What about this other card that’s not on the list? How do I know if it’s a good deal or not?
This will happen. In fact, because inventory levels and prices change quickly, it’s guaranteed to happen. So how do you know if that card you’ve got your eye on is a good buy in its price range?
Here is a resource to help you judge if a card is a good buy or not. The graphics card hierarchy chart groups graphics cards with similar overall performance levels into tiers. The top tier contains the highest-performing cards available, and performance decreases as you go down the tiers from there.
You can use this hierarchy to compare the pricing between two cards, to see which one is a better deal, and also to determine if an upgrade is worthwhile. I don’t recommend upgrading your graphics card unless the replacement card is at least three tiers higher. Otherwise, the upgrade is somewhat parallel, and you may not even notice any worthwhile difference in performance.
At the request of readers, we've added mobile graphics and integrated chipsets to the hierarchy chart. We want to make it clear that there is very little performance data available for these graphics solutions. While the discrete video cards in the chart are placed in tiers based on a lot of information, many of the mobile and integrated devices in the chart are guesstimates based on their specifications. At worst, we doubt they’re more than one tier away from their actual performance, but this is something to keep in mind when considering mobile graphics chipsets.
| Graphics Card Hierarchy Chart | ||
|---|---|---|
| GeForce | Radeon | Intel |
| Discrete: GTX Titan Z | Discrete: R9 295X2 | |
| Discrete: GTX 690, Titan X, 980 Ti | Discrete: HD 7990 | |
| Discrete: GTX 780 Ti, 980, Titan Black | ||
| Discrete: GTX 780, 970, Titan | Discrete: R9 290, 290X, 390X, 390 | |
| Discrete: GTX 590, 680, 770 Go (mobile): 980M | Discrete: HD 6990, 7970 GHz Ed, R9 280X, 380 | |
| Discrete: GTX 580, GTX 670, GTX 960 Go (mobile): 970M | Discrete: HD 5970, 7870 LE (XT), 7950, 280, 285 | |
| Discrete: GTX 660 Ti, GTX 760 Go (mobile): 880M | Discrete: HD 7870, R9 270, R9 270X, R7 370 | |
| Discrete: GTX 295, 480, 570, 660 Go (mobile): 680M, 780M | Discrete: HD 4870 X2, 6970, 7850, R7 265, Mobility: 7970M | |
| Discrete: GTX 470, 560 Ti, 560 Ti 448 Core, 650 Ti Boost, 750 Ti | Discrete: HD 4850 X2, 5870, 6950, R7 260X Mobility: 7950M | |
| Discrete: GTX 560, 650 Ti, 750 Go (mobile): 580M, 675M | Discrete: HD 5850, 6870, 7790 Mobility: 6990M | |
| Discrete: 9800 GX2, 285, 460 256-bit, 465 | Discrete: HD 6850, 7770, R7 260, R7 360 Mobility: 6900M | |
| Discrete: GTX 260, 275, 280, 460 192-bit, 460 SE, 550 Ti, 560 SE, GT 650, GT 740 GDDR5 Go (mobile): 570M, 670M | Discrete: HD 4870, 5770, 4890, 5830, 6770, 6790, 7750 (GDDR5), R7 250 (GDDR5), R7 250E Mobility: HD 5870, 6800M | Iris Pro Graphics 6200 |
