• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA 4000 Series

That £800 gpu will be more like £570 though now with the 4070 super, so now we’ll see how much people value better RT and features over 4gb of vram and a £100 saving.
I’ve clearly taken heed of HUB advice on taking the vram over fancy DLSS or path tracing which tbh looks a bit meh in cyberpunk when I’ve seen it.
 
I don’t even notice RT in games. Raster sometimes looks as good :cry:

At this sort of mid range its 'while not completely useless' not far from it. Also because of the difference in the way AMD vs Nvidia do RT Nvidia are only faster in a similar price braket if you crank the RT right up, in other words if you crank RT in Cyberpunk up, which is why a lot of reviewers do that, so if you're getting 20 FPS vs 30 FPS oh sure the Nvidia card is 50% faster but its still unplaybale. Its just "Biger bar better, nothing else to see here...." -Every tech jurno...

With RT at levels where you get playble FPS, in this range the Nvidia cards aren't much faster anyway.

Whats more, i much prefer AMD's Frame Gen and Drivers, its not just about 4GB more memory.
 
Last edited:

61 / 100 = 1.64

Those numbers look a bit off to me - most reviews seem to put the 2070S about 50% faster than the 1070.

Got me thinking though about your original assertion that Turing was a better price/performance bump over Pascal than was Ada Lovelace over Ampere.
Using the stats from TPU's database:

Pascal -> Turing:
1080Ti ($699) -> 2080Ti ($999): 143% of the price for 131% of the performance = 92% of the price/performance (1.31 / 1.43)
1080 ($599) -> 2080 ($699): 117% of the price for 139% of the performance = 119%
1070 ($379) -> 2070 ($499): 132% of the price for 134% of the performance = 102%

Ampere -> Ada Lovelace
3090 ($1499) -> 4090 ($1599): 107% of the price for 166% of the performance = 155%
3080 ($699) -> 4080 ($1199): 172% of the price for 149% of the performance = 87%
3070 ($499) -> 4070 ($599): 120% of the price for 122% of the performance = 102%

Quite interesting actually. The top model obviously had a much larger bump with 30->40 series but that's reversed with the xx80 cards and then the xx70 are identical.
Two things working in favour of Turing here are the $999 MSRP of the 2080Ti and $699 MSRP of the 3080, both of which were virtually unachievable in the real world. With more realistic prices for these two, it swings in favour of 30-40 series a bit.
 
Those numbers look a bit off to me - most reviews seem to put the 2070S about 50% faster than the 1070.

Got me thinking though about your original assertion that Turing was a better price/performance bump over Pascal than was Ada Lovelace over Ampere.
Using the stats from TPU's database:

Pascal -> Turing:
1080Ti ($699) -> 2080Ti ($999): 143% of the price for 131% of the performance = 92% of the price/performance (1.31 / 1.43)
1080 ($599) -> 2080 ($699): 117% of the price for 139% of the performance = 119%
1070 ($379) -> 2070 ($499): 132% of the price for 134% of the performance = 102%

Ampere -> Ada Lovelace
3090 ($1499) -> 4090 ($1599): 107% of the price for 166% of the performance = 155%
3080 ($699) -> 4080 ($1199): 172% of the price for 149% of the performance = 87%
3070 ($499) -> 4070 ($599): 120% of the price for 122% of the performance = 102%

Quite interesting actually. The top model obviously had a much larger bump with 30->40 series but that's reversed with the xx80 cards and then the xx70 are identical.
Two things working in favour of Turing here are the $999 MSRP of the 2080Ti and $699 MSRP of the 3080, both of which were virtually unachievable in the real world. With more realistic prices for these two, it swings in favour of 30-40 series a bit.

I said upgraded from a 1070 to a 2070 Super.

1070 ($379) -> 2070 Super ($499) 132% of the price for 164% of the performance = 122%.
 
Last edited:
Those numbers look a bit off to me - most reviews seem to put the 2070S about 50% faster than the 1070.

Got me thinking though about your original assertion that Turing was a better price/performance bump over Pascal than was Ada Lovelace over Ampere.
Using the stats from TPU's database:

Pascal -> Turing:
1080Ti ($699) -> 2080Ti ($999): 143% of the price for 131% of the performance = 92% of the price/performance (1.31 / 1.43)
1080 ($599) -> 2080 ($699): 117% of the price for 139% of the performance = 119%
1070 ($379) -> 2070 ($499): 132% of the price for 134% of the performance = 102%

Ampere -> Ada Lovelace
3090 ($1499) -> 4090 ($1599): 107% of the price for 166% of the performance = 155%
3080 ($699) -> 4080 ($1199): 172% of the price for 149% of the performance = 87%
3070 ($499) -> 4070 ($599): 120% of the price for 122% of the performance = 102%

Quite interesting actually. The top model obviously had a much larger bump with 30->40 series but that's reversed with the xx80 cards and then the xx70 are identical.
Two things working in favour of Turing here are the $999 MSRP of the 2080Ti and $699 MSRP of the 3080, both of which were virtually unachievable in the real world. With more realistic prices for these two, it swings in favour of 30-40 series a bit.
So you're saying my 4090 was a bargain? :p
 
I don’t even notice RT in games. Raster sometimes looks as good :cry:
But it is very limited...
You can "fake" a lot in raster (since that's the buzz word), but you can only push it so far while losing a lot of time building the game...

Anyway, this has been talked about a lot. If consoles would be strong enough, probably raster would be abandoned in a gen.
 
But it is very limited...
You can "fake" a lot in raster (since that's the buzz word), but you can only push it so far while losing a lot of time building the game...

Anyway, this has been talked about a lot. If consoles would be strong enough, probably raster would be abandoned in a gen.

RT in games is still almost entirley rasterisation with a venear of traced elements pasted over it.

A 4090 would render a fully ray traced scene at a rate of about 1 frame per minute....
 
Back
Top Bottom