I Found a used 3080ti for 100 dollars less than a new 4070. Is that a good deal?

[Smestn]

Hi! I am upgrading from a gtx 1070, and i think i found a decent deal. 

I have a 1440p monitor right now, but i am thinking to dip my toes into 4k when some OLED 4k monitors come into the market.

Would the 3080 ti be a good deal then? Any thoughts are deeply appreciated on this. 

[ki8aras]

8 minutes ago, Smestn said:

Hi! I am upgrading from a gtx 1070, and i think i found a decent deal. 

I have a 1440p monitor right now, but i am thinking to dip my toes into 4k when some OLED 4k monitors come into the market.

Would the 3080 ti be a good deal then? Any thoughts are deeply appreciated on this. 

what price

[DarkSmith2]

i have higher performance expectations for a GPU at those resolutions. A 3080ti/4070 wouldnt cut it for me on 1440p or 4k.

[Wh0_Am_1]

I would personally buy a 6900XT or 6950XT instead. Which on the used market (in the US) can be found for $400 to $600 and are about equivalent.

[DarkSmith2]

2 minutes ago, Wh0_Am_1 said:

I would personally buy a 6900XT or 6950XT instead. Which on the used market (in the US) can be found for $400 to $600 and are about equivalent.

why would you when a brand new 4070 has more features and way better efficiency for $600 aswell.

[Wh0_Am_1]

3 minutes ago, DarkSmith2 said:

why would you when a brand new 4070 has more features and way better efficiency for $600 aswell.

If you get them at the right time I have bought 6900XTs for ~$420 otherwise the 4070 is worth considering, I haven't checked prices for the 40 series in a while and expected them to be higher.

[DarkSmith2]

23 minutes ago, Wh0_Am_1 said:

If you get them at the right time I have bought 6900XTs for ~$420 otherwise the 4070 is worth considering, I haven't checked prices for the 40 series in a while and expected them to be higher.

cheapest 4070 on newegg is $549.99. However ehm all old gen cards are not worth buying anymore since 4070 and 7800xt release. Buying used always comes with a caveat, you probably have to clean the card, have to replace pads, have to repaste, so you put time, little more money and effort in it and still probably have a product with less lifetime, efficiency and features left. 

For it to be a good deal the card would have to be in a really good shape/condition or way cheaper than just $150 bucks less, compared to similar performance when buying brand new. 

Or in other words a used 6900xt or 6950xt aren't worth more than $300/$350 anymore. If you pay more its a ripoff / delusional.

[freeagent]

41 minutes ago, DarkSmith2 said:

i have higher performance expectations for a GPU at those resolutions. A 3080ti/4070 wouldnt cut it for me on 1440p or 4k.

But they both smoke your 2080.

[Bagzie]

A 6950XT is about 20% faster in typical rasterized games than a 4070 , It's closer to a 4070ti than a 4070 and if you could buy it for $300 it would be the best graphics card deal in the history of pc gaming.

 

It has more vram and with a higher bandwidth rate.

 

That being said the 4070 is a solid card and very well suited to 1440p gaming.

[DarkSmith2]

1 minute ago, freeagent said:

But they both smoke your 2080

sure but its a 2080TI and im on 1080p, and the performance of it bugs me, its really annoyingly slow even for 1080p.
So that also contributes to higher expectations on higher resolutions. 

[Ottoman420]

2 minutes ago, freeagent said:

But they both smoke your 2080.

and my 1050 and i have similar expectations of 4k lol

 

*back to memory oc training*

[DarkSmith2]

8 minutes ago, Bagzie said:

A 6950XT is about 20% faster in typical rasterized games than a 4070 , It's closer to a 4070ti than a 4070 and if you could buy it for $300 it would be the best graphics card deal in the history of pc gaming.

 

It has more vram and with a higher bandwidth rate.

6950xt is only 5% faster than a 6900xt.
This is techpowerups review to the 4070 FE.
https://www.techpowerup.com/review/nvidia-geforce-rtx-4070-founders-edition/32.html

A 6950xt is nowhere near 20% faster than a 4070 on a wide average of games.

[Bagzie]

3 minutes ago, DarkSmith2 said:

6950xt is only 5% faster than a 6900xt.
This is techpowerups review to the 4070 FE.
https://www.techpowerup.com/review/nvidia-geforce-rtx-4070-founders-edition/32.html

A 6950xt isnt nowhere near 20% faster than a 4070 on a wide average of games.

It's ironic that you based your data on the same website that has the 6950XT 20% faster than a 4070.

 

Makes it much harder for you to argue my sourcing when I show you this.

 

 

[DarkSmith2]

11 minutes ago, Bagzie said:

It's ironic that you based your data on the same website that has the 6950XT 20% faster than a 4070.