| Discrete: 8800 Ultra, 9800 GTX, 9800 GTX+, GTS 250, GTS 450 Go (mobile): 560M, 660M | Discrete: HD 3870 X2, 4850, 5750, 6750, 7750 (DDR3), R7 250 (DDR3) Mobility: HD 4850, 5850, 7870M | |
| Discrete: 8800 GTX, 8800 GTS 512 MB, GT 545 (GDDR5), GT 730 64-bit GDDR5 Go (mobile): GTX 280M, 285M, 555M (GDDR5) | Discrete: HD 4770 Mobility: HD 4860, 7770M, 7850M | |
| Discrete: 8800 GT 512 MB, 9800 GT, GT 545 (DDR3), GT 640 (DDR3), GT 740 DDR3 Go (mobile): 9800M GTX, GTX 260M (112), GTS 360M (GDDR5), 555M (DDR3) | Discrete: HD 4830, HD 5670, HD 6670 (GDDR5), HD 7730 (GDDR5) Mobility: HD 5770, HD 5750, 6600M/6700M (GDDR5), 7750M | |
| Discrete: 8800 GTS 640 MB, 9600 GT, GT 240 (GDDR5) Go (mobile): 9800M GTS, GTX 160M | Discrete: HD 2900 XT, HD 3870, HD 5570 (GDDR5), HD 6570 (GDDR5) Mobility: 6500M (GDDR5), 6600M/6700M (DDR3), 7730M | |
| Discrete: 8800 GS, 9600 GSO, GT 240 (DDR3) Go (mobile): GTX 260M (96), GTS 150M, GTS 360M (DDR3) | Discrete: HD 3850 512 MB, HD 4670, HD 5570 (DDR3), HD 6570 (DDR3), HD 6670 (DDR3), HD 7730 (DDR3), R7 240 Mobility: HD 3870, HD 5730, HD 5650, 6500M (DDR3) | |
| Discrete: 8800 GT 256 MB, 8800 GTS 320 MB, GT 440 GDDR5, GT 630 GDDR5, GT 730 128-bit GDDR5 Go (mobile): 8800M | Discrete: HD 2900 PRO, HD 3850 256 MB, 5550 (GDDR5) Mobility: HD 3850 | |
| Discrete: 7950 GX2, GT 440 DDR3, GT 630 DDR3, GT 730 128-bit DDR3 | Discrete: X1950 XTX, HD 4650 (DDR3), 5550 (DDR3) Integrated: HD 7660D | |
| Discrete: 7800 GTX 512, 7900 GTO, 7900 GTX, GT 430, GT 530 Go (mobile): 550M | Discrete: X1900 XT, X1950 XT, X1900 XTX | |
| Discrete: 7800 GTX, 7900 GT, 7950 G, GT 220 (DDR3) Go (mobile): 525M, 540M | Discrete: X1800 XT, X1900 AIW, X1900 GT, X1950 Pro, HD 2900 GT, HD 5550 (DDR2) Integrated: HD 7560D | |
| Discrete: 7800 GT, 7900 GS, 8600 GTS, 9500 GT (GDDR3), GT 220 (DDR2) Go (mobile): 7950 GTX | Discrete: X1800 XL, X1950 GT, HD 4650 (DDR2), HD 6450, R5 230 Mobility: X1800 XT, HD 4650, HD 5165, 6400M Integrated: HD 6620G, 6550D, 7540D | |
| Discrete: 6800 Ultra, 7600 GT, 7800 GS, 8600 GS, 8600 GT (GDDR3), 9500 GT (DDR2) Go (mobile): 7800 GTX, 7900 GTX | Discrete: X800 XT (& PE), X850 XT (& PE), X1650 XT, X1800 GTO, HD 2600 XT, HD 3650 (DDR3), HD 3670 Mobility: X1900, 3670 Integrated: 6520G, 6530D, 7480D | Integrated: Intel HD Graphics 4000 |
| Discrete: 6800 GT, 6800 GS (PCIe), 8600 GT (DDR2), GT 520 Go (mobile): 7800, Go 7900 GS, 520M, 520MX | Discrete: X800 XL, X800 GTO2/GTO16, HD 2600 Pro, HD 3650 (DDR2), Mobility: X800 XT, HD 2600 XT, 3650 Integrated: 6410D, 6480G | |
| Discrete: 6800 GS (AGP) Go (mobile): 6800 Ultra, 7600 GT, 8600M GT, 8700M GT, 410M | Discrete: X800 GTO 256 MB, X800 PRO, X850 Pro, X1650 GT Mobility: HD 2600 Integrated: 6370D, 6380G | |
