LinusTechTips Graphics Card Value calculation

[x69_collegechipset]

After watching you Scrapyard wars series of videos, which were quite entertaining, I was just wondering if you guys would be able to provide an explanation on how Linus calculates which cards are considered good value. I was trying to do the math based on what Linus had been talking about in on of the episodes but didn't seem to work. 

I think people would be very curious to know based on this calculation which graphics card are the best value.

 

[thekeemo]

I have a formula I use that looks like this

Value of modern equivalent divided by half for every year since the cards release

This is very rough and there are other factors too

[TheRandomness]

This chart helps assess performance: http://www.tomshardware.com/reviews/gpu-hierarchy,4388.html

[Castdeath97]

According to what I remember his calculation had some issues with it.

[Tomic]

Find a recent review of a graphics card that you want take the fps and divide it by the price of the card then you have the fps per dollar or pound and then you have the fps per dollar value

[GER_T4IGA]

After watching you Scrapyard wars series of videos, which were quite entertaining, I was just wondering if you guys would be able to provide an explanation on how Linus calculates which cards are considered good value. I was trying to do the math based on what Linus had been talking about in on of the episodes but didn't seem to work. 

I think people would be very curious to know based on this calculation which graphics card are the best value.

 

The question is what CPU+GPU combo will give you the most fps/dollar. That is about it.

[GER_T4IGA]

I have a formula I use that looks like this

Value of modern equivalent divided by half for every year since the cards release

This is very rough and there are other factors too

I'd deem this too rough but this is me.

[thekeemo]

I'd deem this too rough but this is me.

I also take into account if the card is "rare" or retailed for a higher price because of its form factor

[Bhav]

All I know is that for price / performance GTX 970 is winner.

[Mr.Meerkat]

All I know is that for price / performance GTX 970 is winner.

*390

[Bhav]

But that came out later so I shut my eyes and cover my ears and go LALALALALA!

[Enderman]

Find a recent review of a graphics card that you want take the fps and divide it by the price of the card then you have the fps per dollar or pound and then you have the fps per dollar value

no, this is completely wrong

linus admitted it in one of the scrapyard videos

 

the correct way to calculate value is by dividing the FPS by the total cost of the system, NOT just the gpu

 

because if you get a $10 GPU that gets 10 fps, thats one fps per dollar

but if your entire system costs $990 then thats horrible fps per dollar, only 0.01fps per dollar

 

if the GPU cost $1000 and gets 100fps, thats 0.1fps per dollar

and including a $1000 system cost, thats 100/2000= 0.05 fps per dollar

 

5x better price performance

 

so even though the GPU price performance is worse, the actual system price performance is far better

 

that's how linus is calculating it now (and how he used to, before he forgot in the first scrapyard wars season)

fps/system cost

[Bhav]

*390

I forgot to quote.

[MEC-777]

All I know is that for price / performance GTX 970 is winner.

 

That is debatable.

[PaulKolby]

After watching you Scrapyard wars series of videos, which were quite entertaining, I was just wondering if you guys would be able to provide

I think people would be very curious to know based on this calculation which graphics card are the best value.

 

I'm pretty sure later on in that episode he said his formula was incorrect or something on how its better to get a better gpu I forget exactly what he said but I remember him saying his formula didn't work

[Elapo]

I just look at how a card performs in certain games that I know I'm going to be playing, and then I compare different cards using statements like "Is €50 more worth this performance gain to me". If you can get a 680 or a 970, which are close in performance, the 680 is obviously going to be "better value" as in price/performance, but that doesn't take into account all the new stuff the 970 has to offer.

 

I don't really think a performance/price scale is all that useful, unless you're comparing 2 cards with similar performance and features. Trying to smack a number on everything isn't the right course of action.

 

no, this is completely wrong

linus admitted it in one of the scrapyard videos

 

the correct way to calculate value is by dividing the FPS by the total cost of the system, NOT just the gpu

 

because if you get a $10 GPU that gets 10 fps, thats one fps per dollar

but if your entire system costs $990 then thats horrible fps per dollar, only 0.01fps per dollar

 

if the GPU cost $1000 and gets 100fps, thats 0.1fps per dollar

and including a $1000 system cost, thats 100/2000= 0.05 fps per dollar

 

5x better price performance

 

so even though the GPU price performance is worse, the actual system price performance is far better

 

that's how linus is calculating it now (and how he used to, before he forgot in the first scrapyard wars season)

fps/system cost

I'm sorry, but that's a stupid way of doing it IMHO. If I buy a PC for 1$ (I'm exaggerating, obv), which gets 1 FPS, that's 1FPS/$, but that system still sucks, and is arguably not a very good value because it can't do a lot. You can do this to compare 2 systems with equal-ish components, but not to compare a mid-range PC with a gaming pc. The mid range PC might have amazing value, but that doesn't mean it's better for the person who wants a system.

[Mr.Meerkat]

But that came out later so I shut my eyes and cover my ears and go LALALALALA!

Fine, the 290/290x then

[Bhav]

No,what I bought is the best because I have it, LALALALALALALA!

[x69_collegechipset]

Thank you for all the replies, some of the resources I use often to evaluate value in the 3dmark community and toms hardware gpu charts but they do not account for the used cards. This is why I would like to get an explanation from Linus because his math takes used prices into account. 

Depending on how his math works I would love to use this methodology for future computer parts

Sources I currently use:
Toms Hardware Charts: http://www.tomshardware.com/charts/2015-vga-charts/benchmarks,186.html

Futuremark hardware Community: http://www.futuremark.com/hardware/gpu

[NannerBeans]

What he was doing wasn't calculating the value of a used gpu but the performance of the entire build based on changing gpus but the same components for the rest of the build and looking at benchmarks to see if it's a good overall value or not. Used gpu prices change a lot and you just have to look for price history to see if it's a good deal or not. But if you're looking for the best deal for a gpu, it's probably a 780 ti since you can get them for around $250 sometimes. 

[FruitBasketSilex]

This chart helps assess performance: http://www.tomshardware.com/reviews/gpu-hierarchy,4388.html

No 3dfx voodoo. I am dissapoint.

[TheRandomness]

No 3dfx voodoo. I am dissapoint.

waifoo

[FruitBasketSilex]

waifoo

I'm already someone's waifu, sorry. ^-^

[TheRandomness]

I'm already someone's waifu, sorry. ^-^

I know c: (Bunny?)

[FruitBasketSilex]

I know c: (Bunny?)

Yep. :3 I don't know if my boyfriend counts, though. I'm just Kenji's waifu for shits/giggles.