 Ti Super	2024	16,384 MB	GDDR6X	39,537	285 W	not recorded by source	—	76,300
nvidia	GeForce RTX 4070 GDDR6	2024	12,288 MB	GDDR6	22,610	200 W	not recorded by source	—	35,800
amd	Radeon RX 7800M (Navi 32)	2024	12,288 MB	GDDR6	35,870	180 W	not recorded by source	—	—
amd	Radeon Pro V710 (Navi 32)	2024	28,672 MB	GDDR6	28,314	158 W	not recorded by source	—	—
nvidia	B200 GPU accelerator	2024	184,320 MB	HBM3E	74,450	1,000 W	not recorded by source	—	—
nvidia	H200 GPU accelerator (PCIe card)	2024	144,384 MB	HBM3E	60,320	600 W	not recorded by source	—	—
amd	Radeon Pro W7800 48GB (Navi 31)	2024	49,152 MB	GDDR6	45,250	281 W	not recorded by source	—	—
intel	Arc 5 B580	2024	12,288 MB	GDDR6	13,670	190 W	not recorded by source	—	19,600
intel	Arc 5 B570	2025	10,240 MB	GDDR6	11,520	150 W	not recorded by source	—	19,600
nvidia	GeForce RTX 5080	2025	16,384 MB	GDDR7	49,352	360 W	not recorded by source	—	45,600
nvidia	GeForce RTX 5090	2025	32,768 MB	GDDR7	87,780	575 W	not recorded by source	—	92,200
nvidia	GeForce RTX 5090 D	2025	32,768 MB	GDDR7	87,780	575 W	not recorded by source	—	92,200
nvidia	GeForce RTX 5070 Ti	2025	16,384 MB	GDDR7	41,126	300 W	not recorded by source	—	45,600
nvidia	GeForce RTX 5070	2025	12,288 MB	GDDR7	28,430	250 W	not recorded by source	—	31,100
nvidia	RTX PRO 4000 Blackwell	2025	24,576 MB	GDDR7	46,900	140 W	not recorded by source	—	—
nvidia	RTX PRO 4500 Blackwell	2025	32,768 MB	GDDR7	54,940	200 W	not recorded by source	—	—
nvidia	RTX PRO 5000 Blackwell	2025	—	GDDR7	73,690	300 W	not recorded by source	—	—
nvidia	RTX PRO 6000 Blackwell Max-Q Workstation Edition	2025	98,304 MB	GDDR7	110,100	300 W	not recorded by source	—	—
nvidia	RTX PRO 6000 Blackwell Workstation Edition	2025	98,304 MB	GDDR7	126,000	600 W	not recorded by source	—	—
nvidia	GeForce RTX 5060 Ti	2025	8,192 MB	GDDR7	22,183	180 W	not recorded by source	—	21,900
nvidia	GeForce RTX 5060 Ti 16 GB	2025	16,384 MB	GDDR7	22,183	180 W	not recorded by source	—	21,900
nvidia	GeForce RTX 5060	2025	8,192 MB	GDDR7	17,510	145 W	not recorded by source	—	21,900
intel	Arc Pro B60	2025	24,576 MB	GDDR6	12,800	200 W	not recorded by source	—	19,600
nvidia	GeForce RTX 5050	2025	8,192 MB	GDDR6	11,863	130 W	not recorded by source	—	16,900
nvidia	RTX PRO 2000 Blackwell	2025	16,384 MB	GDDR7	17,030	70 W	not recorded by source	—	—
nvidia	GeForce RTX 5090 D V2	2025	24,576 MB	GDDR7	87,780	575 W	not recorded by source	—	92,200
intel	Arc Pro B50	2025	16,384 MB	GDDR6	10,650	70 W	not recorded by source	—	19,600
nvidia	B300 GPU accelerator	2025	294,912 MB	HBM3E	76,990	1,400 W	not recorded by source	—	—
amd	Radeon AI Pro R9600D (Navi 48)	2025	32,768 MB	GDDR6	24,800	150 W	not recorded by source	—	—
amd	Radeon AI Pro R9600 (Navi 48)	2026	32,768 MB	GDDR6	24,800	150 W	not recorded by source	—	—
intel	Arc Pro B70	2026	32,768 MB	GDDR6	22,940	230 W	not recorded by source	—	—

Source: Wikipedia, revision 1369488585 · catalog loaded 2026-08-15