| Discrete: 6800, 7300 GT GDDR3, 7600 GS, 8600M GS Go (mobile): 6800, 7700 | Discrete: X800, X800 GTO 128 MB, X1600 XT, X1650 Pro Mobility: X1800, HD 5145, HD 5470 (GDDR5) | |
| Discrete: 6600 GT, 6800LE, 6800 XT, 7300 GT (DDR2), 8500 GT, 9400 GT Go (mobile): 7600 (128-bit) | Discrete: 9800 XT, X700 PRO, X800 GT, X800 SE, X1300 XT, X1600 PRO, HD 2400 XT, HD 4350, HD 4550, HD 5450 Mobility: X800, 3470, HD 5470 (DDR3), HD 5450, HD 5430, 6300M Integrated: HD 6310, HD 6320 | Integrated: Intel HD Graphics 3000 |
| Discrete: FX 5900, FX 5900 Ultra, FX 5950 Ultra, 6600 (128-bit) Go (mobile): 6800 (128-bit) Integrated: 9300, 9400 | Discrete: 9700, 9700 Pro, 9800, 9800 Pro, X700, X1300 Pro, X1550, HD 2400 Pro Mobility: X1450, X1600, X1700, 2400 XT, X2500, 3450 Integrated: HD 3200, HD 3300, HD 4200, HD 4250, HD 4290, HD 6250, HD 6290 | |
| Discrete: FX 5800 Ultra, FX 5900 XT Go (mobile): 6600, Go 7600 (64-bit) | Discrete: 9500 Pro, 9600 XT, 9800 Pro (128-bit), X600 XT, X1050 (128-bit) Mobility: 9800, X700, X1350, X1400, X2300, HD 2400 | Integrated: Intel HD Graphics (Core i5-6x1), 2000 |
| Discrete: 4 Ti 4600, 4 Ti 4800, FX 5700 Ultra, 6200, 8300, 8400 G, G 210, G 310 Go (mobile): 315M | Discrete: 9600 PRO, 9800 LE, X600 PRO, HD 2300 Mobility: 9700 (128-bit), X600, X1300 Integrated: Xpress 1250 | Integrated: Intel HD Graphics (Core i3 5x0, Core i5-6x0) |
| Discrete: 4 Ti4200, 4 Ti4400, 4 Ti4800 SE, FX 5600 Ultra, FX 5700, 6600 (64-bit), 7300 GS, 8400M GS, 9300M G, 9300M GS | Discrete: 9500, 9550, 9600, X300, X1050 (64-bit) Mobility: 9600 | Integrated: Intel HD Graphics (Pentium G) |
| Discrete: 3 Ti500, FX 5200 Ultra, FX 5600, FX 5700 LE, 6200 TC, 6600 LE, 7200 GS, 7300 LE Go (mobile): 5700, 8200M, 9200M GS, 9100 Integrated: 8200, 8300 | Discrete: 8500, 9100, 9000 PRO, 9600 LE, X300 SE, X1150 Mobility 9700 (64-bit) | Integrated: GMA X4500 |
| Discrete: 3, 3 Ti200, FX 5200 (128-bit), FX 5500, Go (mobile): 5600, 6200, 6400, 7200, 7300, 7400 (64-bit) | Discrete: 9000, 9200, 9250 Mobility: 9600 (64-bit), X300 | |
| Discrete: FX 5200 (64 bit) Go (mobile): 7200, 7400 (32-bit) Integrated: 6100, 6150, 7025, 7050 | Discrete: 9200 SE Integrated: Xpress 200M, Xpress 1000, Xpress 1150 | Integrated: GMA X3000, X3100, X3500 |
| Discrete: 2 GTS, 4 MX 440, 2 Ultra, 2 Ti, 2 Ti 200 | Discrete: 7500 | Integrated: GMA 3000, 3100 |
| Discrete: 256, 2 MX 200, 4 MX 420, 2 MX 400 | Discrete: SDR, LE, DDR, 7000, 7200 | Integrated: GMA 500, 900, 950 |
| Discrete: Nvidia TNT | Discrete: Rage 128 | Discrete: Intel 740 |
Now all that’s left to do is compare performance to your budget, and you'll be able to confidently decide which board is right for you - we even put in the legwork to help find you the best prices!