 

Makes it much harder for you to argue my sourcing when I show you this.

 

 

well if you want to argue you better put a link of the source, so we know what we are looking at. For what i know this is comparing performance to another GPU as baseline, so f.e. 

4070 is 100% faster than 2080ti,
6950xt is 119% faster than 2080ti 

In the same test, that doesnt mean that 4070 is 19% slower than a 6950xt. 
You just have to understand the purpose of this database, and how its done. 
I think it even compares to a way slower card, iirc.

So its you, not understanding this chart.. that makes you believe that a 6950xt is 20% faster than a 4070. When its absolutely not.

[filpo]

10 minutes ago, DarkSmith2 said:

4070 is 100% faster than 2080ti,
6950xt is 119% faster than 2080ti 

no that was using the 4070 as a baseline so the 4070 had 100% of the performance of a 4070 and the 6950 xt has 120% of the performance of a 4070 (i.e. 20% more)

 

10 minutes ago, DarkSmith2 said:

So its you, not understanding this chart.. that makes you believe that a 6950xt is 20% faster than a 4070. When its absolutely not.

lmao

[DarkSmith2]

4 minutes ago, filpo said:

no that was using the 4070 as a baseline so the 4070 had 100% of the performance of a 4070 and the 6950 xt has 120% of the performance of a 4070 (i.e. 20% more)

When that would be the case, why doesnt this claimed performance of the chart represent actual benchmark results.

If a 6950xt would be 20% faster than a 4070, it would almost be as fast as a 7900xt. OR faster than a 4070ti.
In this chart that would mean that a 7900xt is almost 40% faster than a 4070. I mean where is your common sense?

[filpo]

2 minutes ago, DarkSmith2 said:

represent actual benchmark results.

because Techpowerup is using an average of benchmark results they've done. Otherwise it would be literally too much info to absorb on one site (on something that isn't a video)

 

3 minutes ago, DarkSmith2 said:

If a 6950xt would be 20% faster than a 4070, it would almost be as fast as a 7900xt. OR faster than a 4070ti.

It's an AVERAGE, it could have 40% more on one game and then the same on another

[Bagzie]

7 minutes ago, DarkSmith2 said:

well if you want to argue you better put a link of the source, so we know what we are looking at. For what i know this is comparing performance to another GPU as baseline, so f.e. 

4070 is 100% faster than 2080ti,
6950xt is 119% faster than 2080ti 

In the same test, that doesnt mean that 4070 is 19% slower than a 6950xt. 
You just have to understand the purpose of this database, and how its done.

https://www.techpowerup.com/gpu-specs/geforce-rtx-4070.c3924

 

The 100% is the baseline , You would think that would be obvious.

 

Your test is skewed because you have 2 misconceptions , The first is at 1080p the difference in GPU's is going to be somewhat skewed by the high refresh rate pushing more load onto the CPU , This is why your own source showed a 3% difference at 1080p and 6% at 1440p with the 6900XT.

 

A 6950XT is 6% faster than a 6900XT at 1440p and 7% at 4k.

 

https://www.techpowerup.com/review/amd-radeon-rx-6950-xt-reference-design/29.html

 

Let's stick with your own source for this because I'm not willing to debate over roughly 5% performance delta.

 

That would indicate a 14% performance improvement over a 4070 using your own sources.

 

 

 

 

 

 

[Bagzie]

8 minutes ago, DarkSmith2 said:

When that would be the case, why doesnt this claimed performance of the chart represent actual benchmark results.

If a 6950xt would be 20% faster than a 4070, it would almost be as fast as a 7900xt. OR faster than a 4070ti.
In this chart that would mean that a 7900xt is almost 40% faster than a 4070. I mean where is your common sense?

The 7900XT in the relative testing was done at 4k where it absolutely is 40% faster than a 4070.

 

Its 10% faster than a 4070ti at 4k probably mostly related to its 800GB/S memory bandwidth which becomes much more relevant at 4k resolution.

[Dedayog]

54 minutes ago, DarkSmith2 said:

cheapest 4070 on newegg is $549.99. However ehm all old gen cards are not worth buying anymore since 4070 and 7800xt release. Buying used always comes with a caveat, you probably have to clean the card, have to replace pads, have to repaste, so you put time, little more money and effort in it and still probably have a product with less lifetime, efficiency and features left. 

For it to be a good deal the card would have to be in a really good shape/condition or way cheaper than just $150 bucks less, compared to similar performance when buying brand new. 

Or in other words a used 6900xt or 6950xt aren't worth more than $300/$350 anymore. If you pay more its a ripoff / delusional.