1) Just because you as an AMD user don't like the results of what Tom's comes up with does not make them favorable to Nvidia (and vice versa for Nvidia users). I clearly remember Tom's giving a cool reception to the GTX 770 as nothing but a re-branded 680 with and faster speed (which is what it was, just like the 390x vs. 290x except that unlike AMD's $80 price boost with 8GB VRAM over the current market 4GB 290x pricing, the 2GB 770 had a price drop over the 2GB 680 card at the time of market).
Now let's review their findings of the best card values by US dollar category for this month:
$65 - Nvidia
100 - AMD
115- AMD
150 - Tie
200 - AMD
330 - Tie
650 - Nvidia
Dual GPU - AMD
So removing the ties, we are looking at Tom's picking AMD 2:1 over Nvidia in this month's best GPUs to buy for the money. Any more questions on anti-AMD biases? Sheesh...it's like dealing with children around here sometimes.
2) I would surmise that Tom's has a calendar to adhere to for consistency. Why should they abandon one of their monthly GPU best of list just because a new card has a paper launch and they have to wait until they get drivers for the card, test it, etc. That stuff takes time. That goes for Nvidia as well. They will catch it on next month's review of the card. Deal with it.
Oh wait...it was ALREADY EXPLAINED!
"Even in the face of those price drops, you probably held off on pulling the trigger, knowing full well that AMD’s own ultra-high-end answer was imminent. The company’s launch made it difficult for us to write a Best Graphics Cards for the Money update any sooner. We already delayed for three weeks, and there’s still more excitement coming. But we have to keep moving."
Also, crossfire of 290X in 4K is faster than SLI of 780 Tis/970s/980s. Why is the article still ignoring this ?
Also, crossfire of 290X in 4K is faster than SLI of 780 Tis/970s/980s. Why is the article still ignoring this ?
Didn`t You know that Toms was always anti-amd?
Didn`t You know that Toms was always anti-amd?
Your argument doesn't make much sense seeing as half the cards on the list are AMD, and the 295X2 still tops the list as the flagship...
I'm not upgrading from my GTX 670 until there is a die shrink.
http://www.tweaktown.com/tweakipedia/74/recap-nvidia-geforce-gtx-980-sli-performance-4k/index.html
http://www.tweaktown.com/tweakipedia/91/amd-radeon-r9-390x-crossfire-4k/index.html
Uhm, it doesn't exactly blow away 780Ti SLI...
http://images.anandtech.com/graphs/graph7492/59668.png
http://images.anandtech.com/graphs/graph7492/59664.png
http://images.anandtech.com/graphs/graph7492/59672.png
And at what price, the price of heat and power specifically?
http://images.anandtech.com/graphs/graph7492/59712.png
http://images.anandtech.com/graphs/graph7492/59709.png
1) Just because you as an AMD user don't like the results of what Tom's comes up with does not make them favorable to Nvidia (and vice versa for Nvidia users). I clearly remember Tom's giving a cool reception to the GTX 770 as nothing but a re-branded 680 with and faster speed (which is what it was, just like the 390x vs. 290x except that unlike AMD's $80 price boost with 8GB VRAM over the current market 4GB 290x pricing, the 2GB 770 had a price drop over the 2GB 680 card at the time of market).
Now let's review their findings of the best card values by US dollar category for this month:
$65 - Nvidia
100 - AMD
115- AMD
150 - Tie
200 - AMD
330 - Tie
650 - Nvidia
Dual GPU - AMD
So removing the ties, we are looking at Tom's picking AMD 2:1 over Nvidia in this month's best GPUs to buy for the money. Any more questions on anti-AMD biases? Sheesh...it's like dealing with children around here sometimes.
2) I would surmise that Tom's has a calendar to adhere to for consistency. Why should they abandon one of their monthly GPU best of list just because a new card has a paper launch and they have to wait until they get drivers for the card, test it, etc. That stuff takes time. That goes for Nvidia as well. They will catch it on next month's review of the card. Deal with it.
Oh wait...it was ALREADY EXPLAINED!
"Even in the face of those price drops, you probably held off on pulling the trigger, knowing full well that AMD’s own ultra-high-end answer was imminent. The company’s launch made it difficult for us to write a Best Graphics Cards for the Money update any sooner. We already delayed for three weeks, and there’s still more excitement coming. But we have to keep moving."
Uhm, it doesn't exactly blow away 780Ti SLI...
http://images.anandtech.com/graphs/graph7492/59668.png
http://images.anandtech.com/graphs/graph7492/59664.png
http://images.anandtech.com/graphs/graph7492/59672.png
And at what price, the price of heat and power specifically?
http://images.anandtech.com/graphs/graph7492/59712.png
http://images.anandtech.com/graphs/graph7492/59709.png
You are using benchmarks from the 290X launch. Since then, 21 months ago, drivers have improved performance a lot. In current games with current drivers, two 290X outperform 780 Tis, 970s and 980s.
Though we don't know how Fury X performs yet,if it's near 980 Ti level, it's pretty much a given two 390X will be the fastest setup between what you listed.
They also said they have have Fury reviews in just 3 days. But they could not wait 3 days? So they can ignore the AMD Fury cards for another month. ???
They also said they have have Fury reviews in just 3 days. But they could not wait 3 days? So they can ignore the AMD Fury cards for another month. ???
Toms was taking a lot of flack from a lot of other readers comments about not having a June version out already. There is no pleasing some of you folks. They don't put it out they get flack, they put it out they get flack.
AMD in a massive blunder forbid reviews before the 24th, go complain to them. Had AMD released the reviews ASAP most reviewers would have only had slower reference 980Tis to compare with the FuryX. But it looks like by the 24th, up to 20% higher clocked aftermarket cooled 980Tis will be available for comparison. Tom's reviewer stated he already had a 20% higher clocked Gigabyte 980Ti version.
They also said they have have Fury reviews in just 3 days. But they could not wait 3 days? So they can ignore the AMD Fury cards for another month. ???
Toms was taking a lot of flack from a lot of other readers comments about not having a June version out already. There is no pleasing some of you folks. They don't put it out they get flack, they put it out they get flack.
AMD in a massive blunder forbid reviews before the 24th, go complain to them. Had AMD released the reviews ASAP most reviewers would have only had slower reference 980Tis to compare with the FuryX. But it looks like by the 24th, up to 20% higher clocked aftermarket cooled 980Tis will be available for comparison. Tom's reviewer stated he already had a 20% higher clocked Gigabyte 980Ti version.
Good grief, it is not an AMD issue. Tom's do what it do as do AMD. They know they can do it in three days but that is too far afield, can't do it. Give me a break. Tom's made a point to say they waited and they knew of the 24th for how long? For what, as well? What’s really new over the last many months? Not much. Everyone that cares knows what the 980ti can do.
Okay let's try this again. Note these are single card comparos with RECENT Guru3D charts at 4K on June 1 (I'm assuming they are not using two year old drivers). Let's shake it down:
Hitman Absolution:
290X - 34
980 - 33
Bioshock Infinite:
290X - 30
980 - 43
Tomb Raider:
290X - 42
980 - 46
Metro Last Light:
290X - 26
980 - 31
Thief:
290X - 26
980 - 29
Battlefield Hardline:
290X - 26
980 - 25
GTA V:
290X - 30
980 - 31
Witcher 3:
290X - 24
980 - 28
http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_980_ti_review,25.html
http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_980_ti_review,26.html
Look...I'm all for the rivalry between Team Red and Team Green benefiting us all. But let's not spread FUD around that is easily debunked...okay?
PS: drivers do NOT fix heat and power hog issues in 2013 any more than they do in 2015.