Why would you need to replace pads and paste on a 6900xt?  You don't do maintenance on GPU's like a car.

 

Since a used 6950XT beats the 4070 and a 7800XT, why would you determine it to be $350 when they are $500-$550?  It's a generation back only, age doesn't impact here.

 

I think you're applying some personal viewpoint to  pricing instead of looking at a performance based approach.

 

 

[DarkSmith2]

Just now, Dedayog said:

Why would you need to replace pads and paste on a 6900xt?  You don't do maintenance on GPU's like a car.

 

Since a used 6950XT beats the 4070 and a 7800XT, why would you determine it to be $350 when they are $500-$550?  It's a generation back only, age doesn't impact here.

 

I think you're applying some personal viewpoint to  pricing instead of looking at a performance based approach.

no, i mean its used, its not worth as much as new, thats the point. Its nothing personal.

[DarkSmith2]

11 minutes ago, Bagzie said:

That would indicate a 14% performance improvement over a 4070 using your own sources.

see, thats still not 20%.. ? And this is also where reality kicks in... Just easier to compare the 4070FE relative performance to the 6900xt of the very same testing hardware and review. And +/- whatever the 6950xt you think is faster. (its 5%) 

But yea u start admitting. Thank you

[Bagzie]

10 minutes ago, DarkSmith2 said:

see, thats still not 20%.. ?

Sigh.

 

I actually didn't want to have to argue the difference over 5% but here we are..

 

Techpowerups GPU heirarchy is based on a select group of games that they have used for the last 5 years so they can better correlate performance between older GPU's and newer GPU's to give you a good baseline set of tests for people wishing to upgrade.

 

It's alot easier than retesting 90+ GPU's every time they post a new gpu review.

 

In those tests it's 19%.

 

You cannot argue the source is unreliable when you are using the same source my friend , What you are basically saying is the source is only correct if it agrees with your preconceived notion of what is correct.

 

Either the source is correct or the source is unreliable and cannot be used , Make up your mind.

 

Everything here is moot because the point I was making to begin with is what kindoff company would choose to sell a GPU that's 14% faster than a $550 gpu for $300?

 

EDIT : You edited your post so I'll do the same , This is here for people to read , They will make up there own minds and I'm not willing to engage with you anymore because I believe you are absolutely not willing to change your mind based on evidence so what's the point.

[DarkSmith2]

5 minutes ago, Bagzie said:

Sigh.

 

I actually didn't want to have to argue the difference over 5% but here we are..

 

Techpowerups GPU heirarchy is based on a select group of games that they have used for the last 5 years so they can better correlate performance between older GPU's and newer GPU's to give you a good baseline set of tests for people wishing to upgrade.

 

It's alot easier than retesting 90+ GPU's every time they post a new gpu review.

 

In those tests it's 19%.

 

You cannot argue the source is unreliable when you are using the same source my friend , What you are basically saying is the source is only correct if it agrees with your preconceived notion of what is correct.

 

Either the source is correct or the source is unreliable and cannot be used , Make up your mind.

 

Everything here is moot because the point I was making to begin with is what kindoff company would choose to sell a GPU that's 14% faster than a $550 gpu for $300?

well it depends what part of the site you use as a source. I use the real comparisson of more or less recent real benchmarks using the same testing setup, rather than an artificial number that is clustered together out of various different testing setups where a conclusion is based on an unrealistic approach to compare hardware with each other.

Lets take another look at another benchmark, you claim a 6950xt is 10% faster than a 4070ti at 4k?
What testing results specifically make you think that way? 

That would make a 6950xt 21% faster than a 6900xt aswell.


I mean i cant proof any better how pathetic that is.

[Bagzie]

5 minutes ago, DarkSmith2 said:

well it depends what part of the site you use as a source. I use the real comparisson of more or less recent real benchmarks using the same testing setup, rather than an artificial number that is clustered together out of various different testing setups where a conclusion is based on an unrealistic approach to compare hardware with each other.

Lets take another look at another benchmark, you claim a 6950xt is 10% faster than a 4070ti at 4k?
What testing results specifically make you think that way? 

That would make a 6950xt 21% faster than a 6900xt aswell.


I mean i cant proof any better how pathetic that is.

If you could please refrain from misquoting me that would be awesome.

 

I actually said this.

 

"The 7900XT in the relative testing was done at 4k where it absolutely is 40% faster than a 4070.

 

 

 

Its 10% faster than a 4070ti at 4k probably mostly related to its 800GB/S memory bandwidth which becomes much more relevant at 4k resolution".

 

You then posted a result showing the exact thing I said but changed the 7900XT to a 6950XT.

 

It's important to be accurate during a debate otherwise people might think you are a liar or atleast incompetent.